diff options
| author | Ludovic Courtès <ludo@gnu.org> | 2012-02-03 10:51:46 +0100 |
|---|---|---|
| committer | Ludovic Courtès <ludo@gnu.org> | 2012-02-03 10:51:46 +0100 |
| commit | eb4a14ed47a34e03566261b02c77bfcc02b20134 (patch) | |
| tree | 23d7e5a0df3aa502b1b5802a0e6c75cfbef87519 | |
| parent | 1ba05158ebe033d7d0bdfcfef686eae2c9ea0103 (diff) | |
Use Gnulib's `regex' module.
This should help with regex portability, as reported in
<http://bugs.gnu.org/10684> for Darwin 8.11.
* m4/gnulib-cache.m4 (gl_MODULES): Add `regex'.
* configure.ac: Remove header checks for regex.h, rxposix.h, and
rx/rxposix.h. Remove check for the `regcomp' function. Remove
definition of `HAVE_REGCOMP'. Define `ENABLE_REGEX'.
* libguile/init.c: Check for `ENABLE_REGEX' instead of `HAVE_REGCOMP'.
* libguile/regex-posix.c: Always include <regex.h>. Remove #ifdefs for
rxposix.h and co.
40 files changed, 16040 insertions, 43 deletions
diff --git a/configure.ac b/configure.ac index ccc5fa572..5ee942c3f 100644 --- a/configure.ac +++ b/configure.ac @@ -5,7 +5,7 @@ dnl define(GUILE_CONFIGURE_COPYRIGHT,[[ Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006, - 2007, 2008, 2009, 2010, 2011 Free Software Foundation, Inc. + 2007, 2008, 2009, 2010, 2011, 2012 Free Software Foundation, Inc. This file is part of GUILE @@ -656,7 +656,7 @@ AC_SUBST([SCM_I_GSC_HAVE_STRUCT_DIRENT64]) # sched.h - missing on MinGW # AC_CHECK_HEADERS([complex.h fenv.h io.h libc.h limits.h memory.h process.h string.h \ -regex.h rxposix.h rx/rxposix.h sys/dir.h sys/ioctl.h sys/select.h \ +sys/dir.h sys/ioctl.h sys/select.h \ sys/time.h sys/timeb.h sys/times.h sys/stdtypes.h sys/types.h \ sys/utime.h time.h unistd.h utime.h pwd.h grp.h sys/utsname.h \ direct.h langinfo.h nl_types.h machine/fpu.h poll.h sched.h]) @@ -1112,24 +1112,8 @@ if test $guile_cv_localtime_cache = yes; then fi if test "$enable_regex" = yes; then - if test "$ac_cv_header_regex_h" = yes || - test "$ac_cv_header_rxposix_h" = yes || - test "$ac_cv_header_rx_rxposix_h" = yes; then - GUILE_NAMED_CHECK_FUNC(regcomp, norx, [AC_LIBOBJ([regex-posix])], - [AC_CHECK_LIB(rx, main) - GUILE_NAMED_CHECK_FUNC(regcomp, rx, [AC_LIBOBJ([regex-posix])], - [AC_CHECK_LIB(regex, main) - GUILE_NAMED_CHECK_FUNC(regcomp, regex, [AC_LIBOBJ([regex-posix])])])] - ) - dnl The following should not be necessary, but for some reason - dnl autoheader misses it if we don't include it! - if test "$ac_cv_func_regcomp_norx" = yes || - test "$ac_cv_func_regcomp_regex" = yes || - test "$ac_cv_func_regcomp_rx" = yes; then - AC_DEFINE([HAVE_REGCOMP], 1, - [This is included as part of a workaround for a autoheader bug.]) - fi - fi + AC_LIBOBJ([regex-posix]) + AC_DEFINE([ENABLE_REGEX], 1, [Define when regex support is enabled.]) fi AC_REPLACE_FUNCS([strerror memmove mkstemp]) diff --git a/lib/Makefile.am b/lib/Makefile.am index a4a996039..17fb5b4e9 100644 --- a/lib/Makefile.am +++ b/lib/Makefile.am @@ -21,7 +21,7 @@ # the same distribution terms as the rest of that program. # # Generated by gnulib-tool. -# Reproduce by: gnulib-tool --import --dir=. --local-dir=gnulib-local --lib=libgnu --source-base=lib --m4-base=m4 --doc-base=doc --tests-base=tests --aux-dir=build-aux --lgpl=3 --no-conditional-dependencies --libtool --macro-prefix=gl --no-vc-files accept alignof alloca-opt announce-gen autobuild bind byteswap canonicalize-lgpl ceil close connect dirfd duplocale environ extensions flock floor fpieee frexp full-read full-write func gendocs getaddrinfo getpeername getsockname getsockopt git-version-gen gitlog-to-changelog gnu-web-doc-update gnupload havelib iconv_open-utf inet_ntop inet_pton isinf isnan ldexp lib-symbol-versions lib-symbol-visibility libunistring listen localcharset locale log1p maintainer-makefile malloc-gnu malloca nproc open pipe2 putenv recv recvfrom rename send sendto setenv setsockopt shutdown socket stat-time stdlib strftime striconveh string sys_stat trunc verify vsnprintf warnings wchar +# Reproduce by: gnulib-tool --import --dir=. --local-dir=gnulib-local --lib=libgnu --source-base=lib --m4-base=m4 --doc-base=doc --tests-base=tests --aux-dir=build-aux --lgpl=3 --no-conditional-dependencies --libtool --macro-prefix=gl --no-vc-files accept alignof alloca-opt announce-gen autobuild bind byteswap canonicalize-lgpl ceil close connect dirfd duplocale environ extensions flock floor fpieee frexp full-read full-write func gendocs getaddrinfo getpeername getsockname getsockopt git-version-gen gitlog-to-changelog gnu-web-doc-update gnupload havelib iconv_open-utf inet_ntop inet_pton isinf isnan ldexp lib-symbol-versions lib-symbol-visibility libunistring listen localcharset locale log1p maintainer-makefile malloc-gnu malloca nproc open pipe2 putenv recv recvfrom regex rename send sendto setenv setsockopt shutdown socket stat-time stdlib strftime striconveh string sys_stat trunc verify vsnprintf warnings wchar AUTOMAKE_OPTIONS = 1.5 gnits subdir-objects @@ -165,6 +165,15 @@ EXTRA_libgnu_la_SOURCES += bind.c ## end gnulib module bind +## begin gnulib module btowc + + +EXTRA_DIST += btowc.c + +EXTRA_libgnu_la_SOURCES += btowc.c + +## end gnulib module btowc + ## begin gnulib module byteswap BUILT_SOURCES += $(BYTESWAP_H) @@ -755,6 +764,39 @@ EXTRA_libgnu_la_SOURCES += isnan.c isnanl.c ## end gnulib module isnanl +## begin gnulib module langinfo + +BUILT_SOURCES += langinfo.h + +# We need the following in order to create an empty placeholder for +# <langinfo.h> when the system doesn't have one. +langinfo.h: langinfo.in.h $(top_builddir)/config.status $(CXXDEFS_H) $(WARN_ON_USE_H) + $(AM_V_GEN)rm -f $@-t $@ && \ + { echo '/* DO NOT EDIT! GENERATED AUTOMATICALLY! */'; \ + sed -e 's|@''GUARD_PREFIX''@|GL|g' \ + -e 's|@''HAVE_LANGINFO_H''@|$(HAVE_LANGINFO_H)|g' \ + -e 's|@''INCLUDE_NEXT''@|$(INCLUDE_NEXT)|g' \ + -e 's|@''PRAGMA_SYSTEM_HEADER''@|@PRAGMA_SYSTEM_HEADER@|g' \ + -e 's|@''PRAGMA_COLUMNS''@|@PRAGMA_COLUMNS@|g' \ + -e 's|@''NEXT_LANGINFO_H''@|$(NEXT_LANGINFO_H)|g' \ + -e 's/@''GNULIB_NL_LANGINFO''@/$(GNULIB_NL_LANGINFO)/g' \ + -e 's|@''HAVE_LANGINFO_CODESET''@|$(HAVE_LANGINFO_CODESET)|g' \ + -e 's|@''HAVE_LANGINFO_T_FMT_AMPM''@|$(HAVE_LANGINFO_T_FMT_AMPM)|g' \ + -e 's|@''HAVE_LANGINFO_ERA''@|$(HAVE_LANGINFO_ERA)|g' \ + -e 's|@''HAVE_LANGINFO_YESEXPR''@|$(HAVE_LANGINFO_YESEXPR)|g' \ + -e 's|@''HAVE_NL_LANGINFO''@|$(HAVE_NL_LANGINFO)|g' \ + -e 's|@''REPLACE_NL_LANGINFO''@|$(REPLACE_NL_LANGINFO)|g' \ + -e '/definitions of _GL_FUNCDECL_RPL/r $(CXXDEFS_H)' \ + -e '/definition of _GL_WARN_ON_USE/r $(WARN_ON_USE_H)' \ + < $(srcdir)/langinfo.in.h; \ + } > $@-t && \ + mv $@-t $@ +MOSTLYCLEANFILES += langinfo.h langinfo.h-t + +EXTRA_DIST += langinfo.in.h + +## end gnulib module langinfo + ## begin gnulib module lib-symbol-visibility # The value of $(CFLAG_VISIBILITY) needs to be added to the CFLAGS for the @@ -1106,6 +1148,33 @@ EXTRA_DIST += math.in.h ## end gnulib module math +## begin gnulib module mbrtowc + + +EXTRA_DIST += mbrtowc.c + +EXTRA_libgnu_la_SOURCES += mbrtowc.c + +## end gnulib module mbrtowc + +## begin gnulib module mbsinit + + +EXTRA_DIST += mbsinit.c + +EXTRA_libgnu_la_SOURCES += mbsinit.c + +## end gnulib module mbsinit + +## begin gnulib module mbtowc + + +EXTRA_DIST += mbtowc-impl.h mbtowc.c + +EXTRA_libgnu_la_SOURCES += mbtowc.c + +## end gnulib module mbtowc + ## begin gnulib module memchr @@ -1198,6 +1267,15 @@ EXTRA_DIST += netinet_in.in.h ## end gnulib module netinet_in +## begin gnulib module nl_langinfo + + +EXTRA_DIST += nl_langinfo.c + +EXTRA_libgnu_la_SOURCES += nl_langinfo.c + +## end gnulib module nl_langinfo + ## begin gnulib module nproc libgnu_la_SOURCES += nproc.c @@ -1282,6 +1360,15 @@ EXTRA_libgnu_la_SOURCES += recvfrom.c ## end gnulib module recvfrom +## begin gnulib module regex + + +EXTRA_DIST += regcomp.c regex.c regex.h regex_internal.c regex_internal.h regexec.c + +EXTRA_libgnu_la_SOURCES += regcomp.c regex.c regex_internal.c regexec.c + +## end gnulib module regex + ## begin gnulib module rename @@ -1921,6 +2008,22 @@ EXTRA_DIST += stdlib.in.h ## end gnulib module stdlib +## begin gnulib module strcase + + +EXTRA_DIST += strcasecmp.c strncasecmp.c + +EXTRA_libgnu_la_SOURCES += strcasecmp.c strncasecmp.c + +## end gnulib module strcase + +## begin gnulib module streq + + +EXTRA_DIST += streq.h + +## end gnulib module streq + ## begin gnulib module strftime libgnu_la_SOURCES += strftime.c @@ -2040,6 +2143,37 @@ EXTRA_DIST += string.in.h ## end gnulib module string +## begin gnulib module strings + +BUILT_SOURCES += strings.h + +# We need the following in order to create <strings.h> when the system +# doesn't have one that works with the given compiler. +strings.h: strings.in.h $(top_builddir)/config.status $(CXXDEFS_H) $(WARN_ON_USE_H) $(ARG_NONNULL_H) + $(AM_V_GEN)rm -f $@-t $@ && \ + { echo '/* DO NOT EDIT! GENERATED AUTOMATICALLY! */' && \ + sed -e 's|@''GUARD_PREFIX''@|GL|g' \ + -e 's|@''HAVE_STRINGS_H''@|$(HAVE_STRINGS_H)|g' \ + -e 's|@''INCLUDE_NEXT''@|$(INCLUDE_NEXT)|g' \ + -e 's|@''PRAGMA_SYSTEM_HEADER''@|@PRAGMA_SYSTEM_HEADER@|g' \ + -e 's|@''PRAGMA_COLUMNS''@|@PRAGMA_COLUMNS@|g' \ + -e 's|@''NEXT_STRINGS_H''@|$(NEXT_STRINGS_H)|g' \ + -e 's|@''GNULIB_FFS''@|$(GNULIB_FFS)|g' \ + -e 's|@''HAVE_FFS''@|$(HAVE_FFS)|g' \ + -e 's|@''HAVE_STRCASECMP''@|$(HAVE_STRCASECMP)|g' \ + -e 's|@''HAVE_DECL_STRNCASECMP''@|$(HAVE_DECL_STRNCASECMP)|g' \ + -e '/definitions of _GL_FUNCDECL_RPL/r $(CXXDEFS_H)' \ + -e '/definition of _GL_ARG_NONNULL/r $(ARG_NONNULL_H)' \ + -e '/definition of _GL_WARN_ON_USE/r $(WARN_ON_USE_H)' \ + < $(srcdir)/strings.in.h; \ + } > $@-t && \ + mv $@-t $@ +MOSTLYCLEANFILES += strings.h strings.h-t + +EXTRA_DIST += strings.in.h + +## end gnulib module strings + ## begin gnulib module sys_file BUILT_SOURCES += sys/file.h @@ -2704,6 +2838,54 @@ EXTRA_DIST += wchar.in.h ## end gnulib module wchar +## begin gnulib module wcrtomb + + +EXTRA_DIST += wcrtomb.c + +EXTRA_libgnu_la_SOURCES += wcrtomb.c + +## end gnulib module wcrtomb + +## begin gnulib module wctype-h + +BUILT_SOURCES += wctype.h + +# We need the following in order to create <wctype.h> when the system +# doesn't have one that works with the given compiler. +wctype.h: wctype.in.h $(top_builddir)/config.status $(CXXDEFS_H) $(WARN_ON_USE_H) + $(AM_V_GEN)rm -f $@-t $@ && \ + { echo '/* DO NOT EDIT! GENERATED AUTOMATICALLY! */'; \ + sed -e 's|@''GUARD_PREFIX''@|GL|g' \ + -e 's/@''HAVE_WCTYPE_H''@/$(HAVE_WCTYPE_H)/g' \ + -e 's|@''INCLUDE_NEXT''@|$(INCLUDE_NEXT)|g' \ + -e 's|@''PRAGMA_SYSTEM_HEADER''@|@PRAGMA_SYSTEM_HEADER@|g' \ + -e 's|@''PRAGMA_COLUMNS''@|@PRAGMA_COLUMNS@|g' \ + -e 's|@''NEXT_WCTYPE_H''@|$(NEXT_WCTYPE_H)|g' \ + -e 's/@''GNULIB_ISWBLANK''@/$(GNULIB_ISWBLANK)/g' \ + -e 's/@''GNULIB_WCTYPE''@/$(GNULIB_WCTYPE)/g' \ + -e 's/@''GNULIB_ISWCTYPE''@/$(GNULIB_ISWCTYPE)/g' \ + -e 's/@''GNULIB_WCTRANS''@/$(GNULIB_WCTRANS)/g' \ + -e 's/@''GNULIB_TOWCTRANS''@/$(GNULIB_TOWCTRANS)/g' \ + -e 's/@''HAVE_ISWBLANK''@/$(HAVE_ISWBLANK)/g' \ + -e 's/@''HAVE_ISWCNTRL''@/$(HAVE_ISWCNTRL)/g' \ + -e 's/@''HAVE_WCTYPE_T''@/$(HAVE_WCTYPE_T)/g' \ + -e 's/@''HAVE_WCTRANS_T''@/$(HAVE_WCTRANS_T)/g' \ + -e 's/@''HAVE_WINT_T''@/$(HAVE_WINT_T)/g' \ + -e 's/@''REPLACE_ISWBLANK''@/$(REPLACE_ISWBLANK)/g' \ + -e 's/@''REPLACE_ISWCNTRL''@/$(REPLACE_ISWCNTRL)/g' \ + -e 's/@''REPLACE_TOWLOWER''@/$(REPLACE_TOWLOWER)/g' \ + -e '/definitions of _GL_FUNCDECL_RPL/r $(CXXDEFS_H)' \ + -e '/definition of _GL_WARN_ON_USE/r $(WARN_ON_USE_H)' \ + < $(srcdir)/wctype.in.h; \ + } > $@-t && \ + mv $@-t $@ +MOSTLYCLEANFILES += wctype.h wctype.h-t + +EXTRA_DIST += wctype.in.h + +## end gnulib module wctype-h + ## begin gnulib module write diff --git a/lib/btowc.c b/lib/btowc.c new file mode 100644 index 000000000..485e99554 --- /dev/null +++ b/lib/btowc.c @@ -0,0 +1,39 @@ +/* Convert unibyte character to wide character. + Copyright (C) 2008, 2010-2012 Free Software Foundation, Inc. + Written by Bruno Haible <bruno@clisp.org>, 2008. + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. */ + +#include <config.h> + +/* Specification. */ +#include <wchar.h> + +#include <stdio.h> +#include <stdlib.h> + +wint_t +btowc (int c) +{ + if (c != EOF) + { + char buf[1]; + wchar_t wc; + + buf[0] = c; + if (mbtowc (&wc, buf, 1) >= 0) + return wc; + } + return WEOF; +} diff --git a/lib/langinfo.in.h b/lib/langinfo.in.h new file mode 100644 index 000000000..807b245d2 --- /dev/null +++ b/lib/langinfo.in.h @@ -0,0 +1,177 @@ +/* Substitute for and wrapper around <langinfo.h>. + Copyright (C) 2009-2012 Free Software Foundation, Inc. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2, or (at your option) + any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with this program; if not, write to the Free Software Foundation, + Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ + +/* + * POSIX <langinfo.h> for platforms that lack it or have an incomplete one. + * <http://www.opengroup.org/onlinepubs/9699919799/basedefs/langinfo.h.html> + */ + +#ifndef _@GUARD_PREFIX@_LANGINFO_H + +#if __GNUC__ >= 3 +@PRAGMA_SYSTEM_HEADER@ +#endif +@PRAGMA_COLUMNS@ + +/* The include_next requires a split double-inclusion guard. */ +#if @HAVE_LANGINFO_H@ +# @INCLUDE_NEXT@ @NEXT_LANGINFO_H@ +#endif + +#ifndef _@GUARD_PREFIX@_LANGINFO_H +#define _@GUARD_PREFIX@_LANGINFO_H + + +#if !@HAVE_LANGINFO_H@ + +/* A platform that lacks <langinfo.h>. */ + +/* Assume that it also lacks <nl_types.h> and the nl_item type. */ +# if !GNULIB_defined_nl_item +typedef int nl_item; +# define GNULIB_defined_nl_item 1 +# endif + +/* nl_langinfo items of the LC_CTYPE category */ +# define CODESET 10000 +/* nl_langinfo items of the LC_NUMERIC category */ +# define RADIXCHAR 10001 +# define THOUSEP 10002 +/* nl_langinfo items of the LC_TIME category */ +# define D_T_FMT 10003 +# define D_FMT 10004 +# define T_FMT 10005 +# define T_FMT_AMPM 10006 +# define AM_STR 10007 +# define PM_STR 10008 +# define DAY_1 10009 +# define DAY_2 (DAY_1 + 1) +# define DAY_3 (DAY_1 + 2) +# define DAY_4 (DAY_1 + 3) +# define DAY_5 (DAY_1 + 4) +# define DAY_6 (DAY_1 + 5) +# define DAY_7 (DAY_1 + 6) +# define ABDAY_1 10016 +# define ABDAY_2 (ABDAY_1 + 1) +# define ABDAY_3 (ABDAY_1 + 2) +# define ABDAY_4 (ABDAY_1 + 3) +# define ABDAY_5 (ABDAY_1 + 4) +# define ABDAY_6 (ABDAY_1 + 5) +# define ABDAY_7 (ABDAY_1 + 6) +# define MON_1 10023 +# define MON_2 (MON_1 + 1) +# define MON_3 (MON_1 + 2) +# define MON_4 (MON_1 + 3) +# define MON_5 (MON_1 + 4) +# define MON_6 (MON_1 + 5) +# define MON_7 (MON_1 + 6) +# define MON_8 (MON_1 + 7) +# define MON_9 (MON_1 + 8) +# define MON_10 (MON_1 + 9) +# define MON_11 (MON_1 + 10) +# define MON_12 (MON_1 + 11) +# define ABMON_1 10035 +# define ABMON_2 (ABMON_1 + 1) +# define ABMON_3 (ABMON_1 + 2) +# define ABMON_4 (ABMON_1 + 3) +# define ABMON_5 (ABMON_1 + 4) +# define ABMON_6 (ABMON_1 + 5) +# define ABMON_7 (ABMON_1 + 6) +# define ABMON_8 (ABMON_1 + 7) +# define ABMON_9 (ABMON_1 + 8) +# define ABMON_10 (ABMON_1 + 9) +# define ABMON_11 (ABMON_1 + 10) +# define ABMON_12 (ABMON_1 + 11) +# define ERA 10047 +# define ERA_D_FMT 10048 +# define ERA_D_T_FMT 10049 +# define ERA_T_FMT 10050 +# define ALT_DIGITS 10051 +/* nl_langinfo items of the LC_MONETARY category */ +# define CRNCYSTR 10052 +/* nl_langinfo items of the LC_MESSAGES category */ +# define YESEXPR 10053 +# define NOEXPR 10054 + +#else + +/* A platform that has <langinfo.h>. */ + +# if !@HAVE_LANGINFO_CODESET@ +# define CODESET 10000 +# define GNULIB_defined_CODESET 1 +# endif + +# if !@HAVE_LANGINFO_T_FMT_AMPM@ +# define T_FMT_AMPM 10006 +# define GNULIB_defined_T_FMT_AMPM 1 +# endif + +# if !@HAVE_LANGINFO_ERA@ +# define ERA 10047 +# define ERA_D_FMT 10048 +# define ERA_D_T_FMT 10049 +# define ERA_T_FMT 10050 +# define ALT_DIGITS 10051 +# define GNULIB_defined_ERA 1 +# endif + +# if !@HAVE_LANGINFO_YESEXPR@ +# define YESEXPR 10053 +# define NOEXPR 10054 +# define GNULIB_defined_YESEXPR 1 +# endif + +#endif + +/* The definitions of _GL_FUNCDECL_RPL etc. are copied here. */ + +/* The definition of _GL_WARN_ON_USE is copied here. */ + +/* Declare overridden functions. */ + + +/* Return a piece of locale dependent information. + Note: The difference between nl_langinfo (CODESET) and locale_charset () + is that the latter normalizes the encoding names to GNU conventions. */ + +#if @GNULIB_NL_LANGINFO@ +# if @REPLACE_NL_LANGINFO@ +# if !(defined __cplusplus && defined GNULIB_NAMESPACE) +# undef nl_langinfo +# define nl_langinfo rpl_nl_langinfo +# endif +_GL_FUNCDECL_RPL (nl_langinfo, char *, (nl_item item)); +_GL_CXXALIAS_RPL (nl_langinfo, char *, (nl_item item)); +# else +# if !@HAVE_NL_LANGINFO@ +_GL_FUNCDECL_SYS (nl_langinfo, char *, (nl_item item)); +# endif +_GL_CXXALIAS_SYS (nl_langinfo, char *, (nl_item item)); +# endif +_GL_CXXALIASWARN (nl_langinfo); +#elif defined GNULIB_POSIXCHECK +# undef nl_langinfo +# if HAVE_RAW_DECL_NL_LANGINFO +_GL_WARN_ON_USE (nl_langinfo, "nl_langinfo is not portable - " + "use gnulib module nl_langinfo for portability"); +# endif +#endif + + +#endif /* _@GUARD_PREFIX@_LANGINFO_H */ +#endif /* _@GUARD_PREFIX@_LANGINFO_H */ diff --git a/lib/mbrtowc.c b/lib/mbrtowc.c new file mode 100644 index 000000000..05fb14847 --- /dev/null +++ b/lib/mbrtowc.c @@ -0,0 +1,396 @@ +/* Convert multibyte character to wide character. + Copyright (C) 1999-2002, 2005-2012 Free Software Foundation, Inc. + Written by Bruno Haible <bruno@clisp.org>, 2008. + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. */ + +#include <config.h> + +/* Specification. */ +#include <wchar.h> + +#if GNULIB_defined_mbstate_t +/* Implement mbrtowc() on top of mbtowc(). */ + +# include <errno.h> +# include <stdlib.h> + +# include "localcharset.h" +# include "streq.h" +# include "verify.h" + + +verify (sizeof (mbstate_t) >= 4); + +static char internal_state[4]; + +size_t +mbrtowc (wchar_t *pwc, const char *s, size_t n, mbstate_t *ps) +{ + char *pstate = (char *)ps; + + if (s == NULL) + { + pwc = NULL; + s = ""; + n = 1; + } + + if (n == 0) + return (size_t)(-2); + + /* Here n > 0. */ + + if (pstate == NULL) + pstate = internal_state; + + { + size_t nstate = pstate[0]; + char buf[4]; + const char *p; + size_t m; + + switch (nstate) + { + case 0: + p = s; + m = n; + break; + case 3: + buf[2] = pstate[3]; + /*FALLTHROUGH*/ + case 2: + buf[1] = pstate[2]; + /*FALLTHROUGH*/ + case 1: + buf[0] = pstate[1]; + p = buf; + m = nstate; + buf[m++] = s[0]; + if (n >= 2 && m < 4) + { + buf[m++] = s[1]; + if (n >= 3 && m < 4) + buf[m++] = s[2]; + } + break; + default: + errno = EINVAL; + return (size_t)(-1); + } + + /* Here m > 0. */ + +# if __GLIBC__ || defined __UCLIBC__ + /* Work around bug <http://sourceware.org/bugzilla/show_bug.cgi?id=9674> */ + mbtowc (NULL, NULL, 0); +# endif + { + int res = mbtowc (pwc, p, m); + + if (res >= 0) + { + if (pwc != NULL && ((*pwc == 0) != (res == 0))) + abort (); + if (nstate >= (res > 0 ? res : 1)) + abort (); + res -= nstate; + pstate[0] = 0; + return res; + } + + /* mbtowc does not distinguish between invalid and incomplete multibyte + sequences. But mbrtowc needs to make this distinction. + There are two possible approaches: + - Use iconv() and its return value. + - Use built-in knowledge about the possible encodings. + Given the low quality of implementation of iconv() on the systems that + lack mbrtowc(), we use the second approach. + The possible encodings are: + - 8-bit encodings, + - EUC-JP, EUC-KR, GB2312, EUC-TW, BIG5, GB18030, SJIS, + - UTF-8. + Use specialized code for each. */ + if (m >= 4 || m >= MB_CUR_MAX) + goto invalid; + /* Here MB_CUR_MAX > 1 and 0 < m < 4. */ + { + const char *encoding = locale_charset (); + + if (STREQ (encoding, "UTF-8", 'U', 'T', 'F', '-', '8', 0, 0, 0, 0)) + { + /* Cf. unistr/u8-mblen.c. */ + unsigned char c = (unsigned char) p[0]; + + if (c >= 0xc2) + { + if (c < 0xe0) + { + if (m == 1) + goto incomplete; + } + else if (c < 0xf0) + { + if (m == 1) + goto incomplete; + if (m == 2) + { + unsigned char c2 = (unsigned char) p[1]; + + if ((c2 ^ 0x80) < 0x40 + && (c >= 0xe1 || c2 >= 0xa0) + && (c != 0xed || c2 < 0xa0)) + goto incomplete; + } + } + else if (c <= 0xf4) + { + if (m == 1) + goto incomplete; + else /* m == 2 || m == 3 */ + { + unsigned char c2 = (unsigned char) p[1]; + + if ((c2 ^ 0x80) < 0x40 + && (c >= 0xf1 || c2 >= 0x90) + && (c < 0xf4 || (c == 0xf4 && c2 < 0x90))) + { + if (m == 2) + goto incomplete; + else /* m == 3 */ + { + unsigned char c3 = (unsigned char) p[2]; + + if ((c3 ^ 0x80) < 0x40) + goto incomplete; + } + } + } + } + } + goto invalid; + } + + /* As a reference for this code, you can use the GNU libiconv + implementation. Look for uses of the RET_TOOFEW macro. */ + + if (STREQ (encoding, "EUC-JP", 'E', 'U', 'C', '-', 'J', 'P', 0, 0, 0)) + { + if (m == 1) + { + unsigned char c = (unsigned char) p[0]; + + if ((c >= 0xa1 && c < 0xff) || c == 0x8e || c == 0x8f) + goto incomplete; + } + if (m == 2) + { + unsigned char c = (unsigned char) p[0]; + + if (c == 0x8f) + { + unsigned char c2 = (unsigned char) p[1]; + + if (c2 >= 0xa1 && c2 < 0xff) + goto incomplete; + } + } + goto invalid; + } + if (STREQ (encoding, "EUC-KR", 'E', 'U', 'C', '-', 'K', 'R', 0, 0, 0) + || STREQ (encoding, "GB2312", 'G', 'B', '2', '3', '1', '2', 0, 0, 0) + || STREQ (encoding, "BIG5", 'B', 'I', 'G', '5', 0, 0, 0, 0, 0)) + { + if (m == 1) + { + unsigned char c = (unsigned char) p[0]; + + if (c >= 0xa1 && c < 0xff) + goto incomplete; + } + goto invalid; + } + if (STREQ (encoding, "EUC-TW", 'E', 'U', 'C', '-', 'T', 'W', 0, 0, 0)) + { + if (m == 1) + { + unsigned char c = (unsigned char) p[0]; + + if ((c >= 0xa1 && c < 0xff) || c == 0x8e) + goto incomplete; + } + else /* m == 2 || m == 3 */ + { + unsigned char c = (unsigned char) p[0]; + + if (c == 0x8e) + goto incomplete; + } + goto invalid; + } + if (STREQ (encoding, "GB18030", 'G', 'B', '1', '8', '0', '3', '0', 0, 0)) + { + if (m == 1) + { + unsigned char c = (unsigned char) p[0]; + + if ((c >= 0x90 && c <= 0xe3) || (c >= 0xf8 && c <= 0xfe)) + goto incomplete; + } + else /* m == 2 || m == 3 */ + { + unsigned char c = (unsigned char) p[0]; + + if (c >= 0x90 && c <= 0xe3) + { + unsigned char c2 = (unsigned char) p[1]; + + if (c2 >= 0x30 && c2 <= 0x39) + { + if (m == 2) + goto incomplete; + else /* m == 3 */ + { + unsigned char c3 = (unsigned char) p[2]; + + if (c3 >= 0x81 && c3 <= 0xfe) + goto incomplete; + } + } + } + } + goto invalid; + } + if (STREQ (encoding, "SJIS", 'S', 'J', 'I', 'S', 0, 0, 0, 0, 0)) + { + if (m == 1) + { + unsigned char c = (unsigned char) p[0]; + + if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xea) + || (c >= 0xf0 && c <= 0xf9)) + goto incomplete; + } + goto invalid; + } + + /* An unknown multibyte encoding. */ + goto incomplete; + } + + incomplete: + { + size_t k = nstate; + /* Here 0 <= k < m < 4. */ + pstate[++k] = s[0]; + if (k < m) + { + pstate[++k] = s[1]; + if (k < m) + pstate[++k] = s[2]; + } + if (k != m) + abort (); + } + pstate[0] = m; + return (size_t)(-2); + + invalid: + errno = EILSEQ; + /* The conversion state is undefined, says POSIX. */ + return (size_t)(-1); + } + } +} + +#else +/* Override the system's mbrtowc() function. */ + +# undef mbrtowc + +size_t +rpl_mbrtowc (wchar_t *pwc, const char *s, size_t n, mbstate_t *ps) +{ +# if MBRTOWC_NULL_ARG2_BUG || MBRTOWC_RETVAL_BUG + if (s == NULL) + { + pwc = NULL; + s = ""; + n = 1; + } +# endif + +# if MBRTOWC_RETVAL_BUG + { + static mbstate_t internal_state; + + /* Override mbrtowc's internal state. We cannot call mbsinit() on the + hidden internal state, but we can call it on our variable. */ + if (ps == NULL) + ps = &internal_state; + + if (!mbsinit (ps)) + { + /* Parse the rest of the multibyte character byte for byte. */ + size_t count = 0; + for (; n > 0; s++, n--) + { + wchar_t wc; + size_t ret = mbrtowc (&wc, s, 1, ps); + + if (ret == (size_t)(-1)) + return (size_t)(-1); + count++; + if (ret != (size_t)(-2)) + { + /* The multibyte character has been completed. */ + if (pwc != NULL) + *pwc = wc; + return (wc == 0 ? 0 : count); + } + } + return (size_t)(-2); + } + } +# endif + +# if MBRTOWC_NUL_RETVAL_BUG + { + wchar_t wc; + size_t ret = mbrtowc (&wc, s, n, ps); + + if (ret != (size_t)(-1) && ret != (size_t)(-2)) + { + if (pwc != NULL) + *pwc = wc; + if (wc == 0) + ret = 0; + } + return ret; + } +# else + { +# if MBRTOWC_NULL_ARG1_BUG + wchar_t dummy; + + if (pwc == NULL) + pwc = &dummy; +# endif + + return mbrtowc (pwc, s, n, ps); + } +# endif +} + +#endif diff --git a/lib/mbsinit.c b/lib/mbsinit.c new file mode 100644 index 000000000..79278d452 --- /dev/null +++ b/lib/mbsinit.c @@ -0,0 +1,61 @@ +/* Test for initial conversion state. + Copyright (C) 2008-2012 Free Software Foundation, Inc. + Written by Bruno Haible <bruno@clisp.org>, 2008. + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. */ + +#include <config.h> + +/* Specification. */ +#include <wchar.h> + +#include "verify.h" + +#if (defined _WIN32 || defined __WIN32__) && !defined __CYGWIN__ + +/* On native Windows, 'mbstate_t' is defined as 'int'. */ + +int +mbsinit (const mbstate_t *ps) +{ + return ps == NULL || *ps == 0; +} + +#else + +/* Platforms that lack mbsinit() also lack mbrlen(), mbrtowc(), mbsrtowcs() + and wcrtomb(), wcsrtombs(). + We assume that + - sizeof (mbstate_t) >= 4, + - only stateless encodings are supported (such as UTF-8 and EUC-JP, but + not ISO-2022 variants), + - for each encoding, the number of bytes for a wide character is <= 4. + (This maximum is attained for UTF-8, GB18030, EUC-TW.) + We define the meaning of mbstate_t as follows: + - In mb -> wc direction, mbstate_t's first byte contains the number of + buffered bytes (in the range 0..3), followed by up to 3 buffered bytes. + - In wc -> mb direction, mbstate_t contains no information. In other + words, it is always in the initial state. */ + +verify (sizeof (mbstate_t) >= 4); + +int +mbsinit (const mbstate_t *ps) +{ + const char *pstate = (const char *)ps; + + return pstate == NULL || pstate[0] == 0; +} + +#endif diff --git a/lib/mbtowc-impl.h b/lib/mbtowc-impl.h new file mode 100644 index 000000000..3183f918a --- /dev/null +++ b/lib/mbtowc-impl.h @@ -0,0 +1,44 @@ +/* Convert multibyte character to wide character. + Copyright (C) 2011-2012 Free Software Foundation, Inc. + Written by Bruno Haible <bruno@clisp.org>, 2011. + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. */ + +/* We don't need a static internal state, because the encoding is not state + dependent, and when mbrtowc returns (size_t)(-2). we throw the result + away. */ + +int +mbtowc (wchar_t *pwc, const char *s, size_t n) +{ + if (s == NULL) + return 0; + else + { + mbstate_t state; + wchar_t wc; + size_t result; + + memset (&state, 0, sizeof (mbstate_t)); + result = mbrtowc (&wc, s, n, &state); + if (result == (size_t)-1 || result == (size_t)-2) + { + errno = EILSEQ; + return -1; + } + if (pwc != NULL) + *pwc = wc; + return (wc == 0 ? 0 : result); + } +} diff --git a/lib/mbtowc.c b/lib/mbtowc.c new file mode 100644 index 000000000..e48b2f276 --- /dev/null +++ b/lib/mbtowc.c @@ -0,0 +1,26 @@ +/* Convert multibyte character to wide character. + Copyright (C) 2011-2012 Free Software Foundation, Inc. + Written by Bruno Haible <bruno@clisp.org>, 2011. + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. */ + +#include <config.h> + +#include <stdlib.h> + +#include <errno.h> +#include <string.h> +#include <wchar.h> + +#include "mbtowc-impl.h" diff --git a/lib/nl_langinfo.c b/lib/nl_langinfo.c new file mode 100644 index 000000000..4b9bdbe1b --- /dev/null +++ b/lib/nl_langinfo.c @@ -0,0 +1,271 @@ +/* nl_langinfo() replacement: query locale dependent information. + + Copyright (C) 2007-2012 Free Software Foundation, Inc. + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. */ + +#include <config.h> + +/* Specification. */ +#include <langinfo.h> + +#if REPLACE_NL_LANGINFO + +/* Override nl_langinfo with support for added nl_item values. */ + +# include <locale.h> +# include <string.h> + +# undef nl_langinfo + +char * +rpl_nl_langinfo (nl_item item) +{ + switch (item) + { +# if GNULIB_defined_CODESET + case CODESET: + { + const char *locale; + static char buf[2 + 10 + 1]; + + locale = setlocale (LC_CTYPE, NULL); + if (locale != NULL && locale[0] != '\0') + { + /* If the locale name contains an encoding after the dot, return + it. */ + const char *dot = strchr (locale, '.'); + + if (dot != NULL) + { + const char *modifier; + + dot++; + /* Look for the possible @... trailer and remove it, if any. */ + modifier = strchr (dot, '@'); + if (modifier == NULL) + return dot; + if (modifier - dot < sizeof (buf)) + { + memcpy (buf, dot, modifier - dot); + buf [modifier - dot] = '\0'; + return buf; + } + } + } + return ""; + } +# endif +# if GNULIB_defined_T_FMT_AMPM + case T_FMT_AMPM: + return "%I:%M:%S %p"; +# endif +# if GNULIB_defined_ERA + case ERA: + /* The format is not standardized. In glibc it is a sequence of strings + of the form "direction:offset:start_date:end_date:era_name:era_format" + with an empty string at the end. */ + return ""; + case ERA_D_FMT: + /* The %Ex conversion in strftime behaves like %x if the locale does not + have an alternative time format. */ + item = D_FMT; + break; + case ERA_D_T_FMT: + /* The %Ec conversion in strftime behaves like %c if the locale does not + have an alternative time format. */ + item = D_T_FMT; + break; + case ERA_T_FMT: + /* The %EX conversion in strftime behaves like %X if the locale does not + have an alternative time format. */ + item = T_FMT; + break; + case ALT_DIGITS: + /* The format is not standardized. In glibc it is a sequence of 10 + strings, appended in memory. */ + return "\0\0\0\0\0\0\0\0\0\0"; +# endif +# if GNULIB_defined_YESEXPR || !FUNC_NL_LANGINFO_YESEXPR_WORKS + case YESEXPR: + return "^[yY]"; + case NOEXPR: + return "^[nN]"; +# endif + default: + break; + } + return nl_langinfo (item); +} + +#else + +/* Provide nl_langinfo from scratch. */ + +# if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ + +/* Native Windows platforms. */ + +# define WIN32_LEAN_AND_MEAN /* avoid including junk */ +# include <windows.h> + +# include <stdio.h> + +# else + +/* An old Unix platform without locales, such as Linux libc5 or BeOS. */ + +# endif + +# include <locale.h> + +char * +nl_langinfo (nl_item item) +{ + switch (item) + { + /* nl_langinfo items of the LC_CTYPE category */ + case CODESET: +# if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__ + { + static char buf[2 + 10 + 1]; + + /* The Windows API has a function returning the locale's codepage as + a number. */ + sprintf (buf, "CP%u", GetACP ()); + return buf; + } +# elif defined __BEOS__ + return "UTF-8"; +# else + return "ISO-8859-1"; +# endif + /* nl_langinfo items of the LC_NUMERIC category */ + case RADIXCHAR: + return localeconv () ->decimal_point; + case THOUSEP: + return localeconv () ->thousands_sep; + /* nl_langinfo items of the LC_TIME category. + TODO: Really use the locale. */ + case D_T_FMT: + case ERA_D_T_FMT: + return "%a %b %e %H:%M:%S %Y"; + case D_FMT: + case ERA_D_FMT: + return "%m/%d/%y"; + case T_FMT: + case ERA_T_FMT: + return "%H:%M:%S"; + case T_FMT_AMPM: + return "%I:%M:%S %p"; + case AM_STR: + return "AM"; + case PM_STR: + return "PM"; + case DAY_1: + return "Sunday"; + case DAY_2: + return "Monday"; + case DAY_3: + return "Tuesday"; + case DAY_4: + return "Wednesday"; + case DAY_5: + return "Thursday"; + case DAY_6: + return "Friday"; + case DAY_7: + return "Saturday"; + case ABDAY_1: + return "Sun"; + case ABDAY_2: + return "Mon"; + case ABDAY_3: + return "Tue"; + case ABDAY_4: + return "Wed"; + case ABDAY_5: + return "Thu"; + case ABDAY_6: + return "Fri"; + case ABDAY_7: + return "Sat"; + case MON_1: + return "January"; + case MON_2: + return "February"; + case MON_3: + return "March"; + case MON_4: + return "April"; + case MON_5: + return "May"; + case MON_6: + return "June"; + case MON_7: + return "July"; + case MON_8: + return "August"; + case MON_9: + return "September"; + case MON_10: + return "October"; + case MON_11: + return "November"; + case MON_12: + return "December"; + case ABMON_1: + return "Jan"; + case ABMON_2: + return "Feb"; + case ABMON_3: + return "Mar"; + case ABMON_4: + return "Apr"; + case ABMON_5: + return "May"; + case ABMON_6: + return "Jun"; + case ABMON_7: + return "Jul"; + case ABMON_8: + return "Aug"; + case ABMON_9: + return "Sep"; + case ABMON_10: + return "Oct"; + case ABMON_11: + return "Nov"; + case ABMON_12: + return "Dec"; + case ERA: + return ""; + case ALT_DIGITS: + return "\0\0\0\0\0\0\0\0\0\0"; + /* nl_langinfo items of the LC_MONETARY category + TODO: Really use the locale. */ + case CRNCYSTR: + return "-"; + /* nl_langinfo items of the LC_MESSAGES category + TODO: Really use the locale. */ + case YESEXPR: + return "^[yY]"; + case NOEXPR: + return "^[nN]"; + default: + return ""; + } +} + +#endif diff --git a/lib/regcomp.c b/lib/regcomp.c new file mode 100644 index 000000000..ac1330779 --- /dev/null +++ b/lib/regcomp.c @@ -0,0 +1,3876 @@ +/* Extended regular expression matching and search library. + Copyright (C) 2002-2012 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2, or (at your option) + any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License along + with this program; if not, write to the Free Software Foundation, + Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ + +static reg_errcode_t re_compile_internal (regex_t *preg, const char * pattern, + size_t length, reg_syntax_t syntax); +static void re_compile_fastmap_iter (regex_t *bufp, + const re_dfastate_t *init_state, + char *fastmap); +static reg_errcode_t init_dfa (re_dfa_t *dfa, size_t pat_len); +#ifdef RE_ENABLE_I18N +static void free_charset (re_charset_t *cset); +#endif /* RE_ENABLE_I18N */ +static void free_workarea_compile (regex_t *preg); +static reg_errcode_t create_initial_state (re_dfa_t *dfa); +#ifdef RE_ENABLE_I18N +static void optimize_utf8 (re_dfa_t *dfa); +#endif +static reg_errcode_t analyze (regex_t *preg); +static reg_errcode_t preorder (bin_tree_t *root, + reg_errcode_t (fn (void *, bin_tree_t *)), + void *extra); +static reg_errcode_t postorder (bin_tree_t *root, + reg_errcode_t (fn (void *, bin_tree_t *)), + void *extra); +static reg_errcode_t optimize_subexps (void *extra, bin_tree_t *node); +static reg_errcode_t lower_subexps (void *extra, bin_tree_t *node); +static bin_tree_t *lower_subexp (reg_errcode_t *err, regex_t *preg, + bin_tree_t *node); +static reg_errcode_t calc_first (void *extra, bin_tree_t *node); +static reg_errcode_t calc_next (void *extra, bin_tree_t *node); +static reg_errcode_t link_nfa_nodes (void *extra, bin_tree_t *node); +static Idx duplicate_node (re_dfa_t *dfa, Idx org_idx, unsigned int constraint); +static Idx search_duplicated_node (const re_dfa_t *dfa, Idx org_node, + unsigned int constraint); +static reg_errcode_t calc_eclosure (re_dfa_t *dfa); +static reg_errcode_t calc_eclosure_iter (re_node_set *new_set, re_dfa_t *dfa, + Idx node, bool root); +static reg_errcode_t calc_inveclosure (re_dfa_t *dfa); +static Idx fetch_number (re_string_t *input, re_token_t *token, + reg_syntax_t syntax); +static int peek_token (re_token_t *token, re_string_t *input, + reg_syntax_t syntax) internal_function; +static bin_tree_t *parse (re_string_t *regexp, regex_t *preg, + reg_syntax_t syntax, reg_errcode_t *err); +static bin_tree_t *parse_reg_exp (re_string_t *regexp, regex_t *preg, + re_token_t *token, reg_syntax_t syntax, + Idx nest, reg_errcode_t *err); +static bin_tree_t *parse_branch (re_string_t *regexp, regex_t *preg, + re_token_t *token, reg_syntax_t syntax, + Idx nest, reg_errcode_t *err); +static bin_tree_t *parse_expression (re_string_t *regexp, regex_t *preg, + re_token_t *token, reg_syntax_t syntax, + Idx nest, reg_errcode_t *err); +static bin_tree_t *parse_sub_exp (re_string_t *regexp, regex_t *preg, + re_token_t *token, reg_syntax_t syntax, + Idx nest, reg_errcode_t *err); +static bin_tree_t *parse_dup_op (bin_tree_t *dup_elem, re_string_t *regexp, + re_dfa_t *dfa, re_token_t *token, + reg_syntax_t syntax, reg_errcode_t *err); +static bin_tree_t *parse_bracket_exp (re_string_t *regexp, re_dfa_t *dfa, + re_token_t *token, reg_syntax_t syntax, + reg_errcode_t *err); +static reg_errcode_t parse_bracket_element (bracket_elem_t *elem, + re_string_t *regexp, + re_token_t *token, int token_len, + re_dfa_t *dfa, + reg_syntax_t syntax, + bool accept_hyphen); +static reg_errcode_t parse_bracket_symbol (bracket_elem_t *elem, + re_string_t *regexp, + re_token_t *token); +#ifdef RE_ENABLE_I18N +static reg_errcode_t build_equiv_class (bitset_t sbcset, + re_charset_t *mbcset, + Idx *equiv_class_alloc, + const unsigned char *name); +static reg_errcode_t build_charclass (RE_TRANSLATE_TYPE trans, + bitset_t sbcset, + re_charset_t *mbcset, + Idx *char_class_alloc, + const unsigned char *class_name, + reg_syntax_t syntax); +#else /* not RE_ENABLE_I18N */ +static reg_errcode_t build_equiv_class (bitset_t sbcset, + const unsigned char *name); +static reg_errcode_t build_charclass (RE_TRANSLATE_TYPE trans, + bitset_t sbcset, + const unsigned char *class_name, + reg_syntax_t syntax); +#endif /* not RE_ENABLE_I18N */ +static bin_tree_t *build_charclass_op (re_dfa_t *dfa, + RE_TRANSLATE_TYPE trans, + const unsigned char *class_name, + const unsigned char *extra, + bool non_match, reg_errcode_t *err); +static bin_tree_t *create_tree (re_dfa_t *dfa, + bin_tree_t *left, bin_tree_t *right, + re_token_type_t type); +static bin_tree_t *create_token_tree (re_dfa_t *dfa, + bin_tree_t *left, bin_tree_t *right, + const re_token_t *token); +static bin_tree_t *duplicate_tree (const bin_tree_t *src, re_dfa_t *dfa); +static void free_token (re_token_t *node); +static reg_errcode_t free_tree (void *extra, bin_tree_t *node); +static reg_errcode_t mark_opt_subexp (void *extra, bin_tree_t *node); + +/* This table gives an error message for each of the error codes listed + in regex.h. Obviously the order here has to be same as there. + POSIX doesn't require that we do anything for REG_NOERROR, + but why not be nice? */ + +static const char __re_error_msgid[] = + { +#define REG_NOERROR_IDX 0 + gettext_noop ("Success") /* REG_NOERROR */ + "\0" +#define REG_NOMATCH_IDX (REG_NOERROR_IDX + sizeof "Success") + gettext_noop ("No match") /* REG_NOMATCH */ + "\0" +#define REG_BADPAT_IDX (REG_NOMATCH_IDX + sizeof "No match") + gettext_noop ("Invalid regular expression") /* REG_BADPAT */ + "\0" +#define REG_ECOLLATE_IDX (REG_BADPAT_IDX + sizeof "Invalid regular expression") + gettext_noop ("Invalid collation character") /* REG_ECOLLATE */ + "\0" +#define REG_ECTYPE_IDX (REG_ECOLLATE_IDX + sizeof "Invalid collation character") + gettext_noop ("Invalid character class name") /* REG_ECTYPE */ + "\0" +#define REG_EESCAPE_IDX (REG_ECTYPE_IDX + sizeof "Invalid character class name") + gettext_noop ("Trailing backslash") /* REG_EESCAPE */ + "\0" +#define REG_ESUBREG_IDX (REG_EESCAPE_IDX + sizeof "Trailing backslash") + gettext_noop ("Invalid back reference") /* REG_ESUBREG */ + "\0" +#define REG_EBRACK_IDX (REG_ESUBREG_IDX + sizeof "Invalid back reference") + gettext_noop ("Unmatched [ or [^") /* REG_EBRACK */ + "\0" +#define REG_EPAREN_IDX (REG_EBRACK_IDX + sizeof "Unmatched [ or [^") + gettext_noop ("Unmatched ( or \\(") /* REG_EPAREN */ + "\0" +#define REG_EBRACE_IDX (REG_EPAREN_IDX + sizeof "Unmatched ( or \\(") + gettext_noop ("Unmatched \\{") /* REG_EBRACE */ + "\0" +#define REG_BADBR_IDX (REG_EBRACE_IDX + sizeof "Unmatched \\{") + gettext_noop ("Invalid content of \\{\\}") /* REG_BADBR */ + "\0" +#define REG_ERANGE_IDX (REG_BADBR_IDX + sizeof "Invalid content of \\{\\}") + gettext_noop ("Invalid range end") /* REG_ERANGE */ + "\0" +#define REG_ESPACE_IDX (REG_ERANGE_IDX + sizeof "Invalid range end") + gettext_noop ("Memory exhausted") /* REG_ESPACE */ + "\0" +#define REG_BADRPT_IDX (REG_ESPACE_IDX + sizeof "Memory exhausted") + gettext_noop ("Invalid preceding regular expression") /* REG_BADRPT */ + "\0" +#define REG_EEND_IDX (REG_BADRPT_IDX + sizeof "Invalid preceding regular expression") + gettext_noop ("Premature end of regular expression") /* REG_EEND */ + "\0" +#define REG_ESIZE_IDX (REG_EEND_IDX + sizeof "Premature end of regular expression") + gettext_noop ("Regular expression too big") /* REG_ESIZE */ + "\0" +#define REG_ERPAREN_IDX (REG_ESIZE_IDX + sizeof "Regular expression too big") + gettext_noop ("Unmatched ) or \\)") /* REG_ERPAREN */ + }; + +static const size_t __re_error_msgid_idx[] = + { + REG_NOERROR_IDX, + REG_NOMATCH_IDX, + REG_BADPAT_IDX, + REG_ECOLLATE_IDX, + REG_ECTYPE_IDX, + REG_EESCAPE_IDX, + REG_ESUBREG_IDX, + REG_EBRACK_IDX, + REG_EPAREN_IDX, + REG_EBRACE_IDX, + REG_BADBR_IDX, + REG_ERANGE_IDX, + REG_ESPACE_IDX, + REG_BADRPT_IDX, + REG_EEND_IDX, + REG_ESIZE_IDX, + REG_ERPAREN_IDX + }; + +/* Entry points for GNU code. */ + +/* re_compile_pattern is the GNU regular expression compiler: it + compiles PATTERN (of length LENGTH) and puts the result in BUFP. + Returns 0 if the pattern was valid, otherwise an error string. + + Assumes the 'allocated' (and perhaps 'buffer') and 'translate' fields + are set in BUFP on entry. */ + +#ifdef _LIBC +const char * +re_compile_pattern (pattern, length, bufp) + const char *pattern; + size_t length; + struct re_pattern_buffer *bufp; +#else /* size_t might promote */ +const char * +re_compile_pattern (const char *pattern, size_t length, + struct re_pattern_buffer *bufp) +#endif +{ + reg_errcode_t ret; + + /* And GNU code determines whether or not to get register information + by passing null for the REGS argument to re_match, etc., not by + setting no_sub, unless RE_NO_SUB is set. */ + bufp->no_sub = !!(re_syntax_options & RE_NO_SUB); + + /* Match anchors at newline. */ + bufp->newline_anchor = 1; + + ret = re_compile_internal (bufp, pattern, length, re_syntax_options); + + if (!ret) + return NULL; + return gettext (__re_error_msgid + __re_error_msgid_idx[(int) ret]); +} +#ifdef _LIBC +weak_alias (__re_compile_pattern, re_compile_pattern) +#endif + +/* Set by 're_set_syntax' to the current regexp syntax to recognize. Can + also be assigned to arbitrarily: each pattern buffer stores its own + syntax, so it can be changed between regex compilations. */ +/* This has no initializer because initialized variables in Emacs + become read-only after dumping. */ +reg_syntax_t re_syntax_options; + + +/* Specify the precise syntax of regexps for compilation. This provides + for compatibility for various utilities which historically have + different, incompatible syntaxes. + + The argument SYNTAX is a bit mask comprised of the various bits + defined in regex.h. We return the old syntax. */ + +reg_syntax_t +re_set_syntax (syntax) + reg_syntax_t syntax; +{ + reg_syntax_t ret = re_syntax_options; + + re_syntax_options = syntax; + return ret; +} +#ifdef _LIBC +weak_alias (__re_set_syntax, re_set_syntax) +#endif + +int +re_compile_fastmap (bufp) + struct re_pattern_buffer *bufp; +{ + re_dfa_t *dfa = (re_dfa_t *) bufp->buffer; + char *fastmap = bufp->fastmap; + + memset (fastmap, '\0', sizeof (char) * SBC_MAX); + re_compile_fastmap_iter (bufp, dfa->init_state, fastmap); + if (dfa->init_state != dfa->init_state_word) + re_compile_fastmap_iter (bufp, dfa->init_state_word, fastmap); + if (dfa->init_state != dfa->init_state_nl) + re_compile_fastmap_iter (bufp, dfa->init_state_nl, fastmap); + if (dfa->init_state != dfa->init_state_begbuf) + re_compile_fastmap_iter (bufp, dfa->init_state_begbuf, fastmap); + bufp->fastmap_accurate = 1; + return 0; +} +#ifdef _LIBC +weak_alias (__re_compile_fastmap, re_compile_fastmap) +#endif + +static inline void +__attribute ((always_inline)) +re_set_fastmap (char *fastmap, bool icase, int ch) +{ + fastmap[ch] = 1; + if (icase) + fastmap[tolower (ch)] = 1; +} + +/* Helper function for re_compile_fastmap. + Compile fastmap for the initial_state INIT_STATE. */ + +static void +re_compile_fastmap_iter (regex_t *bufp, const re_dfastate_t *init_state, + char *fastmap) +{ + re_dfa_t *dfa = (re_dfa_t *) bufp->buffer; + Idx node_cnt; + bool icase = (dfa->mb_cur_max == 1 && (bufp->syntax & RE_ICASE)); + for (node_cnt = 0; node_cnt < init_state->nodes.nelem; ++node_cnt) + { + Idx node = init_state->nodes.elems[node_cnt]; + re_token_type_t type = dfa->nodes[node].type; + + if (type == CHARACTER) + { + re_set_fastmap (fastmap, icase, dfa->nodes[node].opr.c); +#ifdef RE_ENABLE_I18N + if ((bufp->syntax & RE_ICASE) && dfa->mb_cur_max > 1) + { + unsigned char buf[MB_LEN_MAX]; + unsigned char *p; + wchar_t wc; + mbstate_t state; + + p = buf; + *p++ = dfa->nodes[node].opr.c; + while (++node < dfa->nodes_len + && dfa->nodes[node].type == CHARACTER + && dfa->nodes[node].mb_partial) + *p++ = dfa->nodes[node].opr.c; + memset (&state, '\0', sizeof (state)); + if (__mbrtowc (&wc, (const char *) buf, p - buf, + &state) == p - buf + && (__wcrtomb ((char *) buf, towlower (wc), &state) + != (size_t) -1)) + re_set_fastmap (fastmap, false, buf[0]); + } +#endif + } + else if (type == SIMPLE_BRACKET) + { + int i, ch; + for (i = 0, ch = 0; i < BITSET_WORDS; ++i) + { + int j; + bitset_word_t w = dfa->nodes[node].opr.sbcset[i]; + for (j = 0; j < BITSET_WORD_BITS; ++j, ++ch) + if (w & ((bitset_word_t) 1 << j)) + re_set_fastmap (fastmap, icase, ch); + } + } +#ifdef RE_ENABLE_I18N + else if (type == COMPLEX_BRACKET) + { + re_charset_t *cset = dfa->nodes[node].opr.mbcset; + Idx i; + +# ifdef _LIBC + /* See if we have to try all bytes which start multiple collation + elements. + e.g. In da_DK, we want to catch 'a' since "aa" is a valid + collation element, and don't catch 'b' since 'b' is + the only collation element which starts from 'b' (and + it is caught by SIMPLE_BRACKET). */ + if (_NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES) != 0 + && (cset->ncoll_syms || cset->nranges)) + { + const int32_t *table = (const int32_t *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); + for (i = 0; i < SBC_MAX; ++i) + if (table[i] < 0) + re_set_fastmap (fastmap, icase, i); + } +# endif /* _LIBC */ + + /* See if we have to start the match at all multibyte characters, + i.e. where we would not find an invalid sequence. This only + applies to multibyte character sets; for single byte character + sets, the SIMPLE_BRACKET again suffices. */ + if (dfa->mb_cur_max > 1 + && (cset->nchar_classes || cset->non_match || cset->nranges +# ifdef _LIBC + || cset->nequiv_classes +# endif /* _LIBC */ + )) + { + unsigned char c = 0; + do + { + mbstate_t mbs; + memset (&mbs, 0, sizeof (mbs)); + if (__mbrtowc (NULL, (char *) &c, 1, &mbs) == (size_t) -2) + re_set_fastmap (fastmap, false, (int) c); + } + while (++c != 0); + } + + else + { + /* ... Else catch all bytes which can start the mbchars. */ + for (i = 0; i < cset->nmbchars; ++i) + { + char buf[256]; + mbstate_t state; + memset (&state, '\0', sizeof (state)); + if (__wcrtomb (buf, cset->mbchars[i], &state) != (size_t) -1) + re_set_fastmap (fastmap, icase, *(unsigned char *) buf); + if ((bufp->syntax & RE_ICASE) && dfa->mb_cur_max > 1) + { + if (__wcrtomb (buf, towlower (cset->mbchars[i]), &state) + != (size_t) -1) + re_set_fastmap (fastmap, false, *(unsigned char *) buf); + } + } + } + } +#endif /* RE_ENABLE_I18N */ + else if (type == OP_PERIOD +#ifdef RE_ENABLE_I18N + || type == OP_UTF8_PERIOD +#endif /* RE_ENABLE_I18N */ + || type == END_OF_RE) + { + memset (fastmap, '\1', sizeof (char) * SBC_MAX); + if (type == END_OF_RE) + bufp->can_be_null = 1; + return; + } + } +} + +/* Entry point for POSIX code. */ +/* regcomp takes a regular expression as a string and compiles it. + + PREG is a regex_t *. We do not expect any fields to be initialized, + since POSIX says we shouldn't. Thus, we set + + 'buffer' to the compiled pattern; + 'used' to the length of the compiled pattern; + 'syntax' to RE_SYNTAX_POSIX_EXTENDED if the + REG_EXTENDED bit in CFLAGS is set; otherwise, to + RE_SYNTAX_POSIX_BASIC; + 'newline_anchor' to REG_NEWLINE being set in CFLAGS; + 'fastmap' to an allocated space for the fastmap; + 'fastmap_accurate' to zero; + 're_nsub' to the number of subexpressions in PATTERN. + + PATTERN is the address of the pattern string. + + CFLAGS is a series of bits which affect compilation. + + If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we + use POSIX basic syntax. + + If REG_NEWLINE is set, then . and [^...] don't match newline. + Also, regexec will try a match beginning after every newline. + + If REG_ICASE is set, then we considers upper- and lowercase + versions of letters to be equivalent when matching. + + If REG_NOSUB is set, then when PREG is passed to regexec, that + routine will report only success or failure, and nothing about the + registers. + + It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for + the return codes and their meanings.) */ + +int +regcomp (preg, pattern, cflags) + regex_t *_Restrict_ preg; + const char *_Restrict_ pattern; + int cflags; +{ + reg_errcode_t ret; + reg_syntax_t syntax = ((cflags & REG_EXTENDED) ? RE_SYNTAX_POSIX_EXTENDED + : RE_SYNTAX_POSIX_BASIC); + + preg->buffer = NULL; + preg->allocated = 0; + preg->used = 0; + + /* Try to allocate space for the fastmap. */ + preg->fastmap = re_malloc (char, SBC_MAX); + if (BE (preg->fastmap == NULL, 0)) + return REG_ESPACE; + + syntax |= (cflags & REG_ICASE) ? RE_ICASE : 0; + + /* If REG_NEWLINE is set, newlines are treated differently. */ + if (cflags & REG_NEWLINE) + { /* REG_NEWLINE implies neither . nor [^...] match newline. */ + syntax &= ~RE_DOT_NEWLINE; + syntax |= RE_HAT_LISTS_NOT_NEWLINE; + /* It also changes the matching behavior. */ + preg->newline_anchor = 1; + } + else + preg->newline_anchor = 0; + preg->no_sub = !!(cflags & REG_NOSUB); + preg->translate = NULL; + + ret = re_compile_internal (preg, pattern, strlen (pattern), syntax); + + /* POSIX doesn't distinguish between an unmatched open-group and an + unmatched close-group: both are REG_EPAREN. */ + if (ret == REG_ERPAREN) + ret = REG_EPAREN; + + /* We have already checked preg->fastmap != NULL. */ + if (BE (ret == REG_NOERROR, 1)) + /* Compute the fastmap now, since regexec cannot modify the pattern + buffer. This function never fails in this implementation. */ + (void) re_compile_fastmap (preg); + else + { + /* Some error occurred while compiling the expression. */ + re_free (preg->fastmap); + preg->fastmap = NULL; + } + + return (int) ret; +} +#ifdef _LIBC +weak_alias (__regcomp, regcomp) +#endif + +/* Returns a message corresponding to an error code, ERRCODE, returned + from either regcomp or regexec. We don't use PREG here. */ + +#ifdef _LIBC +size_t +regerror (errcode, preg, errbuf, errbuf_size) + int errcode; + const regex_t *_Restrict_ preg; + char *_Restrict_ errbuf; + size_t errbuf_size; +#else /* size_t might promote */ +size_t +regerror (int errcode, const regex_t *_Restrict_ preg, + char *_Restrict_ errbuf, size_t errbuf_size) +#endif +{ + const char *msg; + size_t msg_size; + + if (BE (errcode < 0 + || errcode >= (int) (sizeof (__re_error_msgid_idx) + / sizeof (__re_error_msgid_idx[0])), 0)) + /* Only error codes returned by the rest of the code should be passed + to this routine. If we are given anything else, or if other regex + code generates an invalid error code, then the program has a bug. + Dump core so we can fix it. */ + abort (); + + msg = gettext (__re_error_msgid + __re_error_msgid_idx[errcode]); + + msg_size = strlen (msg) + 1; /* Includes the null. */ + + if (BE (errbuf_size != 0, 1)) + { + size_t cpy_size = msg_size; + if (BE (msg_size > errbuf_size, 0)) + { + cpy_size = errbuf_size - 1; + errbuf[cpy_size] = '\0'; + } + memcpy (errbuf, msg, cpy_size); + } + + return msg_size; +} +#ifdef _LIBC +weak_alias (__regerror, regerror) +#endif + + +#ifdef RE_ENABLE_I18N +/* This static array is used for the map to single-byte characters when + UTF-8 is used. Otherwise we would allocate memory just to initialize + it the same all the time. UTF-8 is the preferred encoding so this is + a worthwhile optimization. */ +static const bitset_t utf8_sb_map = +{ + /* Set the first 128 bits. */ +# if 4 * BITSET_WORD_BITS < ASCII_CHARS +# error "bitset_word_t is narrower than 32 bits" +# elif 3 * BITSET_WORD_BITS < ASCII_CHARS + BITSET_WORD_MAX, BITSET_WORD_MAX, BITSET_WORD_MAX, +# elif 2 * BITSET_WORD_BITS < ASCII_CHARS + BITSET_WORD_MAX, BITSET_WORD_MAX, +# elif 1 * BITSET_WORD_BITS < ASCII_CHARS + BITSET_WORD_MAX, +# endif + (BITSET_WORD_MAX + >> (SBC_MAX % BITSET_WORD_BITS == 0 + ? 0 + : BITSET_WORD_BITS - SBC_MAX % BITSET_WORD_BITS)) +}; +#endif + + +static void +free_dfa_content (re_dfa_t *dfa) +{ + Idx i, j; + + if (dfa->nodes) + for (i = 0; i < dfa->nodes_len; ++i) + free_token (dfa->nodes + i); + re_free (dfa->nexts); + for (i = 0; i < dfa->nodes_len; ++i) + { + if (dfa->eclosures != NULL) + re_node_set_free (dfa->eclosures + i); + if (dfa->inveclosures != NULL) + re_node_set_free (dfa->inveclosures + i); + if (dfa->edests != NULL) + re_node_set_free (dfa->edests + i); + } + re_free (dfa->edests); + re_free (dfa->eclosures); + re_free (dfa->inveclosures); + re_free (dfa->nodes); + + if (dfa->state_table) + for (i = 0; i <= dfa->state_hash_mask; ++i) + { + struct re_state_table_entry *entry = dfa->state_table + i; + for (j = 0; j < entry->num; ++j) + { + re_dfastate_t *state = entry->array[j]; + free_state (state); + } + re_free (entry->array); + } + re_free (dfa->state_table); +#ifdef RE_ENABLE_I18N + if (dfa->sb_char != utf8_sb_map) + re_free (dfa->sb_char); +#endif + re_free (dfa->subexp_map); +#ifdef DEBUG + re_free (dfa->re_str); +#endif + + re_free (dfa); +} + + +/* Free dynamically allocated space used by PREG. */ + +void +regfree (preg) + regex_t *preg; +{ + re_dfa_t *dfa = (re_dfa_t *) preg->buffer; + if (BE (dfa != NULL, 1)) + free_dfa_content (dfa); + preg->buffer = NULL; + preg->allocated = 0; + + re_free (preg->fastmap); + preg->fastmap = NULL; + + re_free (preg->translate); + preg->translate = NULL; +} +#ifdef _LIBC +weak_alias (__regfree, regfree) +#endif + +/* Entry points compatible with 4.2 BSD regex library. We don't define + them unless specifically requested. */ + +#if defined _REGEX_RE_COMP || defined _LIBC + +/* BSD has one and only one pattern buffer. */ +static struct re_pattern_buffer re_comp_buf; + +char * +# ifdef _LIBC +/* Make these definitions weak in libc, so POSIX programs can redefine + these names if they don't use our functions, and still use + regcomp/regexec above without link errors. */ +weak_function +# endif +re_comp (s) + const char *s; +{ + reg_errcode_t ret; + char *fastmap; + + if (!s) + { + if (!re_comp_buf.buffer) + return gettext ("No previous regular expression"); + return 0; + } + + if (re_comp_buf.buffer) + { + fastmap = re_comp_buf.fastmap; + re_comp_buf.fastmap = NULL; + __regfree (&re_comp_buf); + memset (&re_comp_buf, '\0', sizeof (re_comp_buf)); + re_comp_buf.fastmap = fastmap; + } + + if (re_comp_buf.fastmap == NULL) + { + re_comp_buf.fastmap = (char *) malloc (SBC_MAX); + if (re_comp_buf.fastmap == NULL) + return (char *) gettext (__re_error_msgid + + __re_error_msgid_idx[(int) REG_ESPACE]); + } + + /* Since 're_exec' always passes NULL for the 'regs' argument, we + don't need to initialize the pattern buffer fields which affect it. */ + + /* Match anchors at newlines. */ + re_comp_buf.newline_anchor = 1; + + ret = re_compile_internal (&re_comp_buf, s, strlen (s), re_syntax_options); + + if (!ret) + return NULL; + + /* Yes, we're discarding 'const' here if !HAVE_LIBINTL. */ + return (char *) gettext (__re_error_msgid + __re_error_msgid_idx[(int) ret]); +} + +#ifdef _LIBC +libc_freeres_fn (free_mem) +{ + __regfree (&re_comp_buf); +} +#endif + +#endif /* _REGEX_RE_COMP */ + +/* Internal entry point. + Compile the regular expression PATTERN, whose length is LENGTH. + SYNTAX indicate regular expression's syntax. */ + +static reg_errcode_t +re_compile_internal (regex_t *preg, const char * pattern, size_t length, + reg_syntax_t syntax) +{ + reg_errcode_t err = REG_NOERROR; + re_dfa_t *dfa; + re_string_t regexp; + + /* Initialize the pattern buffer. */ + preg->fastmap_accurate = 0; + preg->syntax = syntax; + preg->not_bol = preg->not_eol = 0; + preg->used = 0; + preg->re_nsub = 0; + preg->can_be_null = 0; + preg->regs_allocated = REGS_UNALLOCATED; + + /* Initialize the dfa. */ + dfa = (re_dfa_t *) preg->buffer; + if (BE (preg->allocated < sizeof (re_dfa_t), 0)) + { + /* If zero allocated, but buffer is non-null, try to realloc + enough space. This loses if buffer's address is bogus, but + that is the user's responsibility. If ->buffer is NULL this + is a simple allocation. */ + dfa = re_realloc (preg->buffer, re_dfa_t, 1); + if (dfa == NULL) + return REG_ESPACE; + preg->allocated = sizeof (re_dfa_t); + preg->buffer = (unsigned char *) dfa; + } + preg->used = sizeof (re_dfa_t); + + err = init_dfa (dfa, length); + if (BE (err != REG_NOERROR, 0)) + { + free_dfa_content (dfa); + preg->buffer = NULL; + preg->allocated = 0; + return err; + } +#ifdef DEBUG + /* Note: length+1 will not overflow since it is checked in init_dfa. */ + dfa->re_str = re_malloc (char, length + 1); + strncpy (dfa->re_str, pattern, length + 1); +#endif + + __libc_lock_init (dfa->lock); + + err = re_string_construct (®exp, pattern, length, preg->translate, + (syntax & RE_ICASE) != 0, dfa); + if (BE (err != REG_NOERROR, 0)) + { + re_compile_internal_free_return: + free_workarea_compile (preg); + re_string_destruct (®exp); + free_dfa_content (dfa); + preg->buffer = NULL; + preg->allocated = 0; + return err; + } + + /* Parse the regular expression, and build a structure tree. */ + preg->re_nsub = 0; + dfa->str_tree = parse (®exp, preg, syntax, &err); + if (BE (dfa->str_tree == NULL, 0)) + goto re_compile_internal_free_return; + + /* Analyze the tree and create the nfa. */ + err = analyze (preg); + if (BE (err != REG_NOERROR, 0)) + goto re_compile_internal_free_return; + +#ifdef RE_ENABLE_I18N + /* If possible, do searching in single byte encoding to speed things up. */ + if (dfa->is_utf8 && !(syntax & RE_ICASE) && preg->translate == NULL) + optimize_utf8 (dfa); +#endif + + /* Then create the initial state of the dfa. */ + err = create_initial_state (dfa); + + /* Release work areas. */ + free_workarea_compile (preg); + re_string_destruct (®exp); + + if (BE (err != REG_NOERROR, 0)) + { + free_dfa_content (dfa); + preg->buffer = NULL; + preg->allocated = 0; + } + + return err; +} + +/* Initialize DFA. We use the length of the regular expression PAT_LEN + as the initial length of some arrays. */ + +static reg_errcode_t +init_dfa (re_dfa_t *dfa, size_t pat_len) +{ + __re_size_t table_size; +#ifndef _LIBC + char *codeset_name; +#endif +#ifdef RE_ENABLE_I18N + size_t max_i18n_object_size = MAX (sizeof (wchar_t), sizeof (wctype_t)); +#else + size_t max_i18n_object_size = 0; +#endif + size_t max_object_size = + MAX (sizeof (struct re_state_table_entry), + MAX (sizeof (re_token_t), + MAX (sizeof (re_node_set), + MAX (sizeof (regmatch_t), + max_i18n_object_size)))); + + memset (dfa, '\0', sizeof (re_dfa_t)); + + /* Force allocation of str_tree_storage the first time. */ + dfa->str_tree_storage_idx = BIN_TREE_STORAGE_SIZE; + + /* Avoid overflows. The extra "/ 2" is for the table_size doubling + calculation below, and for similar doubling calculations + elsewhere. And it's <= rather than <, because some of the + doubling calculations add 1 afterwards. */ + if (BE (SIZE_MAX / max_object_size / 2 <= pat_len, 0)) + return REG_ESPACE; + + dfa->nodes_alloc = pat_len + 1; + dfa->nodes = re_malloc (re_token_t, dfa->nodes_alloc); + + /* table_size = 2 ^ ceil(log pat_len) */ + for (table_size = 1; ; table_size <<= 1) + if (table_size > pat_len) + break; + + dfa->state_table = calloc (sizeof (struct re_state_table_entry), table_size); + dfa->state_hash_mask = table_size - 1; + + dfa->mb_cur_max = MB_CUR_MAX; +#ifdef _LIBC + if (dfa->mb_cur_max == 6 + && strcmp (_NL_CURRENT (LC_CTYPE, _NL_CTYPE_CODESET_NAME), "UTF-8") == 0) + dfa->is_utf8 = 1; + dfa->map_notascii = (_NL_CURRENT_WORD (LC_CTYPE, _NL_CTYPE_MAP_TO_NONASCII) + != 0); +#else + codeset_name = nl_langinfo (CODESET); + if (strcasecmp (codeset_name, "UTF-8") == 0 + || strcasecmp (codeset_name, "UTF8") == 0) + dfa->is_utf8 = 1; + + /* We check exhaustively in the loop below if this charset is a + superset of ASCII. */ + dfa->map_notascii = 0; +#endif + +#ifdef RE_ENABLE_I18N + if (dfa->mb_cur_max > 1) + { + if (dfa->is_utf8) + dfa->sb_char = (re_bitset_ptr_t) utf8_sb_map; + else + { + int i, j, ch; + + dfa->sb_char = (re_bitset_ptr_t) calloc (sizeof (bitset_t), 1); + if (BE (dfa->sb_char == NULL, 0)) + return REG_ESPACE; + + /* Set the bits corresponding to single byte chars. */ + for (i = 0, ch = 0; i < BITSET_WORDS; ++i) + for (j = 0; j < BITSET_WORD_BITS; ++j, ++ch) + { + wint_t wch = __btowc (ch); + if (wch != WEOF) + dfa->sb_char[i] |= (bitset_word_t) 1 << j; +# ifndef _LIBC + if (isascii (ch) && wch != ch) + dfa->map_notascii = 1; +# endif + } + } + } +#endif + + if (BE (dfa->nodes == NULL || dfa->state_table == NULL, 0)) + return REG_ESPACE; + return REG_NOERROR; +} + +/* Initialize WORD_CHAR table, which indicate which character is + "word". In this case "word" means that it is the word construction + character used by some operators like "\<", "\>", etc. */ + +static void +internal_function +init_word_char (re_dfa_t *dfa) +{ + int i, j, ch; + dfa->word_ops_used = 1; + for (i = 0, ch = 0; i < BITSET_WORDS; ++i) + for (j = 0; j < BITSET_WORD_BITS; ++j, ++ch) + if (isalnum (ch) || ch == '_') + dfa->word_char[i] |= (bitset_word_t) 1 << j; +} + +/* Free the work area which are only used while compiling. */ + +static void +free_workarea_compile (regex_t *preg) +{ + re_dfa_t *dfa = (re_dfa_t *) preg->buffer; + bin_tree_storage_t *storage, *next; + for (storage = dfa->str_tree_storage; storage; storage = next) + { + next = storage->next; + re_free (storage); + } + dfa->str_tree_storage = NULL; + dfa->str_tree_storage_idx = BIN_TREE_STORAGE_SIZE; + dfa->str_tree = NULL; + re_free (dfa->org_indices); + dfa->org_indices = NULL; +} + +/* Create initial states for all contexts. */ + +static reg_errcode_t +create_initial_state (re_dfa_t *dfa) +{ + Idx first, i; + reg_errcode_t err; + re_node_set init_nodes; + + /* Initial states have the epsilon closure of the node which is + the first node of the regular expression. */ + first = dfa->str_tree->first->node_idx; + dfa->init_node = first; + err = re_node_set_init_copy (&init_nodes, dfa->eclosures + first); + if (BE (err != REG_NOERROR, 0)) + return err; + + /* The back-references which are in initial states can epsilon transit, + since in this case all of the subexpressions can be null. + Then we add epsilon closures of the nodes which are the next nodes of + the back-references. */ + if (dfa->nbackref > 0) + for (i = 0; i < init_nodes.nelem; ++i) + { + Idx node_idx = init_nodes.elems[i]; + re_token_type_t type = dfa->nodes[node_idx].type; + + Idx clexp_idx; + if (type != OP_BACK_REF) + continue; + for (clexp_idx = 0; clexp_idx < init_nodes.nelem; ++clexp_idx) + { + re_token_t *clexp_node; + clexp_node = dfa->nodes + init_nodes.elems[clexp_idx]; + if (clexp_node->type == OP_CLOSE_SUBEXP + && clexp_node->opr.idx == dfa->nodes[node_idx].opr.idx) + break; + } + if (clexp_idx == init_nodes.nelem) + continue; + + if (type == OP_BACK_REF) + { + Idx dest_idx = dfa->edests[node_idx].elems[0]; + if (!re_node_set_contains (&init_nodes, dest_idx)) + { + reg_errcode_t merge_err + = re_node_set_merge (&init_nodes, dfa->eclosures + dest_idx); + if (merge_err != REG_NOERROR) + return merge_err; + i = 0; + } + } + } + + /* It must be the first time to invoke acquire_state. */ + dfa->init_state = re_acquire_state_context (&err, dfa, &init_nodes, 0); + /* We don't check ERR here, since the initial state must not be NULL. */ + if (BE (dfa->init_state == NULL, 0)) + return err; + if (dfa->init_state->has_constraint) + { + dfa->init_state_word = re_acquire_state_context (&err, dfa, &init_nodes, + CONTEXT_WORD); + dfa->init_state_nl = re_acquire_state_context (&err, dfa, &init_nodes, + CONTEXT_NEWLINE); + dfa->init_state_begbuf = re_acquire_state_context (&err, dfa, + &init_nodes, + CONTEXT_NEWLINE + | CONTEXT_BEGBUF); + if (BE (dfa->init_state_word == NULL || dfa->init_state_nl == NULL + || dfa->init_state_begbuf == NULL, 0)) + return err; + } + else + dfa->init_state_word = dfa->init_state_nl + = dfa->init_state_begbuf = dfa->init_state; + + re_node_set_free (&init_nodes); + return REG_NOERROR; +} + +#ifdef RE_ENABLE_I18N +/* If it is possible to do searching in single byte encoding instead of UTF-8 + to speed things up, set dfa->mb_cur_max to 1, clear is_utf8 and change + DFA nodes where needed. */ + +static void +optimize_utf8 (re_dfa_t *dfa) +{ + Idx node; + int i; + bool mb_chars = false; + bool has_period = false; + + for (node = 0; node < dfa->nodes_len; ++node) + switch (dfa->nodes[node].type) + { + case CHARACTER: + if (dfa->nodes[node].opr.c >= ASCII_CHARS) + mb_chars = true; + break; + case ANCHOR: + switch (dfa->nodes[node].opr.ctx_type) + { + case LINE_FIRST: + case LINE_LAST: + case BUF_FIRST: + case BUF_LAST: + break; + default: + /* Word anchors etc. cannot be handled. It's okay to test + opr.ctx_type since constraints (for all DFA nodes) are + created by ORing one or more opr.ctx_type values. */ + return; + } + break; + case OP_PERIOD: + has_period = true; + break; + case OP_BACK_REF: + case OP_ALT: + case END_OF_RE: + case OP_DUP_ASTERISK: + case OP_OPEN_SUBEXP: + case OP_CLOSE_SUBEXP: + break; + case COMPLEX_BRACKET: + return; + case SIMPLE_BRACKET: + /* Just double check. */ + { + int rshift = (ASCII_CHARS % BITSET_WORD_BITS == 0 + ? 0 + : BITSET_WORD_BITS - ASCII_CHARS % BITSET_WORD_BITS); + for (i = ASCII_CHARS / BITSET_WORD_BITS; i < BITSET_WORDS; ++i) + { + if (dfa->nodes[node].opr.sbcset[i] >> rshift != 0) + return; + rshift = 0; + } + } + break; + default: + abort (); + } + + if (mb_chars || has_period) + for (node = 0; node < dfa->nodes_len; ++node) + { + if (dfa->nodes[node].type == CHARACTER + && dfa->nodes[node].opr.c >= ASCII_CHARS) + dfa->nodes[node].mb_partial = 0; + else if (dfa->nodes[node].type == OP_PERIOD) + dfa->nodes[node].type = OP_UTF8_PERIOD; + } + + /* The search can be in single byte locale. */ + dfa->mb_cur_max = 1; + dfa->is_utf8 = 0; + dfa->has_mb_node = dfa->nbackref > 0 || has_period; +} +#endif + +/* Analyze the structure tree, and calculate "first", "next", "edest", + "eclosure", and "inveclosure". */ + +static reg_errcode_t +analyze (regex_t *preg) +{ + re_dfa_t *dfa = (re_dfa_t *) preg->buffer; + reg_errcode_t ret; + + /* Allocate arrays. */ + dfa->nexts = re_malloc (Idx, dfa->nodes_alloc); + dfa->org_indices = re_malloc (Idx, dfa->nodes_alloc); + dfa->edests = re_malloc (re_node_set, dfa->nodes_alloc); + dfa->eclosures = re_malloc (re_node_set, dfa->nodes_alloc); + if (BE (dfa->nexts == NULL || dfa->org_indices == NULL || dfa->edests == NULL + || dfa->eclosures == NULL, 0)) + return REG_ESPACE; + + dfa->subexp_map = re_malloc (Idx, preg->re_nsub); + if (dfa->subexp_map != NULL) + { + Idx i; + for (i = 0; i < preg->re_nsub; i++) + dfa->subexp_map[i] = i; + preorder (dfa->str_tree, optimize_subexps, dfa); + for (i = 0; i < preg->re_nsub; i++) + if (dfa->subexp_map[i] != i) + break; + if (i == preg->re_nsub) + { + free (dfa->subexp_map); + dfa->subexp_map = NULL; + } + } + + ret = postorder (dfa->str_tree, lower_subexps, preg); + if (BE (ret != REG_NOERROR, 0)) + return ret; + ret = postorder (dfa->str_tree, calc_first, dfa); + if (BE (ret != REG_NOERROR, 0)) + return ret; + preorder (dfa->str_tree, calc_next, dfa); + ret = preorder (dfa->str_tree, link_nfa_nodes, dfa); + if (BE (ret != REG_NOERROR, 0)) + return ret; + ret = calc_eclosure (dfa); + if (BE (ret != REG_NOERROR, 0)) + return ret; + + /* We only need this during the prune_impossible_nodes pass in regexec.c; + skip it if p_i_n will not run, as calc_inveclosure can be quadratic. */ + if ((!preg->no_sub && preg->re_nsub > 0 && dfa->has_plural_match) + || dfa->nbackref) + { + dfa->inveclosures = re_malloc (re_node_set, dfa->nodes_len); + if (BE (dfa->inveclosures == NULL, 0)) + return REG_ESPACE; + ret = calc_inveclosure (dfa); + } + + return ret; +} + +/* Our parse trees are very unbalanced, so we cannot use a stack to + implement parse tree visits. Instead, we use parent pointers and + some hairy code in these two functions. */ +static reg_errcode_t +postorder (bin_tree_t *root, reg_errcode_t (fn (void *, bin_tree_t *)), + void *extra) +{ + bin_tree_t *node, *prev; + + for (node = root; ; ) + { + /* Descend down the tree, preferably to the left (or to the right + if that's the only child). */ + while (node->left || node->right) + if (node->left) + node = node->left; + else + node = node->right; + + do + { + reg_errcode_t err = fn (extra, node); + if (BE (err != REG_NOERROR, 0)) + return err; + if (node->parent == NULL) + return REG_NOERROR; + prev = node; + node = node->parent; + } + /* Go up while we have a node that is reached from the right. */ + while (node->right == prev || node->right == NULL); + node = node->right; + } +} + +static reg_errcode_t +preorder (bin_tree_t *root, reg_errcode_t (fn (void *, bin_tree_t *)), + void *extra) +{ + bin_tree_t *node; + + for (node = root; ; ) + { + reg_errcode_t err = fn (extra, node); + if (BE (err != REG_NOERROR, 0)) + return err; + + /* Go to the left node, or up and to the right. */ + if (node->left) + node = node->left; + else + { + bin_tree_t *prev = NULL; + while (node->right == prev || node->right == NULL) + { + prev = node; + node = node->parent; + if (!node) + return REG_NOERROR; + } + node = node->right; + } + } +} + +/* Optimization pass: if a SUBEXP is entirely contained, strip it and tell + re_search_internal to map the inner one's opr.idx to this one's. Adjust + backreferences as well. Requires a preorder visit. */ +static reg_errcode_t +optimize_subexps (void *extra, bin_tree_t *node) +{ + re_dfa_t *dfa = (re_dfa_t *) extra; + + if (node->token.type == OP_BACK_REF && dfa->subexp_map) + { + int idx = node->token.opr.idx; + node->token.opr.idx = dfa->subexp_map[idx]; + dfa->used_bkref_map |= 1 << node->token.opr.idx; + } + + else if (node->token.type == SUBEXP + && node->left && node->left->token.type == SUBEXP) + { + Idx other_idx = node->left->token.opr.idx; + + node->left = node->left->left; + if (node->left) + node->left->parent = node; + + dfa->subexp_map[other_idx] = dfa->subexp_map[node->token.opr.idx]; + if (other_idx < BITSET_WORD_BITS) + dfa->used_bkref_map &= ~((bitset_word_t) 1 << other_idx); + } + + return REG_NOERROR; +} + +/* Lowering pass: Turn each SUBEXP node into the appropriate concatenation + of OP_OPEN_SUBEXP, the body of the SUBEXP (if any) and OP_CLOSE_SUBEXP. */ +static reg_errcode_t +lower_subexps (void *extra, bin_tree_t *node) +{ + regex_t *preg = (regex_t *) extra; + reg_errcode_t err = REG_NOERROR; + + if (node->left && node->left->token.type == SUBEXP) + { + node->left = lower_subexp (&err, preg, node->left); + if (node->left) + node->left->parent = node; + } + if (node->right && node->right->token.type == SUBEXP) + { + node->right = lower_subexp (&err, preg, node->right); + if (node->right) + node->right->parent = node; + } + + return err; +} + +static bin_tree_t * +lower_subexp (reg_errcode_t *err, regex_t *preg, bin_tree_t *node) +{ + re_dfa_t *dfa = (re_dfa_t *) preg->buffer; + bin_tree_t *body = node->left; + bin_tree_t *op, *cls, *tree1, *tree; + + if (preg->no_sub + /* We do not optimize empty subexpressions, because otherwise we may + have bad CONCAT nodes with NULL children. This is obviously not + very common, so we do not lose much. An example that triggers + this case is the sed "script" /\(\)/x. */ + && node->left != NULL + && (node->token.opr.idx >= BITSET_WORD_BITS + || !(dfa->used_bkref_map + & ((bitset_word_t) 1 << node->token.opr.idx)))) + return node->left; + + /* Convert the SUBEXP node to the concatenation of an + OP_OPEN_SUBEXP, the contents, and an OP_CLOSE_SUBEXP. */ + op = create_tree (dfa, NULL, NULL, OP_OPEN_SUBEXP); + cls = create_tree (dfa, NULL, NULL, OP_CLOSE_SUBEXP); + tree1 = body ? create_tree (dfa, body, cls, CONCAT) : cls; + tree = create_tree (dfa, op, tree1, CONCAT); + if (BE (tree == NULL || tree1 == NULL || op == NULL || cls == NULL, 0)) + { + *err = REG_ESPACE; + return NULL; + } + + op->token.opr.idx = cls->token.opr.idx = node->token.opr.idx; + op->token.opt_subexp = cls->token.opt_subexp = node->token.opt_subexp; + return tree; +} + +/* Pass 1 in building the NFA: compute FIRST and create unlinked automaton + nodes. Requires a postorder visit. */ +static reg_errcode_t +calc_first (void *extra, bin_tree_t *node) +{ + re_dfa_t *dfa = (re_dfa_t *) extra; + if (node->token.type == CONCAT) + { + node->first = node->left->first; + node->node_idx = node->left->node_idx; + } + else + { + node->first = node; + node->node_idx = re_dfa_add_node (dfa, node->token); + if (BE (node->node_idx == REG_MISSING, 0)) + return REG_ESPACE; + if (node->token.type == ANCHOR) + dfa->nodes[node->node_idx].constraint = node->token.opr.ctx_type; + } + return REG_NOERROR; +} + +/* Pass 2: compute NEXT on the tree. Preorder visit. */ +static reg_errcode_t +calc_next (void *extra, bin_tree_t *node) +{ + switch (node->token.type) + { + case OP_DUP_ASTERISK: + node->left->next = node; + break; + case CONCAT: + node->left->next = node->right->first; + node->right->next = node->next; + break; + default: + if (node->left) + node->left->next = node->next; + if (node->right) + node->right->next = node->next; + break; + } + return REG_NOERROR; +} + +/* Pass 3: link all DFA nodes to their NEXT node (any order will do). */ +static reg_errcode_t +link_nfa_nodes (void *extra, bin_tree_t *node) +{ + re_dfa_t *dfa = (re_dfa_t *) extra; + Idx idx = node->node_idx; + reg_errcode_t err = REG_NOERROR; + + switch (node->token.type) + { + case CONCAT: + break; + + case END_OF_RE: + assert (node->next == NULL); + break; + + case OP_DUP_ASTERISK: + case OP_ALT: + { + Idx left, right; + dfa->has_plural_match = 1; + if (node->left != NULL) + left = node->left->first->node_idx; + else + left = node->next->node_idx; + if (node->right != NULL) + right = node->right->first->node_idx; + else + right = node->next->node_idx; + assert (REG_VALID_INDEX (left)); + assert (REG_VALID_INDEX (right)); + err = re_node_set_init_2 (dfa->edests + idx, left, right); + } + break; + + case ANCHOR: + case OP_OPEN_SUBEXP: + case OP_CLOSE_SUBEXP: + err = re_node_set_init_1 (dfa->edests + idx, node->next->node_idx); + break; + + case OP_BACK_REF: + dfa->nexts[idx] = node->next->node_idx; + if (node->token.type == OP_BACK_REF) + err = re_node_set_init_1 (dfa->edests + idx, dfa->nexts[idx]); + break; + + default: + assert (!IS_EPSILON_NODE (node->token.type)); + dfa->nexts[idx] = node->next->node_idx; + break; + } + + return err; +} + +/* Duplicate the epsilon closure of the node ROOT_NODE. + Note that duplicated nodes have constraint INIT_CONSTRAINT in addition + to their own constraint. */ + +static reg_errcode_t +internal_function +duplicate_node_closure (re_dfa_t *dfa, Idx top_org_node, Idx top_clone_node, + Idx root_node, unsigned int init_constraint) +{ + Idx org_node, clone_node; + bool ok; + unsigned int constraint = init_constraint; + for (org_node = top_org_node, clone_node = top_clone_node;;) + { + Idx org_dest, clone_dest; + if (dfa->nodes[org_node].type == OP_BACK_REF) + { + /* If the back reference epsilon-transit, its destination must + also have the constraint. Then duplicate the epsilon closure + of the destination of the back reference, and store it in + edests of the back reference. */ + org_dest = dfa->nexts[org_node]; + re_node_set_empty (dfa->edests + clone_node); + clone_dest = duplicate_node (dfa, org_dest, constraint); + if (BE (clone_dest == REG_MISSING, 0)) + return REG_ESPACE; + dfa->nexts[clone_node] = dfa->nexts[org_node]; + ok = re_node_set_insert (dfa->edests + clone_node, clone_dest); + if (BE (! ok, 0)) + return REG_ESPACE; + } + else if (dfa->edests[org_node].nelem == 0) + { + /* In case of the node can't epsilon-transit, don't duplicate the + destination and store the original destination as the + destination of the node. */ + dfa->nexts[clone_node] = dfa->nexts[org_node]; + break; + } + else if (dfa->edests[org_node].nelem == 1) + { + /* In case of the node can epsilon-transit, and it has only one + destination. */ + org_dest = dfa->edests[org_node].elems[0]; + re_node_set_empty (dfa->edests + clone_node); + /* If the node is root_node itself, it means the epsilon closure + has a loop. Then tie it to the destination of the root_node. */ + if (org_node == root_node && clone_node != org_node) + { + ok = re_node_set_insert (dfa->edests + clone_node, org_dest); + if (BE (! ok, 0)) + return REG_ESPACE; + break; + } + /* In case the node has another constraint, append it. */ + constraint |= dfa->nodes[org_node].constraint; + clone_dest = duplicate_node (dfa, org_dest, constraint); + if (BE (clone_dest == REG_MISSING, 0)) + return REG_ESPACE; + ok = re_node_set_insert (dfa->edests + clone_node, clone_dest); + if (BE (! ok, 0)) + return REG_ESPACE; + } + else /* dfa->edests[org_node].nelem == 2 */ + { + /* In case of the node can epsilon-transit, and it has two + destinations. In the bin_tree_t and DFA, that's '|' and '*'. */ + org_dest = dfa->edests[org_node].elems[0]; + re_node_set_empty (dfa->edests + clone_node); + /* Search for a duplicated node which satisfies the constraint. */ + clone_dest = search_duplicated_node (dfa, org_dest, constraint); + if (clone_dest == REG_MISSING) + { + /* There is no such duplicated node, create a new one. */ + reg_errcode_t err; + clone_dest = duplicate_node (dfa, org_dest, constraint); + if (BE (clone_dest == REG_MISSING, 0)) + return REG_ESPACE; + ok = re_node_set_insert (dfa->edests + clone_node, clone_dest); + if (BE (! ok, 0)) + return REG_ESPACE; + err = duplicate_node_closure (dfa, org_dest, clone_dest, + root_node, constraint); + if (BE (err != REG_NOERROR, 0)) + return err; + } + else + { + /* There is a duplicated node which satisfies the constraint, + use it to avoid infinite loop. */ + ok = re_node_set_insert (dfa->edests + clone_node, clone_dest); + if (BE (! ok, 0)) + return REG_ESPACE; + } + + org_dest = dfa->edests[org_node].elems[1]; + clone_dest = duplicate_node (dfa, org_dest, constraint); + if (BE (clone_dest == REG_MISSING, 0)) + return REG_ESPACE; + ok = re_node_set_insert (dfa->edests + clone_node, clone_dest); + if (BE (! ok, 0)) + return REG_ESPACE; + } + org_node = org_dest; + clone_node = clone_dest; + } + return REG_NOERROR; +} + +/* Search for a node which is duplicated from the node ORG_NODE, and + satisfies the constraint CONSTRAINT. */ + +static Idx +search_duplicated_node (const re_dfa_t *dfa, Idx org_node, + unsigned int constraint) +{ + Idx idx; + for (idx = dfa->nodes_len - 1; dfa->nodes[idx].duplicated && idx > 0; --idx) + { + if (org_node == dfa->org_indices[idx] + && constraint == dfa->nodes[idx].constraint) + return idx; /* Found. */ + } + return REG_MISSING; /* Not found. */ +} + +/* Duplicate the node whose index is ORG_IDX and set the constraint CONSTRAINT. + Return the index of the new node, or REG_MISSING if insufficient storage is + available. */ + +static Idx +duplicate_node (re_dfa_t *dfa, Idx org_idx, unsigned int constraint) +{ + Idx dup_idx = re_dfa_add_node (dfa, dfa->nodes[org_idx]); + if (BE (dup_idx != REG_MISSING, 1)) + { + dfa->nodes[dup_idx].constraint = constraint; + dfa->nodes[dup_idx].constraint |= dfa->nodes[org_idx].constraint; + dfa->nodes[dup_idx].duplicated = 1; + + /* Store the index of the original node. */ + dfa->org_indices[dup_idx] = org_idx; + } + return dup_idx; +} + +static reg_errcode_t +calc_inveclosure (re_dfa_t *dfa) +{ + Idx src, idx; + bool ok; + for (idx = 0; idx < dfa->nodes_len; ++idx) + re_node_set_init_empty (dfa->inveclosures + idx); + + for (src = 0; src < dfa->nodes_len; ++src) + { + Idx *elems = dfa->eclosures[src].elems; + for (idx = 0; idx < dfa->eclosures[src].nelem; ++idx) + { + ok = re_node_set_insert_last (dfa->inveclosures + elems[idx], src); + if (BE (! ok, 0)) + return REG_ESPACE; + } + } + + return REG_NOERROR; +} + +/* Calculate "eclosure" for all the node in DFA. */ + +static reg_errcode_t +calc_eclosure (re_dfa_t *dfa) +{ + Idx node_idx; + bool incomplete; +#ifdef DEBUG + assert (dfa->nodes_len > 0); +#endif + incomplete = false; + /* For each nodes, calculate epsilon closure. */ + for (node_idx = 0; ; ++node_idx) + { + reg_errcode_t err; + re_node_set eclosure_elem; + if (node_idx == dfa->nodes_len) + { + if (!incomplete) + break; + incomplete = false; + node_idx = 0; + } + +#ifdef DEBUG + assert (dfa->eclosures[node_idx].nelem != REG_MISSING); +#endif + + /* If we have already calculated, skip it. */ + if (dfa->eclosures[node_idx].nelem != 0) + continue; + /* Calculate epsilon closure of 'node_idx'. */ + err = calc_eclosure_iter (&eclosure_elem, dfa, node_idx, true); + if (BE (err != REG_NOERROR, 0)) + return err; + + if (dfa->eclosures[node_idx].nelem == 0) + { + incomplete = true; + re_node_set_free (&eclosure_elem); + } + } + return REG_NOERROR; +} + +/* Calculate epsilon closure of NODE. */ + +static reg_errcode_t +calc_eclosure_iter (re_node_set *new_set, re_dfa_t *dfa, Idx node, bool root) +{ + reg_errcode_t err; + Idx i; + re_node_set eclosure; + bool ok; + bool incomplete = false; + err = re_node_set_alloc (&eclosure, dfa->edests[node].nelem + 1); + if (BE (err != REG_NOERROR, 0)) + return err; + + /* This indicates that we are calculating this node now. + We reference this value to avoid infinite loop. */ + dfa->eclosures[node].nelem = REG_MISSING; + + /* If the current node has constraints, duplicate all nodes + since they must inherit the constraints. */ + if (dfa->nodes[node].constraint + && dfa->edests[node].nelem + && !dfa->nodes[dfa->edests[node].elems[0]].duplicated) + { + err = duplicate_node_closure (dfa, node, node, node, + dfa->nodes[node].constraint); + if (BE (err != REG_NOERROR, 0)) + return err; + } + + /* Expand each epsilon destination nodes. */ + if (IS_EPSILON_NODE(dfa->nodes[node].type)) + for (i = 0; i < dfa->edests[node].nelem; ++i) + { + re_node_set eclosure_elem; + Idx edest = dfa->edests[node].elems[i]; + /* If calculating the epsilon closure of 'edest' is in progress, + return intermediate result. */ + if (dfa->eclosures[edest].nelem == REG_MISSING) + { + incomplete = true; + continue; + } + /* If we haven't calculated the epsilon closure of 'edest' yet, + calculate now. Otherwise use calculated epsilon closure. */ + if (dfa->eclosures[edest].nelem == 0) + { + err = calc_eclosure_iter (&eclosure_elem, dfa, edest, false); + if (BE (err != REG_NOERROR, 0)) + return err; + } + else + eclosure_elem = dfa->eclosures[edest]; + /* Merge the epsilon closure of 'edest'. */ + err = re_node_set_merge (&eclosure, &eclosure_elem); + if (BE (err != REG_NOERROR, 0)) + return err; + /* If the epsilon closure of 'edest' is incomplete, + the epsilon closure of this node is also incomplete. */ + if (dfa->eclosures[edest].nelem == 0) + { + incomplete = true; + re_node_set_free (&eclosure_elem); + } + } + + /* An epsilon closure includes itself. */ + ok = re_node_set_insert (&eclosure, node); + if (BE (! ok, 0)) + return REG_ESPACE; + if (incomplete && !root) + dfa->eclosures[node].nelem = 0; + else + dfa->eclosures[node] = eclosure; + *new_set = eclosure; + return REG_NOERROR; +} + +/* Functions for token which are used in the parser. */ + +/* Fetch a token from INPUT. + We must not use this function inside bracket expressions. */ + +static void +internal_function +fetch_token (re_token_t *result, re_string_t *input, reg_syntax_t syntax) +{ + re_string_skip_bytes (input, peek_token (result, input, syntax)); +} + +/* Peek a token from INPUT, and return the length of the token. + We must not use this function inside bracket expressions. */ + +static int +internal_function +peek_token (re_token_t *token, re_string_t *input, reg_syntax_t syntax) +{ + unsigned char c; + + if (re_string_eoi (input)) + { + token->type = END_OF_RE; + return 0; + } + + c = re_string_peek_byte (input, 0); + token->opr.c = c; + + token->word_char = 0; +#ifdef RE_ENABLE_I18N + token->mb_partial = 0; + if (input->mb_cur_max > 1 && + !re_string_first_byte (input, re_string_cur_idx (input))) + { + token->type = CHARACTER; + token->mb_partial = 1; + return 1; + } +#endif + if (c == '\\') + { + unsigned char c2; + if (re_string_cur_idx (input) + 1 >= re_string_length (input)) + { + token->type = BACK_SLASH; + return 1; + } + + c2 = re_string_peek_byte_case (input, 1); + token->opr.c = c2; + token->type = CHARACTER; +#ifdef RE_ENABLE_I18N + if (input->mb_cur_max > 1) + { + wint_t wc = re_string_wchar_at (input, + re_string_cur_idx (input) + 1); + token->word_char = IS_WIDE_WORD_CHAR (wc) != 0; + } + else +#endif + token->word_char = IS_WORD_CHAR (c2) != 0; + + switch (c2) + { + case '|': + if (!(syntax & RE_LIMITED_OPS) && !(syntax & RE_NO_BK_VBAR)) + token->type = OP_ALT; + break; + case '1': case '2': case '3': case '4': case '5': + case '6': case '7': case '8': case '9': + if (!(syntax & RE_NO_BK_REFS)) + { + token->type = OP_BACK_REF; + token->opr.idx = c2 - '1'; + } + break; + case '<': + if (!(syntax & RE_NO_GNU_OPS)) + { + token->type = ANCHOR; + token->opr.ctx_type = WORD_FIRST; + } + break; + case '>': + if (!(syntax & RE_NO_GNU_OPS)) + { + token->type = ANCHOR; + token->opr.ctx_type = WORD_LAST; + } + break; + case 'b': + if (!(syntax & RE_NO_GNU_OPS)) + { + token->type = ANCHOR; + token->opr.ctx_type = WORD_DELIM; + } + break; + case 'B': + if (!(syntax & RE_NO_GNU_OPS)) + { + token->type = ANCHOR; + token->opr.ctx_type = NOT_WORD_DELIM; + } + break; + case 'w': + if (!(syntax & RE_NO_GNU_OPS)) + token->type = OP_WORD; + break; + case 'W': + if (!(syntax & RE_NO_GNU_OPS)) + token->type = OP_NOTWORD; + break; + case 's': + if (!(syntax & RE_NO_GNU_OPS)) + token->type = OP_SPACE; + break; + case 'S': + if (!(syntax & RE_NO_GNU_OPS)) + token->type = OP_NOTSPACE; + break; + case '`': + if (!(syntax & RE_NO_GNU_OPS)) + { + token->type = ANCHOR; + token->opr.ctx_type = BUF_FIRST; + } + break; + case '\'': + if (!(syntax & RE_NO_GNU_OPS)) + { + token->type = ANCHOR; + token->opr.ctx_type = BUF_LAST; + } + break; + case '(': + if (!(syntax & RE_NO_BK_PARENS)) + token->type = OP_OPEN_SUBEXP; + break; + case ')': + if (!(syntax & RE_NO_BK_PARENS)) + token->type = OP_CLOSE_SUBEXP; + break; + case '+': + if (!(syntax & RE_LIMITED_OPS) && (syntax & RE_BK_PLUS_QM)) + token->type = OP_DUP_PLUS; + break; + case '?': + if (!(syntax & RE_LIMITED_OPS) && (syntax & RE_BK_PLUS_QM)) + token->type = OP_DUP_QUESTION; + break; + case '{': + if ((syntax & RE_INTERVALS) && (!(syntax & RE_NO_BK_BRACES))) + token->type = OP_OPEN_DUP_NUM; + break; + case '}': + if ((syntax & RE_INTERVALS) && (!(syntax & RE_NO_BK_BRACES))) + token->type = OP_CLOSE_DUP_NUM; + break; + default: + break; + } + return 2; + } + + token->type = CHARACTER; +#ifdef RE_ENABLE_I18N + if (input->mb_cur_max > 1) + { + wint_t wc = re_string_wchar_at (input, re_string_cur_idx (input)); + token->word_char = IS_WIDE_WORD_CHAR (wc) != 0; + } + else +#endif + token->word_char = IS_WORD_CHAR (token->opr.c); + + switch (c) + { + case '\n': + if (syntax & RE_NEWLINE_ALT) + token->type = OP_ALT; + break; + case '|': + if (!(syntax & RE_LIMITED_OPS) && (syntax & RE_NO_BK_VBAR)) + token->type = OP_ALT; + break; + case '*': + token->type = OP_DUP_ASTERISK; + break; + case '+': + if (!(syntax & RE_LIMITED_OPS) && !(syntax & RE_BK_PLUS_QM)) + token->type = OP_DUP_PLUS; + break; + case '?': + if (!(syntax & RE_LIMITED_OPS) && !(syntax & RE_BK_PLUS_QM)) + token->type = OP_DUP_QUESTION; + break; + case '{': + if ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) + token->type = OP_OPEN_DUP_NUM; + break; + case '}': + if ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) + token->type = OP_CLOSE_DUP_NUM; + break; + case '(': + if (syntax & RE_NO_BK_PARENS) + token->type = OP_OPEN_SUBEXP; + break; + case ')': + if (syntax & RE_NO_BK_PARENS) + token->type = OP_CLOSE_SUBEXP; + break; + case '[': + token->type = OP_OPEN_BRACKET; + break; + case '.': + token->type = OP_PERIOD; + break; + case '^': + if (!(syntax & (RE_CONTEXT_INDEP_ANCHORS | RE_CARET_ANCHORS_HERE)) && + re_string_cur_idx (input) != 0) + { + char prev = re_string_peek_byte (input, -1); + if (!(syntax & RE_NEWLINE_ALT) || prev != '\n') + break; + } + token->type = ANCHOR; + token->opr.ctx_type = LINE_FIRST; + break; + case '$': + if (!(syntax & RE_CONTEXT_INDEP_ANCHORS) && + re_string_cur_idx (input) + 1 != re_string_length (input)) + { + re_token_t next; + re_string_skip_bytes (input, 1); + peek_token (&next, input, syntax); + re_string_skip_bytes (input, -1); + if (next.type != OP_ALT && next.type != OP_CLOSE_SUBEXP) + break; + } + token->type = ANCHOR; + token->opr.ctx_type = LINE_LAST; + break; + default: + break; + } + return 1; +} + +/* Peek a token from INPUT, and return the length of the token. + We must not use this function out of bracket expressions. */ + +static int +internal_function +peek_token_bracket (re_token_t *token, re_string_t *input, reg_syntax_t syntax) +{ + unsigned char c; + if (re_string_eoi (input)) + { + token->type = END_OF_RE; + return 0; + } + c = re_string_peek_byte (input, 0); + token->opr.c = c; + +#ifdef RE_ENABLE_I18N + if (input->mb_cur_max > 1 && + !re_string_first_byte (input, re_string_cur_idx (input))) + { + token->type = CHARACTER; + return 1; + } +#endif /* RE_ENABLE_I18N */ + + if (c == '\\' && (syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) + && re_string_cur_idx (input) + 1 < re_string_length (input)) + { + /* In this case, '\' escape a character. */ + unsigned char c2; + re_string_skip_bytes (input, 1); + c2 = re_string_peek_byte (input, 0); + token->opr.c = c2; + token->type = CHARACTER; + return 1; + } + if (c == '[') /* '[' is a special char in a bracket exps. */ + { + unsigned char c2; + int token_len; + if (re_string_cur_idx (input) + 1 < re_string_length (input)) + c2 = re_string_peek_byte (input, 1); + else + c2 = 0; + token->opr.c = c2; + token_len = 2; + switch (c2) + { + case '.': + token->type = OP_OPEN_COLL_ELEM; + break; + case '=': + token->type = OP_OPEN_EQUIV_CLASS; + break; + case ':': + if (syntax & RE_CHAR_CLASSES) + { + token->type = OP_OPEN_CHAR_CLASS; + break; + } + /* else fall through. */ + default: + token->type = CHARACTER; + token->opr.c = c; + token_len = 1; + break; + } + return token_len; + } + switch (c) + { + case '-': + token->type = OP_CHARSET_RANGE; + break; + case ']': + token->type = OP_CLOSE_BRACKET; + break; + case '^': + token->type = OP_NON_MATCH_LIST; + break; + default: + token->type = CHARACTER; + } + return 1; +} + +/* Functions for parser. */ + +/* Entry point of the parser. + Parse the regular expression REGEXP and return the structure tree. + If an error is occured, ERR is set by error code, and return NULL. + This function build the following tree, from regular expression <reg_exp>: + CAT + / \ + / \ + <reg_exp> EOR + + CAT means concatenation. + EOR means end of regular expression. */ + +static bin_tree_t * +parse (re_string_t *regexp, regex_t *preg, reg_syntax_t syntax, + reg_errcode_t *err) +{ + re_dfa_t *dfa = (re_dfa_t *) preg->buffer; + bin_tree_t *tree, *eor, *root; + re_token_t current_token; + dfa->syntax = syntax; + fetch_token (¤t_token, regexp, syntax | RE_CARET_ANCHORS_HERE); + tree = parse_reg_exp (regexp, preg, ¤t_token, syntax, 0, err); + if (BE (*err != REG_NOERROR && tree == NULL, 0)) + return NULL; + eor = create_tree (dfa, NULL, NULL, END_OF_RE); + if (tree != NULL) + root = create_tree (dfa, tree, eor, CONCAT); + else + root = eor; + if (BE (eor == NULL || root == NULL, 0)) + { + *err = REG_ESPACE; + return NULL; + } + return root; +} + +/* This function build the following tree, from regular expression + <branch1>|<branch2>: + ALT + / \ + / \ + <branch1> <branch2> + + ALT means alternative, which represents the operator '|'. */ + +static bin_tree_t * +parse_reg_exp (re_string_t *regexp, regex_t *preg, re_token_t *token, + reg_syntax_t syntax, Idx nest, reg_errcode_t *err) +{ + re_dfa_t *dfa = (re_dfa_t *) preg->buffer; + bin_tree_t *tree, *branch = NULL; + tree = parse_branch (regexp, preg, token, syntax, nest, err); + if (BE (*err != REG_NOERROR && tree == NULL, 0)) + return NULL; + + while (token->type == OP_ALT) + { + fetch_token (token, regexp, syntax | RE_CARET_ANCHORS_HERE); + if (token->type != OP_ALT && token->type != END_OF_RE + && (nest == 0 || token->type != OP_CLOSE_SUBEXP)) + { + branch = parse_branch (regexp, preg, token, syntax, nest, err); + if (BE (*err != REG_NOERROR && branch == NULL, 0)) + return NULL; + } + else + branch = NULL; + tree = create_tree (dfa, tree, branch, OP_ALT); + if (BE (tree == NULL, 0)) + { + *err = REG_ESPACE; + return NULL; + } + } + return tree; +} + +/* This function build the following tree, from regular expression + <exp1><exp2>: + CAT + / \ + / \ + <exp1> <exp2> + + CAT means concatenation. */ + +static bin_tree_t * +parse_branch (re_string_t *regexp, regex_t *preg, re_token_t *token, + reg_syntax_t syntax, Idx nest, reg_errcode_t *err) +{ + bin_tree_t *tree, *expr; + re_dfa_t *dfa = (re_dfa_t *) preg->buffer; + tree = parse_expression (regexp, preg, token, syntax, nest, err); + if (BE (*err != REG_NOERROR && tree == NULL, 0)) + return NULL; + + while (token->type != OP_ALT && token->type != END_OF_RE + && (nest == 0 || token->type != OP_CLOSE_SUBEXP)) + { + expr = parse_expression (regexp, preg, token, syntax, nest, err); + if (BE (*err != REG_NOERROR && expr == NULL, 0)) + { + return NULL; + } + if (tree != NULL && expr != NULL) + { + tree = create_tree (dfa, tree, expr, CONCAT); + if (tree == NULL) + { + *err = REG_ESPACE; + return NULL; + } + } + else if (tree == NULL) + tree = expr; + /* Otherwise expr == NULL, we don't need to create new tree. */ + } + return tree; +} + +/* This function build the following tree, from regular expression a*: + * + | + a +*/ + +static bin_tree_t * +parse_expression (re_string_t *regexp, regex_t *preg, re_token_t *token, + reg_syntax_t syntax, Idx nest, reg_errcode_t *err) +{ + re_dfa_t *dfa = (re_dfa_t *) preg->buffer; + bin_tree_t *tree; + switch (token->type) + { + case CHARACTER: + tree = create_token_tree (dfa, NULL, NULL, token); + if (BE (tree == NULL, 0)) + { + *err = REG_ESPACE; + return NULL; + } +#ifdef RE_ENABLE_I18N + if (dfa->mb_cur_max > 1) + { + while (!re_string_eoi (regexp) + && !re_string_first_byte (regexp, re_string_cur_idx (regexp))) + { + bin_tree_t *mbc_remain; + fetch_token (token, regexp, syntax); + mbc_remain = create_token_tree (dfa, NULL, NULL, token); + tree = create_tree (dfa, tree, mbc_remain, CONCAT); + if (BE (mbc_remain == NULL || tree == NULL, 0)) + { + *err = REG_ESPACE; + return NULL; + } + } + } +#endif + break; + case OP_OPEN_SUBEXP: + tree = parse_sub_exp (regexp, preg, token, syntax, nest + 1, err); + if (BE (*err != REG_NOERROR && tree == NULL, 0)) + return NULL; + break; + case OP_OPEN_BRACKET: + tree = parse_bracket_exp (regexp, dfa, token, syntax, err); + if (BE (*err != REG_NOERROR && tree == NULL, 0)) + return NULL; + break; + case OP_BACK_REF: + if (!BE (dfa->completed_bkref_map & (1 << token->opr.idx), 1)) + { + *err = REG_ESUBREG; + return NULL; + } + dfa->used_bkref_map |= 1 << token->opr.idx; + tree = create_token_tree (dfa, NULL, NULL, token); + if (BE (tree == NULL, 0)) + { + *err = REG_ESPACE; + return NULL; + } + ++dfa->nbackref; + dfa->has_mb_node = 1; + break; + case OP_OPEN_DUP_NUM: + if (syntax & RE_CONTEXT_INVALID_DUP) + { + *err = REG_BADRPT; + return NULL; + } + /* FALLTHROUGH */ + case OP_DUP_ASTERISK: + case OP_DUP_PLUS: + case OP_DUP_QUESTION: + if (syntax & RE_CONTEXT_INVALID_OPS) + { + *err = REG_BADRPT; + return NULL; + } + else if (syntax & RE_CONTEXT_INDEP_OPS) + { + fetch_token (token, regexp, syntax); + return parse_expression (regexp, preg, token, syntax, nest, err); + } + /* else fall through */ + case OP_CLOSE_SUBEXP: + if ((token->type == OP_CLOSE_SUBEXP) && + !(syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)) + { + *err = REG_ERPAREN; + return NULL; + } + /* else fall through */ + case OP_CLOSE_DUP_NUM: + /* We treat it as a normal character. */ + + /* Then we can these characters as normal characters. */ + token->type = CHARACTER; + /* mb_partial and word_char bits should be initialized already + by peek_token. */ + tree = create_token_tree (dfa, NULL, NULL, token); + if (BE (tree == NULL, 0)) + { + *err = REG_ESPACE; + return NULL; + } + break; + case ANCHOR: + if ((token->opr.ctx_type + & (WORD_DELIM | NOT_WORD_DELIM | WORD_FIRST | WORD_LAST)) + && dfa->word_ops_used == 0) + init_word_char (dfa); + if (token->opr.ctx_type == WORD_DELIM + || token->opr.ctx_type == NOT_WORD_DELIM) + { + bin_tree_t *tree_first, *tree_last; + if (token->opr.ctx_type == WORD_DELIM) + { + token->opr.ctx_type = WORD_FIRST; + tree_first = create_token_tree (dfa, NULL, NULL, token); + token->opr.ctx_type = WORD_LAST; + } + else + { + token->opr.ctx_type = INSIDE_WORD; + tree_first = create_token_tree (dfa, NULL, NULL, token); + token->opr.ctx_type = INSIDE_NOTWORD; + } + tree_last = create_token_tree (dfa, NULL, NULL, token); + tree = create_tree (dfa, tree_first, tree_last, OP_ALT); + if (BE (tree_first == NULL || tree_last == NULL || tree == NULL, 0)) + { + *err = REG_ESPACE; + return NULL; + } + } + else + { + tree = create_token_tree (dfa, NULL, NULL, token); + if (BE (tree == NULL, 0)) + { + *err = REG_ESPACE; + return NULL; + } + } + /* We must return here, since ANCHORs can't be followed + by repetition operators. + eg. RE"^*" is invalid or "<ANCHOR(^)><CHAR(*)>", + it must not be "<ANCHOR(^)><REPEAT(*)>". */ + fetch_token (token, regexp, syntax); + return tree; + case OP_PERIOD: + tree = create_token_tree (dfa, NULL, NULL, token); + if (BE (tree == NULL, 0)) + { + *err = REG_ESPACE; + return NULL; + } + if (dfa->mb_cur_max > 1) + dfa->has_mb_node = 1; + break; + case OP_WORD: + case OP_NOTWORD: + tree = build_charclass_op (dfa, regexp->trans, + (const unsigned char *) "alnum", + (const unsigned char *) "_", + token->type == OP_NOTWORD, err); + if (BE (*err != REG_NOERROR && tree == NULL, 0)) + return NULL; + break; + case OP_SPACE: + case OP_NOTSPACE: + tree = build_charclass_op (dfa, regexp->trans, + (const unsigned char *) "space", + (const unsigned char *) "", + token->type == OP_NOTSPACE, err); + if (BE (*err != REG_NOERROR && tree == NULL, 0)) + return NULL; + break; + case OP_ALT: + case END_OF_RE: + return NULL; + case BACK_SLASH: + *err = REG_EESCAPE; + return NULL; + default: + /* Must not happen? */ +#ifdef DEBUG + assert (0); +#endif + return NULL; + } + fetch_token (token, regexp, syntax); + + while (token->type == OP_DUP_ASTERISK || token->type == OP_DUP_PLUS + || token->type == OP_DUP_QUESTION || token->type == OP_OPEN_DUP_NUM) + { + tree = parse_dup_op (tree, regexp, dfa, token, syntax, err); + if (BE (*err != REG_NOERROR && tree == NULL, 0)) + return NULL; + /* In BRE consecutive duplications are not allowed. */ + if ((syntax & RE_CONTEXT_INVALID_DUP) + && (token->type == OP_DUP_ASTERISK + || token->type == OP_OPEN_DUP_NUM)) + { + *err = REG_BADRPT; + return NULL; + } + } + + return tree; +} + +/* This function build the following tree, from regular expression + (<reg_exp>): + SUBEXP + | + <reg_exp> +*/ + +static bin_tree_t * +parse_sub_exp (re_string_t *regexp, regex_t *preg, re_token_t *token, + reg_syntax_t syntax, Idx nest, reg_errcode_t *err) +{ + re_dfa_t *dfa = (re_dfa_t *) preg->buffer; + bin_tree_t *tree; + size_t cur_nsub; + cur_nsub = preg->re_nsub++; + + fetch_token (token, regexp, syntax | RE_CARET_ANCHORS_HERE); + + /* The subexpression may be a null string. */ + if (token->type == OP_CLOSE_SUBEXP) + tree = NULL; + else + { + tree = parse_reg_exp (regexp, preg, token, syntax, nest, err); + if (BE (*err == REG_NOERROR && token->type != OP_CLOSE_SUBEXP, 0)) + *err = REG_EPAREN; + if (BE (*err != REG_NOERROR, 0)) + return NULL; + } + + if (cur_nsub <= '9' - '1') + dfa->completed_bkref_map |= 1 << cur_nsub; + + tree = create_tree (dfa, tree, NULL, SUBEXP); + if (BE (tree == NULL, 0)) + { + *err = REG_ESPACE; + return NULL; + } + tree->token.opr.idx = cur_nsub; + return tree; +} + +/* This function parse repetition operators like "*", "+", "{1,3}" etc. */ + +static bin_tree_t * +parse_dup_op (bin_tree_t *elem, re_string_t *regexp, re_dfa_t *dfa, + re_token_t *token, reg_syntax_t syntax, reg_errcode_t *err) +{ + bin_tree_t *tree = NULL, *old_tree = NULL; + Idx i, start, end, start_idx = re_string_cur_idx (regexp); + re_token_t start_token = *token; + + if (token->type == OP_OPEN_DUP_NUM) + { + end = 0; + start = fetch_number (regexp, token, syntax); + if (start == REG_MISSING) + { + if (token->type == CHARACTER && token->opr.c == ',') + start = 0; /* We treat "{,m}" as "{0,m}". */ + else + { + *err = REG_BADBR; /* <re>{} is invalid. */ + return NULL; + } + } + if (BE (start != REG_ERROR, 1)) + { + /* We treat "{n}" as "{n,n}". */ + end = ((token->type == OP_CLOSE_DUP_NUM) ? start + : ((token->type == CHARACTER && token->opr.c == ',') + ? fetch_number (regexp, token, syntax) : REG_ERROR)); + } + if (BE (start == REG_ERROR || end == REG_ERROR, 0)) + { + /* Invalid sequence. */ + if (BE (!(syntax & RE_INVALID_INTERVAL_ORD), 0)) + { + if (token->type == END_OF_RE) + *err = REG_EBRACE; + else + *err = REG_BADBR; + + return NULL; + } + + /* If the syntax bit is set, rollback. */ + re_string_set_index (regexp, start_idx); + *token = start_token; + token->type = CHARACTER; + /* mb_partial and word_char bits should be already initialized by + peek_token. */ + return elem; + } + + if (BE ((end != REG_MISSING && start > end) + || token->type != OP_CLOSE_DUP_NUM, 0)) + { + /* First number greater than second. */ + *err = REG_BADBR; + return NULL; + } + } + else + { + start = (token->type == OP_DUP_PLUS) ? 1 : 0; + end = (token->type == OP_DUP_QUESTION) ? 1 : REG_MISSING; + } + + fetch_token (token, regexp, syntax); + + if (BE (elem == NULL, 0)) + return NULL; + if (BE (start == 0 && end == 0, 0)) + { + postorder (elem, free_tree, NULL); + return NULL; + } + + /* Extract "<re>{n,m}" to "<re><re>...<re><re>{0,<m-n>}". */ + if (BE (start > 0, 0)) + { + tree = elem; + for (i = 2; i <= start; ++i) + { + elem = duplicate_tree (elem, dfa); + tree = create_tree (dfa, tree, elem, CONCAT); + if (BE (elem == NULL || tree == NULL, 0)) + goto parse_dup_op_espace; + } + + if (start == end) + return tree; + + /* Duplicate ELEM before it is marked optional. */ + elem = duplicate_tree (elem, dfa); + old_tree = tree; + } + else + old_tree = NULL; + + if (elem->token.type == SUBEXP) + postorder (elem, mark_opt_subexp, (void *) (long) elem->token.opr.idx); + + tree = create_tree (dfa, elem, NULL, + (end == REG_MISSING ? OP_DUP_ASTERISK : OP_ALT)); + if (BE (tree == NULL, 0)) + goto parse_dup_op_espace; + +/* From gnulib's "intprops.h": + True if the arithmetic type T is signed. */ +#define TYPE_SIGNED(t) (! ((t) 0 < (t) -1)) + + /* This loop is actually executed only when end != REG_MISSING, + to rewrite <re>{0,n} as (<re>(<re>...<re>?)?)?... We have + already created the start+1-th copy. */ + if (TYPE_SIGNED (Idx) || end != REG_MISSING) + for (i = start + 2; i <= end; ++i) + { + elem = duplicate_tree (elem, dfa); + tree = create_tree (dfa, tree, elem, CONCAT); + if (BE (elem == NULL || tree == NULL, 0)) + goto parse_dup_op_espace; + + tree = create_tree (dfa, tree, NULL, OP_ALT); + if (BE (tree == NULL, 0)) + goto parse_dup_op_espace; + } + + if (old_tree) + tree = create_tree (dfa, old_tree, tree, CONCAT); + + return tree; + + parse_dup_op_espace: + *err = REG_ESPACE; + return NULL; +} + +/* Size of the names for collating symbol/equivalence_class/character_class. + I'm not sure, but maybe enough. */ +#define BRACKET_NAME_BUF_SIZE 32 + +#ifndef _LIBC + /* Local function for parse_bracket_exp only used in case of NOT _LIBC. + Build the range expression which starts from START_ELEM, and ends + at END_ELEM. The result are written to MBCSET and SBCSET. + RANGE_ALLOC is the allocated size of mbcset->range_starts, and + mbcset->range_ends, is a pointer argument sinse we may + update it. */ + +static reg_errcode_t +internal_function +# ifdef RE_ENABLE_I18N +build_range_exp (const reg_syntax_t syntax, + bitset_t sbcset, + re_charset_t *mbcset, + Idx *range_alloc, + const bracket_elem_t *start_elem, + const bracket_elem_t *end_elem) +# else /* not RE_ENABLE_I18N */ +build_range_exp (const reg_syntax_t syntax, + bitset_t sbcset, + const bracket_elem_t *start_elem, + const bracket_elem_t *end_elem) +# endif /* not RE_ENABLE_I18N */ +{ + unsigned int start_ch, end_ch; + /* Equivalence Classes and Character Classes can't be a range start/end. */ + if (BE (start_elem->type == EQUIV_CLASS || start_elem->type == CHAR_CLASS + || end_elem->type == EQUIV_CLASS || end_elem->type == CHAR_CLASS, + 0)) + return REG_ERANGE; + + /* We can handle no multi character collating elements without libc + support. */ + if (BE ((start_elem->type == COLL_SYM + && strlen ((char *) start_elem->opr.name) > 1) + || (end_elem->type == COLL_SYM + && strlen ((char *) end_elem->opr.name) > 1), 0)) + return REG_ECOLLATE; + +# ifdef RE_ENABLE_I18N + { + wchar_t wc; + wint_t start_wc; + wint_t end_wc; + wchar_t cmp_buf[6] = {L'\0', L'\0', L'\0', L'\0', L'\0', L'\0'}; + + start_ch = ((start_elem->type == SB_CHAR) ? start_elem->opr.ch + : ((start_elem->type == COLL_SYM) ? start_elem->opr.name[0] + : 0)); + end_ch = ((end_elem->type == SB_CHAR) ? end_elem->opr.ch + : ((end_elem->type == COLL_SYM) ? end_elem->opr.name[0] + : 0)); + start_wc = ((start_elem->type == SB_CHAR || start_elem->type == COLL_SYM) + ? __btowc (start_ch) : start_elem->opr.wch); + end_wc = ((end_elem->type == SB_CHAR || end_elem->type == COLL_SYM) + ? __btowc (end_ch) : end_elem->opr.wch); + if (start_wc == WEOF || end_wc == WEOF) + return REG_ECOLLATE; + cmp_buf[0] = start_wc; + cmp_buf[4] = end_wc; + + if (BE ((syntax & RE_NO_EMPTY_RANGES) + && wcscoll (cmp_buf, cmp_buf + 4) > 0, 0)) + return REG_ERANGE; + + /* Got valid collation sequence values, add them as a new entry. + However, for !_LIBC we have no collation elements: if the + character set is single byte, the single byte character set + that we build below suffices. parse_bracket_exp passes + no MBCSET if dfa->mb_cur_max == 1. */ + if (mbcset) + { + /* Check the space of the arrays. */ + if (BE (*range_alloc == mbcset->nranges, 0)) + { + /* There is not enough space, need realloc. */ + wchar_t *new_array_start, *new_array_end; + Idx new_nranges; + + /* +1 in case of mbcset->nranges is 0. */ + new_nranges = 2 * mbcset->nranges + 1; + /* Use realloc since mbcset->range_starts and mbcset->range_ends + are NULL if *range_alloc == 0. */ + new_array_start = re_realloc (mbcset->range_starts, wchar_t, + new_nranges); + new_array_end = re_realloc (mbcset->range_ends, wchar_t, + new_nranges); + + if (BE (new_array_start == NULL || new_array_end == NULL, 0)) + return REG_ESPACE; + + mbcset->range_starts = new_array_start; + mbcset->range_ends = new_array_end; + *range_alloc = new_nranges; + } + + mbcset->range_starts[mbcset->nranges] = start_wc; + mbcset->range_ends[mbcset->nranges++] = end_wc; + } + + /* Build the table for single byte characters. */ + for (wc = 0; wc < SBC_MAX; ++wc) + { + cmp_buf[2] = wc; + if (wcscoll (cmp_buf, cmp_buf + 2) <= 0 + && wcscoll (cmp_buf + 2, cmp_buf + 4) <= 0) + bitset_set (sbcset, wc); + } + } +# else /* not RE_ENABLE_I18N */ + { + unsigned int ch; + start_ch = ((start_elem->type == SB_CHAR ) ? start_elem->opr.ch + : ((start_elem->type == COLL_SYM) ? start_elem->opr.name[0] + : 0)); + end_ch = ((end_elem->type == SB_CHAR ) ? end_elem->opr.ch + : ((end_elem->type == COLL_SYM) ? end_elem->opr.name[0] + : 0)); + if (start_ch > end_ch) + return REG_ERANGE; + /* Build the table for single byte characters. */ + for (ch = 0; ch < SBC_MAX; ++ch) + if (start_ch <= ch && ch <= end_ch) + bitset_set (sbcset, ch); + } +# endif /* not RE_ENABLE_I18N */ + return REG_NOERROR; +} +#endif /* not _LIBC */ + +#ifndef _LIBC +/* Helper function for parse_bracket_exp only used in case of NOT _LIBC.. + Build the collating element which is represented by NAME. + The result are written to MBCSET and SBCSET. + COLL_SYM_ALLOC is the allocated size of mbcset->coll_sym, is a + pointer argument since we may update it. */ + +static reg_errcode_t +internal_function +build_collating_symbol (bitset_t sbcset, +# ifdef RE_ENABLE_I18N + re_charset_t *mbcset, Idx *coll_sym_alloc, +# endif + const unsigned char *name) +{ + size_t name_len = strlen ((const char *) name); + if (BE (name_len != 1, 0)) + return REG_ECOLLATE; + else + { + bitset_set (sbcset, name[0]); + return REG_NOERROR; + } +} +#endif /* not _LIBC */ + +/* This function parse bracket expression like "[abc]", "[a-c]", + "[[.a-a.]]" etc. */ + +static bin_tree_t * +parse_bracket_exp (re_string_t *regexp, re_dfa_t *dfa, re_token_t *token, + reg_syntax_t syntax, reg_errcode_t *err) +{ +#ifdef _LIBC + const unsigned char *collseqmb; + const char *collseqwc; + uint32_t nrules; + int32_t table_size; + const int32_t *symb_table; + const unsigned char *extra; + + /* Local function for parse_bracket_exp used in _LIBC environement. + Seek the collating symbol entry correspondings to NAME. + Return the index of the symbol in the SYMB_TABLE. */ + + auto inline int32_t + __attribute ((always_inline)) + seek_collating_symbol_entry (name, name_len) + const unsigned char *name; + size_t name_len; + { + int32_t hash = elem_hash ((const char *) name, name_len); + int32_t elem = hash % table_size; + if (symb_table[2 * elem] != 0) + { + int32_t second = hash % (table_size - 2) + 1; + + do + { + /* First compare the hashing value. */ + if (symb_table[2 * elem] == hash + /* Compare the length of the name. */ + && name_len == extra[symb_table[2 * elem + 1]] + /* Compare the name. */ + && memcmp (name, &extra[symb_table[2 * elem + 1] + 1], + name_len) == 0) + { + /* Yep, this is the entry. */ + break; + } + + /* Next entry. */ + elem += second; + } + while (symb_table[2 * elem] != 0); + } + return elem; + } + + /* Local function for parse_bracket_exp used in _LIBC environment. + Look up the collation sequence value of BR_ELEM. + Return the value if succeeded, UINT_MAX otherwise. */ + + auto inline unsigned int + __attribute ((always_inline)) + lookup_collation_sequence_value (br_elem) + bracket_elem_t *br_elem; + { + if (br_elem->type == SB_CHAR) + { + /* + if (MB_CUR_MAX == 1) + */ + if (nrules == 0) + return collseqmb[br_elem->opr.ch]; + else + { + wint_t wc = __btowc (br_elem->opr.ch); + return __collseq_table_lookup (collseqwc, wc); + } + } + else if (br_elem->type == MB_CHAR) + { + if (nrules != 0) + return __collseq_table_lookup (collseqwc, br_elem->opr.wch); + } + else if (br_elem->type == COLL_SYM) + { + size_t sym_name_len = strlen ((char *) br_elem->opr.name); + if (nrules != 0) + { + int32_t elem, idx; + elem = seek_collating_symbol_entry (br_elem->opr.name, + sym_name_len); + if (symb_table[2 * elem] != 0) + { + /* We found the entry. */ + idx = symb_table[2 * elem + 1]; + /* Skip the name of collating element name. */ + idx += 1 + extra[idx]; + /* Skip the byte sequence of the collating element. */ + idx += 1 + extra[idx]; + /* Adjust for the alignment. */ + idx = (idx + 3) & ~3; + /* Skip the multibyte collation sequence value. */ + idx += sizeof (unsigned int); + /* Skip the wide char sequence of the collating element. */ + idx += sizeof (unsigned int) * + (1 + *(unsigned int *) (extra + idx)); + /* Return the collation sequence value. */ + return *(unsigned int *) (extra + idx); + } + else if (symb_table[2 * elem] == 0 && sym_name_len == 1) + { + /* No valid character. Match it as a single byte + character. */ + return collseqmb[br_elem->opr.name[0]]; + } + } + else if (sym_name_len == 1) + return collseqmb[br_elem->opr.name[0]]; + } + return UINT_MAX; + } + + /* Local function for parse_bracket_exp used in _LIBC environement. + Build the range expression which starts from START_ELEM, and ends + at END_ELEM. The result are written to MBCSET and SBCSET. + RANGE_ALLOC is the allocated size of mbcset->range_starts, and + mbcset->range_ends, is a pointer argument sinse we may + update it. */ + + auto inline reg_errcode_t + __attribute ((always_inline)) + build_range_exp (sbcset, mbcset, range_alloc, start_elem, end_elem) + re_charset_t *mbcset; + Idx *range_alloc; + bitset_t sbcset; + bracket_elem_t *start_elem, *end_elem; + { + unsigned int ch; + uint32_t start_collseq; + uint32_t end_collseq; + + /* Equivalence Classes and Character Classes can't be a range + start/end. */ + if (BE (start_elem->type == EQUIV_CLASS || start_elem->type == CHAR_CLASS + || end_elem->type == EQUIV_CLASS || end_elem->type == CHAR_CLASS, + 0)) + return REG_ERANGE; + + start_collseq = lookup_collation_sequence_value (start_elem); + end_collseq = lookup_collation_sequence_value (end_elem); + /* Check start/end collation sequence values. */ + if (BE (start_collseq == UINT_MAX || end_collseq == UINT_MAX, 0)) + return REG_ECOLLATE; + if (BE ((syntax & RE_NO_EMPTY_RANGES) && start_collseq > end_collseq, 0)) + return REG_ERANGE; + + /* Got valid collation sequence values, add them as a new entry. + However, if we have no collation elements, and the character set + is single byte, the single byte character set that we + build below suffices. */ + if (nrules > 0 || dfa->mb_cur_max > 1) + { + /* Check the space of the arrays. */ + if (BE (*range_alloc == mbcset->nranges, 0)) + { + /* There is not enough space, need realloc. */ + uint32_t *new_array_start; + uint32_t *new_array_end; + Idx new_nranges; + + /* +1 in case of mbcset->nranges is 0. */ + new_nranges = 2 * mbcset->nranges + 1; + new_array_start = re_realloc (mbcset->range_starts, uint32_t, + new_nranges); + new_array_end = re_realloc (mbcset->range_ends, uint32_t, + new_nranges); + + if (BE (new_array_start == NULL || new_array_end == NULL, 0)) + return REG_ESPACE; + + mbcset->range_starts = new_array_start; + mbcset->range_ends = new_array_end; + *range_alloc = new_nranges; + } + + mbcset->range_starts[mbcset->nranges] = start_collseq; + mbcset->range_ends[mbcset->nranges++] = end_collseq; + } + + /* Build the table for single byte characters. */ + for (ch = 0; ch < SBC_MAX; ch++) + { + uint32_t ch_collseq; + /* + if (MB_CUR_MAX == 1) + */ + if (nrules == 0) + ch_collseq = collseqmb[ch]; + else + ch_collseq = __collseq_table_lookup (collseqwc, __btowc (ch)); + if (start_collseq <= ch_collseq && ch_collseq <= end_collseq) + bitset_set (sbcset, ch); + } + return REG_NOERROR; + } + + /* Local function for parse_bracket_exp used in _LIBC environement. + Build the collating element which is represented by NAME. + The result are written to MBCSET and SBCSET. + COLL_SYM_ALLOC is the allocated size of mbcset->coll_sym, is a + pointer argument sinse we may update it. */ + + auto inline reg_errcode_t + __attribute ((always_inline)) + build_collating_symbol (sbcset, mbcset, coll_sym_alloc, name) + re_charset_t *mbcset; + Idx *coll_sym_alloc; + bitset_t sbcset; + const unsigned char *name; + { + int32_t elem, idx; + size_t name_len = strlen ((const char *) name); + if (nrules != 0) + { + elem = seek_collating_symbol_entry (name, name_len); + if (symb_table[2 * elem] != 0) + { + /* We found the entry. */ + idx = symb_table[2 * elem + 1]; + /* Skip the name of collating element name. */ + idx += 1 + extra[idx]; + } + else if (symb_table[2 * elem] == 0 && name_len == 1) + { + /* No valid character, treat it as a normal + character. */ + bitset_set (sbcset, name[0]); + return REG_NOERROR; + } + else + return REG_ECOLLATE; + + /* Got valid collation sequence, add it as a new entry. */ + /* Check the space of the arrays. */ + if (BE (*coll_sym_alloc == mbcset->ncoll_syms, 0)) + { + /* Not enough, realloc it. */ + /* +1 in case of mbcset->ncoll_syms is 0. */ + Idx new_coll_sym_alloc = 2 * mbcset->ncoll_syms + 1; + /* Use realloc since mbcset->coll_syms is NULL + if *alloc == 0. */ + int32_t *new_coll_syms = re_realloc (mbcset->coll_syms, int32_t, + new_coll_sym_alloc); + if (BE (new_coll_syms == NULL, 0)) + return REG_ESPACE; + mbcset->coll_syms = new_coll_syms; + *coll_sym_alloc = new_coll_sym_alloc; + } + mbcset->coll_syms[mbcset->ncoll_syms++] = idx; + return REG_NOERROR; + } + else + { + if (BE (name_len != 1, 0)) + return REG_ECOLLATE; + else + { + bitset_set (sbcset, name[0]); + return REG_NOERROR; + } + } + } +#endif + + re_token_t br_token; + re_bitset_ptr_t sbcset; +#ifdef RE_ENABLE_I18N + re_charset_t *mbcset; + Idx coll_sym_alloc = 0, range_alloc = 0, mbchar_alloc = 0; + Idx equiv_class_alloc = 0, char_class_alloc = 0; +#endif /* not RE_ENABLE_I18N */ + bool non_match = false; + bin_tree_t *work_tree; + int token_len; + bool first_round = true; +#ifdef _LIBC + collseqmb = (const unsigned char *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_COLLSEQMB); + nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); + if (nrules) + { + /* + if (MB_CUR_MAX > 1) + */ + collseqwc = _NL_CURRENT (LC_COLLATE, _NL_COLLATE_COLLSEQWC); + table_size = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_SYMB_HASH_SIZEMB); + symb_table = (const int32_t *) _NL_CURRENT (LC_COLLATE, + _NL_COLLATE_SYMB_TABLEMB); + extra = (const unsigned char *) _NL_CURRENT (LC_COLLATE, + _NL_COLLATE_SYMB_EXTRAMB); + } +#endif + sbcset = (re_bitset_ptr_t) calloc (sizeof (bitset_t), 1); +#ifdef RE_ENABLE_I18N + mbcset = (re_charset_t *) calloc (sizeof (re_charset_t), 1); +#endif /* RE_ENABLE_I18N */ +#ifdef RE_ENABLE_I18N + if (BE (sbcset == NULL || mbcset == NULL, 0)) +#else + if (BE (sbcset == NULL, 0)) +#endif /* RE_ENABLE_I18N */ + { + *err = REG_ESPACE; + return NULL; + } + + token_len = peek_token_bracket (token, regexp, syntax); + if (BE (token->type == END_OF_RE, 0)) + { + *err = REG_BADPAT; + goto parse_bracket_exp_free_return; + } + if (token->type == OP_NON_MATCH_LIST) + { +#ifdef RE_ENABLE_I18N + mbcset->non_match = 1; +#endif /* not RE_ENABLE_I18N */ + non_match = true; + if (syntax & RE_HAT_LISTS_NOT_NEWLINE) + bitset_set (sbcset, '\n'); + re_string_skip_bytes (regexp, token_len); /* Skip a token. */ + token_len = peek_token_bracket (token, regexp, syntax); + if (BE (token->type == END_OF_RE, 0)) + { + *err = REG_BADPAT; + goto parse_bracket_exp_free_return; + } + } + + /* We treat the first ']' as a normal character. */ + if (token->type == OP_CLOSE_BRACKET) + token->type = CHARACTER; + + while (1) + { + bracket_elem_t start_elem, end_elem; + unsigned char start_name_buf[BRACKET_NAME_BUF_SIZE]; + unsigned char end_name_buf[BRACKET_NAME_BUF_SIZE]; + reg_errcode_t ret; + int token_len2 = 0; + bool is_range_exp = false; + re_token_t token2; + + start_elem.opr.name = start_name_buf; + ret = parse_bracket_element (&start_elem, regexp, token, token_len, dfa, + syntax, first_round); + if (BE (ret != REG_NOERROR, 0)) + { + *err = ret; + goto parse_bracket_exp_free_return; + } + first_round = false; + + /* Get information about the next token. We need it in any case. */ + token_len = peek_token_bracket (token, regexp, syntax); + + /* Do not check for ranges if we know they are not allowed. */ + if (start_elem.type != CHAR_CLASS && start_elem.type != EQUIV_CLASS) + { + if (BE (token->type == END_OF_RE, 0)) + { + *err = REG_EBRACK; + goto parse_bracket_exp_free_return; + } + if (token->type == OP_CHARSET_RANGE) + { + re_string_skip_bytes (regexp, token_len); /* Skip '-'. */ + token_len2 = peek_token_bracket (&token2, regexp, syntax); + if (BE (token2.type == END_OF_RE, 0)) + { + *err = REG_EBRACK; + goto parse_bracket_exp_free_return; + } + if (token2.type == OP_CLOSE_BRACKET) + { + /* We treat the last '-' as a normal character. */ + re_string_skip_bytes (regexp, -token_len); + token->type = CHARACTER; + } + else + is_range_exp = true; + } + } + + if (is_range_exp == true) + { + end_elem.opr.name = end_name_buf; + ret = parse_bracket_element (&end_elem, regexp, &token2, token_len2, + dfa, syntax, true); + if (BE (ret != REG_NOERROR, 0)) + { + *err = ret; + goto parse_bracket_exp_free_return; + } + + token_len = peek_token_bracket (token, regexp, syntax); + +#ifdef _LIBC + *err = build_range_exp (sbcset, mbcset, &range_alloc, + &start_elem, &end_elem); +#else +# ifdef RE_ENABLE_I18N + *err = build_range_exp (syntax, sbcset, + dfa->mb_cur_max > 1 ? mbcset : NULL, + &range_alloc, &start_elem, &end_elem); +# else + *err = build_range_exp (syntax, sbcset, &start_elem, &end_elem); +# endif +#endif /* RE_ENABLE_I18N */ + if (BE (*err != REG_NOERROR, 0)) + goto parse_bracket_exp_free_return; + } + else + { + switch (start_elem.type) + { + case SB_CHAR: + bitset_set (sbcset, start_elem.opr.ch); + break; +#ifdef RE_ENABLE_I18N + case MB_CHAR: + /* Check whether the array has enough space. */ + if (BE (mbchar_alloc == mbcset->nmbchars, 0)) + { + wchar_t *new_mbchars; + /* Not enough, realloc it. */ + /* +1 in case of mbcset->nmbchars is 0. */ + mbchar_alloc = 2 * mbcset->nmbchars + 1; + /* Use realloc since array is NULL if *alloc == 0. */ + new_mbchars = re_realloc (mbcset->mbchars, wchar_t, + mbchar_alloc); + if (BE (new_mbchars == NULL, 0)) + goto parse_bracket_exp_espace; + mbcset->mbchars = new_mbchars; + } + mbcset->mbchars[mbcset->nmbchars++] = start_elem.opr.wch; + break; +#endif /* RE_ENABLE_I18N */ + case EQUIV_CLASS: + *err = build_equiv_class (sbcset, +#ifdef RE_ENABLE_I18N + mbcset, &equiv_class_alloc, +#endif /* RE_ENABLE_I18N */ + start_elem.opr.name); + if (BE (*err != REG_NOERROR, 0)) + goto parse_bracket_exp_free_return; + break; + case COLL_SYM: + *err = build_collating_symbol (sbcset, +#ifdef RE_ENABLE_I18N + mbcset, &coll_sym_alloc, +#endif /* RE_ENABLE_I18N */ + start_elem.opr.name); + if (BE (*err != REG_NOERROR, 0)) + goto parse_bracket_exp_free_return; + break; + case CHAR_CLASS: + *err = build_charclass (regexp->trans, sbcset, +#ifdef RE_ENABLE_I18N + mbcset, &char_class_alloc, +#endif /* RE_ENABLE_I18N */ + start_elem.opr.name, syntax); + if (BE (*err != REG_NOERROR, 0)) + goto parse_bracket_exp_free_return; + break; + default: + assert (0); + break; + } + } + if (BE (token->type == END_OF_RE, 0)) + { + *err = REG_EBRACK; + goto parse_bracket_exp_free_return; + } + if (token->type == OP_CLOSE_BRACKET) + break; + } + + re_string_skip_bytes (regexp, token_len); /* Skip a token. */ + + /* If it is non-matching list. */ + if (non_match) + bitset_not (sbcset); + +#ifdef RE_ENABLE_I18N + /* Ensure only single byte characters are set. */ + if (dfa->mb_cur_max > 1) + bitset_mask (sbcset, dfa->sb_char); + + if (mbcset->nmbchars || mbcset->ncoll_syms || mbcset->nequiv_classes + || mbcset->nranges || (dfa->mb_cur_max > 1 && (mbcset->nchar_classes + || mbcset->non_match))) + { + bin_tree_t *mbc_tree; + int sbc_idx; + /* Build a tree for complex bracket. */ + dfa->has_mb_node = 1; + br_token.type = COMPLEX_BRACKET; + br_token.opr.mbcset = mbcset; + mbc_tree = create_token_tree (dfa, NULL, NULL, &br_token); + if (BE (mbc_tree == NULL, 0)) + goto parse_bracket_exp_espace; + for (sbc_idx = 0; sbc_idx < BITSET_WORDS; ++sbc_idx) + if (sbcset[sbc_idx]) + break; + /* If there are no bits set in sbcset, there is no point + of having both SIMPLE_BRACKET and COMPLEX_BRACKET. */ + if (sbc_idx < BITSET_WORDS) + { + /* Build a tree for simple bracket. */ + br_token.type = SIMPLE_BRACKET; + br_token.opr.sbcset = sbcset; + work_tree = create_token_tree (dfa, NULL, NULL, &br_token); + if (BE (work_tree == NULL, 0)) + goto parse_bracket_exp_espace; + + /* Then join them by ALT node. */ + work_tree = create_tree (dfa, work_tree, mbc_tree, OP_ALT); + if (BE (work_tree == NULL, 0)) + goto parse_bracket_exp_espace; + } + else + { + re_free (sbcset); + work_tree = mbc_tree; + } + } + else +#endif /* not RE_ENABLE_I18N */ + { +#ifdef RE_ENABLE_I18N + free_charset (mbcset); +#endif + /* Build a tree for simple bracket. */ + br_token.type = SIMPLE_BRACKET; + br_token.opr.sbcset = sbcset; + work_tree = create_token_tree (dfa, NULL, NULL, &br_token); + if (BE (work_tree == NULL, 0)) + goto parse_bracket_exp_espace; + } + return work_tree; + + parse_bracket_exp_espace: + *err = REG_ESPACE; + parse_bracket_exp_free_return: + re_free (sbcset); +#ifdef RE_ENABLE_I18N + free_charset (mbcset); +#endif /* RE_ENABLE_I18N */ + return NULL; +} + +/* Parse an element in the bracket expression. */ + +static reg_errcode_t +parse_bracket_element (bracket_elem_t *elem, re_string_t *regexp, + re_token_t *token, int token_len, re_dfa_t *dfa, + reg_syntax_t syntax, bool accept_hyphen) +{ +#ifdef RE_ENABLE_I18N + int cur_char_size; + cur_char_size = re_string_char_size_at (regexp, re_string_cur_idx (regexp)); + if (cur_char_size > 1) + { + elem->type = MB_CHAR; + elem->opr.wch = re_string_wchar_at (regexp, re_string_cur_idx (regexp)); + re_string_skip_bytes (regexp, cur_char_size); + return REG_NOERROR; + } +#endif /* RE_ENABLE_I18N */ + re_string_skip_bytes (regexp, token_len); /* Skip a token. */ + if (token->type == OP_OPEN_COLL_ELEM || token->type == OP_OPEN_CHAR_CLASS + || token->type == OP_OPEN_EQUIV_CLASS) + return parse_bracket_symbol (elem, regexp, token); + if (BE (token->type == OP_CHARSET_RANGE, 0) && !accept_hyphen) + { + /* A '-' must only appear as anything but a range indicator before + the closing bracket. Everything else is an error. */ + re_token_t token2; + (void) peek_token_bracket (&token2, regexp, syntax); + if (token2.type != OP_CLOSE_BRACKET) + /* The actual error value is not standardized since this whole + case is undefined. But ERANGE makes good sense. */ + return REG_ERANGE; + } + elem->type = SB_CHAR; + elem->opr.ch = token->opr.c; + return REG_NOERROR; +} + +/* Parse a bracket symbol in the bracket expression. Bracket symbols are + such as [:<character_class>:], [.<collating_element>.], and + [=<equivalent_class>=]. */ + +static reg_errcode_t +parse_bracket_symbol (bracket_elem_t *elem, re_string_t *regexp, + re_token_t *token) +{ + unsigned char ch, delim = token->opr.c; + int i = 0; + if (re_string_eoi(regexp)) + return REG_EBRACK; + for (;; ++i) + { + if (i >= BRACKET_NAME_BUF_SIZE) + return REG_EBRACK; + if (token->type == OP_OPEN_CHAR_CLASS) + ch = re_string_fetch_byte_case (regexp); + else + ch = re_string_fetch_byte (regexp); + if (re_string_eoi(regexp)) + return REG_EBRACK; + if (ch == delim && re_string_peek_byte (regexp, 0) == ']') + break; + elem->opr.name[i] = ch; + } + re_string_skip_bytes (regexp, 1); + elem->opr.name[i] = '\0'; + switch (token->type) + { + case OP_OPEN_COLL_ELEM: + elem->type = COLL_SYM; + break; + case OP_OPEN_EQUIV_CLASS: + elem->type = EQUIV_CLASS; + break; + case OP_OPEN_CHAR_CLASS: + elem->type = CHAR_CLASS; + break; + default: + break; + } + return REG_NOERROR; +} + + /* Helper function for parse_bracket_exp. + Build the equivalence class which is represented by NAME. + The result are written to MBCSET and SBCSET. + EQUIV_CLASS_ALLOC is the allocated size of mbcset->equiv_classes, + is a pointer argument sinse we may update it. */ + +static reg_errcode_t +#ifdef RE_ENABLE_I18N +build_equiv_class (bitset_t sbcset, re_charset_t *mbcset, + Idx *equiv_class_alloc, const unsigned char *name) +#else /* not RE_ENABLE_I18N */ +build_equiv_class (bitset_t sbcset, const unsigned char *name) +#endif /* not RE_ENABLE_I18N */ +{ +#ifdef _LIBC + uint32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); + if (nrules != 0) + { + const int32_t *table, *indirect; + const unsigned char *weights, *extra, *cp; + unsigned char char_buf[2]; + int32_t idx1, idx2; + unsigned int ch; + size_t len; + /* This #include defines a local function! */ +# include <locale/weight.h> + /* Calculate the index for equivalence class. */ + cp = name; + table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); + weights = (const unsigned char *) _NL_CURRENT (LC_COLLATE, + _NL_COLLATE_WEIGHTMB); + extra = (const unsigned char *) _NL_CURRENT (LC_COLLATE, + _NL_COLLATE_EXTRAMB); + indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, + _NL_COLLATE_INDIRECTMB); + idx1 = findidx (&cp); + if (BE (idx1 == 0 || cp < name + strlen ((const char *) name), 0)) + /* This isn't a valid character. */ + return REG_ECOLLATE; + + /* Build single byte matcing table for this equivalence class. */ + char_buf[1] = (unsigned char) '\0'; + len = weights[idx1 & 0xffffff]; + for (ch = 0; ch < SBC_MAX; ++ch) + { + char_buf[0] = ch; + cp = char_buf; + idx2 = findidx (&cp); +/* + idx2 = table[ch]; +*/ + if (idx2 == 0) + /* This isn't a valid character. */ + continue; + /* Compare only if the length matches and the collation rule + index is the same. */ + if (len == weights[idx2 & 0xffffff] && (idx1 >> 24) == (idx2 >> 24)) + { + int cnt = 0; + + while (cnt <= len && + weights[(idx1 & 0xffffff) + 1 + cnt] + == weights[(idx2 & 0xffffff) + 1 + cnt]) + ++cnt; + + if (cnt > len) + bitset_set (sbcset, ch); + } + } + /* Check whether the array has enough space. */ + if (BE (*equiv_class_alloc == mbcset->nequiv_classes, 0)) + { + /* Not enough, realloc it. */ + /* +1 in case of mbcset->nequiv_classes is 0. */ + Idx new_equiv_class_alloc = 2 * mbcset->nequiv_classes + 1; + /* Use realloc since the array is NULL if *alloc == 0. */ + int32_t *new_equiv_classes = re_realloc (mbcset->equiv_classes, + int32_t, + new_equiv_class_alloc); + if (BE (new_equiv_classes == NULL, 0)) + return REG_ESPACE; + mbcset->equiv_classes = new_equiv_classes; + *equiv_class_alloc = new_equiv_class_alloc; + } + mbcset->equiv_classes[mbcset->nequiv_classes++] = idx1; + } + else +#endif /* _LIBC */ + { + if (BE (strlen ((const char *) name) != 1, 0)) + return REG_ECOLLATE; + bitset_set (sbcset, *name); + } + return REG_NOERROR; +} + + /* Helper function for parse_bracket_exp. + Build the character class which is represented by NAME. + The result are written to MBCSET and SBCSET. + CHAR_CLASS_ALLOC is the allocated size of mbcset->char_classes, + is a pointer argument sinse we may update it. */ + +static reg_errcode_t +#ifdef RE_ENABLE_I18N +build_charclass (RE_TRANSLATE_TYPE trans, bitset_t sbcset, + re_charset_t *mbcset, Idx *char_class_alloc, + const unsigned char *class_name, reg_syntax_t syntax) +#else /* not RE_ENABLE_I18N */ +build_charclass (RE_TRANSLATE_TYPE trans, bitset_t sbcset, + const unsigned char *class_name, reg_syntax_t syntax) +#endif /* not RE_ENABLE_I18N */ +{ + int i; + const char *name = (const char *) class_name; + + /* In case of REG_ICASE "upper" and "lower" match the both of + upper and lower cases. */ + if ((syntax & RE_ICASE) + && (strcmp (name, "upper") == 0 || strcmp (name, "lower") == 0)) + name = "alpha"; + +#ifdef RE_ENABLE_I18N + /* Check the space of the arrays. */ + if (BE (*char_class_alloc == mbcset->nchar_classes, 0)) + { + /* Not enough, realloc it. */ + /* +1 in case of mbcset->nchar_classes is 0. */ + Idx new_char_class_alloc = 2 * mbcset->nchar_classes + 1; + /* Use realloc since array is NULL if *alloc == 0. */ + wctype_t *new_char_classes = re_realloc (mbcset->char_classes, wctype_t, + new_char_class_alloc); + if (BE (new_char_classes == NULL, 0)) + return REG_ESPACE; + mbcset->char_classes = new_char_classes; + *char_class_alloc = new_char_class_alloc; + } + mbcset->char_classes[mbcset->nchar_classes++] = __wctype (name); +#endif /* RE_ENABLE_I18N */ + +#define BUILD_CHARCLASS_LOOP(ctype_func) \ + do { \ + if (BE (trans != NULL, 0)) \ + { \ + for (i = 0; i < SBC_MAX; ++i) \ + if (ctype_func (i)) \ + bitset_set (sbcset, trans[i]); \ + } \ + else \ + { \ + for (i = 0; i < SBC_MAX; ++i) \ + if (ctype_func (i)) \ + bitset_set (sbcset, i); \ + } \ + } while (0) + + if (strcmp (name, "alnum") == 0) + BUILD_CHARCLASS_LOOP (isalnum); + else if (strcmp (name, "cntrl") == 0) + BUILD_CHARCLASS_LOOP (iscntrl); + else if (strcmp (name, "lower") == 0) + BUILD_CHARCLASS_LOOP (islower); + else if (strcmp (name, "space") == 0) + BUILD_CHARCLASS_LOOP (isspace); + else if (strcmp (name, "alpha") == 0) + BUILD_CHARCLASS_LOOP (isalpha); + else if (strcmp (name, "digit") == 0) + BUILD_CHARCLASS_LOOP (isdigit); + else if (strcmp (name, "print") == 0) + BUILD_CHARCLASS_LOOP (isprint); + else if (strcmp (name, "upper") == 0) + BUILD_CHARCLASS_LOOP (isupper); + else if (strcmp (name, "blank") == 0) + BUILD_CHARCLASS_LOOP (isblank); + else if (strcmp (name, "graph") == 0) + BUILD_CHARCLASS_LOOP (isgraph); + else if (strcmp (name, "punct") == 0) + BUILD_CHARCLASS_LOOP (ispunct); + else if (strcmp (name, "xdigit") == 0) + BUILD_CHARCLASS_LOOP (isxdigit); + else + return REG_ECTYPE; + + return REG_NOERROR; +} + +static bin_tree_t * +build_charclass_op (re_dfa_t *dfa, RE_TRANSLATE_TYPE trans, + const unsigned char *class_name, + const unsigned char *extra, bool non_match, + reg_errcode_t *err) +{ + re_bitset_ptr_t sbcset; +#ifdef RE_ENABLE_I18N + re_charset_t *mbcset; + Idx alloc = 0; +#endif /* not RE_ENABLE_I18N */ + reg_errcode_t ret; + re_token_t br_token; + bin_tree_t *tree; + + sbcset = (re_bitset_ptr_t) calloc (sizeof (bitset_t), 1); +#ifdef RE_ENABLE_I18N + mbcset = (re_charset_t *) calloc (sizeof (re_charset_t), 1); +#endif /* RE_ENABLE_I18N */ + +#ifdef RE_ENABLE_I18N + if (BE (sbcset == NULL || mbcset == NULL, 0)) +#else /* not RE_ENABLE_I18N */ + if (BE (sbcset == NULL, 0)) +#endif /* not RE_ENABLE_I18N */ + { + *err = REG_ESPACE; + return NULL; + } + + if (non_match) + { +#ifdef RE_ENABLE_I18N + mbcset->non_match = 1; +#endif /* not RE_ENABLE_I18N */ + } + + /* We don't care the syntax in this case. */ + ret = build_charclass (trans, sbcset, +#ifdef RE_ENABLE_I18N + mbcset, &alloc, +#endif /* RE_ENABLE_I18N */ + class_name, 0); + + if (BE (ret != REG_NOERROR, 0)) + { + re_free (sbcset); +#ifdef RE_ENABLE_I18N + free_charset (mbcset); +#endif /* RE_ENABLE_I18N */ + *err = ret; + return NULL; + } + /* \w match '_' also. */ + for (; *extra; extra++) + bitset_set (sbcset, *extra); + + /* If it is non-matching list. */ + if (non_match) + bitset_not (sbcset); + +#ifdef RE_ENABLE_I18N + /* Ensure only single byte characters are set. */ + if (dfa->mb_cur_max > 1) + bitset_mask (sbcset, dfa->sb_char); +#endif + + /* Build a tree for simple bracket. */ + br_token.type = SIMPLE_BRACKET; + br_token.opr.sbcset = sbcset; + tree = create_token_tree (dfa, NULL, NULL, &br_token); + if (BE (tree == NULL, 0)) + goto build_word_op_espace; + +#ifdef RE_ENABLE_I18N + if (dfa->mb_cur_max > 1) + { + bin_tree_t *mbc_tree; + /* Build a tree for complex bracket. */ + br_token.type = COMPLEX_BRACKET; + br_token.opr.mbcset = mbcset; + dfa->has_mb_node = 1; + mbc_tree = create_token_tree (dfa, NULL, NULL, &br_token); + if (BE (mbc_tree == NULL, 0)) + goto build_word_op_espace; + /* Then join them by ALT node. */ + tree = create_tree (dfa, tree, mbc_tree, OP_ALT); + if (BE (mbc_tree != NULL, 1)) + return tree; + } + else + { + free_charset (mbcset); + return tree; + } +#else /* not RE_ENABLE_I18N */ + return tree; +#endif /* not RE_ENABLE_I18N */ + + build_word_op_espace: + re_free (sbcset); +#ifdef RE_ENABLE_I18N + free_charset (mbcset); +#endif /* RE_ENABLE_I18N */ + *err = REG_ESPACE; + return NULL; +} + +/* This is intended for the expressions like "a{1,3}". + Fetch a number from 'input', and return the number. + Return REG_MISSING if the number field is empty like "{,1}". + Return REG_ERROR if an error occurred. */ + +static Idx +fetch_number (re_string_t *input, re_token_t *token, reg_syntax_t syntax) +{ + Idx num = REG_MISSING; + unsigned char c; + while (1) + { + fetch_token (token, input, syntax); + c = token->opr.c; + if (BE (token->type == END_OF_RE, 0)) + return REG_ERROR; + if (token->type == OP_CLOSE_DUP_NUM || c == ',') + break; + num = ((token->type != CHARACTER || c < '0' || '9' < c + || num == REG_ERROR) + ? REG_ERROR + : ((num == REG_MISSING) ? c - '0' : num * 10 + c - '0')); + num = (num > RE_DUP_MAX) ? REG_ERROR : num; + } + return num; +} + +#ifdef RE_ENABLE_I18N +static void +free_charset (re_charset_t *cset) +{ + re_free (cset->mbchars); +# ifdef _LIBC + re_free (cset->coll_syms); + re_free (cset->equiv_classes); + re_free (cset->range_starts); + re_free (cset->range_ends); +# endif + re_free (cset->char_classes); + re_free (cset); +} +#endif /* RE_ENABLE_I18N */ + +/* Functions for binary tree operation. */ + +/* Create a tree node. */ + +static bin_tree_t * +create_tree (re_dfa_t *dfa, bin_tree_t *left, bin_tree_t *right, + re_token_type_t type) +{ + re_token_t t; + t.type = type; + return create_token_tree (dfa, left, right, &t); +} + +static bin_tree_t * +create_token_tree (re_dfa_t *dfa, bin_tree_t *left, bin_tree_t *right, + const re_token_t *token) +{ + bin_tree_t *tree; + if (BE (dfa->str_tree_storage_idx == BIN_TREE_STORAGE_SIZE, 0)) + { + bin_tree_storage_t *storage = re_malloc (bin_tree_storage_t, 1); + + if (storage == NULL) + return NULL; + storage->next = dfa->str_tree_storage; + dfa->str_tree_storage = storage; + dfa->str_tree_storage_idx = 0; + } + tree = &dfa->str_tree_storage->data[dfa->str_tree_storage_idx++]; + + tree->parent = NULL; + tree->left = left; + tree->right = right; + tree->token = *token; + tree->token.duplicated = 0; + tree->token.opt_subexp = 0; + tree->first = NULL; + tree->next = NULL; + tree->node_idx = REG_MISSING; + + if (left != NULL) + left->parent = tree; + if (right != NULL) + right->parent = tree; + return tree; +} + +/* Mark the tree SRC as an optional subexpression. + To be called from preorder or postorder. */ + +static reg_errcode_t +mark_opt_subexp (void *extra, bin_tree_t *node) +{ + Idx idx = (Idx) (long) extra; + if (node->token.type == SUBEXP && node->token.opr.idx == idx) + node->token.opt_subexp = 1; + + return REG_NOERROR; +} + +/* Free the allocated memory inside NODE. */ + +static void +free_token (re_token_t *node) +{ +#ifdef RE_ENABLE_I18N + if (node->type == COMPLEX_BRACKET && node->duplicated == 0) + free_charset (node->opr.mbcset); + else +#endif /* RE_ENABLE_I18N */ + if (node->type == SIMPLE_BRACKET && node->duplicated == 0) + re_free (node->opr.sbcset); +} + +/* Worker function for tree walking. Free the allocated memory inside NODE + and its children. */ + +static reg_errcode_t +free_tree (void *extra, bin_tree_t *node) +{ + free_token (&node->token); + return REG_NOERROR; +} + + +/* Duplicate the node SRC, and return new node. This is a preorder + visit similar to the one implemented by the generic visitor, but + we need more infrastructure to maintain two parallel trees --- so, + it's easier to duplicate. */ + +static bin_tree_t * +duplicate_tree (const bin_tree_t *root, re_dfa_t *dfa) +{ + const bin_tree_t *node; + bin_tree_t *dup_root; + bin_tree_t **p_new = &dup_root, *dup_node = root->parent; + + for (node = root; ; ) + { + /* Create a new tree and link it back to the current parent. */ + *p_new = create_token_tree (dfa, NULL, NULL, &node->token); + if (*p_new == NULL) + return NULL; + (*p_new)->parent = dup_node; + (*p_new)->token.duplicated = 1; + dup_node = *p_new; + + /* Go to the left node, or up and to the right. */ + if (node->left) + { + node = node->left; + p_new = &dup_node->left; + } + else + { + const bin_tree_t *prev = NULL; + while (node->right == prev || node->right == NULL) + { + prev = node; + node = node->parent; + dup_node = dup_node->parent; + if (!node) + return dup_root; + } + node = node->right; + p_new = &dup_node->right; + } + } +} diff --git a/lib/regex.c b/lib/regex.c new file mode 100644 index 000000000..6e0f752b3 --- /dev/null +++ b/lib/regex.c @@ -0,0 +1,72 @@ +/* Extended regular expression matching and search library. + Copyright (C) 2002-2003, 2005-2006, 2009-2012 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2, or (at your option) + any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License along + with this program; if not, write to the Free Software Foundation, + Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ + +#include <config.h> + +/* Make sure noone compiles this code with a C++ compiler. */ +#if defined __cplusplus && defined _LIBC +# error "This is C code, use a C compiler" +#endif + +#ifdef _LIBC +/* We have to keep the namespace clean. */ +# define regfree(preg) __regfree (preg) +# define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef) +# define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags) +# define regerror(errcode, preg, errbuf, errbuf_size) \ + __regerror(errcode, preg, errbuf, errbuf_size) +# define re_set_registers(bu, re, nu, st, en) \ + __re_set_registers (bu, re, nu, st, en) +# define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \ + __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop) +# define re_match(bufp, string, size, pos, regs) \ + __re_match (bufp, string, size, pos, regs) +# define re_search(bufp, string, size, startpos, range, regs) \ + __re_search (bufp, string, size, startpos, range, regs) +# define re_compile_pattern(pattern, length, bufp) \ + __re_compile_pattern (pattern, length, bufp) +# define re_set_syntax(syntax) __re_set_syntax (syntax) +# define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \ + __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop) +# define re_compile_fastmap(bufp) __re_compile_fastmap (bufp) + +# include "../locale/localeinfo.h" +#endif + +/* On some systems, limits.h sets RE_DUP_MAX to a lower value than + GNU regex allows. Include it before <regex.h>, which correctly + #undefs RE_DUP_MAX and sets it to the right value. */ +#include <limits.h> +#include <strings.h> + +#include <regex.h> +#include "regex_internal.h" + +#include "regex_internal.c" +#include "regcomp.c" +#include "regexec.c" + +/* Binary backward compatibility. */ +#if _LIBC +# include <shlib-compat.h> +# if SHLIB_COMPAT (libc, GLIBC_2_0, GLIBC_2_3) +link_warning (re_max_failures, "the 're_max_failures' variable is obsolete and will go away.") +int re_max_failures = 2000; +# endif +#endif diff --git a/lib/regex.h b/lib/regex.h new file mode 100644 index 000000000..b612adbf5 --- /dev/null +++ b/lib/regex.h @@ -0,0 +1,675 @@ +/* Definitions for data structures and routines for the regular + expression library. + Copyright (C) 1985, 1989-1993, 1995-1998, 2000-2003, 2005-2006, 2009-2012 + Free Software Foundation, Inc. + This file is part of the GNU C Library. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2, or (at your option) + any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License along + with this program; if not, write to the Free Software Foundation, + Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ + +#ifndef _REGEX_H +#define _REGEX_H 1 + +#include <sys/types.h> + +/* Allow the use in C++ code. */ +#ifdef __cplusplus +extern "C" { +#endif + +/* Define __USE_GNU_REGEX to declare GNU extensions that violate the + POSIX name space rules. */ +#undef __USE_GNU_REGEX +#if (defined _GNU_SOURCE \ + || (!defined _POSIX_C_SOURCE && !defined _POSIX_SOURCE \ + && !defined _XOPEN_SOURCE)) +# define __USE_GNU_REGEX 1 +#endif + +#ifdef _REGEX_LARGE_OFFSETS + +/* Use types and values that are wide enough to represent signed and + unsigned byte offsets in memory. This currently works only when + the regex code is used outside of the GNU C library; it is not yet + supported within glibc itself, and glibc users should not define + _REGEX_LARGE_OFFSETS. */ + +/* The type of the offset of a byte within a string. + For historical reasons POSIX 1003.1-2004 requires that regoff_t be + at least as wide as off_t. However, many common POSIX platforms set + regoff_t to the more-sensible ssize_t and the Open Group has + signalled its intention to change the requirement to be that + regoff_t be at least as wide as ptrdiff_t and ssize_t; see XBD ERN + 60 (2005-08-25). We don't know of any hosts where ssize_t or + ptrdiff_t is wider than ssize_t, so ssize_t is safe. */ +typedef ssize_t regoff_t; + +/* The type of nonnegative object indexes. Traditionally, GNU regex + uses 'int' for these. Code that uses __re_idx_t should work + regardless of whether the type is signed. */ +typedef size_t __re_idx_t; + +/* The type of object sizes. */ +typedef size_t __re_size_t; + +/* The type of object sizes, in places where the traditional code + uses unsigned long int. */ +typedef size_t __re_long_size_t; + +#else + +/* Use types that are binary-compatible with the traditional GNU regex + implementation, which mishandles strings longer than INT_MAX. */ + +typedef int regoff_t; +typedef int __re_idx_t; +typedef unsigned int __re_size_t; +typedef unsigned long int __re_long_size_t; + +#endif + +/* The following two types have to be signed and unsigned integer type + wide enough to hold a value of a pointer. For most ANSI compilers + ptrdiff_t and size_t should be likely OK. Still size of these two + types is 2 for Microsoft C. Ugh... */ +typedef long int s_reg_t; +typedef unsigned long int active_reg_t; + +/* The following bits are used to determine the regexp syntax we + recognize. The set/not-set meanings are chosen so that Emacs syntax + remains the value 0. The bits are given in alphabetical order, and + the definitions shifted by one from the previous bit; thus, when we + add or remove a bit, only one other definition need change. */ +typedef unsigned long int reg_syntax_t; + +#ifdef __USE_GNU_REGEX + +/* If this bit is not set, then \ inside a bracket expression is literal. + If set, then such a \ quotes the following character. */ +# define RE_BACKSLASH_ESCAPE_IN_LISTS ((unsigned long int) 1) + +/* If this bit is not set, then + and ? are operators, and \+ and \? are + literals. + If set, then \+ and \? are operators and + and ? are literals. */ +# define RE_BK_PLUS_QM (RE_BACKSLASH_ESCAPE_IN_LISTS << 1) + +/* If this bit is set, then character classes are supported. They are: + [:alpha:], [:upper:], [:lower:], [:digit:], [:alnum:], [:xdigit:], + [:space:], [:print:], [:punct:], [:graph:], and [:cntrl:]. + If not set, then character classes are not supported. */ +# define RE_CHAR_CLASSES (RE_BK_PLUS_QM << 1) + +/* If this bit is set, then ^ and $ are always anchors (outside bracket + expressions, of course). + If this bit is not set, then it depends: + ^ is an anchor if it is at the beginning of a regular + expression or after an open-group or an alternation operator; + $ is an anchor if it is at the end of a regular expression, or + before a close-group or an alternation operator. + + This bit could be (re)combined with RE_CONTEXT_INDEP_OPS, because + POSIX draft 11.2 says that * etc. in leading positions is undefined. + We already implemented a previous draft which made those constructs + invalid, though, so we haven't changed the code back. */ +# define RE_CONTEXT_INDEP_ANCHORS (RE_CHAR_CLASSES << 1) + +/* If this bit is set, then special characters are always special + regardless of where they are in the pattern. + If this bit is not set, then special characters are special only in + some contexts; otherwise they are ordinary. Specifically, + * + ? and intervals are only special when not after the beginning, + open-group, or alternation operator. */ +# define RE_CONTEXT_INDEP_OPS (RE_CONTEXT_INDEP_ANCHORS << 1) + +/* If this bit is set, then *, +, ?, and { cannot be first in an re or + immediately after an alternation or begin-group operator. */ +# define RE_CONTEXT_INVALID_OPS (RE_CONTEXT_INDEP_OPS << 1) + +/* If this bit is set, then . matches newline. + If not set, then it doesn't. */ +# define RE_DOT_NEWLINE (RE_CONTEXT_INVALID_OPS << 1) + +/* If this bit is set, then . doesn't match NUL. + If not set, then it does. */ +# define RE_DOT_NOT_NULL (RE_DOT_NEWLINE << 1) + +/* If this bit is set, nonmatching lists [^...] do not match newline. + If not set, they do. */ +# define RE_HAT_LISTS_NOT_NEWLINE (RE_DOT_NOT_NULL << 1) + +/* If this bit is set, either \{...\} or {...} defines an + interval, depending on RE_NO_BK_BRACES. + If not set, \{, \}, {, and } are literals. */ +# define RE_INTERVALS (RE_HAT_LISTS_NOT_NEWLINE << 1) + +/* If this bit is set, +, ? and | aren't recognized as operators. + If not set, they are. */ +# define RE_LIMITED_OPS (RE_INTERVALS << 1) + +/* If this bit is set, newline is an alternation operator. + If not set, newline is literal. */ +# define RE_NEWLINE_ALT (RE_LIMITED_OPS << 1) + +/* If this bit is set, then '{...}' defines an interval, and \{ and \} + are literals. + If not set, then '\{...\}' defines an interval. */ +# define RE_NO_BK_BRACES (RE_NEWLINE_ALT << 1) + +/* If this bit is set, (...) defines a group, and \( and \) are literals. + If not set, \(...\) defines a group, and ( and ) are literals. */ +# define RE_NO_BK_PARENS (RE_NO_BK_BRACES << 1) + +/* If this bit is set, then \<digit> matches <digit>. + If not set, then \<digit> is a back-reference. */ +# define RE_NO_BK_REFS (RE_NO_BK_PARENS << 1) + +/* If this bit is set, then | is an alternation operator, and \| is literal. + If not set, then \| is an alternation operator, and | is literal. */ +# define RE_NO_BK_VBAR (RE_NO_BK_REFS << 1) + +/* If this bit is set, then an ending range point collating higher + than the starting range point, as in [z-a], is invalid. + If not set, then when ending range point collates higher than the + starting range point, the range is ignored. */ +# define RE_NO_EMPTY_RANGES (RE_NO_BK_VBAR << 1) + +/* If this bit is set, then an unmatched ) is ordinary. + If not set, then an unmatched ) is invalid. */ +# define RE_UNMATCHED_RIGHT_PAREN_ORD (RE_NO_EMPTY_RANGES << 1) + +/* If this bit is set, succeed as soon as we match the whole pattern, + without further backtracking. */ +# define RE_NO_POSIX_BACKTRACKING (RE_UNMATCHED_RIGHT_PAREN_ORD << 1) + +/* If this bit is set, do not process the GNU regex operators. + If not set, then the GNU regex operators are recognized. */ +# define RE_NO_GNU_OPS (RE_NO_POSIX_BACKTRACKING << 1) + +/* If this bit is set, turn on internal regex debugging. + If not set, and debugging was on, turn it off. + This only works if regex.c is compiled -DDEBUG. + We define this bit always, so that all that's needed to turn on + debugging is to recompile regex.c; the calling code can always have + this bit set, and it won't affect anything in the normal case. */ +# define RE_DEBUG (RE_NO_GNU_OPS << 1) + +/* If this bit is set, a syntactically invalid interval is treated as + a string of ordinary characters. For example, the ERE 'a{1' is + treated as 'a\{1'. */ +# define RE_INVALID_INTERVAL_ORD (RE_DEBUG << 1) + +/* If this bit is set, then ignore case when matching. + If not set, then case is significant. */ +# define RE_ICASE (RE_INVALID_INTERVAL_ORD << 1) + +/* This bit is used internally like RE_CONTEXT_INDEP_ANCHORS but only + for ^, because it is difficult to scan the regex backwards to find + whether ^ should be special. */ +# define RE_CARET_ANCHORS_HERE (RE_ICASE << 1) + +/* If this bit is set, then \{ cannot be first in a regex or + immediately after an alternation, open-group or \} operator. */ +# define RE_CONTEXT_INVALID_DUP (RE_CARET_ANCHORS_HERE << 1) + +/* If this bit is set, then no_sub will be set to 1 during + re_compile_pattern. */ +# define RE_NO_SUB (RE_CONTEXT_INVALID_DUP << 1) + +#endif /* defined __USE_GNU_REGEX */ + +/* This global variable defines the particular regexp syntax to use (for + some interfaces). When a regexp is compiled, the syntax used is + stored in the pattern buffer, so changing this does not affect + already-compiled regexps. */ +extern reg_syntax_t re_syntax_options; + +#ifdef __USE_GNU_REGEX +/* Define combinations of the above bits for the standard possibilities. + (The [[[ comments delimit what gets put into the Texinfo file, so + don't delete them!) */ +/* [[[begin syntaxes]]] */ +# define RE_SYNTAX_EMACS 0 + +# define RE_SYNTAX_AWK \ + (RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DOT_NOT_NULL \ + | RE_NO_BK_PARENS | RE_NO_BK_REFS \ + | RE_NO_BK_VBAR | RE_NO_EMPTY_RANGES \ + | RE_DOT_NEWLINE | RE_CONTEXT_INDEP_ANCHORS \ + | RE_UNMATCHED_RIGHT_PAREN_ORD | RE_NO_GNU_OPS) + +# define RE_SYNTAX_GNU_AWK \ + ((RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS | RE_DEBUG) \ + & ~(RE_DOT_NOT_NULL | RE_INTERVALS | RE_CONTEXT_INDEP_OPS \ + | RE_CONTEXT_INVALID_OPS )) + +# define RE_SYNTAX_POSIX_AWK \ + (RE_SYNTAX_POSIX_EXTENDED | RE_BACKSLASH_ESCAPE_IN_LISTS \ + | RE_INTERVALS | RE_NO_GNU_OPS) + +# define RE_SYNTAX_GREP \ + (RE_BK_PLUS_QM | RE_CHAR_CLASSES \ + | RE_HAT_LISTS_NOT_NEWLINE | RE_INTERVALS \ + | RE_NEWLINE_ALT) + +# define RE_SYNTAX_EGREP \ + (RE_CHAR_CLASSES | RE_CONTEXT_INDEP_ANCHORS \ + | RE_CONTEXT_INDEP_OPS | RE_HAT_LISTS_NOT_NEWLINE \ + | RE_NEWLINE_ALT | RE_NO_BK_PARENS \ + | RE_NO_BK_VBAR) + +# define RE_SYNTAX_POSIX_EGREP \ + (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES \ + | RE_INVALID_INTERVAL_ORD) + +/* P1003.2/D11.2, section 4.20.7.1, lines 5078ff. */ +# define RE_SYNTAX_ED RE_SYNTAX_POSIX_BASIC + +# define RE_SYNTAX_SED RE_SYNTAX_POSIX_BASIC + +/* Syntax bits common to both basic and extended POSIX regex syntax. */ +# define _RE_SYNTAX_POSIX_COMMON \ + (RE_CHAR_CLASSES | RE_DOT_NEWLINE | RE_DOT_NOT_NULL \ + | RE_INTERVALS | RE_NO_EMPTY_RANGES) + +# define RE_SYNTAX_POSIX_BASIC \ + (_RE_SYNTAX_POSIX_COMMON | RE_BK_PLUS_QM | RE_CONTEXT_INVALID_DUP) + +/* Differs from ..._POSIX_BASIC only in that RE_BK_PLUS_QM becomes + RE_LIMITED_OPS, i.e., \? \+ \| are not recognized. Actually, this + isn't minimal, since other operators, such as \`, aren't disabled. */ +# define RE_SYNTAX_POSIX_MINIMAL_BASIC \ + (_RE_SYNTAX_POSIX_COMMON | RE_LIMITED_OPS) + +# define RE_SYNTAX_POSIX_EXTENDED \ + (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \ + | RE_CONTEXT_INDEP_OPS | RE_NO_BK_BRACES \ + | RE_NO_BK_PARENS | RE_NO_BK_VBAR \ + | RE_CONTEXT_INVALID_OPS | RE_UNMATCHED_RIGHT_PAREN_ORD) + +/* Differs from ..._POSIX_EXTENDED in that RE_CONTEXT_INDEP_OPS is + removed and RE_NO_BK_REFS is added. */ +# define RE_SYNTAX_POSIX_MINIMAL_EXTENDED \ + (_RE_SYNTAX_POSIX_COMMON | RE_CONTEXT_INDEP_ANCHORS \ + | RE_CONTEXT_INVALID_OPS | RE_NO_BK_BRACES \ + | RE_NO_BK_PARENS | RE_NO_BK_REFS \ + | RE_NO_BK_VBAR | RE_UNMATCHED_RIGHT_PAREN_ORD) +/* [[[end syntaxes]]] */ + +#endif /* defined __USE_GNU_REGEX */ + +#ifdef __USE_GNU_REGEX + +/* Maximum number of duplicates an interval can allow. POSIX-conforming + systems might define this in <limits.h>, but we want our + value, so remove any previous define. */ +# ifdef RE_DUP_MAX +# undef RE_DUP_MAX +# endif + +/* RE_DUP_MAX is 2**15 - 1 because an earlier implementation stored + the counter as a 2-byte signed integer. This is no longer true, so + RE_DUP_MAX could be increased to (INT_MAX / 10 - 1), or to + ((SIZE_MAX - 2) / 10 - 1) if _REGEX_LARGE_OFFSETS is defined. + However, there would be a huge performance problem if someone + actually used a pattern like a\{214748363\}, so RE_DUP_MAX retains + its historical value. */ +# define RE_DUP_MAX (0x7fff) + +#endif /* defined __USE_GNU_REGEX */ + + +/* POSIX 'cflags' bits (i.e., information for 'regcomp'). */ + +/* If this bit is set, then use extended regular expression syntax. + If not set, then use basic regular expression syntax. */ +#define REG_EXTENDED 1 + +/* If this bit is set, then ignore case when matching. + If not set, then case is significant. */ +#define REG_ICASE (1 << 1) + +/* If this bit is set, then anchors do not match at newline + characters in the string. + If not set, then anchors do match at newlines. */ +#define REG_NEWLINE (1 << 2) + +/* If this bit is set, then report only success or fail in regexec. + If not set, then returns differ between not matching and errors. */ +#define REG_NOSUB (1 << 3) + + +/* POSIX 'eflags' bits (i.e., information for regexec). */ + +/* If this bit is set, then the beginning-of-line operator doesn't match + the beginning of the string (presumably because it's not the + beginning of a line). + If not set, then the beginning-of-line operator does match the + beginning of the string. */ +#define REG_NOTBOL 1 + +/* Like REG_NOTBOL, except for the end-of-line. */ +#define REG_NOTEOL (1 << 1) + +/* Use PMATCH[0] to delimit the start and end of the search in the + buffer. */ +#define REG_STARTEND (1 << 2) + + +/* If any error codes are removed, changed, or added, update the + '__re_error_msgid' table in regcomp.c. */ + +typedef enum +{ + _REG_ENOSYS = -1, /* This will never happen for this implementation. */ + _REG_NOERROR = 0, /* Success. */ + _REG_NOMATCH, /* Didn't find a match (for regexec). */ + + /* POSIX regcomp return error codes. (In the order listed in the + standard.) */ + _REG_BADPAT, /* Invalid pattern. */ + _REG_ECOLLATE, /* Invalid collating element. */ + _REG_ECTYPE, /* Invalid character class name. */ + _REG_EESCAPE, /* Trailing backslash. */ + _REG_ESUBREG, /* Invalid back reference. */ + _REG_EBRACK, /* Unmatched left bracket. */ + _REG_EPAREN, /* Parenthesis imbalance. */ + _REG_EBRACE, /* Unmatched \{. */ + _REG_BADBR, /* Invalid contents of \{\}. */ + _REG_ERANGE, /* Invalid range end. */ + _REG_ESPACE, /* Ran out of memory. */ + _REG_BADRPT, /* No preceding re for repetition op. */ + + /* Error codes we've added. */ + _REG_EEND, /* Premature end. */ + _REG_ESIZE, /* Compiled pattern bigger than 2^16 bytes. */ + _REG_ERPAREN /* Unmatched ) or \); not returned from regcomp. */ +} reg_errcode_t; + +#ifdef _XOPEN_SOURCE +# define REG_ENOSYS _REG_ENOSYS +#endif +#define REG_NOERROR _REG_NOERROR +#define REG_NOMATCH _REG_NOMATCH +#define REG_BADPAT _REG_BADPAT +#define REG_ECOLLATE _REG_ECOLLATE +#define REG_ECTYPE _REG_ECTYPE +#define REG_EESCAPE _REG_EESCAPE +#define REG_ESUBREG _REG_ESUBREG +#define REG_EBRACK _REG_EBRACK +#define REG_EPAREN _REG_EPAREN +#define REG_EBRACE _REG_EBRACE +#define REG_BADBR _REG_BADBR +#define REG_ERANGE _REG_ERANGE +#define REG_ESPACE _REG_ESPACE +#define REG_BADRPT _REG_BADRPT +#define REG_EEND _REG_EEND +#define REG_ESIZE _REG_ESIZE +#define REG_ERPAREN _REG_ERPAREN + +/* struct re_pattern_buffer normally uses member names like 'buffer' + that POSIX does not allow. In POSIX mode these members have names + with leading 're_' (e.g., 're_buffer'). */ +#ifdef __USE_GNU_REGEX +# define _REG_RE_NAME(id) id +# define _REG_RM_NAME(id) id +#else +# define _REG_RE_NAME(id) re_##id +# define _REG_RM_NAME(id) rm_##id +#endif + +/* The user can specify the type of the re_translate member by + defining the macro RE_TRANSLATE_TYPE, which defaults to unsigned + char *. This pollutes the POSIX name space, so in POSIX mode just + use unsigned char *. */ +#ifdef __USE_GNU_REGEX +# ifndef RE_TRANSLATE_TYPE +# define RE_TRANSLATE_TYPE unsigned char * +# endif +# define REG_TRANSLATE_TYPE RE_TRANSLATE_TYPE +#else +# define REG_TRANSLATE_TYPE unsigned char * +#endif + +/* This data structure represents a compiled pattern. Before calling + the pattern compiler, the fields 'buffer', 'allocated', 'fastmap', + 'translate', and 'no_sub' can be set. After the pattern has been + compiled, the 're_nsub' field is available. All other fields are + private to the regex routines. */ + +struct re_pattern_buffer +{ + /* Space that holds the compiled pattern. It is declared as + 'unsigned char *' because its elements are sometimes used as + array indexes. */ + unsigned char *_REG_RE_NAME (buffer); + + /* Number of bytes to which 'buffer' points. */ + __re_long_size_t _REG_RE_NAME (allocated); + + /* Number of bytes actually used in 'buffer'. */ + __re_long_size_t _REG_RE_NAME (used); + + /* Syntax setting with which the pattern was compiled. */ + reg_syntax_t _REG_RE_NAME (syntax); + + /* Pointer to a fastmap, if any, otherwise zero. re_search uses the + fastmap, if there is one, to skip over impossible starting points + for matches. */ + char *_REG_RE_NAME (fastmap); + + /* Either a translate table to apply to all characters before + comparing them, or zero for no translation. The translation is + applied to a pattern when it is compiled and to a string when it + is matched. */ + REG_TRANSLATE_TYPE _REG_RE_NAME (translate); + + /* Number of subexpressions found by the compiler. */ + size_t re_nsub; + + /* Zero if this pattern cannot match the empty string, one else. + Well, in truth it's used only in 're_search_2', to see whether or + not we should use the fastmap, so we don't set this absolutely + perfectly; see 're_compile_fastmap' (the "duplicate" case). */ + unsigned int _REG_RE_NAME (can_be_null) : 1; + + /* If REGS_UNALLOCATED, allocate space in the 'regs' structure + for 'max (RE_NREGS, re_nsub + 1)' groups. + If REGS_REALLOCATE, reallocate space if necessary. + If REGS_FIXED, use what's there. */ +#ifdef __USE_GNU_REGEX +# define REGS_UNALLOCATED 0 +# define REGS_REALLOCATE 1 +# define REGS_FIXED 2 +#endif + unsigned int _REG_RE_NAME (regs_allocated) : 2; + + /* Set to zero when 're_compile_pattern' compiles a pattern; set to + one by 're_compile_fastmap' if it updates the fastmap. */ + unsigned int _REG_RE_NAME (fastmap_accurate) : 1; + + /* If set, 're_match_2' does not return information about + subexpressions. */ + unsigned int _REG_RE_NAME (no_sub) : 1; + + /* If set, a beginning-of-line anchor doesn't match at the beginning + of the string. */ + unsigned int _REG_RE_NAME (not_bol) : 1; + + /* Similarly for an end-of-line anchor. */ + unsigned int _REG_RE_NAME (not_eol) : 1; + + /* If true, an anchor at a newline matches. */ + unsigned int _REG_RE_NAME (newline_anchor) : 1; + +/* [[[end pattern_buffer]]] */ +}; + +typedef struct re_pattern_buffer regex_t; + +/* This is the structure we store register match data in. See + regex.texinfo for a full description of what registers match. */ +struct re_registers +{ + __re_size_t _REG_RM_NAME (num_regs); + regoff_t *_REG_RM_NAME (start); + regoff_t *_REG_RM_NAME (end); +}; + + +/* If 'regs_allocated' is REGS_UNALLOCATED in the pattern buffer, + 're_match_2' returns information about at least this many registers + the first time a 'regs' structure is passed. */ +#if !defined RE_NREGS && defined __USE_GNU_REGEX +# define RE_NREGS 30 +#endif + + +/* POSIX specification for registers. Aside from the different names than + 're_registers', POSIX uses an array of structures, instead of a + structure of arrays. */ +typedef struct +{ + regoff_t rm_so; /* Byte offset from string's start to substring's start. */ + regoff_t rm_eo; /* Byte offset from string's start to substring's end. */ +} regmatch_t; + +/* Declarations for routines. */ + +/* Sets the current default syntax to SYNTAX, and return the old syntax. + You can also simply assign to the 're_syntax_options' variable. */ +extern reg_syntax_t re_set_syntax (reg_syntax_t __syntax); + +/* Compile the regular expression PATTERN, with length LENGTH + and syntax given by the global 're_syntax_options', into the buffer + BUFFER. Return NULL if successful, and an error string if not. */ +extern const char *re_compile_pattern (const char *__pattern, size_t __length, + struct re_pattern_buffer *__buffer); + + +/* Compile a fastmap for the compiled pattern in BUFFER; used to + accelerate searches. Return 0 if successful and -2 if was an + internal error. */ +extern int re_compile_fastmap (struct re_pattern_buffer *__buffer); + + +/* Search in the string STRING (with length LENGTH) for the pattern + compiled into BUFFER. Start searching at position START, for RANGE + characters. Return the starting position of the match, -1 for no + match, or -2 for an internal error. Also return register + information in REGS (if REGS and BUFFER->no_sub are nonzero). */ +extern regoff_t re_search (struct re_pattern_buffer *__buffer, + const char *__string, __re_idx_t __length, + __re_idx_t __start, regoff_t __range, + struct re_registers *__regs); + + +/* Like 're_search', but search in the concatenation of STRING1 and + STRING2. Also, stop searching at index START + STOP. */ +extern regoff_t re_search_2 (struct re_pattern_buffer *__buffer, + const char *__string1, __re_idx_t __length1, + const char *__string2, __re_idx_t __length2, + __re_idx_t __start, regoff_t __range, + struct re_registers *__regs, + __re_idx_t __stop); + + +/* Like 're_search', but return how many characters in STRING the regexp + in BUFFER matched, starting at position START. */ +extern regoff_t re_match (struct re_pattern_buffer *__buffer, + const char *__string, __re_idx_t __length, + __re_idx_t __start, struct re_registers *__regs); + + +/* Relates to 're_match' as 're_search_2' relates to 're_search'. */ +extern regoff_t re_match_2 (struct re_pattern_buffer *__buffer, + const char *__string1, __re_idx_t __length1, + const char *__string2, __re_idx_t __length2, + __re_idx_t __start, struct re_registers *__regs, + __re_idx_t __stop); + + +/* Set REGS to hold NUM_REGS registers, storing them in STARTS and + ENDS. Subsequent matches using BUFFER and REGS will use this memory + for recording register information. STARTS and ENDS must be + allocated with malloc, and must each be at least 'NUM_REGS * sizeof + (regoff_t)' bytes long. + + If NUM_REGS == 0, then subsequent matches should allocate their own + register data. + + Unless this function is called, the first search or match using + BUFFER will allocate its own register data, without freeing the old + data. */ +extern void re_set_registers (struct re_pattern_buffer *__buffer, + struct re_registers *__regs, + __re_size_t __num_regs, + regoff_t *__starts, regoff_t *__ends); + +#if defined _REGEX_RE_COMP || defined _LIBC +# ifndef _CRAY +/* 4.2 bsd compatibility. */ +extern char *re_comp (const char *); +extern int re_exec (const char *); +# endif +#endif + +/* GCC 2.95 and later have "__restrict"; C99 compilers have + "restrict", and "configure" may have defined "restrict". + Other compilers use __restrict, __restrict__, and _Restrict, and + 'configure' might #define 'restrict' to those words, so pick a + different name. */ +#ifndef _Restrict_ +# if 199901L <= __STDC_VERSION__ +# define _Restrict_ restrict +# elif 2 < __GNUC__ || (2 == __GNUC__ && 95 <= __GNUC_MINOR__) +# define _Restrict_ __restrict +# else +# define _Restrict_ +# endif +#endif +/* gcc 3.1 and up support the [restrict] syntax. Don't trust + sys/cdefs.h's definition of __restrict_arr, though, as it + mishandles gcc -ansi -pedantic. */ +#ifndef _Restrict_arr_ +# if ((199901L <= __STDC_VERSION__ \ + || ((3 < __GNUC__ || (3 == __GNUC__ && 1 <= __GNUC_MINOR__)) \ + && !defined __STRICT_ANSI__)) \ + && !defined __GNUG__) +# define _Restrict_arr_ _Restrict_ +# else +# define _Restrict_arr_ +# endif +#endif + +/* POSIX compatibility. */ +extern int regcomp (regex_t *_Restrict_ __preg, + const char *_Restrict_ __pattern, + int __cflags); + +extern int regexec (const regex_t *_Restrict_ __preg, + const char *_Restrict_ __string, size_t __nmatch, + regmatch_t __pmatch[_Restrict_arr_], + int __eflags); + +extern size_t regerror (int __errcode, const regex_t *_Restrict_ __preg, + char *_Restrict_ __errbuf, size_t __errbuf_size); + +extern void regfree (regex_t *__preg); + + +#ifdef __cplusplus +} +#endif /* C++ */ + +#endif /* regex.h */ diff --git a/lib/regex_internal.c b/lib/regex_internal.c new file mode 100644 index 000000000..c029c23a9 --- /dev/null +++ b/lib/regex_internal.c @@ -0,0 +1,1741 @@ +/* Extended regular expression matching and search library. + Copyright (C) 2002-2012 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2, or (at your option) + any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License along + with this program; if not, write to the Free Software Foundation, + Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ + +static void re_string_construct_common (const char *str, Idx len, + re_string_t *pstr, + RE_TRANSLATE_TYPE trans, bool icase, + const re_dfa_t *dfa) internal_function; +static re_dfastate_t *create_ci_newstate (const re_dfa_t *dfa, + const re_node_set *nodes, + re_hashval_t hash) internal_function; +static re_dfastate_t *create_cd_newstate (const re_dfa_t *dfa, + const re_node_set *nodes, + unsigned int context, + re_hashval_t hash) internal_function; + +/* Functions for string operation. */ + +/* This function allocate the buffers. It is necessary to call + re_string_reconstruct before using the object. */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +re_string_allocate (re_string_t *pstr, const char *str, Idx len, Idx init_len, + RE_TRANSLATE_TYPE trans, bool icase, const re_dfa_t *dfa) +{ + reg_errcode_t ret; + Idx init_buf_len; + + /* Ensure at least one character fits into the buffers. */ + if (init_len < dfa->mb_cur_max) + init_len = dfa->mb_cur_max; + init_buf_len = (len + 1 < init_len) ? len + 1: init_len; + re_string_construct_common (str, len, pstr, trans, icase, dfa); + + ret = re_string_realloc_buffers (pstr, init_buf_len); + if (BE (ret != REG_NOERROR, 0)) + return ret; + + pstr->word_char = dfa->word_char; + pstr->word_ops_used = dfa->word_ops_used; + pstr->mbs = pstr->mbs_allocated ? pstr->mbs : (unsigned char *) str; + pstr->valid_len = (pstr->mbs_allocated || dfa->mb_cur_max > 1) ? 0 : len; + pstr->valid_raw_len = pstr->valid_len; + return REG_NOERROR; +} + +/* This function allocate the buffers, and initialize them. */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +re_string_construct (re_string_t *pstr, const char *str, Idx len, + RE_TRANSLATE_TYPE trans, bool icase, const re_dfa_t *dfa) +{ + reg_errcode_t ret; + memset (pstr, '\0', sizeof (re_string_t)); + re_string_construct_common (str, len, pstr, trans, icase, dfa); + + if (len > 0) + { + ret = re_string_realloc_buffers (pstr, len + 1); + if (BE (ret != REG_NOERROR, 0)) + return ret; + } + pstr->mbs = pstr->mbs_allocated ? pstr->mbs : (unsigned char *) str; + + if (icase) + { +#ifdef RE_ENABLE_I18N + if (dfa->mb_cur_max > 1) + { + while (1) + { + ret = build_wcs_upper_buffer (pstr); + if (BE (ret != REG_NOERROR, 0)) + return ret; + if (pstr->valid_raw_len >= len) + break; + if (pstr->bufs_len > pstr->valid_len + dfa->mb_cur_max) + break; + ret = re_string_realloc_buffers (pstr, pstr->bufs_len * 2); + if (BE (ret != REG_NOERROR, 0)) + return ret; + } + } + else +#endif /* RE_ENABLE_I18N */ + build_upper_buffer (pstr); + } + else + { +#ifdef RE_ENABLE_I18N + if (dfa->mb_cur_max > 1) + build_wcs_buffer (pstr); + else +#endif /* RE_ENABLE_I18N */ + { + if (trans != NULL) + re_string_translate_buffer (pstr); + else + { + pstr->valid_len = pstr->bufs_len; + pstr->valid_raw_len = pstr->bufs_len; + } + } + } + + return REG_NOERROR; +} + +/* Helper functions for re_string_allocate, and re_string_construct. */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +re_string_realloc_buffers (re_string_t *pstr, Idx new_buf_len) +{ +#ifdef RE_ENABLE_I18N + if (pstr->mb_cur_max > 1) + { + wint_t *new_wcs; + + /* Avoid overflow. */ + size_t max_object_size = MAX (sizeof (wint_t), sizeof (Idx)); + if (BE (SIZE_MAX / max_object_size < new_buf_len, 0)) + return REG_ESPACE; + + new_wcs = re_realloc (pstr->wcs, wint_t, new_buf_len); + if (BE (new_wcs == NULL, 0)) + return REG_ESPACE; + pstr->wcs = new_wcs; + if (pstr->offsets != NULL) + { + Idx *new_offsets = re_realloc (pstr->offsets, Idx, new_buf_len); + if (BE (new_offsets == NULL, 0)) + return REG_ESPACE; + pstr->offsets = new_offsets; + } + } +#endif /* RE_ENABLE_I18N */ + if (pstr->mbs_allocated) + { + unsigned char *new_mbs = re_realloc (pstr->mbs, unsigned char, + new_buf_len); + if (BE (new_mbs == NULL, 0)) + return REG_ESPACE; + pstr->mbs = new_mbs; + } + pstr->bufs_len = new_buf_len; + return REG_NOERROR; +} + + +static void +internal_function +re_string_construct_common (const char *str, Idx len, re_string_t *pstr, + RE_TRANSLATE_TYPE trans, bool icase, + const re_dfa_t *dfa) +{ + pstr->raw_mbs = (const unsigned char *) str; + pstr->len = len; + pstr->raw_len = len; + pstr->trans = trans; + pstr->icase = icase; + pstr->mbs_allocated = (trans != NULL || icase); + pstr->mb_cur_max = dfa->mb_cur_max; + pstr->is_utf8 = dfa->is_utf8; + pstr->map_notascii = dfa->map_notascii; + pstr->stop = pstr->len; + pstr->raw_stop = pstr->stop; +} + +#ifdef RE_ENABLE_I18N + +/* Build wide character buffer PSTR->WCS. + If the byte sequence of the string are: + <mb1>(0), <mb1>(1), <mb2>(0), <mb2>(1), <sb3> + Then wide character buffer will be: + <wc1> , WEOF , <wc2> , WEOF , <wc3> + We use WEOF for padding, they indicate that the position isn't + a first byte of a multibyte character. + + Note that this function assumes PSTR->VALID_LEN elements are already + built and starts from PSTR->VALID_LEN. */ + +static void +internal_function +build_wcs_buffer (re_string_t *pstr) +{ +#ifdef _LIBC + unsigned char buf[MB_LEN_MAX]; + assert (MB_LEN_MAX >= pstr->mb_cur_max); +#else + unsigned char buf[64]; +#endif + mbstate_t prev_st; + Idx byte_idx, end_idx, remain_len; + size_t mbclen; + + /* Build the buffers from pstr->valid_len to either pstr->len or + pstr->bufs_len. */ + end_idx = (pstr->bufs_len > pstr->len) ? pstr->len : pstr->bufs_len; + for (byte_idx = pstr->valid_len; byte_idx < end_idx;) + { + wchar_t wc; + const char *p; + + remain_len = end_idx - byte_idx; + prev_st = pstr->cur_state; + /* Apply the translation if we need. */ + if (BE (pstr->trans != NULL, 0)) + { + int i, ch; + + for (i = 0; i < pstr->mb_cur_max && i < remain_len; ++i) + { + ch = pstr->raw_mbs [pstr->raw_mbs_idx + byte_idx + i]; + buf[i] = pstr->mbs[byte_idx + i] = pstr->trans[ch]; + } + p = (const char *) buf; + } + else + p = (const char *) pstr->raw_mbs + pstr->raw_mbs_idx + byte_idx; + mbclen = __mbrtowc (&wc, p, remain_len, &pstr->cur_state); + if (BE (mbclen == (size_t) -2, 0)) + { + /* The buffer doesn't have enough space, finish to build. */ + pstr->cur_state = prev_st; + break; + } + else if (BE (mbclen == (size_t) -1 || mbclen == 0, 0)) + { + /* We treat these cases as a singlebyte character. */ + mbclen = 1; + wc = (wchar_t) pstr->raw_mbs[pstr->raw_mbs_idx + byte_idx]; + if (BE (pstr->trans != NULL, 0)) + wc = pstr->trans[wc]; + pstr->cur_state = prev_st; + } + + /* Write wide character and padding. */ + pstr->wcs[byte_idx++] = wc; + /* Write paddings. */ + for (remain_len = byte_idx + mbclen - 1; byte_idx < remain_len ;) + pstr->wcs[byte_idx++] = WEOF; + } + pstr->valid_len = byte_idx; + pstr->valid_raw_len = byte_idx; +} + +/* Build wide character buffer PSTR->WCS like build_wcs_buffer, + but for REG_ICASE. */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +build_wcs_upper_buffer (re_string_t *pstr) +{ + mbstate_t prev_st; + Idx src_idx, byte_idx, end_idx, remain_len; + size_t mbclen; +#ifdef _LIBC + char buf[MB_LEN_MAX]; + assert (MB_LEN_MAX >= pstr->mb_cur_max); +#else + char buf[64]; +#endif + + byte_idx = pstr->valid_len; + end_idx = (pstr->bufs_len > pstr->len) ? pstr->len : pstr->bufs_len; + + /* The following optimization assumes that ASCII characters can be + mapped to wide characters with a simple cast. */ + if (! pstr->map_notascii && pstr->trans == NULL && !pstr->offsets_needed) + { + while (byte_idx < end_idx) + { + wchar_t wc; + + if (isascii (pstr->raw_mbs[pstr->raw_mbs_idx + byte_idx]) + && mbsinit (&pstr->cur_state)) + { + /* In case of a singlebyte character. */ + pstr->mbs[byte_idx] + = toupper (pstr->raw_mbs[pstr->raw_mbs_idx + byte_idx]); + /* The next step uses the assumption that wchar_t is encoded + ASCII-safe: all ASCII values can be converted like this. */ + pstr->wcs[byte_idx] = (wchar_t) pstr->mbs[byte_idx]; + ++byte_idx; + continue; + } + + remain_len = end_idx - byte_idx; + prev_st = pstr->cur_state; + mbclen = __mbrtowc (&wc, + ((const char *) pstr->raw_mbs + pstr->raw_mbs_idx + + byte_idx), remain_len, &pstr->cur_state); + if (BE (mbclen < (size_t) -2, 1)) + { + wchar_t wcu = wc; + if (iswlower (wc)) + { + size_t mbcdlen; + + wcu = towupper (wc); + mbcdlen = wcrtomb (buf, wcu, &prev_st); + if (BE (mbclen == mbcdlen, 1)) + memcpy (pstr->mbs + byte_idx, buf, mbclen); + else + { + src_idx = byte_idx; + goto offsets_needed; + } + } + else + memcpy (pstr->mbs + byte_idx, + pstr->raw_mbs + pstr->raw_mbs_idx + byte_idx, mbclen); + pstr->wcs[byte_idx++] = wcu; + /* Write paddings. */ + for (remain_len = byte_idx + mbclen - 1; byte_idx < remain_len ;) + pstr->wcs[byte_idx++] = WEOF; + } + else if (mbclen == (size_t) -1 || mbclen == 0) + { + /* It is an invalid character or '\0'. Just use the byte. */ + int ch = pstr->raw_mbs[pstr->raw_mbs_idx + byte_idx]; + pstr->mbs[byte_idx] = ch; + /* And also cast it to wide char. */ + pstr->wcs[byte_idx++] = (wchar_t) ch; + if (BE (mbclen == (size_t) -1, 0)) + pstr->cur_state = prev_st; + } + else + { + /* The buffer doesn't have enough space, finish to build. */ + pstr->cur_state = prev_st; + break; + } + } + pstr->valid_len = byte_idx; + pstr->valid_raw_len = byte_idx; + return REG_NOERROR; + } + else + for (src_idx = pstr->valid_raw_len; byte_idx < end_idx;) + { + wchar_t wc; + const char *p; + offsets_needed: + remain_len = end_idx - byte_idx; + prev_st = pstr->cur_state; + if (BE (pstr->trans != NULL, 0)) + { + int i, ch; + + for (i = 0; i < pstr->mb_cur_max && i < remain_len; ++i) + { + ch = pstr->raw_mbs [pstr->raw_mbs_idx + src_idx + i]; + buf[i] = pstr->trans[ch]; + } + p = (const char *) buf; + } + else + p = (const char *) pstr->raw_mbs + pstr->raw_mbs_idx + src_idx; + mbclen = __mbrtowc (&wc, p, remain_len, &pstr->cur_state); + if (BE (mbclen < (size_t) -2, 1)) + { + wchar_t wcu = wc; + if (iswlower (wc)) + { + size_t mbcdlen; + + wcu = towupper (wc); + mbcdlen = wcrtomb ((char *) buf, wcu, &prev_st); + if (BE (mbclen == mbcdlen, 1)) + memcpy (pstr->mbs + byte_idx, buf, mbclen); + else if (mbcdlen != (size_t) -1) + { + size_t i; + + if (byte_idx + mbcdlen > pstr->bufs_len) + { + pstr->cur_state = prev_st; + break; + } + + if (pstr->offsets == NULL) + { + pstr->offsets = re_malloc (Idx, pstr->bufs_len); + + if (pstr->offsets == NULL) + return REG_ESPACE; + } + if (!pstr->offsets_needed) + { + for (i = 0; i < (size_t) byte_idx; ++i) + pstr->offsets[i] = i; + pstr->offsets_needed = 1; + } + + memcpy (pstr->mbs + byte_idx, buf, mbcdlen); + pstr->wcs[byte_idx] = wcu; + pstr->offsets[byte_idx] = src_idx; + for (i = 1; i < mbcdlen; ++i) + { + pstr->offsets[byte_idx + i] + = src_idx + (i < mbclen ? i : mbclen - 1); + pstr->wcs[byte_idx + i] = WEOF; + } + pstr->len += mbcdlen - mbclen; + if (pstr->raw_stop > src_idx) + pstr->stop += mbcdlen - mbclen; + end_idx = (pstr->bufs_len > pstr->len) + ? pstr->len : pstr->bufs_len; + byte_idx += mbcdlen; + src_idx += mbclen; + continue; + } + else + memcpy (pstr->mbs + byte_idx, p, mbclen); + } + else + memcpy (pstr->mbs + byte_idx, p, mbclen); + + if (BE (pstr->offsets_needed != 0, 0)) + { + size_t i; + for (i = 0; i < mbclen; ++i) + pstr->offsets[byte_idx + i] = src_idx + i; + } + src_idx += mbclen; + + pstr->wcs[byte_idx++] = wcu; + /* Write paddings. */ + for (remain_len = byte_idx + mbclen - 1; byte_idx < remain_len ;) + pstr->wcs[byte_idx++] = WEOF; + } + else if (mbclen == (size_t) -1 || mbclen == 0) + { + /* It is an invalid character or '\0'. Just use the byte. */ + int ch = pstr->raw_mbs[pstr->raw_mbs_idx + src_idx]; + + if (BE (pstr->trans != NULL, 0)) + ch = pstr->trans [ch]; + pstr->mbs[byte_idx] = ch; + + if (BE (pstr->offsets_needed != 0, 0)) + pstr->offsets[byte_idx] = src_idx; + ++src_idx; + + /* And also cast it to wide char. */ + pstr->wcs[byte_idx++] = (wchar_t) ch; + if (BE (mbclen == (size_t) -1, 0)) + pstr->cur_state = prev_st; + } + else + { + /* The buffer doesn't have enough space, finish to build. */ + pstr->cur_state = prev_st; + break; + } + } + pstr->valid_len = byte_idx; + pstr->valid_raw_len = src_idx; + return REG_NOERROR; +} + +/* Skip characters until the index becomes greater than NEW_RAW_IDX. + Return the index. */ + +static Idx +internal_function +re_string_skip_chars (re_string_t *pstr, Idx new_raw_idx, wint_t *last_wc) +{ + mbstate_t prev_st; + Idx rawbuf_idx; + size_t mbclen; + wint_t wc = WEOF; + + /* Skip the characters which are not necessary to check. */ + for (rawbuf_idx = pstr->raw_mbs_idx + pstr->valid_raw_len; + rawbuf_idx < new_raw_idx;) + { + wchar_t wc2; + Idx remain_len; + remain_len = pstr->len - rawbuf_idx; + prev_st = pstr->cur_state; + mbclen = __mbrtowc (&wc2, (const char *) pstr->raw_mbs + rawbuf_idx, + remain_len, &pstr->cur_state); + if (BE (mbclen == (size_t) -2 || mbclen == (size_t) -1 || mbclen == 0, 0)) + { + /* We treat these cases as a single byte character. */ + if (mbclen == 0 || remain_len == 0) + wc = L'\0'; + else + wc = *(unsigned char *) (pstr->raw_mbs + rawbuf_idx); + mbclen = 1; + pstr->cur_state = prev_st; + } + else + wc = wc2; + /* Then proceed the next character. */ + rawbuf_idx += mbclen; + } + *last_wc = wc; + return rawbuf_idx; +} +#endif /* RE_ENABLE_I18N */ + +/* Build the buffer PSTR->MBS, and apply the translation if we need. + This function is used in case of REG_ICASE. */ + +static void +internal_function +build_upper_buffer (re_string_t *pstr) +{ + Idx char_idx, end_idx; + end_idx = (pstr->bufs_len > pstr->len) ? pstr->len : pstr->bufs_len; + + for (char_idx = pstr->valid_len; char_idx < end_idx; ++char_idx) + { + int ch = pstr->raw_mbs[pstr->raw_mbs_idx + char_idx]; + if (BE (pstr->trans != NULL, 0)) + ch = pstr->trans[ch]; + if (islower (ch)) + pstr->mbs[char_idx] = toupper (ch); + else + pstr->mbs[char_idx] = ch; + } + pstr->valid_len = char_idx; + pstr->valid_raw_len = char_idx; +} + +/* Apply TRANS to the buffer in PSTR. */ + +static void +internal_function +re_string_translate_buffer (re_string_t *pstr) +{ + Idx buf_idx, end_idx; + end_idx = (pstr->bufs_len > pstr->len) ? pstr->len : pstr->bufs_len; + + for (buf_idx = pstr->valid_len; buf_idx < end_idx; ++buf_idx) + { + int ch = pstr->raw_mbs[pstr->raw_mbs_idx + buf_idx]; + pstr->mbs[buf_idx] = pstr->trans[ch]; + } + + pstr->valid_len = buf_idx; + pstr->valid_raw_len = buf_idx; +} + +/* This function re-construct the buffers. + Concretely, convert to wide character in case of pstr->mb_cur_max > 1, + convert to upper case in case of REG_ICASE, apply translation. */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +re_string_reconstruct (re_string_t *pstr, Idx idx, int eflags) +{ + Idx offset; + + if (BE (pstr->raw_mbs_idx <= idx, 0)) + offset = idx - pstr->raw_mbs_idx; + else + { + /* Reset buffer. */ +#ifdef RE_ENABLE_I18N + if (pstr->mb_cur_max > 1) + memset (&pstr->cur_state, '\0', sizeof (mbstate_t)); +#endif /* RE_ENABLE_I18N */ + pstr->len = pstr->raw_len; + pstr->stop = pstr->raw_stop; + pstr->valid_len = 0; + pstr->raw_mbs_idx = 0; + pstr->valid_raw_len = 0; + pstr->offsets_needed = 0; + pstr->tip_context = ((eflags & REG_NOTBOL) ? CONTEXT_BEGBUF + : CONTEXT_NEWLINE | CONTEXT_BEGBUF); + if (!pstr->mbs_allocated) + pstr->mbs = (unsigned char *) pstr->raw_mbs; + offset = idx; + } + + if (BE (offset != 0, 1)) + { + /* Should the already checked characters be kept? */ + if (BE (offset < pstr->valid_raw_len, 1)) + { + /* Yes, move them to the front of the buffer. */ +#ifdef RE_ENABLE_I18N + if (BE (pstr->offsets_needed, 0)) + { + Idx low = 0, high = pstr->valid_len, mid; + do + { + mid = (high + low) / 2; + if (pstr->offsets[mid] > offset) + high = mid; + else if (pstr->offsets[mid] < offset) + low = mid + 1; + else + break; + } + while (low < high); + if (pstr->offsets[mid] < offset) + ++mid; + pstr->tip_context = re_string_context_at (pstr, mid - 1, + eflags); + /* This can be quite complicated, so handle specially + only the common and easy case where the character with + different length representation of lower and upper + case is present at or after offset. */ + if (pstr->valid_len > offset + && mid == offset && pstr->offsets[mid] == offset) + { + memmove (pstr->wcs, pstr->wcs + offset, + (pstr->valid_len - offset) * sizeof (wint_t)); + memmove (pstr->mbs, pstr->mbs + offset, pstr->valid_len - offset); + pstr->valid_len -= offset; + pstr->valid_raw_len -= offset; + for (low = 0; low < pstr->valid_len; low++) + pstr->offsets[low] = pstr->offsets[low + offset] - offset; + } + else + { + /* Otherwise, just find out how long the partial multibyte + character at offset is and fill it with WEOF/255. */ + pstr->len = pstr->raw_len - idx + offset; + pstr->stop = pstr->raw_stop - idx + offset; + pstr->offsets_needed = 0; + while (mid > 0 && pstr->offsets[mid - 1] == offset) + --mid; + while (mid < pstr->valid_len) + if (pstr->wcs[mid] != WEOF) + break; + else + ++mid; + if (mid == pstr->valid_len) + pstr->valid_len = 0; + else + { + pstr->valid_len = pstr->offsets[mid] - offset; + if (pstr->valid_len) + { + for (low = 0; low < pstr->valid_len; ++low) + pstr->wcs[low] = WEOF; + memset (pstr->mbs, 255, pstr->valid_len); + } + } + pstr->valid_raw_len = pstr->valid_len; + } + } + else +#endif + { + pstr->tip_context = re_string_context_at (pstr, offset - 1, + eflags); +#ifdef RE_ENABLE_I18N + if (pstr->mb_cur_max > 1) + memmove (pstr->wcs, pstr->wcs + offset, + (pstr->valid_len - offset) * sizeof (wint_t)); +#endif /* RE_ENABLE_I18N */ + if (BE (pstr->mbs_allocated, 0)) + memmove (pstr->mbs, pstr->mbs + offset, + pstr->valid_len - offset); + pstr->valid_len -= offset; + pstr->valid_raw_len -= offset; +#if DEBUG + assert (pstr->valid_len > 0); +#endif + } + } + else + { +#ifdef RE_ENABLE_I18N + /* No, skip all characters until IDX. */ + Idx prev_valid_len = pstr->valid_len; + + if (BE (pstr->offsets_needed, 0)) + { + pstr->len = pstr->raw_len - idx + offset; + pstr->stop = pstr->raw_stop - idx + offset; + pstr->offsets_needed = 0; + } +#endif + pstr->valid_len = 0; +#ifdef RE_ENABLE_I18N + if (pstr->mb_cur_max > 1) + { + Idx wcs_idx; + wint_t wc = WEOF; + + if (pstr->is_utf8) + { + const unsigned char *raw, *p, *end; + + /* Special case UTF-8. Multi-byte chars start with any + byte other than 0x80 - 0xbf. */ + raw = pstr->raw_mbs + pstr->raw_mbs_idx; + end = raw + (offset - pstr->mb_cur_max); + if (end < pstr->raw_mbs) + end = pstr->raw_mbs; + p = raw + offset - 1; +#ifdef _LIBC + /* We know the wchar_t encoding is UCS4, so for the simple + case, ASCII characters, skip the conversion step. */ + if (isascii (*p) && BE (pstr->trans == NULL, 1)) + { + memset (&pstr->cur_state, '\0', sizeof (mbstate_t)); + /* pstr->valid_len = 0; */ + wc = (wchar_t) *p; + } + else +#endif + for (; p >= end; --p) + if ((*p & 0xc0) != 0x80) + { + mbstate_t cur_state; + wchar_t wc2; + Idx mlen = raw + pstr->len - p; + size_t mbclen; + +#if 0 /* dead code: buf is set but never used */ + unsigned char buf[6]; + if (BE (pstr->trans != NULL, 0)) + { + int i = mlen < 6 ? mlen : 6; + while (--i >= 0) + buf[i] = pstr->trans[p[i]]; + } +#endif + /* XXX Don't use mbrtowc, we know which conversion + to use (UTF-8 -> UCS4). */ + memset (&cur_state, 0, sizeof (cur_state)); + mbclen = __mbrtowc (&wc2, (const char *) p, mlen, + &cur_state); + if (raw + offset - p <= mbclen + && mbclen < (size_t) -2) + { + memset (&pstr->cur_state, '\0', + sizeof (mbstate_t)); + pstr->valid_len = mbclen - (raw + offset - p); + wc = wc2; + } + break; + } + } + + if (wc == WEOF) + pstr->valid_len = re_string_skip_chars (pstr, idx, &wc) - idx; + if (wc == WEOF) + pstr->tip_context + = re_string_context_at (pstr, prev_valid_len - 1, eflags); + else + pstr->tip_context = ((BE (pstr->word_ops_used != 0, 0) + && IS_WIDE_WORD_CHAR (wc)) + ? CONTEXT_WORD + : ((IS_WIDE_NEWLINE (wc) + && pstr->newline_anchor) + ? CONTEXT_NEWLINE : 0)); + if (BE (pstr->valid_len, 0)) + { + for (wcs_idx = 0; wcs_idx < pstr->valid_len; ++wcs_idx) + pstr->wcs[wcs_idx] = WEOF; + if (pstr->mbs_allocated) + memset (pstr->mbs, 255, pstr->valid_len); + } + pstr->valid_raw_len = pstr->valid_len; + } + else +#endif /* RE_ENABLE_I18N */ + { + int c = pstr->raw_mbs[pstr->raw_mbs_idx + offset - 1]; + pstr->valid_raw_len = 0; + if (pstr->trans) + c = pstr->trans[c]; + pstr->tip_context = (bitset_contain (pstr->word_char, c) + ? CONTEXT_WORD + : ((IS_NEWLINE (c) && pstr->newline_anchor) + ? CONTEXT_NEWLINE : 0)); + } + } + if (!BE (pstr->mbs_allocated, 0)) + pstr->mbs += offset; + } + pstr->raw_mbs_idx = idx; + pstr->len -= offset; + pstr->stop -= offset; + + /* Then build the buffers. */ +#ifdef RE_ENABLE_I18N + if (pstr->mb_cur_max > 1) + { + if (pstr->icase) + { + reg_errcode_t ret = build_wcs_upper_buffer (pstr); + if (BE (ret != REG_NOERROR, 0)) + return ret; + } + else + build_wcs_buffer (pstr); + } + else +#endif /* RE_ENABLE_I18N */ + if (BE (pstr->mbs_allocated, 0)) + { + if (pstr->icase) + build_upper_buffer (pstr); + else if (pstr->trans != NULL) + re_string_translate_buffer (pstr); + } + else + pstr->valid_len = pstr->len; + + pstr->cur_idx = 0; + return REG_NOERROR; +} + +static unsigned char +internal_function __attribute ((pure)) +re_string_peek_byte_case (const re_string_t *pstr, Idx idx) +{ + int ch; + Idx off; + + /* Handle the common (easiest) cases first. */ + if (BE (!pstr->mbs_allocated, 1)) + return re_string_peek_byte (pstr, idx); + +#ifdef RE_ENABLE_I18N + if (pstr->mb_cur_max > 1 + && ! re_string_is_single_byte_char (pstr, pstr->cur_idx + idx)) + return re_string_peek_byte (pstr, idx); +#endif + + off = pstr->cur_idx + idx; +#ifdef RE_ENABLE_I18N + if (pstr->offsets_needed) + off = pstr->offsets[off]; +#endif + + ch = pstr->raw_mbs[pstr->raw_mbs_idx + off]; + +#ifdef RE_ENABLE_I18N + /* Ensure that e.g. for tr_TR.UTF-8 BACKSLASH DOTLESS SMALL LETTER I + this function returns CAPITAL LETTER I instead of first byte of + DOTLESS SMALL LETTER I. The latter would confuse the parser, + since peek_byte_case doesn't advance cur_idx in any way. */ + if (pstr->offsets_needed && !isascii (ch)) + return re_string_peek_byte (pstr, idx); +#endif + + return ch; +} + +static unsigned char +internal_function __attribute ((pure)) +re_string_fetch_byte_case (re_string_t *pstr) +{ + if (BE (!pstr->mbs_allocated, 1)) + return re_string_fetch_byte (pstr); + +#ifdef RE_ENABLE_I18N + if (pstr->offsets_needed) + { + Idx off; + int ch; + + /* For tr_TR.UTF-8 [[:islower:]] there is + [[: CAPITAL LETTER I WITH DOT lower:]] in mbs. Skip + in that case the whole multi-byte character and return + the original letter. On the other side, with + [[: DOTLESS SMALL LETTER I return [[:I, as doing + anything else would complicate things too much. */ + + if (!re_string_first_byte (pstr, pstr->cur_idx)) + return re_string_fetch_byte (pstr); + + off = pstr->offsets[pstr->cur_idx]; + ch = pstr->raw_mbs[pstr->raw_mbs_idx + off]; + + if (! isascii (ch)) + return re_string_fetch_byte (pstr); + + re_string_skip_bytes (pstr, + re_string_char_size_at (pstr, pstr->cur_idx)); + return ch; + } +#endif + + return pstr->raw_mbs[pstr->raw_mbs_idx + pstr->cur_idx++]; +} + +static void +internal_function +re_string_destruct (re_string_t *pstr) +{ +#ifdef RE_ENABLE_I18N + re_free (pstr->wcs); + re_free (pstr->offsets); +#endif /* RE_ENABLE_I18N */ + if (pstr->mbs_allocated) + re_free (pstr->mbs); +} + +/* Return the context at IDX in INPUT. */ + +static unsigned int +internal_function +re_string_context_at (const re_string_t *input, Idx idx, int eflags) +{ + int c; + if (BE (! REG_VALID_INDEX (idx), 0)) + /* In this case, we use the value stored in input->tip_context, + since we can't know the character in input->mbs[-1] here. */ + return input->tip_context; + if (BE (idx == input->len, 0)) + return ((eflags & REG_NOTEOL) ? CONTEXT_ENDBUF + : CONTEXT_NEWLINE | CONTEXT_ENDBUF); +#ifdef RE_ENABLE_I18N + if (input->mb_cur_max > 1) + { + wint_t wc; + Idx wc_idx = idx; + while(input->wcs[wc_idx] == WEOF) + { +#ifdef DEBUG + /* It must not happen. */ + assert (REG_VALID_INDEX (wc_idx)); +#endif + --wc_idx; + if (! REG_VALID_INDEX (wc_idx)) + return input->tip_context; + } + wc = input->wcs[wc_idx]; + if (BE (input->word_ops_used != 0, 0) && IS_WIDE_WORD_CHAR (wc)) + return CONTEXT_WORD; + return (IS_WIDE_NEWLINE (wc) && input->newline_anchor + ? CONTEXT_NEWLINE : 0); + } + else +#endif + { + c = re_string_byte_at (input, idx); + if (bitset_contain (input->word_char, c)) + return CONTEXT_WORD; + return IS_NEWLINE (c) && input->newline_anchor ? CONTEXT_NEWLINE : 0; + } +} + +/* Functions for set operation. */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +re_node_set_alloc (re_node_set *set, Idx size) +{ + set->alloc = size; + set->nelem = 0; + set->elems = re_malloc (Idx, size); + if (BE (set->elems == NULL, 0)) + return REG_ESPACE; + return REG_NOERROR; +} + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +re_node_set_init_1 (re_node_set *set, Idx elem) +{ + set->alloc = 1; + set->nelem = 1; + set->elems = re_malloc (Idx, 1); + if (BE (set->elems == NULL, 0)) + { + set->alloc = set->nelem = 0; + return REG_ESPACE; + } + set->elems[0] = elem; + return REG_NOERROR; +} + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +re_node_set_init_2 (re_node_set *set, Idx elem1, Idx elem2) +{ + set->alloc = 2; + set->elems = re_malloc (Idx, 2); + if (BE (set->elems == NULL, 0)) + return REG_ESPACE; + if (elem1 == elem2) + { + set->nelem = 1; + set->elems[0] = elem1; + } + else + { + set->nelem = 2; + if (elem1 < elem2) + { + set->elems[0] = elem1; + set->elems[1] = elem2; + } + else + { + set->elems[0] = elem2; + set->elems[1] = elem1; + } + } + return REG_NOERROR; +} + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +re_node_set_init_copy (re_node_set *dest, const re_node_set *src) +{ + dest->nelem = src->nelem; + if (src->nelem > 0) + { + dest->alloc = dest->nelem; + dest->elems = re_malloc (Idx, dest->alloc); + if (BE (dest->elems == NULL, 0)) + { + dest->alloc = dest->nelem = 0; + return REG_ESPACE; + } + memcpy (dest->elems, src->elems, src->nelem * sizeof (Idx)); + } + else + re_node_set_init_empty (dest); + return REG_NOERROR; +} + +/* Calculate the intersection of the sets SRC1 and SRC2. And merge it to + DEST. Return value indicate the error code or REG_NOERROR if succeeded. + Note: We assume dest->elems is NULL, when dest->alloc is 0. */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +re_node_set_add_intersect (re_node_set *dest, const re_node_set *src1, + const re_node_set *src2) +{ + Idx i1, i2, is, id, delta, sbase; + if (src1->nelem == 0 || src2->nelem == 0) + return REG_NOERROR; + + /* We need dest->nelem + 2 * elems_in_intersection; this is a + conservative estimate. */ + if (src1->nelem + src2->nelem + dest->nelem > dest->alloc) + { + Idx new_alloc = src1->nelem + src2->nelem + dest->alloc; + Idx *new_elems = re_realloc (dest->elems, Idx, new_alloc); + if (BE (new_elems == NULL, 0)) + return REG_ESPACE; + dest->elems = new_elems; + dest->alloc = new_alloc; + } + + /* Find the items in the intersection of SRC1 and SRC2, and copy + into the top of DEST those that are not already in DEST itself. */ + sbase = dest->nelem + src1->nelem + src2->nelem; + i1 = src1->nelem - 1; + i2 = src2->nelem - 1; + id = dest->nelem - 1; + for (;;) + { + if (src1->elems[i1] == src2->elems[i2]) + { + /* Try to find the item in DEST. Maybe we could binary search? */ + while (REG_VALID_INDEX (id) && dest->elems[id] > src1->elems[i1]) + --id; + + if (! REG_VALID_INDEX (id) || dest->elems[id] != src1->elems[i1]) + dest->elems[--sbase] = src1->elems[i1]; + + if (! REG_VALID_INDEX (--i1) || ! REG_VALID_INDEX (--i2)) + break; + } + + /* Lower the highest of the two items. */ + else if (src1->elems[i1] < src2->elems[i2]) + { + if (! REG_VALID_INDEX (--i2)) + break; + } + else + { + if (! REG_VALID_INDEX (--i1)) + break; + } + } + + id = dest->nelem - 1; + is = dest->nelem + src1->nelem + src2->nelem - 1; + delta = is - sbase + 1; + + /* Now copy. When DELTA becomes zero, the remaining + DEST elements are already in place; this is more or + less the same loop that is in re_node_set_merge. */ + dest->nelem += delta; + if (delta > 0 && REG_VALID_INDEX (id)) + for (;;) + { + if (dest->elems[is] > dest->elems[id]) + { + /* Copy from the top. */ + dest->elems[id + delta--] = dest->elems[is--]; + if (delta == 0) + break; + } + else + { + /* Slide from the bottom. */ + dest->elems[id + delta] = dest->elems[id]; + if (! REG_VALID_INDEX (--id)) + break; + } + } + + /* Copy remaining SRC elements. */ + memcpy (dest->elems, dest->elems + sbase, delta * sizeof (Idx)); + + return REG_NOERROR; +} + +/* Calculate the union set of the sets SRC1 and SRC2. And store it to + DEST. Return value indicate the error code or REG_NOERROR if succeeded. */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +re_node_set_init_union (re_node_set *dest, const re_node_set *src1, + const re_node_set *src2) +{ + Idx i1, i2, id; + if (src1 != NULL && src1->nelem > 0 && src2 != NULL && src2->nelem > 0) + { + dest->alloc = src1->nelem + src2->nelem; + dest->elems = re_malloc (Idx, dest->alloc); + if (BE (dest->elems == NULL, 0)) + return REG_ESPACE; + } + else + { + if (src1 != NULL && src1->nelem > 0) + return re_node_set_init_copy (dest, src1); + else if (src2 != NULL && src2->nelem > 0) + return re_node_set_init_copy (dest, src2); + else + re_node_set_init_empty (dest); + return REG_NOERROR; + } + for (i1 = i2 = id = 0 ; i1 < src1->nelem && i2 < src2->nelem ;) + { + if (src1->elems[i1] > src2->elems[i2]) + { + dest->elems[id++] = src2->elems[i2++]; + continue; + } + if (src1->elems[i1] == src2->elems[i2]) + ++i2; + dest->elems[id++] = src1->elems[i1++]; + } + if (i1 < src1->nelem) + { + memcpy (dest->elems + id, src1->elems + i1, + (src1->nelem - i1) * sizeof (Idx)); + id += src1->nelem - i1; + } + else if (i2 < src2->nelem) + { + memcpy (dest->elems + id, src2->elems + i2, + (src2->nelem - i2) * sizeof (Idx)); + id += src2->nelem - i2; + } + dest->nelem = id; + return REG_NOERROR; +} + +/* Calculate the union set of the sets DEST and SRC. And store it to + DEST. Return value indicate the error code or REG_NOERROR if succeeded. */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +re_node_set_merge (re_node_set *dest, const re_node_set *src) +{ + Idx is, id, sbase, delta; + if (src == NULL || src->nelem == 0) + return REG_NOERROR; + if (dest->alloc < 2 * src->nelem + dest->nelem) + { + Idx new_alloc = 2 * (src->nelem + dest->alloc); + Idx *new_buffer = re_realloc (dest->elems, Idx, new_alloc); + if (BE (new_buffer == NULL, 0)) + return REG_ESPACE; + dest->elems = new_buffer; + dest->alloc = new_alloc; + } + + if (BE (dest->nelem == 0, 0)) + { + dest->nelem = src->nelem; + memcpy (dest->elems, src->elems, src->nelem * sizeof (Idx)); + return REG_NOERROR; + } + + /* Copy into the top of DEST the items of SRC that are not + found in DEST. Maybe we could binary search in DEST? */ + for (sbase = dest->nelem + 2 * src->nelem, + is = src->nelem - 1, id = dest->nelem - 1; + REG_VALID_INDEX (is) && REG_VALID_INDEX (id); ) + { + if (dest->elems[id] == src->elems[is]) + is--, id--; + else if (dest->elems[id] < src->elems[is]) + dest->elems[--sbase] = src->elems[is--]; + else /* if (dest->elems[id] > src->elems[is]) */ + --id; + } + + if (REG_VALID_INDEX (is)) + { + /* If DEST is exhausted, the remaining items of SRC must be unique. */ + sbase -= is + 1; + memcpy (dest->elems + sbase, src->elems, (is + 1) * sizeof (Idx)); + } + + id = dest->nelem - 1; + is = dest->nelem + 2 * src->nelem - 1; + delta = is - sbase + 1; + if (delta == 0) + return REG_NOERROR; + + /* Now copy. When DELTA becomes zero, the remaining + DEST elements are already in place. */ + dest->nelem += delta; + for (;;) + { + if (dest->elems[is] > dest->elems[id]) + { + /* Copy from the top. */ + dest->elems[id + delta--] = dest->elems[is--]; + if (delta == 0) + break; + } + else + { + /* Slide from the bottom. */ + dest->elems[id + delta] = dest->elems[id]; + if (! REG_VALID_INDEX (--id)) + { + /* Copy remaining SRC elements. */ + memcpy (dest->elems, dest->elems + sbase, + delta * sizeof (Idx)); + break; + } + } + } + + return REG_NOERROR; +} + +/* Insert the new element ELEM to the re_node_set* SET. + SET should not already have ELEM. + Return true if successful. */ + +static bool +internal_function __attribute_warn_unused_result__ +re_node_set_insert (re_node_set *set, Idx elem) +{ + Idx idx; + /* In case the set is empty. */ + if (set->alloc == 0) + return BE (re_node_set_init_1 (set, elem) == REG_NOERROR, 1); + + if (BE (set->nelem, 0) == 0) + { + /* We already guaranteed above that set->alloc != 0. */ + set->elems[0] = elem; + ++set->nelem; + return true; + } + + /* Realloc if we need. */ + if (set->alloc == set->nelem) + { + Idx *new_elems; + set->alloc = set->alloc * 2; + new_elems = re_realloc (set->elems, Idx, set->alloc); + if (BE (new_elems == NULL, 0)) + return false; + set->elems = new_elems; + } + + /* Move the elements which follows the new element. Test the + first element separately to skip a check in the inner loop. */ + if (elem < set->elems[0]) + { + idx = 0; + for (idx = set->nelem; idx > 0; idx--) + set->elems[idx] = set->elems[idx - 1]; + } + else + { + for (idx = set->nelem; set->elems[idx - 1] > elem; idx--) + set->elems[idx] = set->elems[idx - 1]; + } + + /* Insert the new element. */ + set->elems[idx] = elem; + ++set->nelem; + return true; +} + +/* Insert the new element ELEM to the re_node_set* SET. + SET should not already have any element greater than or equal to ELEM. + Return true if successful. */ + +static bool +internal_function __attribute_warn_unused_result__ +re_node_set_insert_last (re_node_set *set, Idx elem) +{ + /* Realloc if we need. */ + if (set->alloc == set->nelem) + { + Idx *new_elems; + set->alloc = (set->alloc + 1) * 2; + new_elems = re_realloc (set->elems, Idx, set->alloc); + if (BE (new_elems == NULL, 0)) + return false; + set->elems = new_elems; + } + + /* Insert the new element. */ + set->elems[set->nelem++] = elem; + return true; +} + +/* Compare two node sets SET1 and SET2. + Return true if SET1 and SET2 are equivalent. */ + +static bool +internal_function __attribute ((pure)) +re_node_set_compare (const re_node_set *set1, const re_node_set *set2) +{ + Idx i; + if (set1 == NULL || set2 == NULL || set1->nelem != set2->nelem) + return false; + for (i = set1->nelem ; REG_VALID_INDEX (--i) ; ) + if (set1->elems[i] != set2->elems[i]) + return false; + return true; +} + +/* Return (idx + 1) if SET contains the element ELEM, return 0 otherwise. */ + +static Idx +internal_function __attribute ((pure)) +re_node_set_contains (const re_node_set *set, Idx elem) +{ + __re_size_t idx, right, mid; + if (! REG_VALID_NONZERO_INDEX (set->nelem)) + return 0; + + /* Binary search the element. */ + idx = 0; + right = set->nelem - 1; + while (idx < right) + { + mid = (idx + right) / 2; + if (set->elems[mid] < elem) + idx = mid + 1; + else + right = mid; + } + return set->elems[idx] == elem ? idx + 1 : 0; +} + +static void +internal_function +re_node_set_remove_at (re_node_set *set, Idx idx) +{ + if (idx < 0 || idx >= set->nelem) + return; + --set->nelem; + for (; idx < set->nelem; idx++) + set->elems[idx] = set->elems[idx + 1]; +} + + +/* Add the token TOKEN to dfa->nodes, and return the index of the token. + Or return REG_MISSING if an error occurred. */ + +static Idx +internal_function +re_dfa_add_node (re_dfa_t *dfa, re_token_t token) +{ + if (BE (dfa->nodes_len >= dfa->nodes_alloc, 0)) + { + size_t new_nodes_alloc = dfa->nodes_alloc * 2; + Idx *new_nexts, *new_indices; + re_node_set *new_edests, *new_eclosures; + re_token_t *new_nodes; + size_t max_object_size = + MAX (sizeof (re_token_t), + MAX (sizeof (re_node_set), + sizeof (Idx))); + + /* Avoid overflows. */ + if (BE (SIZE_MAX / 2 / max_object_size < dfa->nodes_alloc, 0)) + return REG_MISSING; + + new_nodes = re_realloc (dfa->nodes, re_token_t, new_nodes_alloc); + if (BE (new_nodes == NULL, 0)) + return REG_MISSING; + dfa->nodes = new_nodes; + new_nexts = re_realloc (dfa->nexts, Idx, new_nodes_alloc); + new_indices = re_realloc (dfa->org_indices, Idx, new_nodes_alloc); + new_edests = re_realloc (dfa->edests, re_node_set, new_nodes_alloc); + new_eclosures = re_realloc (dfa->eclosures, re_node_set, new_nodes_alloc); + if (BE (new_nexts == NULL || new_indices == NULL + || new_edests == NULL || new_eclosures == NULL, 0)) + return REG_MISSING; + dfa->nexts = new_nexts; + dfa->org_indices = new_indices; + dfa->edests = new_edests; + dfa->eclosures = new_eclosures; + dfa->nodes_alloc = new_nodes_alloc; + } + dfa->nodes[dfa->nodes_len] = token; + dfa->nodes[dfa->nodes_len].constraint = 0; +#ifdef RE_ENABLE_I18N + { + int type = token.type; + dfa->nodes[dfa->nodes_len].accept_mb = + (type == OP_PERIOD && dfa->mb_cur_max > 1) || type == COMPLEX_BRACKET; + } +#endif + dfa->nexts[dfa->nodes_len] = REG_MISSING; + re_node_set_init_empty (dfa->edests + dfa->nodes_len); + re_node_set_init_empty (dfa->eclosures + dfa->nodes_len); + return dfa->nodes_len++; +} + +static inline re_hashval_t +internal_function +calc_state_hash (const re_node_set *nodes, unsigned int context) +{ + re_hashval_t hash = nodes->nelem + context; + Idx i; + for (i = 0 ; i < nodes->nelem ; i++) + hash += nodes->elems[i]; + return hash; +} + +/* Search for the state whose node_set is equivalent to NODES. + Return the pointer to the state, if we found it in the DFA. + Otherwise create the new one and return it. In case of an error + return NULL and set the error code in ERR. + Note: - We assume NULL as the invalid state, then it is possible that + return value is NULL and ERR is REG_NOERROR. + - We never return non-NULL value in case of any errors, it is for + optimization. */ + +static re_dfastate_t * +internal_function __attribute_warn_unused_result__ +re_acquire_state (reg_errcode_t *err, const re_dfa_t *dfa, + const re_node_set *nodes) +{ + re_hashval_t hash; + re_dfastate_t *new_state; + struct re_state_table_entry *spot; + Idx i; +#ifdef lint + /* Suppress bogus uninitialized-variable warnings. */ + *err = REG_NOERROR; +#endif + if (BE (nodes->nelem == 0, 0)) + { + *err = REG_NOERROR; + return NULL; + } + hash = calc_state_hash (nodes, 0); + spot = dfa->state_table + (hash & dfa->state_hash_mask); + + for (i = 0 ; i < spot->num ; i++) + { + re_dfastate_t *state = spot->array[i]; + if (hash != state->hash) + continue; + if (re_node_set_compare (&state->nodes, nodes)) + return state; + } + + /* There are no appropriate state in the dfa, create the new one. */ + new_state = create_ci_newstate (dfa, nodes, hash); + if (BE (new_state == NULL, 0)) + *err = REG_ESPACE; + + return new_state; +} + +/* Search for the state whose node_set is equivalent to NODES and + whose context is equivalent to CONTEXT. + Return the pointer to the state, if we found it in the DFA. + Otherwise create the new one and return it. In case of an error + return NULL and set the error code in ERR. + Note: - We assume NULL as the invalid state, then it is possible that + return value is NULL and ERR is REG_NOERROR. + - We never return non-NULL value in case of any errors, it is for + optimization. */ + +static re_dfastate_t * +internal_function __attribute_warn_unused_result__ +re_acquire_state_context (reg_errcode_t *err, const re_dfa_t *dfa, + const re_node_set *nodes, unsigned int context) +{ + re_hashval_t hash; + re_dfastate_t *new_state; + struct re_state_table_entry *spot; + Idx i; +#ifdef lint + /* Suppress bogus uninitialized-variable warnings. */ + *err = REG_NOERROR; +#endif + if (nodes->nelem == 0) + { + *err = REG_NOERROR; + return NULL; + } + hash = calc_state_hash (nodes, context); + spot = dfa->state_table + (hash & dfa->state_hash_mask); + + for (i = 0 ; i < spot->num ; i++) + { + re_dfastate_t *state = spot->array[i]; + if (state->hash == hash + && state->context == context + && re_node_set_compare (state->entrance_nodes, nodes)) + return state; + } + /* There are no appropriate state in 'dfa', create the new one. */ + new_state = create_cd_newstate (dfa, nodes, context, hash); + if (BE (new_state == NULL, 0)) + *err = REG_ESPACE; + + return new_state; +} + +/* Finish initialization of the new state NEWSTATE, and using its hash value + HASH put in the appropriate bucket of DFA's state table. Return value + indicates the error code if failed. */ + +static reg_errcode_t +__attribute_warn_unused_result__ +register_state (const re_dfa_t *dfa, re_dfastate_t *newstate, + re_hashval_t hash) +{ + struct re_state_table_entry *spot; + reg_errcode_t err; + Idx i; + + newstate->hash = hash; + err = re_node_set_alloc (&newstate->non_eps_nodes, newstate->nodes.nelem); + if (BE (err != REG_NOERROR, 0)) + return REG_ESPACE; + for (i = 0; i < newstate->nodes.nelem; i++) + { + Idx elem = newstate->nodes.elems[i]; + if (!IS_EPSILON_NODE (dfa->nodes[elem].type)) + if (BE (! re_node_set_insert_last (&newstate->non_eps_nodes, elem), 0)) + return REG_ESPACE; + } + + spot = dfa->state_table + (hash & dfa->state_hash_mask); + if (BE (spot->alloc <= spot->num, 0)) + { + Idx new_alloc = 2 * spot->num + 2; + re_dfastate_t **new_array = re_realloc (spot->array, re_dfastate_t *, + new_alloc); + if (BE (new_array == NULL, 0)) + return REG_ESPACE; + spot->array = new_array; + spot->alloc = new_alloc; + } + spot->array[spot->num++] = newstate; + return REG_NOERROR; +} + +static void +free_state (re_dfastate_t *state) +{ + re_node_set_free (&state->non_eps_nodes); + re_node_set_free (&state->inveclosure); + if (state->entrance_nodes != &state->nodes) + { + re_node_set_free (state->entrance_nodes); + re_free (state->entrance_nodes); + } + re_node_set_free (&state->nodes); + re_free (state->word_trtable); + re_free (state->trtable); + re_free (state); +} + +/* Create the new state which is independ of contexts. + Return the new state if succeeded, otherwise return NULL. */ + +static re_dfastate_t * +internal_function __attribute_warn_unused_result__ +create_ci_newstate (const re_dfa_t *dfa, const re_node_set *nodes, + re_hashval_t hash) +{ + Idx i; + reg_errcode_t err; + re_dfastate_t *newstate; + + newstate = (re_dfastate_t *) calloc (sizeof (re_dfastate_t), 1); + if (BE (newstate == NULL, 0)) + return NULL; + err = re_node_set_init_copy (&newstate->nodes, nodes); + if (BE (err != REG_NOERROR, 0)) + { + re_free (newstate); + return NULL; + } + + newstate->entrance_nodes = &newstate->nodes; + for (i = 0 ; i < nodes->nelem ; i++) + { + re_token_t *node = dfa->nodes + nodes->elems[i]; + re_token_type_t type = node->type; + if (type == CHARACTER && !node->constraint) + continue; +#ifdef RE_ENABLE_I18N + newstate->accept_mb |= node->accept_mb; +#endif /* RE_ENABLE_I18N */ + + /* If the state has the halt node, the state is a halt state. */ + if (type == END_OF_RE) + newstate->halt = 1; + else if (type == OP_BACK_REF) + newstate->has_backref = 1; + else if (type == ANCHOR || node->constraint) + newstate->has_constraint = 1; + } + err = register_state (dfa, newstate, hash); + if (BE (err != REG_NOERROR, 0)) + { + free_state (newstate); + newstate = NULL; + } + return newstate; +} + +/* Create the new state which is depend on the context CONTEXT. + Return the new state if succeeded, otherwise return NULL. */ + +static re_dfastate_t * +internal_function __attribute_warn_unused_result__ +create_cd_newstate (const re_dfa_t *dfa, const re_node_set *nodes, + unsigned int context, re_hashval_t hash) +{ + Idx i, nctx_nodes = 0; + reg_errcode_t err; + re_dfastate_t *newstate; + + newstate = (re_dfastate_t *) calloc (sizeof (re_dfastate_t), 1); + if (BE (newstate == NULL, 0)) + return NULL; + err = re_node_set_init_copy (&newstate->nodes, nodes); + if (BE (err != REG_NOERROR, 0)) + { + re_free (newstate); + return NULL; + } + + newstate->context = context; + newstate->entrance_nodes = &newstate->nodes; + + for (i = 0 ; i < nodes->nelem ; i++) + { + re_token_t *node = dfa->nodes + nodes->elems[i]; + re_token_type_t type = node->type; + unsigned int constraint = node->constraint; + + if (type == CHARACTER && !constraint) + continue; +#ifdef RE_ENABLE_I18N + newstate->accept_mb |= node->accept_mb; +#endif /* RE_ENABLE_I18N */ + + /* If the state has the halt node, the state is a halt state. */ + if (type == END_OF_RE) + newstate->halt = 1; + else if (type == OP_BACK_REF) + newstate->has_backref = 1; + + if (constraint) + { + if (newstate->entrance_nodes == &newstate->nodes) + { + newstate->entrance_nodes = re_malloc (re_node_set, 1); + if (BE (newstate->entrance_nodes == NULL, 0)) + { + free_state (newstate); + return NULL; + } + if (re_node_set_init_copy (newstate->entrance_nodes, nodes) + != REG_NOERROR) + return NULL; + nctx_nodes = 0; + newstate->has_constraint = 1; + } + + if (NOT_SATISFY_PREV_CONSTRAINT (constraint,context)) + { + re_node_set_remove_at (&newstate->nodes, i - nctx_nodes); + ++nctx_nodes; + } + } + } + err = register_state (dfa, newstate, hash); + if (BE (err != REG_NOERROR, 0)) + { + free_state (newstate); + newstate = NULL; + } + return newstate; +} diff --git a/lib/regex_internal.h b/lib/regex_internal.h new file mode 100644 index 000000000..726119203 --- /dev/null +++ b/lib/regex_internal.h @@ -0,0 +1,866 @@ +/* Extended regular expression matching and search library. + Copyright (C) 2002-2012 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2, or (at your option) + any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License along + with this program; if not, write to the Free Software Foundation, + Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ + +#ifndef _REGEX_INTERNAL_H +#define _REGEX_INTERNAL_H 1 + +#include <assert.h> +#include <ctype.h> +#include <stdbool.h> +#include <stdio.h> +#include <stdlib.h> +#include <string.h> + +#include <langinfo.h> +#ifndef _LIBC +# include "localcharset.h" +#endif +#include <locale.h> + +#include <wchar.h> +#include <wctype.h> +#include <stdint.h> +#if defined _LIBC +# include <bits/libc-lock.h> +#else +# define __libc_lock_init(NAME) do { } while (0) +# define __libc_lock_lock(NAME) do { } while (0) +# define __libc_lock_unlock(NAME) do { } while (0) +#endif + +/* In case that the system doesn't have isblank(). */ +#if !defined _LIBC && ! (defined isblank || (HAVE_ISBLANK && HAVE_DECL_ISBLANK)) +# define isblank(ch) ((ch) == ' ' || (ch) == '\t') +#endif + +#ifdef _LIBC +# ifndef _RE_DEFINE_LOCALE_FUNCTIONS +# define _RE_DEFINE_LOCALE_FUNCTIONS 1 +# include <locale/localeinfo.h> +# include <locale/elem-hash.h> +# include <locale/coll-lookup.h> +# endif +#endif + +/* This is for other GNU distributions with internationalized messages. */ +#if (HAVE_LIBINTL_H && ENABLE_NLS) || defined _LIBC +# include <libintl.h> +# ifdef _LIBC +# undef gettext +# define gettext(msgid) \ + INTUSE(__dcgettext) (_libc_intl_domainname, msgid, LC_MESSAGES) +# endif +#else +# define gettext(msgid) (msgid) +#endif + +#ifndef gettext_noop +/* This define is so xgettext can find the internationalizable + strings. */ +# define gettext_noop(String) String +#endif + +/* For loser systems without the definition. */ +#ifndef SIZE_MAX +# define SIZE_MAX ((size_t) -1) +#endif + +#if (defined MB_CUR_MAX && HAVE_WCTYPE_H && HAVE_ISWCTYPE && HAVE_WCSCOLL) || _LIBC +# define RE_ENABLE_I18N +#endif + +#if __GNUC__ >= 3 +# define BE(expr, val) __builtin_expect (expr, val) +#else +# define BE(expr, val) (expr) +# ifdef _LIBC +# define inline +# endif +#endif + +/* Number of ASCII characters. */ +#define ASCII_CHARS 0x80 + +/* Number of single byte characters. */ +#define SBC_MAX (UCHAR_MAX + 1) + +#define COLL_ELEM_LEN_MAX 8 + +/* The character which represents newline. */ +#define NEWLINE_CHAR '\n' +#define WIDE_NEWLINE_CHAR L'\n' + +/* Rename to standard API for using out of glibc. */ +#ifndef _LIBC +# define __wctype wctype +# define __iswctype iswctype +# define __btowc btowc +# define __wcrtomb wcrtomb +# define __mbrtowc mbrtowc +# define __regfree regfree +# define attribute_hidden +#endif /* not _LIBC */ + +#if __GNUC__ >= 4 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 1) +# define __attribute(arg) __attribute__ (arg) +#else +# define __attribute(arg) +#endif + +typedef __re_idx_t Idx; + +/* Special return value for failure to match. */ +#define REG_MISSING ((Idx) -1) + +/* Special return value for internal error. */ +#define REG_ERROR ((Idx) -2) + +/* Test whether N is a valid index, and is not one of the above. */ +#ifdef _REGEX_LARGE_OFFSETS +# define REG_VALID_INDEX(n) ((Idx) (n) < REG_ERROR) +#else +# define REG_VALID_INDEX(n) (0 <= (n)) +#endif + +/* Test whether N is a valid nonzero index. */ +#ifdef _REGEX_LARGE_OFFSETS +# define REG_VALID_NONZERO_INDEX(n) ((Idx) ((n) - 1) < (Idx) (REG_ERROR - 1)) +#else +# define REG_VALID_NONZERO_INDEX(n) (0 < (n)) +#endif + +/* A hash value, suitable for computing hash tables. */ +typedef __re_size_t re_hashval_t; + +/* An integer used to represent a set of bits. It must be unsigned, + and must be at least as wide as unsigned int. */ +typedef unsigned long int bitset_word_t; +/* All bits set in a bitset_word_t. */ +#define BITSET_WORD_MAX ULONG_MAX + +/* Number of bits in a bitset_word_t. For portability to hosts with + padding bits, do not use '(sizeof (bitset_word_t) * CHAR_BIT)'; + instead, deduce it directly from BITSET_WORD_MAX. Avoid + greater-than-32-bit integers and unconditional shifts by more than + 31 bits, as they're not portable. */ +#if BITSET_WORD_MAX == 0xffffffffUL +# define BITSET_WORD_BITS 32 +#elif BITSET_WORD_MAX >> 31 >> 4 == 1 +# define BITSET_WORD_BITS 36 +#elif BITSET_WORD_MAX >> 31 >> 16 == 1 +# define BITSET_WORD_BITS 48 +#elif BITSET_WORD_MAX >> 31 >> 28 == 1 +# define BITSET_WORD_BITS 60 +#elif BITSET_WORD_MAX >> 31 >> 31 >> 1 == 1 +# define BITSET_WORD_BITS 64 +#elif BITSET_WORD_MAX >> 31 >> 31 >> 9 == 1 +# define BITSET_WORD_BITS 72 +#elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 3 == 1 +# define BITSET_WORD_BITS 128 +#elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 == 1 +# define BITSET_WORD_BITS 256 +#elif BITSET_WORD_MAX >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 31 >> 7 > 1 +# define BITSET_WORD_BITS 257 /* any value > SBC_MAX will do here */ +# if BITSET_WORD_BITS <= SBC_MAX +# error "Invalid SBC_MAX" +# endif +#else +# error "Add case for new bitset_word_t size" +#endif + +/* Number of bitset_word_t values in a bitset_t. */ +#define BITSET_WORDS ((SBC_MAX + BITSET_WORD_BITS - 1) / BITSET_WORD_BITS) + +typedef bitset_word_t bitset_t[BITSET_WORDS]; +typedef bitset_word_t *re_bitset_ptr_t; +typedef const bitset_word_t *re_const_bitset_ptr_t; + +#define PREV_WORD_CONSTRAINT 0x0001 +#define PREV_NOTWORD_CONSTRAINT 0x0002 +#define NEXT_WORD_CONSTRAINT 0x0004 +#define NEXT_NOTWORD_CONSTRAINT 0x0008 +#define PREV_NEWLINE_CONSTRAINT 0x0010 +#define NEXT_NEWLINE_CONSTRAINT 0x0020 +#define PREV_BEGBUF_CONSTRAINT 0x0040 +#define NEXT_ENDBUF_CONSTRAINT 0x0080 +#define WORD_DELIM_CONSTRAINT 0x0100 +#define NOT_WORD_DELIM_CONSTRAINT 0x0200 + +typedef enum +{ + INSIDE_WORD = PREV_WORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, + WORD_FIRST = PREV_NOTWORD_CONSTRAINT | NEXT_WORD_CONSTRAINT, + WORD_LAST = PREV_WORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, + INSIDE_NOTWORD = PREV_NOTWORD_CONSTRAINT | NEXT_NOTWORD_CONSTRAINT, + LINE_FIRST = PREV_NEWLINE_CONSTRAINT, + LINE_LAST = NEXT_NEWLINE_CONSTRAINT, + BUF_FIRST = PREV_BEGBUF_CONSTRAINT, + BUF_LAST = NEXT_ENDBUF_CONSTRAINT, + WORD_DELIM = WORD_DELIM_CONSTRAINT, + NOT_WORD_DELIM = NOT_WORD_DELIM_CONSTRAINT +} re_context_type; + +typedef struct +{ + Idx alloc; + Idx nelem; + Idx *elems; +} re_node_set; + +typedef enum +{ + NON_TYPE = 0, + + /* Node type, These are used by token, node, tree. */ + CHARACTER = 1, + END_OF_RE = 2, + SIMPLE_BRACKET = 3, + OP_BACK_REF = 4, + OP_PERIOD = 5, +#ifdef RE_ENABLE_I18N + COMPLEX_BRACKET = 6, + OP_UTF8_PERIOD = 7, +#endif /* RE_ENABLE_I18N */ + + /* We define EPSILON_BIT as a macro so that OP_OPEN_SUBEXP is used + when the debugger shows values of this enum type. */ +#define EPSILON_BIT 8 + OP_OPEN_SUBEXP = EPSILON_BIT | 0, + OP_CLOSE_SUBEXP = EPSILON_BIT | 1, + OP_ALT = EPSILON_BIT | 2, + OP_DUP_ASTERISK = EPSILON_BIT | 3, + ANCHOR = EPSILON_BIT | 4, + + /* Tree type, these are used only by tree. */ + CONCAT = 16, + SUBEXP = 17, + + /* Token type, these are used only by token. */ + OP_DUP_PLUS = 18, + OP_DUP_QUESTION, + OP_OPEN_BRACKET, + OP_CLOSE_BRACKET, + OP_CHARSET_RANGE, + OP_OPEN_DUP_NUM, + OP_CLOSE_DUP_NUM, + OP_NON_MATCH_LIST, + OP_OPEN_COLL_ELEM, + OP_CLOSE_COLL_ELEM, + OP_OPEN_EQUIV_CLASS, + OP_CLOSE_EQUIV_CLASS, + OP_OPEN_CHAR_CLASS, + OP_CLOSE_CHAR_CLASS, + OP_WORD, + OP_NOTWORD, + OP_SPACE, + OP_NOTSPACE, + BACK_SLASH + +} re_token_type_t; + +#ifdef RE_ENABLE_I18N +typedef struct +{ + /* Multibyte characters. */ + wchar_t *mbchars; + + /* Collating symbols. */ +# ifdef _LIBC + int32_t *coll_syms; +# endif + + /* Equivalence classes. */ +# ifdef _LIBC + int32_t *equiv_classes; +# endif + + /* Range expressions. */ +# ifdef _LIBC + uint32_t *range_starts; + uint32_t *range_ends; +# else /* not _LIBC */ + wchar_t *range_starts; + wchar_t *range_ends; +# endif /* not _LIBC */ + + /* Character classes. */ + wctype_t *char_classes; + + /* If this character set is the non-matching list. */ + unsigned int non_match : 1; + + /* # of multibyte characters. */ + Idx nmbchars; + + /* # of collating symbols. */ + Idx ncoll_syms; + + /* # of equivalence classes. */ + Idx nequiv_classes; + + /* # of range expressions. */ + Idx nranges; + + /* # of character classes. */ + Idx nchar_classes; +} re_charset_t; +#endif /* RE_ENABLE_I18N */ + +typedef struct +{ + union + { + unsigned char c; /* for CHARACTER */ + re_bitset_ptr_t sbcset; /* for SIMPLE_BRACKET */ +#ifdef RE_ENABLE_I18N + re_charset_t *mbcset; /* for COMPLEX_BRACKET */ +#endif /* RE_ENABLE_I18N */ + Idx idx; /* for BACK_REF */ + re_context_type ctx_type; /* for ANCHOR */ + } opr; +#if __GNUC__ >= 2 && !defined __STRICT_ANSI__ + re_token_type_t type : 8; +#else + re_token_type_t type; +#endif + unsigned int constraint : 10; /* context constraint */ + unsigned int duplicated : 1; + unsigned int opt_subexp : 1; +#ifdef RE_ENABLE_I18N + unsigned int accept_mb : 1; + /* These 2 bits can be moved into the union if needed (e.g. if running out + of bits; move opr.c to opr.c.c and move the flags to opr.c.flags). */ + unsigned int mb_partial : 1; +#endif + unsigned int word_char : 1; +} re_token_t; + +#define IS_EPSILON_NODE(type) ((type) & EPSILON_BIT) + +struct re_string_t +{ + /* Indicate the raw buffer which is the original string passed as an + argument of regexec(), re_search(), etc.. */ + const unsigned char *raw_mbs; + /* Store the multibyte string. In case of "case insensitive mode" like + REG_ICASE, upper cases of the string are stored, otherwise MBS points + the same address that RAW_MBS points. */ + unsigned char *mbs; +#ifdef RE_ENABLE_I18N + /* Store the wide character string which is corresponding to MBS. */ + wint_t *wcs; + Idx *offsets; + mbstate_t cur_state; +#endif + /* Index in RAW_MBS. Each character mbs[i] corresponds to + raw_mbs[raw_mbs_idx + i]. */ + Idx raw_mbs_idx; + /* The length of the valid characters in the buffers. */ + Idx valid_len; + /* The corresponding number of bytes in raw_mbs array. */ + Idx valid_raw_len; + /* The length of the buffers MBS and WCS. */ + Idx bufs_len; + /* The index in MBS, which is updated by re_string_fetch_byte. */ + Idx cur_idx; + /* length of RAW_MBS array. */ + Idx raw_len; + /* This is RAW_LEN - RAW_MBS_IDX + VALID_LEN - VALID_RAW_LEN. */ + Idx len; + /* End of the buffer may be shorter than its length in the cases such + as re_match_2, re_search_2. Then, we use STOP for end of the buffer + instead of LEN. */ + Idx raw_stop; + /* This is RAW_STOP - RAW_MBS_IDX adjusted through OFFSETS. */ + Idx stop; + + /* The context of mbs[0]. We store the context independently, since + the context of mbs[0] may be different from raw_mbs[0], which is + the beginning of the input string. */ + unsigned int tip_context; + /* The translation passed as a part of an argument of re_compile_pattern. */ + RE_TRANSLATE_TYPE trans; + /* Copy of re_dfa_t's word_char. */ + re_const_bitset_ptr_t word_char; + /* true if REG_ICASE. */ + unsigned char icase; + unsigned char is_utf8; + unsigned char map_notascii; + unsigned char mbs_allocated; + unsigned char offsets_needed; + unsigned char newline_anchor; + unsigned char word_ops_used; + int mb_cur_max; +}; +typedef struct re_string_t re_string_t; + + +struct re_dfa_t; +typedef struct re_dfa_t re_dfa_t; + +#ifndef _LIBC +# define internal_function +#endif + +static reg_errcode_t re_string_realloc_buffers (re_string_t *pstr, + Idx new_buf_len) + internal_function; +#ifdef RE_ENABLE_I18N +static void build_wcs_buffer (re_string_t *pstr) internal_function; +static reg_errcode_t build_wcs_upper_buffer (re_string_t *pstr) + internal_function; +#endif /* RE_ENABLE_I18N */ +static void build_upper_buffer (re_string_t *pstr) internal_function; +static void re_string_translate_buffer (re_string_t *pstr) internal_function; +static unsigned int re_string_context_at (const re_string_t *input, Idx idx, + int eflags) + internal_function __attribute ((pure)); +#define re_string_peek_byte(pstr, offset) \ + ((pstr)->mbs[(pstr)->cur_idx + offset]) +#define re_string_fetch_byte(pstr) \ + ((pstr)->mbs[(pstr)->cur_idx++]) +#define re_string_first_byte(pstr, idx) \ + ((idx) == (pstr)->valid_len || (pstr)->wcs[idx] != WEOF) +#define re_string_is_single_byte_char(pstr, idx) \ + ((pstr)->wcs[idx] != WEOF && ((pstr)->valid_len == (idx) + 1 \ + || (pstr)->wcs[(idx) + 1] != WEOF)) +#define re_string_eoi(pstr) ((pstr)->stop <= (pstr)->cur_idx) +#define re_string_cur_idx(pstr) ((pstr)->cur_idx) +#define re_string_get_buffer(pstr) ((pstr)->mbs) +#define re_string_length(pstr) ((pstr)->len) +#define re_string_byte_at(pstr,idx) ((pstr)->mbs[idx]) +#define re_string_skip_bytes(pstr,idx) ((pstr)->cur_idx += (idx)) +#define re_string_set_index(pstr,idx) ((pstr)->cur_idx = (idx)) + +#include <alloca.h> + +#ifndef _LIBC +# if HAVE_ALLOCA +/* The OS usually guarantees only one guard page at the bottom of the stack, + and a page size can be as small as 4096 bytes. So we cannot safely + allocate anything larger than 4096 bytes. Also care for the possibility + of a few compiler-allocated temporary stack slots. */ +# define __libc_use_alloca(n) ((n) < 4032) +# else +/* alloca is implemented with malloc, so just use malloc. */ +# define __libc_use_alloca(n) 0 +# undef alloca +# define alloca(n) malloc (n) +# endif +#endif + +#ifndef MAX +# define MAX(a,b) ((a) < (b) ? (b) : (a)) +#endif + +#define re_malloc(t,n) ((t *) malloc ((n) * sizeof (t))) +#define re_realloc(p,t,n) ((t *) realloc (p, (n) * sizeof (t))) +#define re_free(p) free (p) + +struct bin_tree_t +{ + struct bin_tree_t *parent; + struct bin_tree_t *left; + struct bin_tree_t *right; + struct bin_tree_t *first; + struct bin_tree_t *next; + + re_token_t token; + + /* 'node_idx' is the index in dfa->nodes, if 'type' == 0. + Otherwise 'type' indicate the type of this node. */ + Idx node_idx; +}; +typedef struct bin_tree_t bin_tree_t; + +#define BIN_TREE_STORAGE_SIZE \ + ((1024 - sizeof (void *)) / sizeof (bin_tree_t)) + +struct bin_tree_storage_t +{ + struct bin_tree_storage_t *next; + bin_tree_t data[BIN_TREE_STORAGE_SIZE]; +}; +typedef struct bin_tree_storage_t bin_tree_storage_t; + +#define CONTEXT_WORD 1 +#define CONTEXT_NEWLINE (CONTEXT_WORD << 1) +#define CONTEXT_BEGBUF (CONTEXT_NEWLINE << 1) +#define CONTEXT_ENDBUF (CONTEXT_BEGBUF << 1) + +#define IS_WORD_CONTEXT(c) ((c) & CONTEXT_WORD) +#define IS_NEWLINE_CONTEXT(c) ((c) & CONTEXT_NEWLINE) +#define IS_BEGBUF_CONTEXT(c) ((c) & CONTEXT_BEGBUF) +#define IS_ENDBUF_CONTEXT(c) ((c) & CONTEXT_ENDBUF) +#define IS_ORDINARY_CONTEXT(c) ((c) == 0) + +#define IS_WORD_CHAR(ch) (isalnum (ch) || (ch) == '_') +#define IS_NEWLINE(ch) ((ch) == NEWLINE_CHAR) +#define IS_WIDE_WORD_CHAR(ch) (iswalnum (ch) || (ch) == L'_') +#define IS_WIDE_NEWLINE(ch) ((ch) == WIDE_NEWLINE_CHAR) + +#define NOT_SATISFY_PREV_CONSTRAINT(constraint,context) \ + ((((constraint) & PREV_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ + || ((constraint & PREV_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ + || ((constraint & PREV_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context))\ + || ((constraint & PREV_BEGBUF_CONSTRAINT) && !IS_BEGBUF_CONTEXT (context))) + +#define NOT_SATISFY_NEXT_CONSTRAINT(constraint,context) \ + ((((constraint) & NEXT_WORD_CONSTRAINT) && !IS_WORD_CONTEXT (context)) \ + || (((constraint) & NEXT_NOTWORD_CONSTRAINT) && IS_WORD_CONTEXT (context)) \ + || (((constraint) & NEXT_NEWLINE_CONSTRAINT) && !IS_NEWLINE_CONTEXT (context)) \ + || (((constraint) & NEXT_ENDBUF_CONSTRAINT) && !IS_ENDBUF_CONTEXT (context))) + +struct re_dfastate_t +{ + re_hashval_t hash; + re_node_set nodes; + re_node_set non_eps_nodes; + re_node_set inveclosure; + re_node_set *entrance_nodes; + struct re_dfastate_t **trtable, **word_trtable; + unsigned int context : 4; + unsigned int halt : 1; + /* If this state can accept "multi byte". + Note that we refer to multibyte characters, and multi character + collating elements as "multi byte". */ + unsigned int accept_mb : 1; + /* If this state has backreference node(s). */ + unsigned int has_backref : 1; + unsigned int has_constraint : 1; +}; +typedef struct re_dfastate_t re_dfastate_t; + +struct re_state_table_entry +{ + Idx num; + Idx alloc; + re_dfastate_t **array; +}; + +/* Array type used in re_sub_match_last_t and re_sub_match_top_t. */ + +typedef struct +{ + Idx next_idx; + Idx alloc; + re_dfastate_t **array; +} state_array_t; + +/* Store information about the node NODE whose type is OP_CLOSE_SUBEXP. */ + +typedef struct +{ + Idx node; + Idx str_idx; /* The position NODE match at. */ + state_array_t path; +} re_sub_match_last_t; + +/* Store information about the node NODE whose type is OP_OPEN_SUBEXP. + And information about the node, whose type is OP_CLOSE_SUBEXP, + corresponding to NODE is stored in LASTS. */ + +typedef struct +{ + Idx str_idx; + Idx node; + state_array_t *path; + Idx alasts; /* Allocation size of LASTS. */ + Idx nlasts; /* The number of LASTS. */ + re_sub_match_last_t **lasts; +} re_sub_match_top_t; + +struct re_backref_cache_entry +{ + Idx node; + Idx str_idx; + Idx subexp_from; + Idx subexp_to; + char more; + char unused; + unsigned short int eps_reachable_subexps_map; +}; + +typedef struct +{ + /* The string object corresponding to the input string. */ + re_string_t input; +#if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L) + const re_dfa_t *const dfa; +#else + const re_dfa_t *dfa; +#endif + /* EFLAGS of the argument of regexec. */ + int eflags; + /* Where the matching ends. */ + Idx match_last; + Idx last_node; + /* The state log used by the matcher. */ + re_dfastate_t **state_log; + Idx state_log_top; + /* Back reference cache. */ + Idx nbkref_ents; + Idx abkref_ents; + struct re_backref_cache_entry *bkref_ents; + int max_mb_elem_len; + Idx nsub_tops; + Idx asub_tops; + re_sub_match_top_t **sub_tops; +} re_match_context_t; + +typedef struct +{ + re_dfastate_t **sifted_states; + re_dfastate_t **limited_states; + Idx last_node; + Idx last_str_idx; + re_node_set limits; +} re_sift_context_t; + +struct re_fail_stack_ent_t +{ + Idx idx; + Idx node; + regmatch_t *regs; + re_node_set eps_via_nodes; +}; + +struct re_fail_stack_t +{ + Idx num; + Idx alloc; + struct re_fail_stack_ent_t *stack; +}; + +struct re_dfa_t +{ + re_token_t *nodes; + size_t nodes_alloc; + size_t nodes_len; + Idx *nexts; + Idx *org_indices; + re_node_set *edests; + re_node_set *eclosures; + re_node_set *inveclosures; + struct re_state_table_entry *state_table; + re_dfastate_t *init_state; + re_dfastate_t *init_state_word; + re_dfastate_t *init_state_nl; + re_dfastate_t *init_state_begbuf; + bin_tree_t *str_tree; + bin_tree_storage_t *str_tree_storage; + re_bitset_ptr_t sb_char; + int str_tree_storage_idx; + + /* number of subexpressions 're_nsub' is in regex_t. */ + re_hashval_t state_hash_mask; + Idx init_node; + Idx nbackref; /* The number of backreference in this dfa. */ + + /* Bitmap expressing which backreference is used. */ + bitset_word_t used_bkref_map; + bitset_word_t completed_bkref_map; + + unsigned int has_plural_match : 1; + /* If this dfa has "multibyte node", which is a backreference or + a node which can accept multibyte character or multi character + collating element. */ + unsigned int has_mb_node : 1; + unsigned int is_utf8 : 1; + unsigned int map_notascii : 1; + unsigned int word_ops_used : 1; + int mb_cur_max; + bitset_t word_char; + reg_syntax_t syntax; + Idx *subexp_map; +#ifdef DEBUG + char* re_str; +#endif +#ifdef _LIBC + __libc_lock_define (, lock) +#endif +}; + +#define re_node_set_init_empty(set) memset (set, '\0', sizeof (re_node_set)) +#define re_node_set_remove(set,id) \ + (re_node_set_remove_at (set, re_node_set_contains (set, id) - 1)) +#define re_node_set_empty(p) ((p)->nelem = 0) +#define re_node_set_free(set) re_free ((set)->elems) + + +typedef enum +{ + SB_CHAR, + MB_CHAR, + EQUIV_CLASS, + COLL_SYM, + CHAR_CLASS +} bracket_elem_type; + +typedef struct +{ + bracket_elem_type type; + union + { + unsigned char ch; + unsigned char *name; + wchar_t wch; + } opr; +} bracket_elem_t; + + +/* Inline functions for bitset_t operation. */ + +static inline void +bitset_set (bitset_t set, Idx i) +{ + set[i / BITSET_WORD_BITS] |= (bitset_word_t) 1 << i % BITSET_WORD_BITS; +} + +static inline void +bitset_clear (bitset_t set, Idx i) +{ + set[i / BITSET_WORD_BITS] &= ~ ((bitset_word_t) 1 << i % BITSET_WORD_BITS); +} + +static inline bool +bitset_contain (const bitset_t set, Idx i) +{ + return (set[i / BITSET_WORD_BITS] >> i % BITSET_WORD_BITS) & 1; +} + +static inline void +bitset_empty (bitset_t set) +{ + memset (set, '\0', sizeof (bitset_t)); +} + +static inline void +bitset_set_all (bitset_t set) +{ + memset (set, -1, sizeof (bitset_word_t) * (SBC_MAX / BITSET_WORD_BITS)); + if (SBC_MAX % BITSET_WORD_BITS != 0) + set[BITSET_WORDS - 1] = + ((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1; +} + +static inline void +bitset_copy (bitset_t dest, const bitset_t src) +{ + memcpy (dest, src, sizeof (bitset_t)); +} + +static inline void +bitset_not (bitset_t set) +{ + int bitset_i; + for (bitset_i = 0; bitset_i < SBC_MAX / BITSET_WORD_BITS; ++bitset_i) + set[bitset_i] = ~set[bitset_i]; + if (SBC_MAX % BITSET_WORD_BITS != 0) + set[BITSET_WORDS - 1] = + ((((bitset_word_t) 1 << SBC_MAX % BITSET_WORD_BITS) - 1) + & ~set[BITSET_WORDS - 1]); +} + +static inline void +bitset_merge (bitset_t dest, const bitset_t src) +{ + int bitset_i; + for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) + dest[bitset_i] |= src[bitset_i]; +} + +static inline void +bitset_mask (bitset_t dest, const bitset_t src) +{ + int bitset_i; + for (bitset_i = 0; bitset_i < BITSET_WORDS; ++bitset_i) + dest[bitset_i] &= src[bitset_i]; +} + +#ifdef RE_ENABLE_I18N +/* Inline functions for re_string. */ +static inline int +internal_function __attribute ((pure)) +re_string_char_size_at (const re_string_t *pstr, Idx idx) +{ + int byte_idx; + if (pstr->mb_cur_max == 1) + return 1; + for (byte_idx = 1; idx + byte_idx < pstr->valid_len; ++byte_idx) + if (pstr->wcs[idx + byte_idx] != WEOF) + break; + return byte_idx; +} + +static inline wint_t +internal_function __attribute ((pure)) +re_string_wchar_at (const re_string_t *pstr, Idx idx) +{ + if (pstr->mb_cur_max == 1) + return (wint_t) pstr->mbs[idx]; + return (wint_t) pstr->wcs[idx]; +} + +static int +internal_function __attribute ((pure)) +re_string_elem_size_at (const re_string_t *pstr, Idx idx) +{ +# ifdef _LIBC + const unsigned char *p, *extra; + const int32_t *table, *indirect; + int32_t tmp; +# include <locale/weight.h> + uint_fast32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); + + if (nrules != 0) + { + table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); + extra = (const unsigned char *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB); + indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, + _NL_COLLATE_INDIRECTMB); + p = pstr->mbs + idx; + tmp = findidx (&p); + return p - pstr->mbs - idx; + } + else +# endif /* _LIBC */ + return 1; +} +#endif /* RE_ENABLE_I18N */ + +#ifndef __GNUC_PREREQ +# if defined __GNUC__ && defined __GNUC_MINOR__ +# define __GNUC_PREREQ(maj, min) \ + ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min)) +# else +# define __GNUC_PREREQ(maj, min) 0 +# endif +#endif + +#if __GNUC_PREREQ (3,4) +# undef __attribute_warn_unused_result__ +# define __attribute_warn_unused_result__ \ + __attribute__ ((__warn_unused_result__)) +#else +# define __attribute_warn_unused_result__ /* empty */ +#endif + +#endif /* _REGEX_INTERNAL_H */ diff --git a/lib/regexec.c b/lib/regexec.c new file mode 100644 index 000000000..7130f2a92 --- /dev/null +++ b/lib/regexec.c @@ -0,0 +1,4417 @@ +/* Extended regular expression matching and search library. + Copyright (C) 2002-2012 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2, or (at your option) + any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License along + with this program; if not, write to the Free Software Foundation, + Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ + +static reg_errcode_t match_ctx_init (re_match_context_t *cache, int eflags, + Idx n) internal_function; +static void match_ctx_clean (re_match_context_t *mctx) internal_function; +static void match_ctx_free (re_match_context_t *cache) internal_function; +static reg_errcode_t match_ctx_add_entry (re_match_context_t *cache, Idx node, + Idx str_idx, Idx from, Idx to) + internal_function; +static Idx search_cur_bkref_entry (const re_match_context_t *mctx, Idx str_idx) + internal_function; +static reg_errcode_t match_ctx_add_subtop (re_match_context_t *mctx, Idx node, + Idx str_idx) internal_function; +static re_sub_match_last_t * match_ctx_add_sublast (re_sub_match_top_t *subtop, + Idx node, Idx str_idx) + internal_function; +static void sift_ctx_init (re_sift_context_t *sctx, re_dfastate_t **sifted_sts, + re_dfastate_t **limited_sts, Idx last_node, + Idx last_str_idx) + internal_function; +static reg_errcode_t re_search_internal (const regex_t *preg, + const char *string, Idx length, + Idx start, Idx last_start, Idx stop, + size_t nmatch, regmatch_t pmatch[], + int eflags) internal_function; +static regoff_t re_search_2_stub (struct re_pattern_buffer *bufp, + const char *string1, Idx length1, + const char *string2, Idx length2, + Idx start, regoff_t range, + struct re_registers *regs, + Idx stop, bool ret_len) internal_function; +static regoff_t re_search_stub (struct re_pattern_buffer *bufp, + const char *string, Idx length, Idx start, + regoff_t range, Idx stop, + struct re_registers *regs, + bool ret_len) internal_function; +static unsigned int re_copy_regs (struct re_registers *regs, regmatch_t *pmatch, + Idx nregs, int regs_allocated) + internal_function; +static reg_errcode_t prune_impossible_nodes (re_match_context_t *mctx) + internal_function; +static Idx check_matching (re_match_context_t *mctx, bool fl_longest_match, + Idx *p_match_first) internal_function; +static Idx check_halt_state_context (const re_match_context_t *mctx, + const re_dfastate_t *state, Idx idx) + internal_function; +static void update_regs (const re_dfa_t *dfa, regmatch_t *pmatch, + regmatch_t *prev_idx_match, Idx cur_node, + Idx cur_idx, Idx nmatch) internal_function; +static reg_errcode_t push_fail_stack (struct re_fail_stack_t *fs, + Idx str_idx, Idx dest_node, Idx nregs, + regmatch_t *regs, + re_node_set *eps_via_nodes) + internal_function; +static reg_errcode_t set_regs (const regex_t *preg, + const re_match_context_t *mctx, + size_t nmatch, regmatch_t *pmatch, + bool fl_backtrack) internal_function; +static reg_errcode_t free_fail_stack_return (struct re_fail_stack_t *fs) + internal_function; + +#ifdef RE_ENABLE_I18N +static int sift_states_iter_mb (const re_match_context_t *mctx, + re_sift_context_t *sctx, + Idx node_idx, Idx str_idx, Idx max_str_idx) + internal_function; +#endif /* RE_ENABLE_I18N */ +static reg_errcode_t sift_states_backward (const re_match_context_t *mctx, + re_sift_context_t *sctx) + internal_function; +static reg_errcode_t build_sifted_states (const re_match_context_t *mctx, + re_sift_context_t *sctx, Idx str_idx, + re_node_set *cur_dest) + internal_function; +static reg_errcode_t update_cur_sifted_state (const re_match_context_t *mctx, + re_sift_context_t *sctx, + Idx str_idx, + re_node_set *dest_nodes) + internal_function; +static reg_errcode_t add_epsilon_src_nodes (const re_dfa_t *dfa, + re_node_set *dest_nodes, + const re_node_set *candidates) + internal_function; +static bool check_dst_limits (const re_match_context_t *mctx, + const re_node_set *limits, + Idx dst_node, Idx dst_idx, Idx src_node, + Idx src_idx) internal_function; +static int check_dst_limits_calc_pos_1 (const re_match_context_t *mctx, + int boundaries, Idx subexp_idx, + Idx from_node, Idx bkref_idx) + internal_function; +static int check_dst_limits_calc_pos (const re_match_context_t *mctx, + Idx limit, Idx subexp_idx, + Idx node, Idx str_idx, + Idx bkref_idx) internal_function; +static reg_errcode_t check_subexp_limits (const re_dfa_t *dfa, + re_node_set *dest_nodes, + const re_node_set *candidates, + re_node_set *limits, + struct re_backref_cache_entry *bkref_ents, + Idx str_idx) internal_function; +static reg_errcode_t sift_states_bkref (const re_match_context_t *mctx, + re_sift_context_t *sctx, + Idx str_idx, const re_node_set *candidates) + internal_function; +static reg_errcode_t merge_state_array (const re_dfa_t *dfa, + re_dfastate_t **dst, + re_dfastate_t **src, Idx num) + internal_function; +static re_dfastate_t *find_recover_state (reg_errcode_t *err, + re_match_context_t *mctx) internal_function; +static re_dfastate_t *transit_state (reg_errcode_t *err, + re_match_context_t *mctx, + re_dfastate_t *state) internal_function; +static re_dfastate_t *merge_state_with_log (reg_errcode_t *err, + re_match_context_t *mctx, + re_dfastate_t *next_state) + internal_function; +static reg_errcode_t check_subexp_matching_top (re_match_context_t *mctx, + re_node_set *cur_nodes, + Idx str_idx) internal_function; +#if 0 +static re_dfastate_t *transit_state_sb (reg_errcode_t *err, + re_match_context_t *mctx, + re_dfastate_t *pstate) + internal_function; +#endif +#ifdef RE_ENABLE_I18N +static reg_errcode_t transit_state_mb (re_match_context_t *mctx, + re_dfastate_t *pstate) + internal_function; +#endif /* RE_ENABLE_I18N */ +static reg_errcode_t transit_state_bkref (re_match_context_t *mctx, + const re_node_set *nodes) + internal_function; +static reg_errcode_t get_subexp (re_match_context_t *mctx, + Idx bkref_node, Idx bkref_str_idx) + internal_function; +static reg_errcode_t get_subexp_sub (re_match_context_t *mctx, + const re_sub_match_top_t *sub_top, + re_sub_match_last_t *sub_last, + Idx bkref_node, Idx bkref_str) + internal_function; +static Idx find_subexp_node (const re_dfa_t *dfa, const re_node_set *nodes, + Idx subexp_idx, int type) internal_function; +static reg_errcode_t check_arrival (re_match_context_t *mctx, + state_array_t *path, Idx top_node, + Idx top_str, Idx last_node, Idx last_str, + int type) internal_function; +static reg_errcode_t check_arrival_add_next_nodes (re_match_context_t *mctx, + Idx str_idx, + re_node_set *cur_nodes, + re_node_set *next_nodes) + internal_function; +static reg_errcode_t check_arrival_expand_ecl (const re_dfa_t *dfa, + re_node_set *cur_nodes, + Idx ex_subexp, int type) + internal_function; +static reg_errcode_t check_arrival_expand_ecl_sub (const re_dfa_t *dfa, + re_node_set *dst_nodes, + Idx target, Idx ex_subexp, + int type) internal_function; +static reg_errcode_t expand_bkref_cache (re_match_context_t *mctx, + re_node_set *cur_nodes, Idx cur_str, + Idx subexp_num, int type) + internal_function; +static bool build_trtable (const re_dfa_t *dfa, + re_dfastate_t *state) internal_function; +#ifdef RE_ENABLE_I18N +static int check_node_accept_bytes (const re_dfa_t *dfa, Idx node_idx, + const re_string_t *input, Idx idx) + internal_function; +# ifdef _LIBC +static unsigned int find_collation_sequence_value (const unsigned char *mbs, + size_t name_len) + internal_function; +# endif /* _LIBC */ +#endif /* RE_ENABLE_I18N */ +static Idx group_nodes_into_DFAstates (const re_dfa_t *dfa, + const re_dfastate_t *state, + re_node_set *states_node, + bitset_t *states_ch) internal_function; +static bool check_node_accept (const re_match_context_t *mctx, + const re_token_t *node, Idx idx) + internal_function; +static reg_errcode_t extend_buffers (re_match_context_t *mctx) + internal_function; + +/* Entry point for POSIX code. */ + +/* regexec searches for a given pattern, specified by PREG, in the + string STRING. + + If NMATCH is zero or REG_NOSUB was set in the cflags argument to + 'regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at + least NMATCH elements, and we set them to the offsets of the + corresponding matched substrings. + + EFLAGS specifies "execution flags" which affect matching: if + REG_NOTBOL is set, then ^ does not match at the beginning of the + string; if REG_NOTEOL is set, then $ does not match at the end. + + We return 0 if we find a match and REG_NOMATCH if not. */ + +int +regexec (preg, string, nmatch, pmatch, eflags) + const regex_t *_Restrict_ preg; + const char *_Restrict_ string; + size_t nmatch; + regmatch_t pmatch[_Restrict_arr_]; + int eflags; +{ + reg_errcode_t err; + Idx start, length; +#ifdef _LIBC + re_dfa_t *dfa = (re_dfa_t *) preg->buffer; +#endif + + if (eflags & ~(REG_NOTBOL | REG_NOTEOL | REG_STARTEND)) + return REG_BADPAT; + + if (eflags & REG_STARTEND) + { + start = pmatch[0].rm_so; + length = pmatch[0].rm_eo; + } + else + { + start = 0; + length = strlen (string); + } + + __libc_lock_lock (dfa->lock); + if (preg->no_sub) + err = re_search_internal (preg, string, length, start, length, + length, 0, NULL, eflags); + else + err = re_search_internal (preg, string, length, start, length, + length, nmatch, pmatch, eflags); + __libc_lock_unlock (dfa->lock); + return err != REG_NOERROR; +} + +#ifdef _LIBC +# include <shlib-compat.h> +versioned_symbol (libc, __regexec, regexec, GLIBC_2_3_4); + +# if SHLIB_COMPAT (libc, GLIBC_2_0, GLIBC_2_3_4) +__typeof__ (__regexec) __compat_regexec; + +int +attribute_compat_text_section +__compat_regexec (const regex_t *_Restrict_ preg, + const char *_Restrict_ string, size_t nmatch, + regmatch_t pmatch[], int eflags) +{ + return regexec (preg, string, nmatch, pmatch, + eflags & (REG_NOTBOL | REG_NOTEOL)); +} +compat_symbol (libc, __compat_regexec, regexec, GLIBC_2_0); +# endif +#endif + +/* Entry points for GNU code. */ + +/* re_match, re_search, re_match_2, re_search_2 + + The former two functions operate on STRING with length LENGTH, + while the later two operate on concatenation of STRING1 and STRING2 + with lengths LENGTH1 and LENGTH2, respectively. + + re_match() matches the compiled pattern in BUFP against the string, + starting at index START. + + re_search() first tries matching at index START, then it tries to match + starting from index START + 1, and so on. The last start position tried + is START + RANGE. (Thus RANGE = 0 forces re_search to operate the same + way as re_match().) + + The parameter STOP of re_{match,search}_2 specifies that no match exceeding + the first STOP characters of the concatenation of the strings should be + concerned. + + If REGS is not NULL, and BUFP->no_sub is not set, the offsets of the match + and all groups is stored in REGS. (For the "_2" variants, the offsets are + computed relative to the concatenation, not relative to the individual + strings.) + + On success, re_match* functions return the length of the match, re_search* + return the position of the start of the match. Return value -1 means no + match was found and -2 indicates an internal error. */ + +regoff_t +re_match (bufp, string, length, start, regs) + struct re_pattern_buffer *bufp; + const char *string; + Idx length, start; + struct re_registers *regs; +{ + return re_search_stub (bufp, string, length, start, 0, length, regs, true); +} +#ifdef _LIBC +weak_alias (__re_match, re_match) +#endif + +regoff_t +re_search (bufp, string, length, start, range, regs) + struct re_pattern_buffer *bufp; + const char *string; + Idx length, start; + regoff_t range; + struct re_registers *regs; +{ + return re_search_stub (bufp, string, length, start, range, length, regs, + false); +} +#ifdef _LIBC +weak_alias (__re_search, re_search) +#endif + +regoff_t +re_match_2 (bufp, string1, length1, string2, length2, start, regs, stop) + struct re_pattern_buffer *bufp; + const char *string1, *string2; + Idx length1, length2, start, stop; + struct re_registers *regs; +{ + return re_search_2_stub (bufp, string1, length1, string2, length2, + start, 0, regs, stop, true); +} +#ifdef _LIBC +weak_alias (__re_match_2, re_match_2) +#endif + +regoff_t +re_search_2 (bufp, string1, length1, string2, length2, start, range, regs, stop) + struct re_pattern_buffer *bufp; + const char *string1, *string2; + Idx length1, length2, start, stop; + regoff_t range; + struct re_registers *regs; +{ + return re_search_2_stub (bufp, string1, length1, string2, length2, + start, range, regs, stop, false); +} +#ifdef _LIBC +weak_alias (__re_search_2, re_search_2) +#endif + +static regoff_t +internal_function +re_search_2_stub (struct re_pattern_buffer *bufp, + const char *string1, Idx length1, + const char *string2, Idx length2, + Idx start, regoff_t range, struct re_registers *regs, + Idx stop, bool ret_len) +{ + const char *str; + regoff_t rval; + Idx len = length1 + length2; + char *s = NULL; + + if (BE (length1 < 0 || length2 < 0 || stop < 0 || len < length1, 0)) + return -2; + + /* Concatenate the strings. */ + if (length2 > 0) + if (length1 > 0) + { + s = re_malloc (char, len); + + if (BE (s == NULL, 0)) + return -2; +#ifdef _LIBC + memcpy (__mempcpy (s, string1, length1), string2, length2); +#else + memcpy (s, string1, length1); + memcpy (s + length1, string2, length2); +#endif + str = s; + } + else + str = string2; + else + str = string1; + + rval = re_search_stub (bufp, str, len, start, range, stop, regs, + ret_len); + re_free (s); + return rval; +} + +/* The parameters have the same meaning as those of re_search. + Additional parameters: + If RET_LEN is true the length of the match is returned (re_match style); + otherwise the position of the match is returned. */ + +static regoff_t +internal_function +re_search_stub (struct re_pattern_buffer *bufp, + const char *string, Idx length, + Idx start, regoff_t range, Idx stop, struct re_registers *regs, + bool ret_len) +{ + reg_errcode_t result; + regmatch_t *pmatch; + Idx nregs; + regoff_t rval; + int eflags = 0; +#ifdef _LIBC + re_dfa_t *dfa = (re_dfa_t *) bufp->buffer; +#endif + Idx last_start = start + range; + + /* Check for out-of-range. */ + if (BE (start < 0 || start > length, 0)) + return -1; + if (BE (length < last_start || (0 <= range && last_start < start), 0)) + last_start = length; + else if (BE (last_start < 0 || (range < 0 && start <= last_start), 0)) + last_start = 0; + + __libc_lock_lock (dfa->lock); + + eflags |= (bufp->not_bol) ? REG_NOTBOL : 0; + eflags |= (bufp->not_eol) ? REG_NOTEOL : 0; + + /* Compile fastmap if we haven't yet. */ + if (start < last_start && bufp->fastmap != NULL && !bufp->fastmap_accurate) + re_compile_fastmap (bufp); + + if (BE (bufp->no_sub, 0)) + regs = NULL; + + /* We need at least 1 register. */ + if (regs == NULL) + nregs = 1; + else if (BE (bufp->regs_allocated == REGS_FIXED + && regs->num_regs <= bufp->re_nsub, 0)) + { + nregs = regs->num_regs; + if (BE (nregs < 1, 0)) + { + /* Nothing can be copied to regs. */ + regs = NULL; + nregs = 1; + } + } + else + nregs = bufp->re_nsub + 1; + pmatch = re_malloc (regmatch_t, nregs); + if (BE (pmatch == NULL, 0)) + { + rval = -2; + goto out; + } + + result = re_search_internal (bufp, string, length, start, last_start, stop, + nregs, pmatch, eflags); + + rval = 0; + + /* I hope we needn't fill ther regs with -1's when no match was found. */ + if (result != REG_NOERROR) + rval = -1; + else if (regs != NULL) + { + /* If caller wants register contents data back, copy them. */ + bufp->regs_allocated = re_copy_regs (regs, pmatch, nregs, + bufp->regs_allocated); + if (BE (bufp->regs_allocated == REGS_UNALLOCATED, 0)) + rval = -2; + } + + if (BE (rval == 0, 1)) + { + if (ret_len) + { + assert (pmatch[0].rm_so == start); + rval = pmatch[0].rm_eo - start; + } + else + rval = pmatch[0].rm_so; + } + re_free (pmatch); + out: + __libc_lock_unlock (dfa->lock); + return rval; +} + +static unsigned int +internal_function +re_copy_regs (struct re_registers *regs, regmatch_t *pmatch, Idx nregs, + int regs_allocated) +{ + int rval = REGS_REALLOCATE; + Idx i; + Idx need_regs = nregs + 1; + /* We need one extra element beyond 'num_regs' for the '-1' marker GNU code + uses. */ + + /* Have the register data arrays been allocated? */ + if (regs_allocated == REGS_UNALLOCATED) + { /* No. So allocate them with malloc. */ + regs->start = re_malloc (regoff_t, need_regs); + if (BE (regs->start == NULL, 0)) + return REGS_UNALLOCATED; + regs->end = re_malloc (regoff_t, need_regs); + if (BE (regs->end == NULL, 0)) + { + re_free (regs->start); + return REGS_UNALLOCATED; + } + regs->num_regs = need_regs; + } + else if (regs_allocated == REGS_REALLOCATE) + { /* Yes. If we need more elements than were already + allocated, reallocate them. If we need fewer, just + leave it alone. */ + if (BE (need_regs > regs->num_regs, 0)) + { + regoff_t *new_start = re_realloc (regs->start, regoff_t, need_regs); + regoff_t *new_end; + if (BE (new_start == NULL, 0)) + return REGS_UNALLOCATED; + new_end = re_realloc (regs->end, regoff_t, need_regs); + if (BE (new_end == NULL, 0)) + { + re_free (new_start); + return REGS_UNALLOCATED; + } + regs->start = new_start; + regs->end = new_end; + regs->num_regs = need_regs; + } + } + else + { + assert (regs_allocated == REGS_FIXED); + /* This function may not be called with REGS_FIXED and nregs too big. */ + assert (regs->num_regs >= nregs); + rval = REGS_FIXED; + } + + /* Copy the regs. */ + for (i = 0; i < nregs; ++i) + { + regs->start[i] = pmatch[i].rm_so; + regs->end[i] = pmatch[i].rm_eo; + } + for ( ; i < regs->num_regs; ++i) + regs->start[i] = regs->end[i] = -1; + + return rval; +} + +/* Set REGS to hold NUM_REGS registers, storing them in STARTS and + ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use + this memory for recording register information. STARTS and ENDS + must be allocated using the malloc library routine, and must each + be at least NUM_REGS * sizeof (regoff_t) bytes long. + + If NUM_REGS == 0, then subsequent matches should allocate their own + register data. + + Unless this function is called, the first search or match using + PATTERN_BUFFER will allocate its own register data, without + freeing the old data. */ + +void +re_set_registers (bufp, regs, num_regs, starts, ends) + struct re_pattern_buffer *bufp; + struct re_registers *regs; + __re_size_t num_regs; + regoff_t *starts, *ends; +{ + if (num_regs) + { + bufp->regs_allocated = REGS_REALLOCATE; + regs->num_regs = num_regs; + regs->start = starts; + regs->end = ends; + } + else + { + bufp->regs_allocated = REGS_UNALLOCATED; + regs->num_regs = 0; + regs->start = regs->end = NULL; + } +} +#ifdef _LIBC +weak_alias (__re_set_registers, re_set_registers) +#endif + +/* Entry points compatible with 4.2 BSD regex library. We don't define + them unless specifically requested. */ + +#if defined _REGEX_RE_COMP || defined _LIBC +int +# ifdef _LIBC +weak_function +# endif +re_exec (s) + const char *s; +{ + return 0 == regexec (&re_comp_buf, s, 0, NULL, 0); +} +#endif /* _REGEX_RE_COMP */ + +/* Internal entry point. */ + +/* Searches for a compiled pattern PREG in the string STRING, whose + length is LENGTH. NMATCH, PMATCH, and EFLAGS have the same + meaning as with regexec. LAST_START is START + RANGE, where + START and RANGE have the same meaning as with re_search. + Return REG_NOERROR if we find a match, and REG_NOMATCH if not, + otherwise return the error code. + Note: We assume front end functions already check ranges. + (0 <= LAST_START && LAST_START <= LENGTH) */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +re_search_internal (const regex_t *preg, + const char *string, Idx length, + Idx start, Idx last_start, Idx stop, + size_t nmatch, regmatch_t pmatch[], + int eflags) +{ + reg_errcode_t err; + const re_dfa_t *dfa = (const re_dfa_t *) preg->buffer; + Idx left_lim, right_lim; + int incr; + bool fl_longest_match; + int match_kind; + Idx match_first; + Idx match_last = REG_MISSING; + Idx extra_nmatch; + bool sb; + int ch; +#if defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L) + re_match_context_t mctx = { .dfa = dfa }; +#else + re_match_context_t mctx; +#endif + char *fastmap = ((preg->fastmap != NULL && preg->fastmap_accurate + && start != last_start && !preg->can_be_null) + ? preg->fastmap : NULL); + RE_TRANSLATE_TYPE t = preg->translate; + +#if !(defined _LIBC || (defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L)) + memset (&mctx, '\0', sizeof (re_match_context_t)); + mctx.dfa = dfa; +#endif + + extra_nmatch = (nmatch > preg->re_nsub) ? nmatch - (preg->re_nsub + 1) : 0; + nmatch -= extra_nmatch; + + /* Check if the DFA haven't been compiled. */ + if (BE (preg->used == 0 || dfa->init_state == NULL + || dfa->init_state_word == NULL || dfa->init_state_nl == NULL + || dfa->init_state_begbuf == NULL, 0)) + return REG_NOMATCH; + +#ifdef DEBUG + /* We assume front-end functions already check them. */ + assert (0 <= last_start && last_start <= length); +#endif + + /* If initial states with non-begbuf contexts have no elements, + the regex must be anchored. If preg->newline_anchor is set, + we'll never use init_state_nl, so do not check it. */ + if (dfa->init_state->nodes.nelem == 0 + && dfa->init_state_word->nodes.nelem == 0 + && (dfa->init_state_nl->nodes.nelem == 0 + || !preg->newline_anchor)) + { + if (start != 0 && last_start != 0) + return REG_NOMATCH; + start = last_start = 0; + } + + /* We must check the longest matching, if nmatch > 0. */ + fl_longest_match = (nmatch != 0 || dfa->nbackref); + + err = re_string_allocate (&mctx.input, string, length, dfa->nodes_len + 1, + preg->translate, (preg->syntax & RE_ICASE) != 0, + dfa); + if (BE (err != REG_NOERROR, 0)) + goto free_return; + mctx.input.stop = stop; + mctx.input.raw_stop = stop; + mctx.input.newline_anchor = preg->newline_anchor; + + err = match_ctx_init (&mctx, eflags, dfa->nbackref * 2); + if (BE (err != REG_NOERROR, 0)) + goto free_return; + + /* We will log all the DFA states through which the dfa pass, + if nmatch > 1, or this dfa has "multibyte node", which is a + back-reference or a node which can accept multibyte character or + multi character collating element. */ + if (nmatch > 1 || dfa->has_mb_node) + { + /* Avoid overflow. */ + if (BE (SIZE_MAX / sizeof (re_dfastate_t *) <= mctx.input.bufs_len, 0)) + { + err = REG_ESPACE; + goto free_return; + } + + mctx.state_log = re_malloc (re_dfastate_t *, mctx.input.bufs_len + 1); + if (BE (mctx.state_log == NULL, 0)) + { + err = REG_ESPACE; + goto free_return; + } + } + else + mctx.state_log = NULL; + + match_first = start; + mctx.input.tip_context = (eflags & REG_NOTBOL) ? CONTEXT_BEGBUF + : CONTEXT_NEWLINE | CONTEXT_BEGBUF; + + /* Check incrementally whether of not the input string match. */ + incr = (last_start < start) ? -1 : 1; + left_lim = (last_start < start) ? last_start : start; + right_lim = (last_start < start) ? start : last_start; + sb = dfa->mb_cur_max == 1; + match_kind = + (fastmap + ? ((sb || !(preg->syntax & RE_ICASE || t) ? 4 : 0) + | (start <= last_start ? 2 : 0) + | (t != NULL ? 1 : 0)) + : 8); + + for (;; match_first += incr) + { + err = REG_NOMATCH; + if (match_first < left_lim || right_lim < match_first) + goto free_return; + + /* Advance as rapidly as possible through the string, until we + find a plausible place to start matching. This may be done + with varying efficiency, so there are various possibilities: + only the most common of them are specialized, in order to + save on code size. We use a switch statement for speed. */ + switch (match_kind) + { + case 8: + /* No fastmap. */ + break; + + case 7: + /* Fastmap with single-byte translation, match forward. */ + while (BE (match_first < right_lim, 1) + && !fastmap[t[(unsigned char) string[match_first]]]) + ++match_first; + goto forward_match_found_start_or_reached_end; + + case 6: + /* Fastmap without translation, match forward. */ + while (BE (match_first < right_lim, 1) + && !fastmap[(unsigned char) string[match_first]]) + ++match_first; + + forward_match_found_start_or_reached_end: + if (BE (match_first == right_lim, 0)) + { + ch = match_first >= length + ? 0 : (unsigned char) string[match_first]; + if (!fastmap[t ? t[ch] : ch]) + goto free_return; + } + break; + + case 4: + case 5: + /* Fastmap without multi-byte translation, match backwards. */ + while (match_first >= left_lim) + { + ch = match_first >= length + ? 0 : (unsigned char) string[match_first]; + if (fastmap[t ? t[ch] : ch]) + break; + --match_first; + } + if (match_first < left_lim) + goto free_return; + break; + + default: + /* In this case, we can't determine easily the current byte, + since it might be a component byte of a multibyte + character. Then we use the constructed buffer instead. */ + for (;;) + { + /* If MATCH_FIRST is out of the valid range, reconstruct the + buffers. */ + __re_size_t offset = match_first - mctx.input.raw_mbs_idx; + if (BE (offset >= (__re_size_t) mctx.input.valid_raw_len, 0)) + { + err = re_string_reconstruct (&mctx.input, match_first, + eflags); + if (BE (err != REG_NOERROR, 0)) + goto free_return; + + offset = match_first - mctx.input.raw_mbs_idx; + } + /* If MATCH_FIRST is out of the buffer, leave it as '\0'. + Note that MATCH_FIRST must not be smaller than 0. */ + ch = (match_first >= length + ? 0 : re_string_byte_at (&mctx.input, offset)); + if (fastmap[ch]) + break; + match_first += incr; + if (match_first < left_lim || match_first > right_lim) + { + err = REG_NOMATCH; + goto free_return; + } + } + break; + } + + /* Reconstruct the buffers so that the matcher can assume that + the matching starts from the beginning of the buffer. */ + err = re_string_reconstruct (&mctx.input, match_first, eflags); + if (BE (err != REG_NOERROR, 0)) + goto free_return; + +#ifdef RE_ENABLE_I18N + /* Don't consider this char as a possible match start if it part, + yet isn't the head, of a multibyte character. */ + if (!sb && !re_string_first_byte (&mctx.input, 0)) + continue; +#endif + + /* It seems to be appropriate one, then use the matcher. */ + /* We assume that the matching starts from 0. */ + mctx.state_log_top = mctx.nbkref_ents = mctx.max_mb_elem_len = 0; + match_last = check_matching (&mctx, fl_longest_match, + start <= last_start ? &match_first : NULL); + if (match_last != REG_MISSING) + { + if (BE (match_last == REG_ERROR, 0)) + { + err = REG_ESPACE; + goto free_return; + } + else + { + mctx.match_last = match_last; + if ((!preg->no_sub && nmatch > 1) || dfa->nbackref) + { + re_dfastate_t *pstate = mctx.state_log[match_last]; + mctx.last_node = check_halt_state_context (&mctx, pstate, + match_last); + } + if ((!preg->no_sub && nmatch > 1 && dfa->has_plural_match) + || dfa->nbackref) + { + err = prune_impossible_nodes (&mctx); + if (err == REG_NOERROR) + break; + if (BE (err != REG_NOMATCH, 0)) + goto free_return; + match_last = REG_MISSING; + } + else + break; /* We found a match. */ + } + } + + match_ctx_clean (&mctx); + } + +#ifdef DEBUG + assert (match_last != REG_MISSING); + assert (err == REG_NOERROR); +#endif + + /* Set pmatch[] if we need. */ + if (nmatch > 0) + { + Idx reg_idx; + + /* Initialize registers. */ + for (reg_idx = 1; reg_idx < nmatch; ++reg_idx) + pmatch[reg_idx].rm_so = pmatch[reg_idx].rm_eo = -1; + + /* Set the points where matching start/end. */ + pmatch[0].rm_so = 0; + pmatch[0].rm_eo = mctx.match_last; + /* FIXME: This function should fail if mctx.match_last exceeds + the maximum possible regoff_t value. We need a new error + code REG_OVERFLOW. */ + + if (!preg->no_sub && nmatch > 1) + { + err = set_regs (preg, &mctx, nmatch, pmatch, + dfa->has_plural_match && dfa->nbackref > 0); + if (BE (err != REG_NOERROR, 0)) + goto free_return; + } + + /* At last, add the offset to the each registers, since we slided + the buffers so that we could assume that the matching starts + from 0. */ + for (reg_idx = 0; reg_idx < nmatch; ++reg_idx) + if (pmatch[reg_idx].rm_so != -1) + { +#ifdef RE_ENABLE_I18N + if (BE (mctx.input.offsets_needed != 0, 0)) + { + pmatch[reg_idx].rm_so = + (pmatch[reg_idx].rm_so == mctx.input.valid_len + ? mctx.input.valid_raw_len + : mctx.input.offsets[pmatch[reg_idx].rm_so]); + pmatch[reg_idx].rm_eo = + (pmatch[reg_idx].rm_eo == mctx.input.valid_len + ? mctx.input.valid_raw_len + : mctx.input.offsets[pmatch[reg_idx].rm_eo]); + } +#else + assert (mctx.input.offsets_needed == 0); +#endif + pmatch[reg_idx].rm_so += match_first; + pmatch[reg_idx].rm_eo += match_first; + } + for (reg_idx = 0; reg_idx < extra_nmatch; ++reg_idx) + { + pmatch[nmatch + reg_idx].rm_so = -1; + pmatch[nmatch + reg_idx].rm_eo = -1; + } + + if (dfa->subexp_map) + for (reg_idx = 0; reg_idx + 1 < nmatch; reg_idx++) + if (dfa->subexp_map[reg_idx] != reg_idx) + { + pmatch[reg_idx + 1].rm_so + = pmatch[dfa->subexp_map[reg_idx] + 1].rm_so; + pmatch[reg_idx + 1].rm_eo + = pmatch[dfa->subexp_map[reg_idx] + 1].rm_eo; + } + } + + free_return: + re_free (mctx.state_log); + if (dfa->nbackref) + match_ctx_free (&mctx); + re_string_destruct (&mctx.input); + return err; +} + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +prune_impossible_nodes (re_match_context_t *mctx) +{ + const re_dfa_t *const dfa = mctx->dfa; + Idx halt_node, match_last; + reg_errcode_t ret; + re_dfastate_t **sifted_states; + re_dfastate_t **lim_states = NULL; + re_sift_context_t sctx; +#ifdef DEBUG + assert (mctx->state_log != NULL); +#endif + match_last = mctx->match_last; + halt_node = mctx->last_node; + + /* Avoid overflow. */ + if (BE (SIZE_MAX / sizeof (re_dfastate_t *) <= match_last, 0)) + return REG_ESPACE; + + sifted_states = re_malloc (re_dfastate_t *, match_last + 1); + if (BE (sifted_states == NULL, 0)) + { + ret = REG_ESPACE; + goto free_return; + } + if (dfa->nbackref) + { + lim_states = re_malloc (re_dfastate_t *, match_last + 1); + if (BE (lim_states == NULL, 0)) + { + ret = REG_ESPACE; + goto free_return; + } + while (1) + { + memset (lim_states, '\0', + sizeof (re_dfastate_t *) * (match_last + 1)); + sift_ctx_init (&sctx, sifted_states, lim_states, halt_node, + match_last); + ret = sift_states_backward (mctx, &sctx); + re_node_set_free (&sctx.limits); + if (BE (ret != REG_NOERROR, 0)) + goto free_return; + if (sifted_states[0] != NULL || lim_states[0] != NULL) + break; + do + { + --match_last; + if (! REG_VALID_INDEX (match_last)) + { + ret = REG_NOMATCH; + goto free_return; + } + } while (mctx->state_log[match_last] == NULL + || !mctx->state_log[match_last]->halt); + halt_node = check_halt_state_context (mctx, + mctx->state_log[match_last], + match_last); + } + ret = merge_state_array (dfa, sifted_states, lim_states, + match_last + 1); + re_free (lim_states); + lim_states = NULL; + if (BE (ret != REG_NOERROR, 0)) + goto free_return; + } + else + { + sift_ctx_init (&sctx, sifted_states, lim_states, halt_node, match_last); + ret = sift_states_backward (mctx, &sctx); + re_node_set_free (&sctx.limits); + if (BE (ret != REG_NOERROR, 0)) + goto free_return; + if (sifted_states[0] == NULL) + { + ret = REG_NOMATCH; + goto free_return; + } + } + re_free (mctx->state_log); + mctx->state_log = sifted_states; + sifted_states = NULL; + mctx->last_node = halt_node; + mctx->match_last = match_last; + ret = REG_NOERROR; + free_return: + re_free (sifted_states); + re_free (lim_states); + return ret; +} + +/* Acquire an initial state and return it. + We must select appropriate initial state depending on the context, + since initial states may have constraints like "\<", "^", etc.. */ + +static inline re_dfastate_t * +__attribute ((always_inline)) internal_function +acquire_init_state_context (reg_errcode_t *err, const re_match_context_t *mctx, + Idx idx) +{ + const re_dfa_t *const dfa = mctx->dfa; + if (dfa->init_state->has_constraint) + { + unsigned int context; + context = re_string_context_at (&mctx->input, idx - 1, mctx->eflags); + if (IS_WORD_CONTEXT (context)) + return dfa->init_state_word; + else if (IS_ORDINARY_CONTEXT (context)) + return dfa->init_state; + else if (IS_BEGBUF_CONTEXT (context) && IS_NEWLINE_CONTEXT (context)) + return dfa->init_state_begbuf; + else if (IS_NEWLINE_CONTEXT (context)) + return dfa->init_state_nl; + else if (IS_BEGBUF_CONTEXT (context)) + { + /* It is relatively rare case, then calculate on demand. */ + return re_acquire_state_context (err, dfa, + dfa->init_state->entrance_nodes, + context); + } + else + /* Must not happen? */ + return dfa->init_state; + } + else + return dfa->init_state; +} + +/* Check whether the regular expression match input string INPUT or not, + and return the index where the matching end. Return REG_MISSING if + there is no match, and return REG_ERROR in case of an error. + FL_LONGEST_MATCH means we want the POSIX longest matching. + If P_MATCH_FIRST is not NULL, and the match fails, it is set to the + next place where we may want to try matching. + Note that the matcher assume that the maching starts from the current + index of the buffer. */ + +static Idx +internal_function __attribute_warn_unused_result__ +check_matching (re_match_context_t *mctx, bool fl_longest_match, + Idx *p_match_first) +{ + const re_dfa_t *const dfa = mctx->dfa; + reg_errcode_t err; + Idx match = 0; + Idx match_last = REG_MISSING; + Idx cur_str_idx = re_string_cur_idx (&mctx->input); + re_dfastate_t *cur_state; + bool at_init_state = p_match_first != NULL; + Idx next_start_idx = cur_str_idx; + + err = REG_NOERROR; + cur_state = acquire_init_state_context (&err, mctx, cur_str_idx); + /* An initial state must not be NULL (invalid). */ + if (BE (cur_state == NULL, 0)) + { + assert (err == REG_ESPACE); + return REG_ERROR; + } + + if (mctx->state_log != NULL) + { + mctx->state_log[cur_str_idx] = cur_state; + + /* Check OP_OPEN_SUBEXP in the initial state in case that we use them + later. E.g. Processing back references. */ + if (BE (dfa->nbackref, 0)) + { + at_init_state = false; + err = check_subexp_matching_top (mctx, &cur_state->nodes, 0); + if (BE (err != REG_NOERROR, 0)) + return err; + + if (cur_state->has_backref) + { + err = transit_state_bkref (mctx, &cur_state->nodes); + if (BE (err != REG_NOERROR, 0)) + return err; + } + } + } + + /* If the RE accepts NULL string. */ + if (BE (cur_state->halt, 0)) + { + if (!cur_state->has_constraint + || check_halt_state_context (mctx, cur_state, cur_str_idx)) + { + if (!fl_longest_match) + return cur_str_idx; + else + { + match_last = cur_str_idx; + match = 1; + } + } + } + + while (!re_string_eoi (&mctx->input)) + { + re_dfastate_t *old_state = cur_state; + Idx next_char_idx = re_string_cur_idx (&mctx->input) + 1; + + if (BE (next_char_idx >= mctx->input.bufs_len, 0) + || (BE (next_char_idx >= mctx->input.valid_len, 0) + && mctx->input.valid_len < mctx->input.len)) + { + err = extend_buffers (mctx); + if (BE (err != REG_NOERROR, 0)) + { + assert (err == REG_ESPACE); + return REG_ERROR; + } + } + + cur_state = transit_state (&err, mctx, cur_state); + if (mctx->state_log != NULL) + cur_state = merge_state_with_log (&err, mctx, cur_state); + + if (cur_state == NULL) + { + /* Reached the invalid state or an error. Try to recover a valid + state using the state log, if available and if we have not + already found a valid (even if not the longest) match. */ + if (BE (err != REG_NOERROR, 0)) + return REG_ERROR; + + if (mctx->state_log == NULL + || (match && !fl_longest_match) + || (cur_state = find_recover_state (&err, mctx)) == NULL) + break; + } + + if (BE (at_init_state, 0)) + { + if (old_state == cur_state) + next_start_idx = next_char_idx; + else + at_init_state = false; + } + + if (cur_state->halt) + { + /* Reached a halt state. + Check the halt state can satisfy the current context. */ + if (!cur_state->has_constraint + || check_halt_state_context (mctx, cur_state, + re_string_cur_idx (&mctx->input))) + { + /* We found an appropriate halt state. */ + match_last = re_string_cur_idx (&mctx->input); + match = 1; + + /* We found a match, do not modify match_first below. */ + p_match_first = NULL; + if (!fl_longest_match) + break; + } + } + } + + if (p_match_first) + *p_match_first += next_start_idx; + + return match_last; +} + +/* Check NODE match the current context. */ + +static bool +internal_function +check_halt_node_context (const re_dfa_t *dfa, Idx node, unsigned int context) +{ + re_token_type_t type = dfa->nodes[node].type; + unsigned int constraint = dfa->nodes[node].constraint; + if (type != END_OF_RE) + return false; + if (!constraint) + return true; + if (NOT_SATISFY_NEXT_CONSTRAINT (constraint, context)) + return false; + return true; +} + +/* Check the halt state STATE match the current context. + Return 0 if not match, if the node, STATE has, is a halt node and + match the context, return the node. */ + +static Idx +internal_function +check_halt_state_context (const re_match_context_t *mctx, + const re_dfastate_t *state, Idx idx) +{ + Idx i; + unsigned int context; +#ifdef DEBUG + assert (state->halt); +#endif + context = re_string_context_at (&mctx->input, idx, mctx->eflags); + for (i = 0; i < state->nodes.nelem; ++i) + if (check_halt_node_context (mctx->dfa, state->nodes.elems[i], context)) + return state->nodes.elems[i]; + return 0; +} + +/* Compute the next node to which "NFA" transit from NODE("NFA" is a NFA + corresponding to the DFA). + Return the destination node, and update EPS_VIA_NODES; + return REG_MISSING in case of errors. */ + +static Idx +internal_function +proceed_next_node (const re_match_context_t *mctx, Idx nregs, regmatch_t *regs, + Idx *pidx, Idx node, re_node_set *eps_via_nodes, + struct re_fail_stack_t *fs) +{ + const re_dfa_t *const dfa = mctx->dfa; + Idx i; + bool ok; + if (IS_EPSILON_NODE (dfa->nodes[node].type)) + { + re_node_set *cur_nodes = &mctx->state_log[*pidx]->nodes; + re_node_set *edests = &dfa->edests[node]; + Idx dest_node; + ok = re_node_set_insert (eps_via_nodes, node); + if (BE (! ok, 0)) + return REG_ERROR; + /* Pick up a valid destination, or return REG_MISSING if none + is found. */ + for (dest_node = REG_MISSING, i = 0; i < edests->nelem; ++i) + { + Idx candidate = edests->elems[i]; + if (!re_node_set_contains (cur_nodes, candidate)) + continue; + if (dest_node == REG_MISSING) + dest_node = candidate; + + else + { + /* In order to avoid infinite loop like "(a*)*", return the second + epsilon-transition if the first was already considered. */ + if (re_node_set_contains (eps_via_nodes, dest_node)) + return candidate; + + /* Otherwise, push the second epsilon-transition on the fail stack. */ + else if (fs != NULL + && push_fail_stack (fs, *pidx, candidate, nregs, regs, + eps_via_nodes)) + return REG_ERROR; + + /* We know we are going to exit. */ + break; + } + } + return dest_node; + } + else + { + Idx naccepted = 0; + re_token_type_t type = dfa->nodes[node].type; + +#ifdef RE_ENABLE_I18N + if (dfa->nodes[node].accept_mb) + naccepted = check_node_accept_bytes (dfa, node, &mctx->input, *pidx); + else +#endif /* RE_ENABLE_I18N */ + if (type == OP_BACK_REF) + { + Idx subexp_idx = dfa->nodes[node].opr.idx + 1; + naccepted = regs[subexp_idx].rm_eo - regs[subexp_idx].rm_so; + if (fs != NULL) + { + if (regs[subexp_idx].rm_so == -1 || regs[subexp_idx].rm_eo == -1) + return REG_MISSING; + else if (naccepted) + { + char *buf = (char *) re_string_get_buffer (&mctx->input); + if (memcmp (buf + regs[subexp_idx].rm_so, buf + *pidx, + naccepted) != 0) + return REG_MISSING; + } + } + + if (naccepted == 0) + { + Idx dest_node; + ok = re_node_set_insert (eps_via_nodes, node); + if (BE (! ok, 0)) + return REG_ERROR; + dest_node = dfa->edests[node].elems[0]; + if (re_node_set_contains (&mctx->state_log[*pidx]->nodes, + dest_node)) + return dest_node; + } + } + + if (naccepted != 0 + || check_node_accept (mctx, dfa->nodes + node, *pidx)) + { + Idx dest_node = dfa->nexts[node]; + *pidx = (naccepted == 0) ? *pidx + 1 : *pidx + naccepted; + if (fs && (*pidx > mctx->match_last || mctx->state_log[*pidx] == NULL + || !re_node_set_contains (&mctx->state_log[*pidx]->nodes, + dest_node))) + return REG_MISSING; + re_node_set_empty (eps_via_nodes); + return dest_node; + } + } + return REG_MISSING; +} + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +push_fail_stack (struct re_fail_stack_t *fs, Idx str_idx, Idx dest_node, + Idx nregs, regmatch_t *regs, re_node_set *eps_via_nodes) +{ + reg_errcode_t err; + Idx num = fs->num++; + if (fs->num == fs->alloc) + { + struct re_fail_stack_ent_t *new_array; + new_array = realloc (fs->stack, (sizeof (struct re_fail_stack_ent_t) + * fs->alloc * 2)); + if (new_array == NULL) + return REG_ESPACE; + fs->alloc *= 2; + fs->stack = new_array; + } + fs->stack[num].idx = str_idx; + fs->stack[num].node = dest_node; + fs->stack[num].regs = re_malloc (regmatch_t, nregs); + if (fs->stack[num].regs == NULL) + return REG_ESPACE; + memcpy (fs->stack[num].regs, regs, sizeof (regmatch_t) * nregs); + err = re_node_set_init_copy (&fs->stack[num].eps_via_nodes, eps_via_nodes); + return err; +} + +static Idx +internal_function +pop_fail_stack (struct re_fail_stack_t *fs, Idx *pidx, Idx nregs, + regmatch_t *regs, re_node_set *eps_via_nodes) +{ + Idx num = --fs->num; + assert (REG_VALID_INDEX (num)); + *pidx = fs->stack[num].idx; + memcpy (regs, fs->stack[num].regs, sizeof (regmatch_t) * nregs); + re_node_set_free (eps_via_nodes); + re_free (fs->stack[num].regs); + *eps_via_nodes = fs->stack[num].eps_via_nodes; + return fs->stack[num].node; +} + +/* Set the positions where the subexpressions are starts/ends to registers + PMATCH. + Note: We assume that pmatch[0] is already set, and + pmatch[i].rm_so == pmatch[i].rm_eo == -1 for 0 < i < nmatch. */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +set_regs (const regex_t *preg, const re_match_context_t *mctx, size_t nmatch, + regmatch_t *pmatch, bool fl_backtrack) +{ + const re_dfa_t *dfa = (const re_dfa_t *) preg->buffer; + Idx idx, cur_node; + re_node_set eps_via_nodes; + struct re_fail_stack_t *fs; + struct re_fail_stack_t fs_body = { 0, 2, NULL }; + regmatch_t *prev_idx_match; + bool prev_idx_match_malloced = false; + +#ifdef DEBUG + assert (nmatch > 1); + assert (mctx->state_log != NULL); +#endif + if (fl_backtrack) + { + fs = &fs_body; + fs->stack = re_malloc (struct re_fail_stack_ent_t, fs->alloc); + if (fs->stack == NULL) + return REG_ESPACE; + } + else + fs = NULL; + + cur_node = dfa->init_node; + re_node_set_init_empty (&eps_via_nodes); + + if (__libc_use_alloca (nmatch * sizeof (regmatch_t))) + prev_idx_match = (regmatch_t *) alloca (nmatch * sizeof (regmatch_t)); + else + { + prev_idx_match = re_malloc (regmatch_t, nmatch); + if (prev_idx_match == NULL) + { + free_fail_stack_return (fs); + return REG_ESPACE; + } + prev_idx_match_malloced = true; + } + memcpy (prev_idx_match, pmatch, sizeof (regmatch_t) * nmatch); + + for (idx = pmatch[0].rm_so; idx <= pmatch[0].rm_eo ;) + { + update_regs (dfa, pmatch, prev_idx_match, cur_node, idx, nmatch); + + if (idx == pmatch[0].rm_eo && cur_node == mctx->last_node) + { + Idx reg_idx; + if (fs) + { + for (reg_idx = 0; reg_idx < nmatch; ++reg_idx) + if (pmatch[reg_idx].rm_so > -1 && pmatch[reg_idx].rm_eo == -1) + break; + if (reg_idx == nmatch) + { + re_node_set_free (&eps_via_nodes); + if (prev_idx_match_malloced) + re_free (prev_idx_match); + return free_fail_stack_return (fs); + } + cur_node = pop_fail_stack (fs, &idx, nmatch, pmatch, + &eps_via_nodes); + } + else + { + re_node_set_free (&eps_via_nodes); + if (prev_idx_match_malloced) + re_free (prev_idx_match); + return REG_NOERROR; + } + } + + /* Proceed to next node. */ + cur_node = proceed_next_node (mctx, nmatch, pmatch, &idx, cur_node, + &eps_via_nodes, fs); + + if (BE (! REG_VALID_INDEX (cur_node), 0)) + { + if (BE (cur_node == REG_ERROR, 0)) + { + re_node_set_free (&eps_via_nodes); + if (prev_idx_match_malloced) + re_free (prev_idx_match); + free_fail_stack_return (fs); + return REG_ESPACE; + } + if (fs) + cur_node = pop_fail_stack (fs, &idx, nmatch, pmatch, + &eps_via_nodes); + else + { + re_node_set_free (&eps_via_nodes); + if (prev_idx_match_malloced) + re_free (prev_idx_match); + return REG_NOMATCH; + } + } + } + re_node_set_free (&eps_via_nodes); + if (prev_idx_match_malloced) + re_free (prev_idx_match); + return free_fail_stack_return (fs); +} + +static reg_errcode_t +internal_function +free_fail_stack_return (struct re_fail_stack_t *fs) +{ + if (fs) + { + Idx fs_idx; + for (fs_idx = 0; fs_idx < fs->num; ++fs_idx) + { + re_node_set_free (&fs->stack[fs_idx].eps_via_nodes); + re_free (fs->stack[fs_idx].regs); + } + re_free (fs->stack); + } + return REG_NOERROR; +} + +static void +internal_function +update_regs (const re_dfa_t *dfa, regmatch_t *pmatch, + regmatch_t *prev_idx_match, Idx cur_node, Idx cur_idx, Idx nmatch) +{ + int type = dfa->nodes[cur_node].type; + if (type == OP_OPEN_SUBEXP) + { + Idx reg_num = dfa->nodes[cur_node].opr.idx + 1; + + /* We are at the first node of this sub expression. */ + if (reg_num < nmatch) + { + pmatch[reg_num].rm_so = cur_idx; + pmatch[reg_num].rm_eo = -1; + } + } + else if (type == OP_CLOSE_SUBEXP) + { + Idx reg_num = dfa->nodes[cur_node].opr.idx + 1; + if (reg_num < nmatch) + { + /* We are at the last node of this sub expression. */ + if (pmatch[reg_num].rm_so < cur_idx) + { + pmatch[reg_num].rm_eo = cur_idx; + /* This is a non-empty match or we are not inside an optional + subexpression. Accept this right away. */ + memcpy (prev_idx_match, pmatch, sizeof (regmatch_t) * nmatch); + } + else + { + if (dfa->nodes[cur_node].opt_subexp + && prev_idx_match[reg_num].rm_so != -1) + /* We transited through an empty match for an optional + subexpression, like (a?)*, and this is not the subexp's + first match. Copy back the old content of the registers + so that matches of an inner subexpression are undone as + well, like in ((a?))*. */ + memcpy (pmatch, prev_idx_match, sizeof (regmatch_t) * nmatch); + else + /* We completed a subexpression, but it may be part of + an optional one, so do not update PREV_IDX_MATCH. */ + pmatch[reg_num].rm_eo = cur_idx; + } + } + } +} + +/* This function checks the STATE_LOG from the SCTX->last_str_idx to 0 + and sift the nodes in each states according to the following rules. + Updated state_log will be wrote to STATE_LOG. + + Rules: We throw away the Node 'a' in the STATE_LOG[STR_IDX] if... + 1. When STR_IDX == MATCH_LAST(the last index in the state_log): + If 'a' isn't the LAST_NODE and 'a' can't epsilon transit to + the LAST_NODE, we throw away the node 'a'. + 2. When 0 <= STR_IDX < MATCH_LAST and 'a' accepts + string 's' and transit to 'b': + i. If 'b' isn't in the STATE_LOG[STR_IDX+strlen('s')], we throw + away the node 'a'. + ii. If 'b' is in the STATE_LOG[STR_IDX+strlen('s')] but 'b' is + thrown away, we throw away the node 'a'. + 3. When 0 <= STR_IDX < MATCH_LAST and 'a' epsilon transit to 'b': + i. If 'b' isn't in the STATE_LOG[STR_IDX], we throw away the + node 'a'. + ii. If 'b' is in the STATE_LOG[STR_IDX] but 'b' is thrown away, + we throw away the node 'a'. */ + +#define STATE_NODE_CONTAINS(state,node) \ + ((state) != NULL && re_node_set_contains (&(state)->nodes, node)) + +static reg_errcode_t +internal_function +sift_states_backward (const re_match_context_t *mctx, re_sift_context_t *sctx) +{ + reg_errcode_t err; + int null_cnt = 0; + Idx str_idx = sctx->last_str_idx; + re_node_set cur_dest; + +#ifdef DEBUG + assert (mctx->state_log != NULL && mctx->state_log[str_idx] != NULL); +#endif + + /* Build sifted state_log[str_idx]. It has the nodes which can epsilon + transit to the last_node and the last_node itself. */ + err = re_node_set_init_1 (&cur_dest, sctx->last_node); + if (BE (err != REG_NOERROR, 0)) + return err; + err = update_cur_sifted_state (mctx, sctx, str_idx, &cur_dest); + if (BE (err != REG_NOERROR, 0)) + goto free_return; + + /* Then check each states in the state_log. */ + while (str_idx > 0) + { + /* Update counters. */ + null_cnt = (sctx->sifted_states[str_idx] == NULL) ? null_cnt + 1 : 0; + if (null_cnt > mctx->max_mb_elem_len) + { + memset (sctx->sifted_states, '\0', + sizeof (re_dfastate_t *) * str_idx); + re_node_set_free (&cur_dest); + return REG_NOERROR; + } + re_node_set_empty (&cur_dest); + --str_idx; + + if (mctx->state_log[str_idx]) + { + err = build_sifted_states (mctx, sctx, str_idx, &cur_dest); + if (BE (err != REG_NOERROR, 0)) + goto free_return; + } + + /* Add all the nodes which satisfy the following conditions: + - It can epsilon transit to a node in CUR_DEST. + - It is in CUR_SRC. + And update state_log. */ + err = update_cur_sifted_state (mctx, sctx, str_idx, &cur_dest); + if (BE (err != REG_NOERROR, 0)) + goto free_return; + } + err = REG_NOERROR; + free_return: + re_node_set_free (&cur_dest); + return err; +} + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +build_sifted_states (const re_match_context_t *mctx, re_sift_context_t *sctx, + Idx str_idx, re_node_set *cur_dest) +{ + const re_dfa_t *const dfa = mctx->dfa; + const re_node_set *cur_src = &mctx->state_log[str_idx]->non_eps_nodes; + Idx i; + + /* Then build the next sifted state. + We build the next sifted state on 'cur_dest', and update + 'sifted_states[str_idx]' with 'cur_dest'. + Note: + 'cur_dest' is the sifted state from 'state_log[str_idx + 1]'. + 'cur_src' points the node_set of the old 'state_log[str_idx]' + (with the epsilon nodes pre-filtered out). */ + for (i = 0; i < cur_src->nelem; i++) + { + Idx prev_node = cur_src->elems[i]; + int naccepted = 0; + bool ok; + +#ifdef DEBUG + re_token_type_t type = dfa->nodes[prev_node].type; + assert (!IS_EPSILON_NODE (type)); +#endif +#ifdef RE_ENABLE_I18N + /* If the node may accept "multi byte". */ + if (dfa->nodes[prev_node].accept_mb) + naccepted = sift_states_iter_mb (mctx, sctx, prev_node, + str_idx, sctx->last_str_idx); +#endif /* RE_ENABLE_I18N */ + + /* We don't check backreferences here. + See update_cur_sifted_state(). */ + if (!naccepted + && check_node_accept (mctx, dfa->nodes + prev_node, str_idx) + && STATE_NODE_CONTAINS (sctx->sifted_states[str_idx + 1], + dfa->nexts[prev_node])) + naccepted = 1; + + if (naccepted == 0) + continue; + + if (sctx->limits.nelem) + { + Idx to_idx = str_idx + naccepted; + if (check_dst_limits (mctx, &sctx->limits, + dfa->nexts[prev_node], to_idx, + prev_node, str_idx)) + continue; + } + ok = re_node_set_insert (cur_dest, prev_node); + if (BE (! ok, 0)) + return REG_ESPACE; + } + + return REG_NOERROR; +} + +/* Helper functions. */ + +static reg_errcode_t +internal_function +clean_state_log_if_needed (re_match_context_t *mctx, Idx next_state_log_idx) +{ + Idx top = mctx->state_log_top; + + if (next_state_log_idx >= mctx->input.bufs_len + || (next_state_log_idx >= mctx->input.valid_len + && mctx->input.valid_len < mctx->input.len)) + { + reg_errcode_t err; + err = extend_buffers (mctx); + if (BE (err != REG_NOERROR, 0)) + return err; + } + + if (top < next_state_log_idx) + { + memset (mctx->state_log + top + 1, '\0', + sizeof (re_dfastate_t *) * (next_state_log_idx - top)); + mctx->state_log_top = next_state_log_idx; + } + return REG_NOERROR; +} + +static reg_errcode_t +internal_function +merge_state_array (const re_dfa_t *dfa, re_dfastate_t **dst, + re_dfastate_t **src, Idx num) +{ + Idx st_idx; + reg_errcode_t err; + for (st_idx = 0; st_idx < num; ++st_idx) + { + if (dst[st_idx] == NULL) + dst[st_idx] = src[st_idx]; + else if (src[st_idx] != NULL) + { + re_node_set merged_set; + err = re_node_set_init_union (&merged_set, &dst[st_idx]->nodes, + &src[st_idx]->nodes); + if (BE (err != REG_NOERROR, 0)) + return err; + dst[st_idx] = re_acquire_state (&err, dfa, &merged_set); + re_node_set_free (&merged_set); + if (BE (err != REG_NOERROR, 0)) + return err; + } + } + return REG_NOERROR; +} + +static reg_errcode_t +internal_function +update_cur_sifted_state (const re_match_context_t *mctx, + re_sift_context_t *sctx, Idx str_idx, + re_node_set *dest_nodes) +{ + const re_dfa_t *const dfa = mctx->dfa; + reg_errcode_t err = REG_NOERROR; + const re_node_set *candidates; + candidates = ((mctx->state_log[str_idx] == NULL) ? NULL + : &mctx->state_log[str_idx]->nodes); + + if (dest_nodes->nelem == 0) + sctx->sifted_states[str_idx] = NULL; + else + { + if (candidates) + { + /* At first, add the nodes which can epsilon transit to a node in + DEST_NODE. */ + err = add_epsilon_src_nodes (dfa, dest_nodes, candidates); + if (BE (err != REG_NOERROR, 0)) + return err; + + /* Then, check the limitations in the current sift_context. */ + if (sctx->limits.nelem) + { + err = check_subexp_limits (dfa, dest_nodes, candidates, &sctx->limits, + mctx->bkref_ents, str_idx); + if (BE (err != REG_NOERROR, 0)) + return err; + } + } + + sctx->sifted_states[str_idx] = re_acquire_state (&err, dfa, dest_nodes); + if (BE (err != REG_NOERROR, 0)) + return err; + } + + if (candidates && mctx->state_log[str_idx]->has_backref) + { + err = sift_states_bkref (mctx, sctx, str_idx, candidates); + if (BE (err != REG_NOERROR, 0)) + return err; + } + return REG_NOERROR; +} + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +add_epsilon_src_nodes (const re_dfa_t *dfa, re_node_set *dest_nodes, + const re_node_set *candidates) +{ + reg_errcode_t err = REG_NOERROR; + Idx i; + + re_dfastate_t *state = re_acquire_state (&err, dfa, dest_nodes); + if (BE (err != REG_NOERROR, 0)) + return err; + + if (!state->inveclosure.alloc) + { + err = re_node_set_alloc (&state->inveclosure, dest_nodes->nelem); + if (BE (err != REG_NOERROR, 0)) + return REG_ESPACE; + for (i = 0; i < dest_nodes->nelem; i++) + { + err = re_node_set_merge (&state->inveclosure, + dfa->inveclosures + dest_nodes->elems[i]); + if (BE (err != REG_NOERROR, 0)) + return REG_ESPACE; + } + } + return re_node_set_add_intersect (dest_nodes, candidates, + &state->inveclosure); +} + +static reg_errcode_t +internal_function +sub_epsilon_src_nodes (const re_dfa_t *dfa, Idx node, re_node_set *dest_nodes, + const re_node_set *candidates) +{ + Idx ecl_idx; + reg_errcode_t err; + re_node_set *inv_eclosure = dfa->inveclosures + node; + re_node_set except_nodes; + re_node_set_init_empty (&except_nodes); + for (ecl_idx = 0; ecl_idx < inv_eclosure->nelem; ++ecl_idx) + { + Idx cur_node = inv_eclosure->elems[ecl_idx]; + if (cur_node == node) + continue; + if (IS_EPSILON_NODE (dfa->nodes[cur_node].type)) + { + Idx edst1 = dfa->edests[cur_node].elems[0]; + Idx edst2 = ((dfa->edests[cur_node].nelem > 1) + ? dfa->edests[cur_node].elems[1] : REG_MISSING); + if ((!re_node_set_contains (inv_eclosure, edst1) + && re_node_set_contains (dest_nodes, edst1)) + || (REG_VALID_NONZERO_INDEX (edst2) + && !re_node_set_contains (inv_eclosure, edst2) + && re_node_set_contains (dest_nodes, edst2))) + { + err = re_node_set_add_intersect (&except_nodes, candidates, + dfa->inveclosures + cur_node); + if (BE (err != REG_NOERROR, 0)) + { + re_node_set_free (&except_nodes); + return err; + } + } + } + } + for (ecl_idx = 0; ecl_idx < inv_eclosure->nelem; ++ecl_idx) + { + Idx cur_node = inv_eclosure->elems[ecl_idx]; + if (!re_node_set_contains (&except_nodes, cur_node)) + { + Idx idx = re_node_set_contains (dest_nodes, cur_node) - 1; + re_node_set_remove_at (dest_nodes, idx); + } + } + re_node_set_free (&except_nodes); + return REG_NOERROR; +} + +static bool +internal_function +check_dst_limits (const re_match_context_t *mctx, const re_node_set *limits, + Idx dst_node, Idx dst_idx, Idx src_node, Idx src_idx) +{ + const re_dfa_t *const dfa = mctx->dfa; + Idx lim_idx, src_pos, dst_pos; + + Idx dst_bkref_idx = search_cur_bkref_entry (mctx, dst_idx); + Idx src_bkref_idx = search_cur_bkref_entry (mctx, src_idx); + for (lim_idx = 0; lim_idx < limits->nelem; ++lim_idx) + { + Idx subexp_idx; + struct re_backref_cache_entry *ent; + ent = mctx->bkref_ents + limits->elems[lim_idx]; + subexp_idx = dfa->nodes[ent->node].opr.idx; + + dst_pos = check_dst_limits_calc_pos (mctx, limits->elems[lim_idx], + subexp_idx, dst_node, dst_idx, + dst_bkref_idx); + src_pos = check_dst_limits_calc_pos (mctx, limits->elems[lim_idx], + subexp_idx, src_node, src_idx, + src_bkref_idx); + + /* In case of: + <src> <dst> ( <subexp> ) + ( <subexp> ) <src> <dst> + ( <subexp1> <src> <subexp2> <dst> <subexp3> ) */ + if (src_pos == dst_pos) + continue; /* This is unrelated limitation. */ + else + return true; + } + return false; +} + +static int +internal_function +check_dst_limits_calc_pos_1 (const re_match_context_t *mctx, int boundaries, + Idx subexp_idx, Idx from_node, Idx bkref_idx) +{ + const re_dfa_t *const dfa = mctx->dfa; + const re_node_set *eclosures = dfa->eclosures + from_node; + Idx node_idx; + + /* Else, we are on the boundary: examine the nodes on the epsilon + closure. */ + for (node_idx = 0; node_idx < eclosures->nelem; ++node_idx) + { + Idx node = eclosures->elems[node_idx]; + switch (dfa->nodes[node].type) + { + case OP_BACK_REF: + if (bkref_idx != REG_MISSING) + { + struct re_backref_cache_entry *ent = mctx->bkref_ents + bkref_idx; + do + { + Idx dst; + int cpos; + + if (ent->node != node) + continue; + + if (subexp_idx < BITSET_WORD_BITS + && !(ent->eps_reachable_subexps_map + & ((bitset_word_t) 1 << subexp_idx))) + continue; + + /* Recurse trying to reach the OP_OPEN_SUBEXP and + OP_CLOSE_SUBEXP cases below. But, if the + destination node is the same node as the source + node, don't recurse because it would cause an + infinite loop: a regex that exhibits this behavior + is ()\1*\1* */ + dst = dfa->edests[node].elems[0]; + if (dst == from_node) + { + if (boundaries & 1) + return -1; + else /* if (boundaries & 2) */ + return 0; + } + + cpos = + check_dst_limits_calc_pos_1 (mctx, boundaries, subexp_idx, + dst, bkref_idx); + if (cpos == -1 /* && (boundaries & 1) */) + return -1; + if (cpos == 0 && (boundaries & 2)) + return 0; + + if (subexp_idx < BITSET_WORD_BITS) + ent->eps_reachable_subexps_map + &= ~((bitset_word_t) 1 << subexp_idx); + } + while (ent++->more); + } + break; + + case OP_OPEN_SUBEXP: + if ((boundaries & 1) && subexp_idx == dfa->nodes[node].opr.idx) + return -1; + break; + + case OP_CLOSE_SUBEXP: + if ((boundaries & 2) && subexp_idx == dfa->nodes[node].opr.idx) + return 0; + break; + + default: + break; + } + } + + return (boundaries & 2) ? 1 : 0; +} + +static int +internal_function +check_dst_limits_calc_pos (const re_match_context_t *mctx, Idx limit, + Idx subexp_idx, Idx from_node, Idx str_idx, + Idx bkref_idx) +{ + struct re_backref_cache_entry *lim = mctx->bkref_ents + limit; + int boundaries; + + /* If we are outside the range of the subexpression, return -1 or 1. */ + if (str_idx < lim->subexp_from) + return -1; + + if (lim->subexp_to < str_idx) + return 1; + + /* If we are within the subexpression, return 0. */ + boundaries = (str_idx == lim->subexp_from); + boundaries |= (str_idx == lim->subexp_to) << 1; + if (boundaries == 0) + return 0; + + /* Else, examine epsilon closure. */ + return check_dst_limits_calc_pos_1 (mctx, boundaries, subexp_idx, + from_node, bkref_idx); +} + +/* Check the limitations of sub expressions LIMITS, and remove the nodes + which are against limitations from DEST_NODES. */ + +static reg_errcode_t +internal_function +check_subexp_limits (const re_dfa_t *dfa, re_node_set *dest_nodes, + const re_node_set *candidates, re_node_set *limits, + struct re_backref_cache_entry *bkref_ents, Idx str_idx) +{ + reg_errcode_t err; + Idx node_idx, lim_idx; + + for (lim_idx = 0; lim_idx < limits->nelem; ++lim_idx) + { + Idx subexp_idx; + struct re_backref_cache_entry *ent; + ent = bkref_ents + limits->elems[lim_idx]; + + if (str_idx <= ent->subexp_from || ent->str_idx < str_idx) + continue; /* This is unrelated limitation. */ + + subexp_idx = dfa->nodes[ent->node].opr.idx; + if (ent->subexp_to == str_idx) + { + Idx ops_node = REG_MISSING; + Idx cls_node = REG_MISSING; + for (node_idx = 0; node_idx < dest_nodes->nelem; ++node_idx) + { + Idx node = dest_nodes->elems[node_idx]; + re_token_type_t type = dfa->nodes[node].type; + if (type == OP_OPEN_SUBEXP + && subexp_idx == dfa->nodes[node].opr.idx) + ops_node = node; + else if (type == OP_CLOSE_SUBEXP + && subexp_idx == dfa->nodes[node].opr.idx) + cls_node = node; + } + + /* Check the limitation of the open subexpression. */ + /* Note that (ent->subexp_to = str_idx != ent->subexp_from). */ + if (REG_VALID_INDEX (ops_node)) + { + err = sub_epsilon_src_nodes (dfa, ops_node, dest_nodes, + candidates); + if (BE (err != REG_NOERROR, 0)) + return err; + } + + /* Check the limitation of the close subexpression. */ + if (REG_VALID_INDEX (cls_node)) + for (node_idx = 0; node_idx < dest_nodes->nelem; ++node_idx) + { + Idx node = dest_nodes->elems[node_idx]; + if (!re_node_set_contains (dfa->inveclosures + node, + cls_node) + && !re_node_set_contains (dfa->eclosures + node, + cls_node)) + { + /* It is against this limitation. + Remove it form the current sifted state. */ + err = sub_epsilon_src_nodes (dfa, node, dest_nodes, + candidates); + if (BE (err != REG_NOERROR, 0)) + return err; + --node_idx; + } + } + } + else /* (ent->subexp_to != str_idx) */ + { + for (node_idx = 0; node_idx < dest_nodes->nelem; ++node_idx) + { + Idx node = dest_nodes->elems[node_idx]; + re_token_type_t type = dfa->nodes[node].type; + if (type == OP_CLOSE_SUBEXP || type == OP_OPEN_SUBEXP) + { + if (subexp_idx != dfa->nodes[node].opr.idx) + continue; + /* It is against this limitation. + Remove it form the current sifted state. */ + err = sub_epsilon_src_nodes (dfa, node, dest_nodes, + candidates); + if (BE (err != REG_NOERROR, 0)) + return err; + } + } + } + } + return REG_NOERROR; +} + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +sift_states_bkref (const re_match_context_t *mctx, re_sift_context_t *sctx, + Idx str_idx, const re_node_set *candidates) +{ + const re_dfa_t *const dfa = mctx->dfa; + reg_errcode_t err; + Idx node_idx, node; + re_sift_context_t local_sctx; + Idx first_idx = search_cur_bkref_entry (mctx, str_idx); + + if (first_idx == REG_MISSING) + return REG_NOERROR; + + local_sctx.sifted_states = NULL; /* Mark that it hasn't been initialized. */ + + for (node_idx = 0; node_idx < candidates->nelem; ++node_idx) + { + Idx enabled_idx; + re_token_type_t type; + struct re_backref_cache_entry *entry; + node = candidates->elems[node_idx]; + type = dfa->nodes[node].type; + /* Avoid infinite loop for the REs like "()\1+". */ + if (node == sctx->last_node && str_idx == sctx->last_str_idx) + continue; + if (type != OP_BACK_REF) + continue; + + entry = mctx->bkref_ents + first_idx; + enabled_idx = first_idx; + do + { + Idx subexp_len; + Idx to_idx; + Idx dst_node; + bool ok; + re_dfastate_t *cur_state; + + if (entry->node != node) + continue; + subexp_len = entry->subexp_to - entry->subexp_from; + to_idx = str_idx + subexp_len; + dst_node = (subexp_len ? dfa->nexts[node] + : dfa->edests[node].elems[0]); + + if (to_idx > sctx->last_str_idx + || sctx->sifted_states[to_idx] == NULL + || !STATE_NODE_CONTAINS (sctx->sifted_states[to_idx], dst_node) + || check_dst_limits (mctx, &sctx->limits, node, + str_idx, dst_node, to_idx)) + continue; + + if (local_sctx.sifted_states == NULL) + { + local_sctx = *sctx; + err = re_node_set_init_copy (&local_sctx.limits, &sctx->limits); + if (BE (err != REG_NOERROR, 0)) + goto free_return; + } + local_sctx.last_node = node; + local_sctx.last_str_idx = str_idx; + ok = re_node_set_insert (&local_sctx.limits, enabled_idx); + if (BE (! ok, 0)) + { + err = REG_ESPACE; + goto free_return; + } + cur_state = local_sctx.sifted_states[str_idx]; + err = sift_states_backward (mctx, &local_sctx); + if (BE (err != REG_NOERROR, 0)) + goto free_return; + if (sctx->limited_states != NULL) + { + err = merge_state_array (dfa, sctx->limited_states, + local_sctx.sifted_states, + str_idx + 1); + if (BE (err != REG_NOERROR, 0)) + goto free_return; + } + local_sctx.sifted_states[str_idx] = cur_state; + re_node_set_remove (&local_sctx.limits, enabled_idx); + + /* mctx->bkref_ents may have changed, reload the pointer. */ + entry = mctx->bkref_ents + enabled_idx; + } + while (enabled_idx++, entry++->more); + } + err = REG_NOERROR; + free_return: + if (local_sctx.sifted_states != NULL) + { + re_node_set_free (&local_sctx.limits); + } + + return err; +} + + +#ifdef RE_ENABLE_I18N +static int +internal_function +sift_states_iter_mb (const re_match_context_t *mctx, re_sift_context_t *sctx, + Idx node_idx, Idx str_idx, Idx max_str_idx) +{ + const re_dfa_t *const dfa = mctx->dfa; + int naccepted; + /* Check the node can accept "multi byte". */ + naccepted = check_node_accept_bytes (dfa, node_idx, &mctx->input, str_idx); + if (naccepted > 0 && str_idx + naccepted <= max_str_idx && + !STATE_NODE_CONTAINS (sctx->sifted_states[str_idx + naccepted], + dfa->nexts[node_idx])) + /* The node can't accept the "multi byte", or the + destination was already thrown away, then the node + could't accept the current input "multi byte". */ + naccepted = 0; + /* Otherwise, it is sure that the node could accept + 'naccepted' bytes input. */ + return naccepted; +} +#endif /* RE_ENABLE_I18N */ + + +/* Functions for state transition. */ + +/* Return the next state to which the current state STATE will transit by + accepting the current input byte, and update STATE_LOG if necessary. + If STATE can accept a multibyte char/collating element/back reference + update the destination of STATE_LOG. */ + +static re_dfastate_t * +internal_function __attribute_warn_unused_result__ +transit_state (reg_errcode_t *err, re_match_context_t *mctx, + re_dfastate_t *state) +{ + re_dfastate_t **trtable; + unsigned char ch; + +#ifdef RE_ENABLE_I18N + /* If the current state can accept multibyte. */ + if (BE (state->accept_mb, 0)) + { + *err = transit_state_mb (mctx, state); + if (BE (*err != REG_NOERROR, 0)) + return NULL; + } +#endif /* RE_ENABLE_I18N */ + + /* Then decide the next state with the single byte. */ +#if 0 + if (0) + /* don't use transition table */ + return transit_state_sb (err, mctx, state); +#endif + + /* Use transition table */ + ch = re_string_fetch_byte (&mctx->input); + for (;;) + { + trtable = state->trtable; + if (BE (trtable != NULL, 1)) + return trtable[ch]; + + trtable = state->word_trtable; + if (BE (trtable != NULL, 1)) + { + unsigned int context; + context + = re_string_context_at (&mctx->input, + re_string_cur_idx (&mctx->input) - 1, + mctx->eflags); + if (IS_WORD_CONTEXT (context)) + return trtable[ch + SBC_MAX]; + else + return trtable[ch]; + } + + if (!build_trtable (mctx->dfa, state)) + { + *err = REG_ESPACE; + return NULL; + } + + /* Retry, we now have a transition table. */ + } +} + +/* Update the state_log if we need */ +static re_dfastate_t * +internal_function +merge_state_with_log (reg_errcode_t *err, re_match_context_t *mctx, + re_dfastate_t *next_state) +{ + const re_dfa_t *const dfa = mctx->dfa; + Idx cur_idx = re_string_cur_idx (&mctx->input); + + if (cur_idx > mctx->state_log_top) + { + mctx->state_log[cur_idx] = next_state; + mctx->state_log_top = cur_idx; + } + else if (mctx->state_log[cur_idx] == 0) + { + mctx->state_log[cur_idx] = next_state; + } + else + { + re_dfastate_t *pstate; + unsigned int context; + re_node_set next_nodes, *log_nodes, *table_nodes = NULL; + /* If (state_log[cur_idx] != 0), it implies that cur_idx is + the destination of a multibyte char/collating element/ + back reference. Then the next state is the union set of + these destinations and the results of the transition table. */ + pstate = mctx->state_log[cur_idx]; + log_nodes = pstate->entrance_nodes; + if (next_state != NULL) + { + table_nodes = next_state->entrance_nodes; + *err = re_node_set_init_union (&next_nodes, table_nodes, + log_nodes); + if (BE (*err != REG_NOERROR, 0)) + return NULL; + } + else + next_nodes = *log_nodes; + /* Note: We already add the nodes of the initial state, + then we don't need to add them here. */ + + context = re_string_context_at (&mctx->input, + re_string_cur_idx (&mctx->input) - 1, + mctx->eflags); + next_state = mctx->state_log[cur_idx] + = re_acquire_state_context (err, dfa, &next_nodes, context); + /* We don't need to check errors here, since the return value of + this function is next_state and ERR is already set. */ + + if (table_nodes != NULL) + re_node_set_free (&next_nodes); + } + + if (BE (dfa->nbackref, 0) && next_state != NULL) + { + /* Check OP_OPEN_SUBEXP in the current state in case that we use them + later. We must check them here, since the back references in the + next state might use them. */ + *err = check_subexp_matching_top (mctx, &next_state->nodes, + cur_idx); + if (BE (*err != REG_NOERROR, 0)) + return NULL; + + /* If the next state has back references. */ + if (next_state->has_backref) + { + *err = transit_state_bkref (mctx, &next_state->nodes); + if (BE (*err != REG_NOERROR, 0)) + return NULL; + next_state = mctx->state_log[cur_idx]; + } + } + + return next_state; +} + +/* Skip bytes in the input that correspond to part of a + multi-byte match, then look in the log for a state + from which to restart matching. */ +static re_dfastate_t * +internal_function +find_recover_state (reg_errcode_t *err, re_match_context_t *mctx) +{ + re_dfastate_t *cur_state; + do + { + Idx max = mctx->state_log_top; + Idx cur_str_idx = re_string_cur_idx (&mctx->input); + + do + { + if (++cur_str_idx > max) + return NULL; + re_string_skip_bytes (&mctx->input, 1); + } + while (mctx->state_log[cur_str_idx] == NULL); + + cur_state = merge_state_with_log (err, mctx, NULL); + } + while (*err == REG_NOERROR && cur_state == NULL); + return cur_state; +} + +/* Helper functions for transit_state. */ + +/* From the node set CUR_NODES, pick up the nodes whose types are + OP_OPEN_SUBEXP and which have corresponding back references in the regular + expression. And register them to use them later for evaluating the + correspoding back references. */ + +static reg_errcode_t +internal_function +check_subexp_matching_top (re_match_context_t *mctx, re_node_set *cur_nodes, + Idx str_idx) +{ + const re_dfa_t *const dfa = mctx->dfa; + Idx node_idx; + reg_errcode_t err; + + /* TODO: This isn't efficient. + Because there might be more than one nodes whose types are + OP_OPEN_SUBEXP and whose index is SUBEXP_IDX, we must check all + nodes. + E.g. RE: (a){2} */ + for (node_idx = 0; node_idx < cur_nodes->nelem; ++node_idx) + { + Idx node = cur_nodes->elems[node_idx]; + if (dfa->nodes[node].type == OP_OPEN_SUBEXP + && dfa->nodes[node].opr.idx < BITSET_WORD_BITS + && (dfa->used_bkref_map + & ((bitset_word_t) 1 << dfa->nodes[node].opr.idx))) + { + err = match_ctx_add_subtop (mctx, node, str_idx); + if (BE (err != REG_NOERROR, 0)) + return err; + } + } + return REG_NOERROR; +} + +#if 0 +/* Return the next state to which the current state STATE will transit by + accepting the current input byte. */ + +static re_dfastate_t * +transit_state_sb (reg_errcode_t *err, re_match_context_t *mctx, + re_dfastate_t *state) +{ + const re_dfa_t *const dfa = mctx->dfa; + re_node_set next_nodes; + re_dfastate_t *next_state; + Idx node_cnt, cur_str_idx = re_string_cur_idx (&mctx->input); + unsigned int context; + + *err = re_node_set_alloc (&next_nodes, state->nodes.nelem + 1); + if (BE (*err != REG_NOERROR, 0)) + return NULL; + for (node_cnt = 0; node_cnt < state->nodes.nelem; ++node_cnt) + { + Idx cur_node = state->nodes.elems[node_cnt]; + if (check_node_accept (mctx, dfa->nodes + cur_node, cur_str_idx)) + { + *err = re_node_set_merge (&next_nodes, + dfa->eclosures + dfa->nexts[cur_node]); + if (BE (*err != REG_NOERROR, 0)) + { + re_node_set_free (&next_nodes); + return NULL; + } + } + } + context = re_string_context_at (&mctx->input, cur_str_idx, mctx->eflags); + next_state = re_acquire_state_context (err, dfa, &next_nodes, context); + /* We don't need to check errors here, since the return value of + this function is next_state and ERR is already set. */ + + re_node_set_free (&next_nodes); + re_string_skip_bytes (&mctx->input, 1); + return next_state; +} +#endif + +#ifdef RE_ENABLE_I18N +static reg_errcode_t +internal_function +transit_state_mb (re_match_context_t *mctx, re_dfastate_t *pstate) +{ + const re_dfa_t *const dfa = mctx->dfa; + reg_errcode_t err; + Idx i; + + for (i = 0; i < pstate->nodes.nelem; ++i) + { + re_node_set dest_nodes, *new_nodes; + Idx cur_node_idx = pstate->nodes.elems[i]; + int naccepted; + Idx dest_idx; + unsigned int context; + re_dfastate_t *dest_state; + + if (!dfa->nodes[cur_node_idx].accept_mb) + continue; + + if (dfa->nodes[cur_node_idx].constraint) + { + context = re_string_context_at (&mctx->input, + re_string_cur_idx (&mctx->input), + mctx->eflags); + if (NOT_SATISFY_NEXT_CONSTRAINT (dfa->nodes[cur_node_idx].constraint, + context)) + continue; + } + + /* How many bytes the node can accept? */ + naccepted = check_node_accept_bytes (dfa, cur_node_idx, &mctx->input, + re_string_cur_idx (&mctx->input)); + if (naccepted == 0) + continue; + + /* The node can accepts 'naccepted' bytes. */ + dest_idx = re_string_cur_idx (&mctx->input) + naccepted; + mctx->max_mb_elem_len = ((mctx->max_mb_elem_len < naccepted) ? naccepted + : mctx->max_mb_elem_len); + err = clean_state_log_if_needed (mctx, dest_idx); + if (BE (err != REG_NOERROR, 0)) + return err; +#ifdef DEBUG + assert (dfa->nexts[cur_node_idx] != REG_MISSING); +#endif + new_nodes = dfa->eclosures + dfa->nexts[cur_node_idx]; + + dest_state = mctx->state_log[dest_idx]; + if (dest_state == NULL) + dest_nodes = *new_nodes; + else + { + err = re_node_set_init_union (&dest_nodes, + dest_state->entrance_nodes, new_nodes); + if (BE (err != REG_NOERROR, 0)) + return err; + } + context = re_string_context_at (&mctx->input, dest_idx - 1, + mctx->eflags); + mctx->state_log[dest_idx] + = re_acquire_state_context (&err, dfa, &dest_nodes, context); + if (dest_state != NULL) + re_node_set_free (&dest_nodes); + if (BE (mctx->state_log[dest_idx] == NULL && err != REG_NOERROR, 0)) + return err; + } + return REG_NOERROR; +} +#endif /* RE_ENABLE_I18N */ + +static reg_errcode_t +internal_function +transit_state_bkref (re_match_context_t *mctx, const re_node_set *nodes) +{ + const re_dfa_t *const dfa = mctx->dfa; + reg_errcode_t err; + Idx i; + Idx cur_str_idx = re_string_cur_idx (&mctx->input); + + for (i = 0; i < nodes->nelem; ++i) + { + Idx dest_str_idx, prev_nelem, bkc_idx; + Idx node_idx = nodes->elems[i]; + unsigned int context; + const re_token_t *node = dfa->nodes + node_idx; + re_node_set *new_dest_nodes; + + /* Check whether 'node' is a backreference or not. */ + if (node->type != OP_BACK_REF) + continue; + + if (node->constraint) + { + context = re_string_context_at (&mctx->input, cur_str_idx, + mctx->eflags); + if (NOT_SATISFY_NEXT_CONSTRAINT (node->constraint, context)) + continue; + } + + /* 'node' is a backreference. + Check the substring which the substring matched. */ + bkc_idx = mctx->nbkref_ents; + err = get_subexp (mctx, node_idx, cur_str_idx); + if (BE (err != REG_NOERROR, 0)) + goto free_return; + + /* And add the epsilon closures (which is 'new_dest_nodes') of + the backreference to appropriate state_log. */ +#ifdef DEBUG + assert (dfa->nexts[node_idx] != REG_MISSING); +#endif + for (; bkc_idx < mctx->nbkref_ents; ++bkc_idx) + { + Idx subexp_len; + re_dfastate_t *dest_state; + struct re_backref_cache_entry *bkref_ent; + bkref_ent = mctx->bkref_ents + bkc_idx; + if (bkref_ent->node != node_idx || bkref_ent->str_idx != cur_str_idx) + continue; + subexp_len = bkref_ent->subexp_to - bkref_ent->subexp_from; + new_dest_nodes = (subexp_len == 0 + ? dfa->eclosures + dfa->edests[node_idx].elems[0] + : dfa->eclosures + dfa->nexts[node_idx]); + dest_str_idx = (cur_str_idx + bkref_ent->subexp_to + - bkref_ent->subexp_from); + context = re_string_context_at (&mctx->input, dest_str_idx - 1, + mctx->eflags); + dest_state = mctx->state_log[dest_str_idx]; + prev_nelem = ((mctx->state_log[cur_str_idx] == NULL) ? 0 + : mctx->state_log[cur_str_idx]->nodes.nelem); + /* Add 'new_dest_node' to state_log. */ + if (dest_state == NULL) + { + mctx->state_log[dest_str_idx] + = re_acquire_state_context (&err, dfa, new_dest_nodes, + context); + if (BE (mctx->state_log[dest_str_idx] == NULL + && err != REG_NOERROR, 0)) + goto free_return; + } + else + { + re_node_set dest_nodes; + err = re_node_set_init_union (&dest_nodes, + dest_state->entrance_nodes, + new_dest_nodes); + if (BE (err != REG_NOERROR, 0)) + { + re_node_set_free (&dest_nodes); + goto free_return; + } + mctx->state_log[dest_str_idx] + = re_acquire_state_context (&err, dfa, &dest_nodes, context); + re_node_set_free (&dest_nodes); + if (BE (mctx->state_log[dest_str_idx] == NULL + && err != REG_NOERROR, 0)) + goto free_return; + } + /* We need to check recursively if the backreference can epsilon + transit. */ + if (subexp_len == 0 + && mctx->state_log[cur_str_idx]->nodes.nelem > prev_nelem) + { + err = check_subexp_matching_top (mctx, new_dest_nodes, + cur_str_idx); + if (BE (err != REG_NOERROR, 0)) + goto free_return; + err = transit_state_bkref (mctx, new_dest_nodes); + if (BE (err != REG_NOERROR, 0)) + goto free_return; + } + } + } + err = REG_NOERROR; + free_return: + return err; +} + +/* Enumerate all the candidates which the backreference BKREF_NODE can match + at BKREF_STR_IDX, and register them by match_ctx_add_entry(). + Note that we might collect inappropriate candidates here. + However, the cost of checking them strictly here is too high, then we + delay these checking for prune_impossible_nodes(). */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +get_subexp (re_match_context_t *mctx, Idx bkref_node, Idx bkref_str_idx) +{ + const re_dfa_t *const dfa = mctx->dfa; + Idx subexp_num, sub_top_idx; + const char *buf = (const char *) re_string_get_buffer (&mctx->input); + /* Return if we have already checked BKREF_NODE at BKREF_STR_IDX. */ + Idx cache_idx = search_cur_bkref_entry (mctx, bkref_str_idx); + if (cache_idx != REG_MISSING) + { + const struct re_backref_cache_entry *entry + = mctx->bkref_ents + cache_idx; + do + if (entry->node == bkref_node) + return REG_NOERROR; /* We already checked it. */ + while (entry++->more); + } + + subexp_num = dfa->nodes[bkref_node].opr.idx; + + /* For each sub expression */ + for (sub_top_idx = 0; sub_top_idx < mctx->nsub_tops; ++sub_top_idx) + { + reg_errcode_t err; + re_sub_match_top_t *sub_top = mctx->sub_tops[sub_top_idx]; + re_sub_match_last_t *sub_last; + Idx sub_last_idx, sl_str, bkref_str_off; + + if (dfa->nodes[sub_top->node].opr.idx != subexp_num) + continue; /* It isn't related. */ + + sl_str = sub_top->str_idx; + bkref_str_off = bkref_str_idx; + /* At first, check the last node of sub expressions we already + evaluated. */ + for (sub_last_idx = 0; sub_last_idx < sub_top->nlasts; ++sub_last_idx) + { + regoff_t sl_str_diff; + sub_last = sub_top->lasts[sub_last_idx]; + sl_str_diff = sub_last->str_idx - sl_str; + /* The matched string by the sub expression match with the substring + at the back reference? */ + if (sl_str_diff > 0) + { + if (BE (bkref_str_off + sl_str_diff > mctx->input.valid_len, 0)) + { + /* Not enough chars for a successful match. */ + if (bkref_str_off + sl_str_diff > mctx->input.len) + break; + + err = clean_state_log_if_needed (mctx, + bkref_str_off + + sl_str_diff); + if (BE (err != REG_NOERROR, 0)) + return err; + buf = (const char *) re_string_get_buffer (&mctx->input); + } + if (memcmp (buf + bkref_str_off, buf + sl_str, sl_str_diff) != 0) + /* We don't need to search this sub expression any more. */ + break; + } + bkref_str_off += sl_str_diff; + sl_str += sl_str_diff; + err = get_subexp_sub (mctx, sub_top, sub_last, bkref_node, + bkref_str_idx); + + /* Reload buf, since the preceding call might have reallocated + the buffer. */ + buf = (const char *) re_string_get_buffer (&mctx->input); + + if (err == REG_NOMATCH) + continue; + if (BE (err != REG_NOERROR, 0)) + return err; + } + + if (sub_last_idx < sub_top->nlasts) + continue; + if (sub_last_idx > 0) + ++sl_str; + /* Then, search for the other last nodes of the sub expression. */ + for (; sl_str <= bkref_str_idx; ++sl_str) + { + Idx cls_node; + regoff_t sl_str_off; + const re_node_set *nodes; + sl_str_off = sl_str - sub_top->str_idx; + /* The matched string by the sub expression match with the substring + at the back reference? */ + if (sl_str_off > 0) + { + if (BE (bkref_str_off >= mctx->input.valid_len, 0)) + { + /* If we are at the end of the input, we cannot match. */ + if (bkref_str_off >= mctx->input.len) + break; + + err = extend_buffers (mctx); + if (BE (err != REG_NOERROR, 0)) + return err; + + buf = (const char *) re_string_get_buffer (&mctx->input); + } + if (buf [bkref_str_off++] != buf[sl_str - 1]) + break; /* We don't need to search this sub expression + any more. */ + } + if (mctx->state_log[sl_str] == NULL) + continue; + /* Does this state have a ')' of the sub expression? */ + nodes = &mctx->state_log[sl_str]->nodes; + cls_node = find_subexp_node (dfa, nodes, subexp_num, + OP_CLOSE_SUBEXP); + if (cls_node == REG_MISSING) + continue; /* No. */ + if (sub_top->path == NULL) + { + sub_top->path = calloc (sizeof (state_array_t), + sl_str - sub_top->str_idx + 1); + if (sub_top->path == NULL) + return REG_ESPACE; + } + /* Can the OP_OPEN_SUBEXP node arrive the OP_CLOSE_SUBEXP node + in the current context? */ + err = check_arrival (mctx, sub_top->path, sub_top->node, + sub_top->str_idx, cls_node, sl_str, + OP_CLOSE_SUBEXP); + if (err == REG_NOMATCH) + continue; + if (BE (err != REG_NOERROR, 0)) + return err; + sub_last = match_ctx_add_sublast (sub_top, cls_node, sl_str); + if (BE (sub_last == NULL, 0)) + return REG_ESPACE; + err = get_subexp_sub (mctx, sub_top, sub_last, bkref_node, + bkref_str_idx); + if (err == REG_NOMATCH) + continue; + } + } + return REG_NOERROR; +} + +/* Helper functions for get_subexp(). */ + +/* Check SUB_LAST can arrive to the back reference BKREF_NODE at BKREF_STR. + If it can arrive, register the sub expression expressed with SUB_TOP + and SUB_LAST. */ + +static reg_errcode_t +internal_function +get_subexp_sub (re_match_context_t *mctx, const re_sub_match_top_t *sub_top, + re_sub_match_last_t *sub_last, Idx bkref_node, Idx bkref_str) +{ + reg_errcode_t err; + Idx to_idx; + /* Can the subexpression arrive the back reference? */ + err = check_arrival (mctx, &sub_last->path, sub_last->node, + sub_last->str_idx, bkref_node, bkref_str, + OP_OPEN_SUBEXP); + if (err != REG_NOERROR) + return err; + err = match_ctx_add_entry (mctx, bkref_node, bkref_str, sub_top->str_idx, + sub_last->str_idx); + if (BE (err != REG_NOERROR, 0)) + return err; + to_idx = bkref_str + sub_last->str_idx - sub_top->str_idx; + return clean_state_log_if_needed (mctx, to_idx); +} + +/* Find the first node which is '(' or ')' and whose index is SUBEXP_IDX. + Search '(' if FL_OPEN, or search ')' otherwise. + TODO: This function isn't efficient... + Because there might be more than one nodes whose types are + OP_OPEN_SUBEXP and whose index is SUBEXP_IDX, we must check all + nodes. + E.g. RE: (a){2} */ + +static Idx +internal_function +find_subexp_node (const re_dfa_t *dfa, const re_node_set *nodes, + Idx subexp_idx, int type) +{ + Idx cls_idx; + for (cls_idx = 0; cls_idx < nodes->nelem; ++cls_idx) + { + Idx cls_node = nodes->elems[cls_idx]; + const re_token_t *node = dfa->nodes + cls_node; + if (node->type == type + && node->opr.idx == subexp_idx) + return cls_node; + } + return REG_MISSING; +} + +/* Check whether the node TOP_NODE at TOP_STR can arrive to the node + LAST_NODE at LAST_STR. We record the path onto PATH since it will be + heavily reused. + Return REG_NOERROR if it can arrive, or REG_NOMATCH otherwise. */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +check_arrival (re_match_context_t *mctx, state_array_t *path, Idx top_node, + Idx top_str, Idx last_node, Idx last_str, int type) +{ + const re_dfa_t *const dfa = mctx->dfa; + reg_errcode_t err = REG_NOERROR; + Idx subexp_num, backup_cur_idx, str_idx, null_cnt; + re_dfastate_t *cur_state = NULL; + re_node_set *cur_nodes, next_nodes; + re_dfastate_t **backup_state_log; + unsigned int context; + + subexp_num = dfa->nodes[top_node].opr.idx; + /* Extend the buffer if we need. */ + if (BE (path->alloc < last_str + mctx->max_mb_elem_len + 1, 0)) + { + re_dfastate_t **new_array; + Idx old_alloc = path->alloc; + Idx new_alloc = old_alloc + last_str + mctx->max_mb_elem_len + 1; + if (BE (new_alloc < old_alloc, 0) + || BE (SIZE_MAX / sizeof (re_dfastate_t *) < new_alloc, 0)) + return REG_ESPACE; + new_array = re_realloc (path->array, re_dfastate_t *, new_alloc); + if (BE (new_array == NULL, 0)) + return REG_ESPACE; + path->array = new_array; + path->alloc = new_alloc; + memset (new_array + old_alloc, '\0', + sizeof (re_dfastate_t *) * (path->alloc - old_alloc)); + } + + str_idx = path->next_idx ? path->next_idx : top_str; + + /* Temporary modify MCTX. */ + backup_state_log = mctx->state_log; + backup_cur_idx = mctx->input.cur_idx; + mctx->state_log = path->array; + mctx->input.cur_idx = str_idx; + + /* Setup initial node set. */ + context = re_string_context_at (&mctx->input, str_idx - 1, mctx->eflags); + if (str_idx == top_str) + { + err = re_node_set_init_1 (&next_nodes, top_node); + if (BE (err != REG_NOERROR, 0)) + return err; + err = check_arrival_expand_ecl (dfa, &next_nodes, subexp_num, type); + if (BE (err != REG_NOERROR, 0)) + { + re_node_set_free (&next_nodes); + return err; + } + } + else + { + cur_state = mctx->state_log[str_idx]; + if (cur_state && cur_state->has_backref) + { + err = re_node_set_init_copy (&next_nodes, &cur_state->nodes); + if (BE (err != REG_NOERROR, 0)) + return err; + } + else + re_node_set_init_empty (&next_nodes); + } + if (str_idx == top_str || (cur_state && cur_state->has_backref)) + { + if (next_nodes.nelem) + { + err = expand_bkref_cache (mctx, &next_nodes, str_idx, + subexp_num, type); + if (BE (err != REG_NOERROR, 0)) + { + re_node_set_free (&next_nodes); + return err; + } + } + cur_state = re_acquire_state_context (&err, dfa, &next_nodes, context); + if (BE (cur_state == NULL && err != REG_NOERROR, 0)) + { + re_node_set_free (&next_nodes); + return err; + } + mctx->state_log[str_idx] = cur_state; + } + + for (null_cnt = 0; str_idx < last_str && null_cnt <= mctx->max_mb_elem_len;) + { + re_node_set_empty (&next_nodes); + if (mctx->state_log[str_idx + 1]) + { + err = re_node_set_merge (&next_nodes, + &mctx->state_log[str_idx + 1]->nodes); + if (BE (err != REG_NOERROR, 0)) + { + re_node_set_free (&next_nodes); + return err; + } + } + if (cur_state) + { + err = check_arrival_add_next_nodes (mctx, str_idx, + &cur_state->non_eps_nodes, + &next_nodes); + if (BE (err != REG_NOERROR, 0)) + { + re_node_set_free (&next_nodes); + return err; + } + } + ++str_idx; + if (next_nodes.nelem) + { + err = check_arrival_expand_ecl (dfa, &next_nodes, subexp_num, type); + if (BE (err != REG_NOERROR, 0)) + { + re_node_set_free (&next_nodes); + return err; + } + err = expand_bkref_cache (mctx, &next_nodes, str_idx, + subexp_num, type); + if (BE (err != REG_NOERROR, 0)) + { + re_node_set_free (&next_nodes); + return err; + } + } + context = re_string_context_at (&mctx->input, str_idx - 1, mctx->eflags); + cur_state = re_acquire_state_context (&err, dfa, &next_nodes, context); + if (BE (cur_state == NULL && err != REG_NOERROR, 0)) + { + re_node_set_free (&next_nodes); + return err; + } + mctx->state_log[str_idx] = cur_state; + null_cnt = cur_state == NULL ? null_cnt + 1 : 0; + } + re_node_set_free (&next_nodes); + cur_nodes = (mctx->state_log[last_str] == NULL ? NULL + : &mctx->state_log[last_str]->nodes); + path->next_idx = str_idx; + + /* Fix MCTX. */ + mctx->state_log = backup_state_log; + mctx->input.cur_idx = backup_cur_idx; + + /* Then check the current node set has the node LAST_NODE. */ + if (cur_nodes != NULL && re_node_set_contains (cur_nodes, last_node)) + return REG_NOERROR; + + return REG_NOMATCH; +} + +/* Helper functions for check_arrival. */ + +/* Calculate the destination nodes of CUR_NODES at STR_IDX, and append them + to NEXT_NODES. + TODO: This function is similar to the functions transit_state*(), + however this function has many additional works. + Can't we unify them? */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +check_arrival_add_next_nodes (re_match_context_t *mctx, Idx str_idx, + re_node_set *cur_nodes, re_node_set *next_nodes) +{ + const re_dfa_t *const dfa = mctx->dfa; + bool ok; + Idx cur_idx; +#ifdef RE_ENABLE_I18N + reg_errcode_t err = REG_NOERROR; +#endif + re_node_set union_set; + re_node_set_init_empty (&union_set); + for (cur_idx = 0; cur_idx < cur_nodes->nelem; ++cur_idx) + { + int naccepted = 0; + Idx cur_node = cur_nodes->elems[cur_idx]; +#ifdef DEBUG + re_token_type_t type = dfa->nodes[cur_node].type; + assert (!IS_EPSILON_NODE (type)); +#endif +#ifdef RE_ENABLE_I18N + /* If the node may accept "multi byte". */ + if (dfa->nodes[cur_node].accept_mb) + { + naccepted = check_node_accept_bytes (dfa, cur_node, &mctx->input, + str_idx); + if (naccepted > 1) + { + re_dfastate_t *dest_state; + Idx next_node = dfa->nexts[cur_node]; + Idx next_idx = str_idx + naccepted; + dest_state = mctx->state_log[next_idx]; + re_node_set_empty (&union_set); + if (dest_state) + { + err = re_node_set_merge (&union_set, &dest_state->nodes); + if (BE (err != REG_NOERROR, 0)) + { + re_node_set_free (&union_set); + return err; + } + } + ok = re_node_set_insert (&union_set, next_node); + if (BE (! ok, 0)) + { + re_node_set_free (&union_set); + return REG_ESPACE; + } + mctx->state_log[next_idx] = re_acquire_state (&err, dfa, + &union_set); + if (BE (mctx->state_log[next_idx] == NULL + && err != REG_NOERROR, 0)) + { + re_node_set_free (&union_set); + return err; + } + } + } +#endif /* RE_ENABLE_I18N */ + if (naccepted + || check_node_accept (mctx, dfa->nodes + cur_node, str_idx)) + { + ok = re_node_set_insert (next_nodes, dfa->nexts[cur_node]); + if (BE (! ok, 0)) + { + re_node_set_free (&union_set); + return REG_ESPACE; + } + } + } + re_node_set_free (&union_set); + return REG_NOERROR; +} + +/* For all the nodes in CUR_NODES, add the epsilon closures of them to + CUR_NODES, however exclude the nodes which are: + - inside the sub expression whose number is EX_SUBEXP, if FL_OPEN. + - out of the sub expression whose number is EX_SUBEXP, if !FL_OPEN. +*/ + +static reg_errcode_t +internal_function +check_arrival_expand_ecl (const re_dfa_t *dfa, re_node_set *cur_nodes, + Idx ex_subexp, int type) +{ + reg_errcode_t err; + Idx idx, outside_node; + re_node_set new_nodes; +#ifdef DEBUG + assert (cur_nodes->nelem); +#endif + err = re_node_set_alloc (&new_nodes, cur_nodes->nelem); + if (BE (err != REG_NOERROR, 0)) + return err; + /* Create a new node set NEW_NODES with the nodes which are epsilon + closures of the node in CUR_NODES. */ + + for (idx = 0; idx < cur_nodes->nelem; ++idx) + { + Idx cur_node = cur_nodes->elems[idx]; + const re_node_set *eclosure = dfa->eclosures + cur_node; + outside_node = find_subexp_node (dfa, eclosure, ex_subexp, type); + if (outside_node == REG_MISSING) + { + /* There are no problematic nodes, just merge them. */ + err = re_node_set_merge (&new_nodes, eclosure); + if (BE (err != REG_NOERROR, 0)) + { + re_node_set_free (&new_nodes); + return err; + } + } + else + { + /* There are problematic nodes, re-calculate incrementally. */ + err = check_arrival_expand_ecl_sub (dfa, &new_nodes, cur_node, + ex_subexp, type); + if (BE (err != REG_NOERROR, 0)) + { + re_node_set_free (&new_nodes); + return err; + } + } + } + re_node_set_free (cur_nodes); + *cur_nodes = new_nodes; + return REG_NOERROR; +} + +/* Helper function for check_arrival_expand_ecl. + Check incrementally the epsilon closure of TARGET, and if it isn't + problematic append it to DST_NODES. */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +check_arrival_expand_ecl_sub (const re_dfa_t *dfa, re_node_set *dst_nodes, + Idx target, Idx ex_subexp, int type) +{ + Idx cur_node; + for (cur_node = target; !re_node_set_contains (dst_nodes, cur_node);) + { + bool ok; + + if (dfa->nodes[cur_node].type == type + && dfa->nodes[cur_node].opr.idx == ex_subexp) + { + if (type == OP_CLOSE_SUBEXP) + { + ok = re_node_set_insert (dst_nodes, cur_node); + if (BE (! ok, 0)) + return REG_ESPACE; + } + break; + } + ok = re_node_set_insert (dst_nodes, cur_node); + if (BE (! ok, 0)) + return REG_ESPACE; + if (dfa->edests[cur_node].nelem == 0) + break; + if (dfa->edests[cur_node].nelem == 2) + { + reg_errcode_t err; + err = check_arrival_expand_ecl_sub (dfa, dst_nodes, + dfa->edests[cur_node].elems[1], + ex_subexp, type); + if (BE (err != REG_NOERROR, 0)) + return err; + } + cur_node = dfa->edests[cur_node].elems[0]; + } + return REG_NOERROR; +} + + +/* For all the back references in the current state, calculate the + destination of the back references by the appropriate entry + in MCTX->BKREF_ENTS. */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +expand_bkref_cache (re_match_context_t *mctx, re_node_set *cur_nodes, + Idx cur_str, Idx subexp_num, int type) +{ + const re_dfa_t *const dfa = mctx->dfa; + reg_errcode_t err; + Idx cache_idx_start = search_cur_bkref_entry (mctx, cur_str); + struct re_backref_cache_entry *ent; + + if (cache_idx_start == REG_MISSING) + return REG_NOERROR; + + restart: + ent = mctx->bkref_ents + cache_idx_start; + do + { + Idx to_idx, next_node; + + /* Is this entry ENT is appropriate? */ + if (!re_node_set_contains (cur_nodes, ent->node)) + continue; /* No. */ + + to_idx = cur_str + ent->subexp_to - ent->subexp_from; + /* Calculate the destination of the back reference, and append it + to MCTX->STATE_LOG. */ + if (to_idx == cur_str) + { + /* The backreference did epsilon transit, we must re-check all the + node in the current state. */ + re_node_set new_dests; + reg_errcode_t err2, err3; + next_node = dfa->edests[ent->node].elems[0]; + if (re_node_set_contains (cur_nodes, next_node)) + continue; + err = re_node_set_init_1 (&new_dests, next_node); + err2 = check_arrival_expand_ecl (dfa, &new_dests, subexp_num, type); + err3 = re_node_set_merge (cur_nodes, &new_dests); + re_node_set_free (&new_dests); + if (BE (err != REG_NOERROR || err2 != REG_NOERROR + || err3 != REG_NOERROR, 0)) + { + err = (err != REG_NOERROR ? err + : (err2 != REG_NOERROR ? err2 : err3)); + return err; + } + /* TODO: It is still inefficient... */ + goto restart; + } + else + { + re_node_set union_set; + next_node = dfa->nexts[ent->node]; + if (mctx->state_log[to_idx]) + { + bool ok; + if (re_node_set_contains (&mctx->state_log[to_idx]->nodes, + next_node)) + continue; + err = re_node_set_init_copy (&union_set, + &mctx->state_log[to_idx]->nodes); + ok = re_node_set_insert (&union_set, next_node); + if (BE (err != REG_NOERROR || ! ok, 0)) + { + re_node_set_free (&union_set); + err = err != REG_NOERROR ? err : REG_ESPACE; + return err; + } + } + else + { + err = re_node_set_init_1 (&union_set, next_node); + if (BE (err != REG_NOERROR, 0)) + return err; + } + mctx->state_log[to_idx] = re_acquire_state (&err, dfa, &union_set); + re_node_set_free (&union_set); + if (BE (mctx->state_log[to_idx] == NULL + && err != REG_NOERROR, 0)) + return err; + } + } + while (ent++->more); + return REG_NOERROR; +} + +/* Build transition table for the state. + Return true if successful. */ + +static bool +internal_function +build_trtable (const re_dfa_t *dfa, re_dfastate_t *state) +{ + reg_errcode_t err; + Idx i, j; + int ch; + bool need_word_trtable = false; + bitset_word_t elem, mask; + bool dests_node_malloced = false; + bool dest_states_malloced = false; + Idx ndests; /* Number of the destination states from 'state'. */ + re_dfastate_t **trtable; + re_dfastate_t **dest_states = NULL, **dest_states_word, **dest_states_nl; + re_node_set follows, *dests_node; + bitset_t *dests_ch; + bitset_t acceptable; + + struct dests_alloc + { + re_node_set dests_node[SBC_MAX]; + bitset_t dests_ch[SBC_MAX]; + } *dests_alloc; + + /* We build DFA states which corresponds to the destination nodes + from 'state'. 'dests_node[i]' represents the nodes which i-th + destination state contains, and 'dests_ch[i]' represents the + characters which i-th destination state accepts. */ + if (__libc_use_alloca (sizeof (struct dests_alloc))) + dests_alloc = (struct dests_alloc *) alloca (sizeof (struct dests_alloc)); + else + { + dests_alloc = re_malloc (struct dests_alloc, 1); + if (BE (dests_alloc == NULL, 0)) + return false; + dests_node_malloced = true; + } + dests_node = dests_alloc->dests_node; + dests_ch = dests_alloc->dests_ch; + + /* Initialize transiton table. */ + state->word_trtable = state->trtable = NULL; + + /* At first, group all nodes belonging to 'state' into several + destinations. */ + ndests = group_nodes_into_DFAstates (dfa, state, dests_node, dests_ch); + if (BE (! REG_VALID_NONZERO_INDEX (ndests), 0)) + { + if (dests_node_malloced) + free (dests_alloc); + if (ndests == 0) + { + state->trtable = (re_dfastate_t **) + calloc (sizeof (re_dfastate_t *), SBC_MAX); + if (BE (state->trtable == NULL, 0)) + return false; + return true; + } + return false; + } + + err = re_node_set_alloc (&follows, ndests + 1); + if (BE (err != REG_NOERROR, 0)) + goto out_free; + + /* Avoid arithmetic overflow in size calculation. */ + if (BE ((((SIZE_MAX - (sizeof (re_node_set) + sizeof (bitset_t)) * SBC_MAX) + / (3 * sizeof (re_dfastate_t *))) + < ndests), + 0)) + goto out_free; + + if (__libc_use_alloca ((sizeof (re_node_set) + sizeof (bitset_t)) * SBC_MAX + + ndests * 3 * sizeof (re_dfastate_t *))) + dest_states = (re_dfastate_t **) + alloca (ndests * 3 * sizeof (re_dfastate_t *)); + else + { + dest_states = (re_dfastate_t **) + malloc (ndests * 3 * sizeof (re_dfastate_t *)); + if (BE (dest_states == NULL, 0)) + { +out_free: + if (dest_states_malloced) + free (dest_states); + re_node_set_free (&follows); + for (i = 0; i < ndests; ++i) + re_node_set_free (dests_node + i); + if (dests_node_malloced) + free (dests_alloc); + return false; + } + dest_states_malloced = true; + } + dest_states_word = dest_states + ndests; + dest_states_nl = dest_states_word + ndests; + bitset_empty (acceptable); + + /* Then build the states for all destinations. */ + for (i = 0; i < ndests; ++i) + { + Idx next_node; + re_node_set_empty (&follows); + /* Merge the follows of this destination states. */ + for (j = 0; j < dests_node[i].nelem; ++j) + { + next_node = dfa->nexts[dests_node[i].elems[j]]; + if (next_node != REG_MISSING) + { + err = re_node_set_merge (&follows, dfa->eclosures + next_node); + if (BE (err != REG_NOERROR, 0)) + goto out_free; + } + } + dest_states[i] = re_acquire_state_context (&err, dfa, &follows, 0); + if (BE (dest_states[i] == NULL && err != REG_NOERROR, 0)) + goto out_free; + /* If the new state has context constraint, + build appropriate states for these contexts. */ + if (dest_states[i]->has_constraint) + { + dest_states_word[i] = re_acquire_state_context (&err, dfa, &follows, + CONTEXT_WORD); + if (BE (dest_states_word[i] == NULL && err != REG_NOERROR, 0)) + goto out_free; + + if (dest_states[i] != dest_states_word[i] && dfa->mb_cur_max > 1) + need_word_trtable = true; + + dest_states_nl[i] = re_acquire_state_context (&err, dfa, &follows, + CONTEXT_NEWLINE); + if (BE (dest_states_nl[i] == NULL && err != REG_NOERROR, 0)) + goto out_free; + } + else + { + dest_states_word[i] = dest_states[i]; + dest_states_nl[i] = dest_states[i]; + } + bitset_merge (acceptable, dests_ch[i]); + } + + if (!BE (need_word_trtable, 0)) + { + /* We don't care about whether the following character is a word + character, or we are in a single-byte character set so we can + discern by looking at the character code: allocate a + 256-entry transition table. */ + trtable = state->trtable = + (re_dfastate_t **) calloc (sizeof (re_dfastate_t *), SBC_MAX); + if (BE (trtable == NULL, 0)) + goto out_free; + + /* For all characters ch...: */ + for (i = 0; i < BITSET_WORDS; ++i) + for (ch = i * BITSET_WORD_BITS, elem = acceptable[i], mask = 1; + elem; + mask <<= 1, elem >>= 1, ++ch) + if (BE (elem & 1, 0)) + { + /* There must be exactly one destination which accepts + character ch. See group_nodes_into_DFAstates. */ + for (j = 0; (dests_ch[j][i] & mask) == 0; ++j) + ; + + /* j-th destination accepts the word character ch. */ + if (dfa->word_char[i] & mask) + trtable[ch] = dest_states_word[j]; + else + trtable[ch] = dest_states[j]; + } + } + else + { + /* We care about whether the following character is a word + character, and we are in a multi-byte character set: discern + by looking at the character code: build two 256-entry + transition tables, one starting at trtable[0] and one + starting at trtable[SBC_MAX]. */ + trtable = state->word_trtable = + (re_dfastate_t **) calloc (sizeof (re_dfastate_t *), 2 * SBC_MAX); + if (BE (trtable == NULL, 0)) + goto out_free; + + /* For all characters ch...: */ + for (i = 0; i < BITSET_WORDS; ++i) + for (ch = i * BITSET_WORD_BITS, elem = acceptable[i], mask = 1; + elem; + mask <<= 1, elem >>= 1, ++ch) + if (BE (elem & 1, 0)) + { + /* There must be exactly one destination which accepts + character ch. See group_nodes_into_DFAstates. */ + for (j = 0; (dests_ch[j][i] & mask) == 0; ++j) + ; + + /* j-th destination accepts the word character ch. */ + trtable[ch] = dest_states[j]; + trtable[ch + SBC_MAX] = dest_states_word[j]; + } + } + + /* new line */ + if (bitset_contain (acceptable, NEWLINE_CHAR)) + { + /* The current state accepts newline character. */ + for (j = 0; j < ndests; ++j) + if (bitset_contain (dests_ch[j], NEWLINE_CHAR)) + { + /* k-th destination accepts newline character. */ + trtable[NEWLINE_CHAR] = dest_states_nl[j]; + if (need_word_trtable) + trtable[NEWLINE_CHAR + SBC_MAX] = dest_states_nl[j]; + /* There must be only one destination which accepts + newline. See group_nodes_into_DFAstates. */ + break; + } + } + + if (dest_states_malloced) + free (dest_states); + + re_node_set_free (&follows); + for (i = 0; i < ndests; ++i) + re_node_set_free (dests_node + i); + + if (dests_node_malloced) + free (dests_alloc); + + return true; +} + +/* Group all nodes belonging to STATE into several destinations. + Then for all destinations, set the nodes belonging to the destination + to DESTS_NODE[i] and set the characters accepted by the destination + to DEST_CH[i]. This function return the number of destinations. */ + +static Idx +internal_function +group_nodes_into_DFAstates (const re_dfa_t *dfa, const re_dfastate_t *state, + re_node_set *dests_node, bitset_t *dests_ch) +{ + reg_errcode_t err; + bool ok; + Idx i, j, k; + Idx ndests; /* Number of the destinations from 'state'. */ + bitset_t accepts; /* Characters a node can accept. */ + const re_node_set *cur_nodes = &state->nodes; + bitset_empty (accepts); + ndests = 0; + + /* For all the nodes belonging to 'state', */ + for (i = 0; i < cur_nodes->nelem; ++i) + { + re_token_t *node = &dfa->nodes[cur_nodes->elems[i]]; + re_token_type_t type = node->type; + unsigned int constraint = node->constraint; + + /* Enumerate all single byte character this node can accept. */ + if (type == CHARACTER) + bitset_set (accepts, node->opr.c); + else if (type == SIMPLE_BRACKET) + { + bitset_merge (accepts, node->opr.sbcset); + } + else if (type == OP_PERIOD) + { +#ifdef RE_ENABLE_I18N + if (dfa->mb_cur_max > 1) + bitset_merge (accepts, dfa->sb_char); + else +#endif + bitset_set_all (accepts); + if (!(dfa->syntax & RE_DOT_NEWLINE)) + bitset_clear (accepts, '\n'); + if (dfa->syntax & RE_DOT_NOT_NULL) + bitset_clear (accepts, '\0'); + } +#ifdef RE_ENABLE_I18N + else if (type == OP_UTF8_PERIOD) + { + if (ASCII_CHARS % BITSET_WORD_BITS == 0) + memset (accepts, -1, ASCII_CHARS / CHAR_BIT); + else + bitset_merge (accepts, utf8_sb_map); + if (!(dfa->syntax & RE_DOT_NEWLINE)) + bitset_clear (accepts, '\n'); + if (dfa->syntax & RE_DOT_NOT_NULL) + bitset_clear (accepts, '\0'); + } +#endif + else + continue; + + /* Check the 'accepts' and sift the characters which are not + match it the context. */ + if (constraint) + { + if (constraint & NEXT_NEWLINE_CONSTRAINT) + { + bool accepts_newline = bitset_contain (accepts, NEWLINE_CHAR); + bitset_empty (accepts); + if (accepts_newline) + bitset_set (accepts, NEWLINE_CHAR); + else + continue; + } + if (constraint & NEXT_ENDBUF_CONSTRAINT) + { + bitset_empty (accepts); + continue; + } + + if (constraint & NEXT_WORD_CONSTRAINT) + { + bitset_word_t any_set = 0; + if (type == CHARACTER && !node->word_char) + { + bitset_empty (accepts); + continue; + } +#ifdef RE_ENABLE_I18N + if (dfa->mb_cur_max > 1) + for (j = 0; j < BITSET_WORDS; ++j) + any_set |= (accepts[j] &= (dfa->word_char[j] | ~dfa->sb_char[j])); + else +#endif + for (j = 0; j < BITSET_WORDS; ++j) + any_set |= (accepts[j] &= dfa->word_char[j]); + if (!any_set) + continue; + } + if (constraint & NEXT_NOTWORD_CONSTRAINT) + { + bitset_word_t any_set = 0; + if (type == CHARACTER && node->word_char) + { + bitset_empty (accepts); + continue; + } +#ifdef RE_ENABLE_I18N + if (dfa->mb_cur_max > 1) + for (j = 0; j < BITSET_WORDS; ++j) + any_set |= (accepts[j] &= ~(dfa->word_char[j] & dfa->sb_char[j])); + else +#endif + for (j = 0; j < BITSET_WORDS; ++j) + any_set |= (accepts[j] &= ~dfa->word_char[j]); + if (!any_set) + continue; + } + } + + /* Then divide 'accepts' into DFA states, or create a new + state. Above, we make sure that accepts is not empty. */ + for (j = 0; j < ndests; ++j) + { + bitset_t intersec; /* Intersection sets, see below. */ + bitset_t remains; + /* Flags, see below. */ + bitset_word_t has_intersec, not_subset, not_consumed; + + /* Optimization, skip if this state doesn't accept the character. */ + if (type == CHARACTER && !bitset_contain (dests_ch[j], node->opr.c)) + continue; + + /* Enumerate the intersection set of this state and 'accepts'. */ + has_intersec = 0; + for (k = 0; k < BITSET_WORDS; ++k) + has_intersec |= intersec[k] = accepts[k] & dests_ch[j][k]; + /* And skip if the intersection set is empty. */ + if (!has_intersec) + continue; + + /* Then check if this state is a subset of 'accepts'. */ + not_subset = not_consumed = 0; + for (k = 0; k < BITSET_WORDS; ++k) + { + not_subset |= remains[k] = ~accepts[k] & dests_ch[j][k]; + not_consumed |= accepts[k] = accepts[k] & ~dests_ch[j][k]; + } + + /* If this state isn't a subset of 'accepts', create a + new group state, which has the 'remains'. */ + if (not_subset) + { + bitset_copy (dests_ch[ndests], remains); + bitset_copy (dests_ch[j], intersec); + err = re_node_set_init_copy (dests_node + ndests, &dests_node[j]); + if (BE (err != REG_NOERROR, 0)) + goto error_return; + ++ndests; + } + + /* Put the position in the current group. */ + ok = re_node_set_insert (&dests_node[j], cur_nodes->elems[i]); + if (BE (! ok, 0)) + goto error_return; + + /* If all characters are consumed, go to next node. */ + if (!not_consumed) + break; + } + /* Some characters remain, create a new group. */ + if (j == ndests) + { + bitset_copy (dests_ch[ndests], accepts); + err = re_node_set_init_1 (dests_node + ndests, cur_nodes->elems[i]); + if (BE (err != REG_NOERROR, 0)) + goto error_return; + ++ndests; + bitset_empty (accepts); + } + } + return ndests; + error_return: + for (j = 0; j < ndests; ++j) + re_node_set_free (dests_node + j); + return REG_MISSING; +} + +#ifdef RE_ENABLE_I18N +/* Check how many bytes the node 'dfa->nodes[node_idx]' accepts. + Return the number of the bytes the node accepts. + STR_IDX is the current index of the input string. + + This function handles the nodes which can accept one character, or + one collating element like '.', '[a-z]', opposite to the other nodes + can only accept one byte. */ + +static int +internal_function +check_node_accept_bytes (const re_dfa_t *dfa, Idx node_idx, + const re_string_t *input, Idx str_idx) +{ + const re_token_t *node = dfa->nodes + node_idx; + int char_len, elem_len; + Idx i; + + if (BE (node->type == OP_UTF8_PERIOD, 0)) + { + unsigned char c = re_string_byte_at (input, str_idx), d; + if (BE (c < 0xc2, 1)) + return 0; + + if (str_idx + 2 > input->len) + return 0; + + d = re_string_byte_at (input, str_idx + 1); + if (c < 0xe0) + return (d < 0x80 || d > 0xbf) ? 0 : 2; + else if (c < 0xf0) + { + char_len = 3; + if (c == 0xe0 && d < 0xa0) + return 0; + } + else if (c < 0xf8) + { + char_len = 4; + if (c == 0xf0 && d < 0x90) + return 0; + } + else if (c < 0xfc) + { + char_len = 5; + if (c == 0xf8 && d < 0x88) + return 0; + } + else if (c < 0xfe) + { + char_len = 6; + if (c == 0xfc && d < 0x84) + return 0; + } + else + return 0; + + if (str_idx + char_len > input->len) + return 0; + + for (i = 1; i < char_len; ++i) + { + d = re_string_byte_at (input, str_idx + i); + if (d < 0x80 || d > 0xbf) + return 0; + } + return char_len; + } + + char_len = re_string_char_size_at (input, str_idx); + if (node->type == OP_PERIOD) + { + if (char_len <= 1) + return 0; + /* FIXME: I don't think this if is needed, as both '\n' + and '\0' are char_len == 1. */ + /* '.' accepts any one character except the following two cases. */ + if ((!(dfa->syntax & RE_DOT_NEWLINE) && + re_string_byte_at (input, str_idx) == '\n') || + ((dfa->syntax & RE_DOT_NOT_NULL) && + re_string_byte_at (input, str_idx) == '\0')) + return 0; + return char_len; + } + + elem_len = re_string_elem_size_at (input, str_idx); + if ((elem_len <= 1 && char_len <= 1) || char_len == 0) + return 0; + + if (node->type == COMPLEX_BRACKET) + { + const re_charset_t *cset = node->opr.mbcset; +# ifdef _LIBC + const unsigned char *pin + = ((const unsigned char *) re_string_get_buffer (input) + str_idx); + Idx j; + uint32_t nrules; +# endif /* _LIBC */ + int match_len = 0; + wchar_t wc = ((cset->nranges || cset->nchar_classes || cset->nmbchars) + ? re_string_wchar_at (input, str_idx) : 0); + + /* match with multibyte character? */ + for (i = 0; i < cset->nmbchars; ++i) + if (wc == cset->mbchars[i]) + { + match_len = char_len; + goto check_node_accept_bytes_match; + } + /* match with character_class? */ + for (i = 0; i < cset->nchar_classes; ++i) + { + wctype_t wt = cset->char_classes[i]; + if (__iswctype (wc, wt)) + { + match_len = char_len; + goto check_node_accept_bytes_match; + } + } + +# ifdef _LIBC + nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); + if (nrules != 0) + { + unsigned int in_collseq = 0; + const int32_t *table, *indirect; + const unsigned char *weights, *extra; + const char *collseqwc; + int32_t idx; + /* This #include defines a local function! */ +# include <locale/weight.h> + + /* match with collating_symbol? */ + if (cset->ncoll_syms) + extra = (const unsigned char *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB); + for (i = 0; i < cset->ncoll_syms; ++i) + { + const unsigned char *coll_sym = extra + cset->coll_syms[i]; + /* Compare the length of input collating element and + the length of current collating element. */ + if (*coll_sym != elem_len) + continue; + /* Compare each bytes. */ + for (j = 0; j < *coll_sym; j++) + if (pin[j] != coll_sym[1 + j]) + break; + if (j == *coll_sym) + { + /* Match if every bytes is equal. */ + match_len = j; + goto check_node_accept_bytes_match; + } + } + + if (cset->nranges) + { + if (elem_len <= char_len) + { + collseqwc = _NL_CURRENT (LC_COLLATE, _NL_COLLATE_COLLSEQWC); + in_collseq = __collseq_table_lookup (collseqwc, wc); + } + else + in_collseq = find_collation_sequence_value (pin, elem_len); + } + /* match with range expression? */ + for (i = 0; i < cset->nranges; ++i) + if (cset->range_starts[i] <= in_collseq + && in_collseq <= cset->range_ends[i]) + { + match_len = elem_len; + goto check_node_accept_bytes_match; + } + + /* match with equivalence_class? */ + if (cset->nequiv_classes) + { + const unsigned char *cp = pin; + table = (const int32_t *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); + weights = (const unsigned char *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTMB); + extra = (const unsigned char *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB); + indirect = (const int32_t *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTMB); + int32_t idx = findidx (&cp); + if (idx > 0) + for (i = 0; i < cset->nequiv_classes; ++i) + { + int32_t equiv_class_idx = cset->equiv_classes[i]; + size_t weight_len = weights[idx & 0xffffff]; + if (weight_len == weights[equiv_class_idx & 0xffffff] + && (idx >> 24) == (equiv_class_idx >> 24)) + { + Idx cnt = 0; + + idx &= 0xffffff; + equiv_class_idx &= 0xffffff; + + while (cnt <= weight_len + && (weights[equiv_class_idx + 1 + cnt] + == weights[idx + 1 + cnt])) + ++cnt; + if (cnt > weight_len) + { + match_len = elem_len; + goto check_node_accept_bytes_match; + } + } + } + } + } + else +# endif /* _LIBC */ + { + /* match with range expression? */ +#if __GNUC__ >= 2 && ! (__STDC_VERSION__ < 199901L && defined __STRICT_ANSI__) + wchar_t cmp_buf[] = {L'\0', L'\0', wc, L'\0', L'\0', L'\0'}; +#else + wchar_t cmp_buf[] = {L'\0', L'\0', L'\0', L'\0', L'\0', L'\0'}; + cmp_buf[2] = wc; +#endif + for (i = 0; i < cset->nranges; ++i) + { + cmp_buf[0] = cset->range_starts[i]; + cmp_buf[4] = cset->range_ends[i]; + if (wcscoll (cmp_buf, cmp_buf + 2) <= 0 + && wcscoll (cmp_buf + 2, cmp_buf + 4) <= 0) + { + match_len = char_len; + goto check_node_accept_bytes_match; + } + } + } + check_node_accept_bytes_match: + if (!cset->non_match) + return match_len; + else + { + if (match_len > 0) + return 0; + else + return (elem_len > char_len) ? elem_len : char_len; + } + } + return 0; +} + +# ifdef _LIBC +static unsigned int +internal_function +find_collation_sequence_value (const unsigned char *mbs, size_t mbs_len) +{ + uint32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); + if (nrules == 0) + { + if (mbs_len == 1) + { + /* No valid character. Match it as a single byte character. */ + const unsigned char *collseq = (const unsigned char *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_COLLSEQMB); + return collseq[mbs[0]]; + } + return UINT_MAX; + } + else + { + int32_t idx; + const unsigned char *extra = (const unsigned char *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB); + int32_t extrasize = (const unsigned char *) + _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB + 1) - extra; + + for (idx = 0; idx < extrasize;) + { + int mbs_cnt; + bool found = false; + int32_t elem_mbs_len; + /* Skip the name of collating element name. */ + idx = idx + extra[idx] + 1; + elem_mbs_len = extra[idx++]; + if (mbs_len == elem_mbs_len) + { + for (mbs_cnt = 0; mbs_cnt < elem_mbs_len; ++mbs_cnt) + if (extra[idx + mbs_cnt] != mbs[mbs_cnt]) + break; + if (mbs_cnt == elem_mbs_len) + /* Found the entry. */ + found = true; + } + /* Skip the byte sequence of the collating element. */ + idx += elem_mbs_len; + /* Adjust for the alignment. */ + idx = (idx + 3) & ~3; + /* Skip the collation sequence value. */ + idx += sizeof (uint32_t); + /* Skip the wide char sequence of the collating element. */ + idx = idx + sizeof (uint32_t) * (extra[idx] + 1); + /* If we found the entry, return the sequence value. */ + if (found) + return *(uint32_t *) (extra + idx); + /* Skip the collation sequence value. */ + idx += sizeof (uint32_t); + } + return UINT_MAX; + } +} +# endif /* _LIBC */ +#endif /* RE_ENABLE_I18N */ + +/* Check whether the node accepts the byte which is IDX-th + byte of the INPUT. */ + +static bool +internal_function +check_node_accept (const re_match_context_t *mctx, const re_token_t *node, + Idx idx) +{ + unsigned char ch; + ch = re_string_byte_at (&mctx->input, idx); + switch (node->type) + { + case CHARACTER: + if (node->opr.c != ch) + return false; + break; + + case SIMPLE_BRACKET: + if (!bitset_contain (node->opr.sbcset, ch)) + return false; + break; + +#ifdef RE_ENABLE_I18N + case OP_UTF8_PERIOD: + if (ch >= ASCII_CHARS) + return false; + /* FALLTHROUGH */ +#endif + case OP_PERIOD: + if ((ch == '\n' && !(mctx->dfa->syntax & RE_DOT_NEWLINE)) + || (ch == '\0' && (mctx->dfa->syntax & RE_DOT_NOT_NULL))) + return false; + break; + + default: + return false; + } + + if (node->constraint) + { + /* The node has constraints. Check whether the current context + satisfies the constraints. */ + unsigned int context = re_string_context_at (&mctx->input, idx, + mctx->eflags); + if (NOT_SATISFY_NEXT_CONSTRAINT (node->constraint, context)) + return false; + } + + return true; +} + +/* Extend the buffers, if the buffers have run out. */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +extend_buffers (re_match_context_t *mctx) +{ + reg_errcode_t ret; + re_string_t *pstr = &mctx->input; + + /* Avoid overflow. */ + if (BE (SIZE_MAX / 2 / sizeof (re_dfastate_t *) <= pstr->bufs_len, 0)) + return REG_ESPACE; + + /* Double the lengthes of the buffers. */ + ret = re_string_realloc_buffers (pstr, pstr->bufs_len * 2); + if (BE (ret != REG_NOERROR, 0)) + return ret; + + if (mctx->state_log != NULL) + { + /* And double the length of state_log. */ + /* XXX We have no indication of the size of this buffer. If this + allocation fail we have no indication that the state_log array + does not have the right size. */ + re_dfastate_t **new_array = re_realloc (mctx->state_log, re_dfastate_t *, + pstr->bufs_len + 1); + if (BE (new_array == NULL, 0)) + return REG_ESPACE; + mctx->state_log = new_array; + } + + /* Then reconstruct the buffers. */ + if (pstr->icase) + { +#ifdef RE_ENABLE_I18N + if (pstr->mb_cur_max > 1) + { + ret = build_wcs_upper_buffer (pstr); + if (BE (ret != REG_NOERROR, 0)) + return ret; + } + else +#endif /* RE_ENABLE_I18N */ + build_upper_buffer (pstr); + } + else + { +#ifdef RE_ENABLE_I18N + if (pstr->mb_cur_max > 1) + build_wcs_buffer (pstr); + else +#endif /* RE_ENABLE_I18N */ + { + if (pstr->trans != NULL) + re_string_translate_buffer (pstr); + } + } + return REG_NOERROR; +} + + +/* Functions for matching context. */ + +/* Initialize MCTX. */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +match_ctx_init (re_match_context_t *mctx, int eflags, Idx n) +{ + mctx->eflags = eflags; + mctx->match_last = REG_MISSING; + if (n > 0) + { + /* Avoid overflow. */ + size_t max_object_size = + MAX (sizeof (struct re_backref_cache_entry), + sizeof (re_sub_match_top_t *)); + if (BE (SIZE_MAX / max_object_size < n, 0)) + return REG_ESPACE; + + mctx->bkref_ents = re_malloc (struct re_backref_cache_entry, n); + mctx->sub_tops = re_malloc (re_sub_match_top_t *, n); + if (BE (mctx->bkref_ents == NULL || mctx->sub_tops == NULL, 0)) + return REG_ESPACE; + } + /* Already zero-ed by the caller. + else + mctx->bkref_ents = NULL; + mctx->nbkref_ents = 0; + mctx->nsub_tops = 0; */ + mctx->abkref_ents = n; + mctx->max_mb_elem_len = 1; + mctx->asub_tops = n; + return REG_NOERROR; +} + +/* Clean the entries which depend on the current input in MCTX. + This function must be invoked when the matcher changes the start index + of the input, or changes the input string. */ + +static void +internal_function +match_ctx_clean (re_match_context_t *mctx) +{ + Idx st_idx; + for (st_idx = 0; st_idx < mctx->nsub_tops; ++st_idx) + { + Idx sl_idx; + re_sub_match_top_t *top = mctx->sub_tops[st_idx]; + for (sl_idx = 0; sl_idx < top->nlasts; ++sl_idx) + { + re_sub_match_last_t *last = top->lasts[sl_idx]; + re_free (last->path.array); + re_free (last); + } + re_free (top->lasts); + if (top->path) + { + re_free (top->path->array); + re_free (top->path); + } + free (top); + } + + mctx->nsub_tops = 0; + mctx->nbkref_ents = 0; +} + +/* Free all the memory associated with MCTX. */ + +static void +internal_function +match_ctx_free (re_match_context_t *mctx) +{ + /* First, free all the memory associated with MCTX->SUB_TOPS. */ + match_ctx_clean (mctx); + re_free (mctx->sub_tops); + re_free (mctx->bkref_ents); +} + +/* Add a new backreference entry to MCTX. + Note that we assume that caller never call this function with duplicate + entry, and call with STR_IDX which isn't smaller than any existing entry. +*/ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +match_ctx_add_entry (re_match_context_t *mctx, Idx node, Idx str_idx, Idx from, + Idx to) +{ + if (mctx->nbkref_ents >= mctx->abkref_ents) + { + struct re_backref_cache_entry* new_entry; + new_entry = re_realloc (mctx->bkref_ents, struct re_backref_cache_entry, + mctx->abkref_ents * 2); + if (BE (new_entry == NULL, 0)) + { + re_free (mctx->bkref_ents); + return REG_ESPACE; + } + mctx->bkref_ents = new_entry; + memset (mctx->bkref_ents + mctx->nbkref_ents, '\0', + sizeof (struct re_backref_cache_entry) * mctx->abkref_ents); + mctx->abkref_ents *= 2; + } + if (mctx->nbkref_ents > 0 + && mctx->bkref_ents[mctx->nbkref_ents - 1].str_idx == str_idx) + mctx->bkref_ents[mctx->nbkref_ents - 1].more = 1; + + mctx->bkref_ents[mctx->nbkref_ents].node = node; + mctx->bkref_ents[mctx->nbkref_ents].str_idx = str_idx; + mctx->bkref_ents[mctx->nbkref_ents].subexp_from = from; + mctx->bkref_ents[mctx->nbkref_ents].subexp_to = to; + + /* This is a cache that saves negative results of check_dst_limits_calc_pos. + If bit N is clear, means that this entry won't epsilon-transition to + an OP_OPEN_SUBEXP or OP_CLOSE_SUBEXP for the N+1-th subexpression. If + it is set, check_dst_limits_calc_pos_1 will recurse and try to find one + such node. + + A backreference does not epsilon-transition unless it is empty, so set + to all zeros if FROM != TO. */ + mctx->bkref_ents[mctx->nbkref_ents].eps_reachable_subexps_map + = (from == to ? -1 : 0); + + mctx->bkref_ents[mctx->nbkref_ents++].more = 0; + if (mctx->max_mb_elem_len < to - from) + mctx->max_mb_elem_len = to - from; + return REG_NOERROR; +} + +/* Return the first entry with the same str_idx, or REG_MISSING if none is + found. Note that MCTX->BKREF_ENTS is already sorted by MCTX->STR_IDX. */ + +static Idx +internal_function +search_cur_bkref_entry (const re_match_context_t *mctx, Idx str_idx) +{ + Idx left, right, mid, last; + last = right = mctx->nbkref_ents; + for (left = 0; left < right;) + { + mid = (left + right) / 2; + if (mctx->bkref_ents[mid].str_idx < str_idx) + left = mid + 1; + else + right = mid; + } + if (left < last && mctx->bkref_ents[left].str_idx == str_idx) + return left; + else + return REG_MISSING; +} + +/* Register the node NODE, whose type is OP_OPEN_SUBEXP, and which matches + at STR_IDX. */ + +static reg_errcode_t +internal_function __attribute_warn_unused_result__ +match_ctx_add_subtop (re_match_context_t *mctx, Idx node, Idx str_idx) +{ +#ifdef DEBUG + assert (mctx->sub_tops != NULL); + assert (mctx->asub_tops > 0); +#endif + if (BE (mctx->nsub_tops == mctx->asub_tops, 0)) + { + Idx new_asub_tops = mctx->asub_tops * 2; + re_sub_match_top_t **new_array = re_realloc (mctx->sub_tops, + re_sub_match_top_t *, + new_asub_tops); + if (BE (new_array == NULL, 0)) + return REG_ESPACE; + mctx->sub_tops = new_array; + mctx->asub_tops = new_asub_tops; + } + mctx->sub_tops[mctx->nsub_tops] = calloc (1, sizeof (re_sub_match_top_t)); + if (BE (mctx->sub_tops[mctx->nsub_tops] == NULL, 0)) + return REG_ESPACE; + mctx->sub_tops[mctx->nsub_tops]->node = node; + mctx->sub_tops[mctx->nsub_tops++]->str_idx = str_idx; + return REG_NOERROR; +} + +/* Register the node NODE, whose type is OP_CLOSE_SUBEXP, and which matches + at STR_IDX, whose corresponding OP_OPEN_SUBEXP is SUB_TOP. */ + +static re_sub_match_last_t * +internal_function +match_ctx_add_sublast (re_sub_match_top_t *subtop, Idx node, Idx str_idx) +{ + re_sub_match_last_t *new_entry; + if (BE (subtop->nlasts == subtop->alasts, 0)) + { + Idx new_alasts = 2 * subtop->alasts + 1; + re_sub_match_last_t **new_array = re_realloc (subtop->lasts, + re_sub_match_last_t *, + new_alasts); + if (BE (new_array == NULL, 0)) + return NULL; + subtop->lasts = new_array; + subtop->alasts = new_alasts; + } + new_entry = calloc (1, sizeof (re_sub_match_last_t)); + if (BE (new_entry != NULL, 1)) + { + subtop->lasts[subtop->nlasts] = new_entry; + new_entry->node = node; + new_entry->str_idx = str_idx; + ++subtop->nlasts; + } + return new_entry; +} + +static void +internal_function +sift_ctx_init (re_sift_context_t *sctx, re_dfastate_t **sifted_sts, + re_dfastate_t **limited_sts, Idx last_node, Idx last_str_idx) +{ + sctx->sifted_states = sifted_sts; + sctx->limited_states = limited_sts; + sctx->last_node = last_node; + sctx->last_str_idx = last_str_idx; + re_node_set_init_empty (&sctx->limits); +} diff --git a/lib/strcasecmp.c b/lib/strcasecmp.c new file mode 100644 index 000000000..1ed517531 --- /dev/null +++ b/lib/strcasecmp.c @@ -0,0 +1,63 @@ +/* Case-insensitive string comparison function. + Copyright (C) 1998-1999, 2005-2007, 2009-2012 Free Software Foundation, Inc. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2, or (at your option) + any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with this program; if not, write to the Free Software Foundation, + Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ + +#include <config.h> + +/* Specification. */ +#include <string.h> + +#include <ctype.h> +#include <limits.h> + +#define TOLOWER(Ch) (isupper (Ch) ? tolower (Ch) : (Ch)) + +/* Compare strings S1 and S2, ignoring case, returning less than, equal to or + greater than zero if S1 is lexicographically less than, equal to or greater + than S2. + Note: This function does not work with multibyte strings! */ + +int +strcasecmp (const char *s1, const char *s2) +{ + const unsigned char *p1 = (const unsigned char *) s1; + const unsigned char *p2 = (const unsigned char *) s2; + unsigned char c1, c2; + + if (p1 == p2) + return 0; + + do + { + c1 = TOLOWER (*p1); + c2 = TOLOWER (*p2); + + if (c1 == '\0') + break; + + ++p1; + ++p2; + } + while (c1 == c2); + + if (UCHAR_MAX <= INT_MAX) + return c1 - c2; + else + /* On machines where 'char' and 'int' are types of the same size, the + difference of two 'unsigned char' values - including the sign bit - + doesn't fit in an 'int'. */ + return (c1 > c2 ? 1 : c1 < c2 ? -1 : 0); +} diff --git a/lib/streq.h b/lib/streq.h new file mode 100644 index 000000000..b64941116 --- /dev/null +++ b/lib/streq.h @@ -0,0 +1,176 @@ +/* Optimized string comparison. + Copyright (C) 2001-2002, 2007, 2009-2012 Free Software Foundation, Inc. + + This program is free software: you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published + by the Free Software Foundation; either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. */ + +/* Written by Bruno Haible <bruno@clisp.org>. */ + +#ifndef _GL_STREQ_H +#define _GL_STREQ_H + +#include <string.h> + +/* STREQ allows to optimize string comparison with a small literal string. + STREQ (s, "EUC-KR", 'E', 'U', 'C', '-', 'K', 'R', 0, 0, 0) + is semantically equivalent to + strcmp (s, "EUC-KR") == 0 + just faster. */ + +/* Help GCC to generate good code for string comparisons with + immediate strings. */ +#if defined (__GNUC__) && defined (__OPTIMIZE__) + +static inline int +streq9 (const char *s1, const char *s2) +{ + return strcmp (s1 + 9, s2 + 9) == 0; +} + +static inline int +streq8 (const char *s1, const char *s2, char s28) +{ + if (s1[8] == s28) + { + if (s28 == 0) + return 1; + else + return streq9 (s1, s2); + } + else + return 0; +} + +static inline int +streq7 (const char *s1, const char *s2, char s27, char s28) +{ + if (s1[7] == s27) + { + if (s27 == 0) + return 1; + else + return streq8 (s1, s2, s28); + } + else + return 0; +} + +static inline int +streq6 (const char *s1, const char *s2, char s26, char s27, char s28) +{ + if (s1[6] == s26) + { + if (s26 == 0) + return 1; + else + return streq7 (s1, s2, s27, s28); + } + else + return 0; +} + +static inline int +streq5 (const char *s1, const char *s2, char s25, char s26, char s27, char s28) +{ + if (s1[5] == s25) + { + if (s25 == 0) + return 1; + else + return streq6 (s1, s2, s26, s27, s28); + } + else + return 0; +} + +static inline int +streq4 (const char *s1, const char *s2, char s24, char s25, char s26, char s27, char s28) +{ + if (s1[4] == s24) + { + if (s24 == 0) + return 1; + else + return streq5 (s1, s2, s25, s26, s27, s28); + } + else + return 0; +} + +static inline int +streq3 (const char *s1, const char *s2, char s23, char s24, char s25, char s26, char s27, char s28) +{ + if (s1[3] == s23) + { + if (s23 == 0) + return 1; + else + return streq4 (s1, s2, s24, s25, s26, s27, s28); + } + else + return 0; +} + +static inline int +streq2 (const char *s1, const char *s2, char s22, char s23, char s24, char s25, char s26, char s27, char s28) +{ + if (s1[2] == s22) + { + if (s22 == 0) + return 1; + else + return streq3 (s1, s2, s23, s24, s25, s26, s27, s28); + } + else + return 0; +} + +static inline int +streq1 (const char *s1, const char *s2, char s21, char s22, char s23, char s24, char s25, char s26, char s27, char s28) +{ + if (s1[1] == s21) + { + if (s21 == 0) + return 1; + else + return streq2 (s1, s2, s22, s23, s24, s25, s26, s27, s28); + } + else + return 0; +} + +static inline int +streq0 (const char *s1, const char *s2, char s20, char s21, char s22, char s23, char s24, char s25, char s26, char s27, char s28) +{ + if (s1[0] == s20) + { + if (s20 == 0) + return 1; + else + return streq1 (s1, s2, s21, s22, s23, s24, s25, s26, s27, s28); + } + else + return 0; +} + +#define STREQ(s1,s2,s20,s21,s22,s23,s24,s25,s26,s27,s28) \ + streq0 (s1, s2, s20, s21, s22, s23, s24, s25, s26, s27, s28) + +#else + +#define STREQ(s1,s2,s20,s21,s22,s23,s24,s25,s26,s27,s28) \ + (strcmp (s1, s2) == 0) + +#endif + +#endif /* _GL_STREQ_H */ diff --git a/lib/strings.in.h b/lib/strings.in.h new file mode 100644 index 000000000..fc60919b7 --- /dev/null +++ b/lib/strings.in.h @@ -0,0 +1,123 @@ +/* A substitute <strings.h>. + + Copyright (C) 2007-2012 Free Software Foundation, Inc. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2, or (at your option) + any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with this program; if not, write to the Free Software Foundation, + Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ + +#ifndef _@GUARD_PREFIX@_STRINGS_H + +#if __GNUC__ >= 3 +@PRAGMA_SYSTEM_HEADER@ +#endif +@PRAGMA_COLUMNS@ + +/* Minix 3.1.8 has a bug: <sys/types.h> must be included before <strings.h>. + But avoid namespace pollution on glibc systems. */ +#if defined __minix && !defined __GLIBC__ +# include <sys/types.h> +#endif + +/* The include_next requires a split double-inclusion guard. */ +#if @HAVE_STRINGS_H@ +# @INCLUDE_NEXT@ @NEXT_STRINGS_H@ +#endif + +#ifndef _@GUARD_PREFIX@_STRINGS_H +#define _@GUARD_PREFIX@_STRINGS_H + +#if ! @HAVE_DECL_STRNCASECMP@ +/* Get size_t. */ +# include <stddef.h> +#endif + + +/* The definitions of _GL_FUNCDECL_RPL etc. are copied here. */ + +/* The definition of _GL_ARG_NONNULL is copied here. */ + +/* The definition of _GL_WARN_ON_USE is copied here. */ + +#ifdef __cplusplus +extern "C" { +#endif + + + /* Find the index of the least-significant set bit. */ +#if @GNULIB_FFS@ +# if !@HAVE_FFS@ +_GL_FUNCDECL_SYS (ffs, int, (int i)); +# endif +_GL_CXXALIAS_SYS (ffs, int, (int i)); +_GL_CXXALIASWARN (ffs); +#elif defined GNULIB_POSIXCHECK +# undef ffs +# if HAVE_RAW_DECL_FFS +_GL_WARN_ON_USE (ffs, "ffs is not portable - use the ffs module"); +# endif +#endif + +/* Compare strings S1 and S2, ignoring case, returning less than, equal to or + greater than zero if S1 is lexicographically less than, equal to or greater + than S2. + Note: This function does not work in multibyte locales. */ +#if ! @HAVE_STRCASECMP@ +extern int strcasecmp (char const *s1, char const *s2) + _GL_ARG_NONNULL ((1, 2)); +#endif +#if defined GNULIB_POSIXCHECK +/* strcasecmp() does not work with multibyte strings: + POSIX says that it operates on "strings", and "string" in POSIX is defined + as a sequence of bytes, not of characters. */ +# undef strcasecmp +# if HAVE_RAW_DECL_STRCASECMP +_GL_WARN_ON_USE (strcasecmp, "strcasecmp cannot work correctly on character " + "strings in multibyte locales - " + "use mbscasecmp if you care about " + "internationalization, or use c_strcasecmp , " + "gnulib module c-strcase) if you want a locale " + "independent function"); +# endif +#endif + +/* Compare no more than N bytes of strings S1 and S2, ignoring case, + returning less than, equal to or greater than zero if S1 is + lexicographically less than, equal to or greater than S2. + Note: This function cannot work correctly in multibyte locales. */ +#if ! @HAVE_DECL_STRNCASECMP@ +extern int strncasecmp (char const *s1, char const *s2, size_t n) + _GL_ARG_NONNULL ((1, 2)); +#endif +#if defined GNULIB_POSIXCHECK +/* strncasecmp() does not work with multibyte strings: + POSIX says that it operates on "strings", and "string" in POSIX is defined + as a sequence of bytes, not of characters. */ +# undef strncasecmp +# if HAVE_RAW_DECL_STRNCASECMP +_GL_WARN_ON_USE (strncasecmp, "strncasecmp cannot work correctly on character " + "strings in multibyte locales - " + "use mbsncasecmp or mbspcasecmp if you care about " + "internationalization, or use c_strncasecmp , " + "gnulib module c-strcase) if you want a locale " + "independent function"); +# endif +#endif + + +#ifdef __cplusplus +} +#endif + +#endif /* _@GUARD_PREFIX@_STRING_H */ +#endif /* _@GUARD_PREFIX@_STRING_H */ diff --git a/lib/strncasecmp.c b/lib/strncasecmp.c new file mode 100644 index 000000000..d330ff678 --- /dev/null +++ b/lib/strncasecmp.c @@ -0,0 +1,63 @@ +/* strncasecmp.c -- case insensitive string comparator + Copyright (C) 1998-1999, 2005-2007, 2009-2012 Free Software Foundation, Inc. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2, or (at your option) + any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with this program; if not, write to the Free Software Foundation, + Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ + +#include <config.h> + +/* Specification. */ +#include <string.h> + +#include <ctype.h> +#include <limits.h> + +#define TOLOWER(Ch) (isupper (Ch) ? tolower (Ch) : (Ch)) + +/* Compare no more than N bytes of strings S1 and S2, ignoring case, + returning less than, equal to or greater than zero if S1 is + lexicographically less than, equal to or greater than S2. + Note: This function cannot work correctly in multibyte locales. */ + +int +strncasecmp (const char *s1, const char *s2, size_t n) +{ + register const unsigned char *p1 = (const unsigned char *) s1; + register const unsigned char *p2 = (const unsigned char *) s2; + unsigned char c1, c2; + + if (p1 == p2 || n == 0) + return 0; + + do + { + c1 = TOLOWER (*p1); + c2 = TOLOWER (*p2); + + if (--n == 0 || c1 == '\0') + break; + + ++p1; + ++p2; + } + while (c1 == c2); + + if (UCHAR_MAX <= INT_MAX) + return c1 - c2; + else + /* On machines where 'char' and 'int' are types of the same size, the + difference of two 'unsigned char' values - including the sign bit - + doesn't fit in an 'int'. */ + return (c1 > c2 ? 1 : c1 < c2 ? -1 : 0); +} diff --git a/lib/wcrtomb.c b/lib/wcrtomb.c new file mode 100644 index 000000000..a4d6bcdbd --- /dev/null +++ b/lib/wcrtomb.c @@ -0,0 +1,53 @@ +/* Convert wide character to multibyte character. + Copyright (C) 2008-2012 Free Software Foundation, Inc. + Written by Bruno Haible <bruno@clisp.org>, 2008. + + This program is free software: you can redistribute it and/or modify + it under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with this program. If not, see <http://www.gnu.org/licenses/>. */ + +#include <config.h> + +/* Specification. */ +#include <wchar.h> + +#include <errno.h> +#include <stdlib.h> + + +size_t +wcrtomb (char *s, wchar_t wc, mbstate_t *ps) +{ + /* This implementation of wcrtomb on top of wctomb() supports only + stateless encodings. ps must be in the initial state. */ + if (ps != NULL && !mbsinit (ps)) + { + errno = EINVAL; + return (size_t)(-1); + } + + if (s == NULL) + /* We know the NUL wide character corresponds to the NUL character. */ + return 1; + else + { + int ret = wctomb (s, wc); + + if (ret >= 0) + return ret; + else + { + errno = EILSEQ; + return (size_t)(-1); + } + } +} diff --git a/lib/wctype.in.h b/lib/wctype.in.h new file mode 100644 index 000000000..f25ad30e2 --- /dev/null +++ b/lib/wctype.in.h @@ -0,0 +1,499 @@ +/* A substitute for ISO C99 <wctype.h>, for platforms that lack it. + + Copyright (C) 2006-2012 Free Software Foundation, Inc. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2, or (at your option) + any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with this program; if not, write to the Free Software Foundation, + Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */ + +/* Written by Bruno Haible and Paul Eggert. */ + +/* + * ISO C 99 <wctype.h> for platforms that lack it. + * <http://www.opengroup.org/susv3xbd/wctype.h.html> + * + * iswctype, towctrans, towlower, towupper, wctrans, wctype, + * wctrans_t, and wctype_t are not yet implemented. + */ + +#ifndef _@GUARD_PREFIX@_WCTYPE_H + +#if __GNUC__ >= 3 +@PRAGMA_SYSTEM_HEADER@ +#endif +@PRAGMA_COLUMNS@ + +#if @HAVE_WINT_T@ +/* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. + Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before + <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> must be + included before <wchar.h>. */ +# include <stddef.h> +# include <stdio.h> +# include <time.h> +# include <wchar.h> +#endif + +/* Include the original <wctype.h> if it exists. + BeOS 5 has the functions but no <wctype.h>. */ +/* The include_next requires a split double-inclusion guard. */ +#if @HAVE_WCTYPE_H@ +# @INCLUDE_NEXT@ @NEXT_WCTYPE_H@ +#endif + +#ifndef _@GUARD_PREFIX@_WCTYPE_H +#define _@GUARD_PREFIX@_WCTYPE_H + +/* The definitions of _GL_FUNCDECL_RPL etc. are copied here. */ + +/* The definition of _GL_WARN_ON_USE is copied here. */ + +/* Solaris 2.6 <wctype.h> includes <widec.h> which includes <euc.h> which + #defines a number of identifiers in the application namespace. Revert + these #defines. */ +#ifdef __sun +# undef multibyte +# undef eucw1 +# undef eucw2 +# undef eucw3 +# undef scrw1 +# undef scrw2 +# undef scrw3 +#endif + +/* Define wint_t and WEOF. (Also done in wchar.in.h.) */ +#if !@HAVE_WINT_T@ && !defined wint_t +# define wint_t int +# ifndef WEOF +# define WEOF -1 +# endif +#else +/* MSVC defines wint_t as 'unsigned short' in <crtdefs.h>. + This is too small: ISO C 99 section 7.24.1.(2) says that wint_t must be + "unchanged by default argument promotions". Override it. */ +# if defined _MSC_VER +# if !GNULIB_defined_wint_t +# include <crtdefs.h> +typedef unsigned int rpl_wint_t; +# undef wint_t +# define wint_t rpl_wint_t +# define GNULIB_defined_wint_t 1 +# endif +# endif +# ifndef WEOF +# define WEOF ((wint_t) -1) +# endif +#endif + + +#if !GNULIB_defined_wctype_functions + +/* FreeBSD 4.4 to 4.11 has <wctype.h> but lacks the functions. + Linux libc5 has <wctype.h> and the functions but they are broken. + Assume all 11 functions (all isw* except iswblank) are implemented the + same way, or not at all. */ +# if ! @HAVE_ISWCNTRL@ || @REPLACE_ISWCNTRL@ + +/* IRIX 5.3 has macros but no functions, its isw* macros refer to an + undefined variable _ctmp_ and to <ctype.h> macros like _P, and they + refer to system functions like _iswctype that are not in the + standard C library. Rather than try to get ancient buggy + implementations like this to work, just disable them. */ +# undef iswalnum +# undef iswalpha +# undef iswblank +# undef iswcntrl +# undef iswdigit +# undef iswgraph +# undef iswlower +# undef iswprint +# undef iswpunct +# undef iswspace +# undef iswupper +# undef iswxdigit +# undef towlower +# undef towupper + +/* Linux libc5 has <wctype.h> and the functions but they are broken. */ +# if @REPLACE_ISWCNTRL@ +# if !(defined __cplusplus && defined GNULIB_NAMESPACE) +# define iswalnum rpl_iswalnum +# define iswalpha rpl_iswalpha +# define iswblank rpl_iswblank +# define iswcntrl rpl_iswcntrl +# define iswdigit rpl_iswdigit +# define iswgraph rpl_iswgraph +# define iswlower rpl_iswlower +# define iswprint rpl_iswprint +# define iswpunct rpl_iswpunct +# define iswspace rpl_iswspace +# define iswupper rpl_iswupper +# define iswxdigit rpl_iswxdigit +# endif +# endif +# if @REPLACE_TOWLOWER@ +# if !(defined __cplusplus && defined GNULIB_NAMESPACE) +# define towlower rpl_towlower +# define towupper rpl_towupper +# endif +# endif + +static inline int +# if @REPLACE_ISWCNTRL@ +rpl_iswalnum +# else +iswalnum +# endif + (wint_t wc) +{ + return ((wc >= '0' && wc <= '9') + || ((wc & ~0x20) >= 'A' && (wc & ~0x20) <= 'Z')); +} + +static inline int +# if @REPLACE_ISWCNTRL@ +rpl_iswalpha +# else +iswalpha +# endif + (wint_t wc) +{ + return (wc & ~0x20) >= 'A' && (wc & ~0x20) <= 'Z'; +} + +static inline int +# if @REPLACE_ISWCNTRL@ +rpl_iswblank +# else +iswblank +# endif + (wint_t wc) +{ + return wc == ' ' || wc == '\t'; +} + +static inline int +# if @REPLACE_ISWCNTRL@ +rpl_iswcntrl +# else +iswcntrl +# endif + (wint_t wc) +{ + return (wc & ~0x1f) == 0 || wc == 0x7f; +} + +static inline int +# if @REPLACE_ISWCNTRL@ +rpl_iswdigit +# else +iswdigit +# endif + (wint_t wc) +{ + return wc >= '0' && wc <= '9'; +} + +static inline int +# if @REPLACE_ISWCNTRL@ +rpl_iswgraph +# else +iswgraph +# endif + (wint_t wc) +{ + return wc >= '!' && wc <= '~'; +} + +static inline int +# if @REPLACE_ISWCNTRL@ +rpl_iswlower +# else +iswlower +# endif + (wint_t wc) +{ + return wc >= 'a' && wc <= 'z'; +} + +static inline int +# if @REPLACE_ISWCNTRL@ +rpl_iswprint +# else +iswprint +# endif + (wint_t wc) +{ + return wc >= ' ' && wc <= '~'; +} + +static inline int +# if @REPLACE_ISWCNTRL@ +rpl_iswpunct +# else +iswpunct +# endif + (wint_t wc) +{ + return (wc >= '!' && wc <= '~' + && !((wc >= '0' && wc <= '9') + || ((wc & ~0x20) >= 'A' && (wc & ~0x20) <= 'Z'))); +} + +static inline int +# if @REPLACE_ISWCNTRL@ +rpl_iswspace +# else +iswspace +# endif + (wint_t wc) +{ + return (wc == ' ' || wc == '\t' + || wc == '\n' || wc == '\v' || wc == '\f' || wc == '\r'); +} + +static inline int +# if @REPLACE_ISWCNTRL@ +rpl_iswupper +# else +iswupper +# endif + (wint_t wc) +{ + return wc >= 'A' && wc <= 'Z'; +} + +static inline int +# if @REPLACE_ISWCNTRL@ +rpl_iswxdigit +# else +iswxdigit +# endif + (wint_t wc) +{ + return ((wc >= '0' && wc <= '9') + || ((wc & ~0x20) >= 'A' && (wc & ~0x20) <= 'F')); +} + +static inline wint_t +# if @REPLACE_TOWLOWER@ +rpl_towlower +# else +towlower +# endif + (wint_t wc) +{ + return (wc >= 'A' && wc <= 'Z' ? wc - 'A' + 'a' : wc); +} + +static inline wint_t +# if @REPLACE_TOWLOWER@ +rpl_towupper +# else +towupper +# endif + (wint_t wc) +{ + return (wc >= 'a' && wc <= 'z' ? wc - 'a' + 'A' : wc); +} + +# elif @GNULIB_ISWBLANK@ && (! @HAVE_ISWBLANK@ || @REPLACE_ISWBLANK@) +/* Only the iswblank function is missing. */ + +# if @REPLACE_ISWBLANK@ +# if !(defined __cplusplus && defined GNULIB_NAMESPACE) +# define iswblank rpl_iswblank +# endif +_GL_FUNCDECL_RPL (iswblank, int, (wint_t wc)); +# else +_GL_FUNCDECL_SYS (iswblank, int, (wint_t wc)); +# endif + +# endif + +# if defined __MINGW32__ + +/* On native Windows, wchar_t is uint16_t, and wint_t is uint32_t. + The functions towlower and towupper are implemented in the MSVCRT library + to take a wchar_t argument and return a wchar_t result. mingw declares + these functions to take a wint_t argument and return a wint_t result. + This means that: + 1. When the user passes an argument outside the range 0x0000..0xFFFF, the + function will look only at the lower 16 bits. This is allowed according + to POSIX. + 2. The return value is returned in the lower 16 bits of the result register. + The upper 16 bits are random: whatever happened to be in that part of the + result register. We need to fix this by adding a zero-extend from + wchar_t to wint_t after the call. */ + +static inline wint_t +rpl_towlower (wint_t wc) +{ + return (wint_t) (wchar_t) towlower (wc); +} +# if !(defined __cplusplus && defined GNULIB_NAMESPACE) +# define towlower rpl_towlower +# endif + +static inline wint_t +rpl_towupper (wint_t wc) +{ + return (wint_t) (wchar_t) towupper (wc); +} +# if !(defined __cplusplus && defined GNULIB_NAMESPACE) +# define towupper rpl_towupper +# endif + +# endif /* __MINGW32__ */ + +# define GNULIB_defined_wctype_functions 1 +#endif + +#if @REPLACE_ISWCNTRL@ +_GL_CXXALIAS_RPL (iswalnum, int, (wint_t wc)); +_GL_CXXALIAS_RPL (iswalpha, int, (wint_t wc)); +_GL_CXXALIAS_RPL (iswcntrl, int, (wint_t wc)); +_GL_CXXALIAS_RPL (iswdigit, int, (wint_t wc)); +_GL_CXXALIAS_RPL (iswgraph, int, (wint_t wc)); +_GL_CXXALIAS_RPL (iswlower, int, (wint_t wc)); +_GL_CXXALIAS_RPL (iswprint, int, (wint_t wc)); +_GL_CXXALIAS_RPL (iswpunct, int, (wint_t wc)); +_GL_CXXALIAS_RPL (iswspace, int, (wint_t wc)); +_GL_CXXALIAS_RPL (iswupper, int, (wint_t wc)); +_GL_CXXALIAS_RPL (iswxdigit, int, (wint_t wc)); +#else +_GL_CXXALIAS_SYS (iswalnum, int, (wint_t wc)); +_GL_CXXALIAS_SYS (iswalpha, int, (wint_t wc)); +_GL_CXXALIAS_SYS (iswcntrl, int, (wint_t wc)); +_GL_CXXALIAS_SYS (iswdigit, int, (wint_t wc)); +_GL_CXXALIAS_SYS (iswgraph, int, (wint_t wc)); +_GL_CXXALIAS_SYS (iswlower, int, (wint_t wc)); +_GL_CXXALIAS_SYS (iswprint, int, (wint_t wc)); +_GL_CXXALIAS_SYS (iswpunct, int, (wint_t wc)); +_GL_CXXALIAS_SYS (iswspace, int, (wint_t wc)); +_GL_CXXALIAS_SYS (iswupper, int, (wint_t wc)); +_GL_CXXALIAS_SYS (iswxdigit, int, (wint_t wc)); +#endif +_GL_CXXALIASWARN (iswalnum); +_GL_CXXALIASWARN (iswalpha); +_GL_CXXALIASWARN (iswcntrl); +_GL_CXXALIASWARN (iswdigit); +_GL_CXXALIASWARN (iswgraph); +_GL_CXXALIASWARN (iswlower); +_GL_CXXALIASWARN (iswprint); +_GL_CXXALIASWARN (iswpunct); +_GL_CXXALIASWARN (iswspace); +_GL_CXXALIASWARN (iswupper); +_GL_CXXALIASWARN (iswxdigit); + +#if @GNULIB_ISWBLANK@ +# if @REPLACE_ISWCNTRL@ || @REPLACE_ISWBLANK@ +_GL_CXXALIAS_RPL (iswblank, int, (wint_t wc)); +# else +_GL_CXXALIAS_SYS (iswblank, int, (wint_t wc)); +# endif +_GL_CXXALIASWARN (iswblank); +#endif + +#if !@HAVE_WCTYPE_T@ +# if !GNULIB_defined_wctype_t +typedef void * wctype_t; +# define GNULIB_defined_wctype_t 1 +# endif +#endif + +/* Get a descriptor for a wide character property. */ +#if @GNULIB_WCTYPE@ +# if !@HAVE_WCTYPE_T@ +_GL_FUNCDECL_SYS (wctype, wctype_t, (const char *name)); +# endif +_GL_CXXALIAS_SYS (wctype, wctype_t, (const char *name)); +_GL_CXXALIASWARN (wctype); +#elif defined GNULIB_POSIXCHECK +# undef wctype +# if HAVE_RAW_DECL_WCTYPE +_GL_WARN_ON_USE (wctype, "wctype is unportable - " + "use gnulib module wctype for portability"); +# endif +#endif + +/* Test whether a wide character has a given property. + The argument WC must be either a wchar_t value or WEOF. + The argument DESC must have been returned by the wctype() function. */ +#if @GNULIB_ISWCTYPE@ +# if !@HAVE_WCTYPE_T@ +_GL_FUNCDECL_SYS (iswctype, int, (wint_t wc, wctype_t desc)); +# endif +_GL_CXXALIAS_SYS (iswctype, int, (wint_t wc, wctype_t desc)); +_GL_CXXALIASWARN (iswctype); +#elif defined GNULIB_POSIXCHECK +# undef iswctype +# if HAVE_RAW_DECL_ISWCTYPE +_GL_WARN_ON_USE (iswctype, "iswctype is unportable - " + "use gnulib module iswctype for portability"); +# endif +#endif + +#if @REPLACE_TOWLOWER@ || defined __MINGW32__ +_GL_CXXALIAS_RPL (towlower, wint_t, (wint_t wc)); +_GL_CXXALIAS_RPL (towupper, wint_t, (wint_t wc)); +#else +_GL_CXXALIAS_SYS (towlower, wint_t, (wint_t wc)); +_GL_CXXALIAS_SYS (towupper, wint_t, (wint_t wc)); +#endif +_GL_CXXALIASWARN (towlower); +_GL_CXXALIASWARN (towupper); + +#if !@HAVE_WCTRANS_T@ +# if !GNULIB_defined_wctrans_t +typedef void * wctrans_t; +# define GNULIB_defined_wctrans_t 1 +# endif +#endif + +/* Get a descriptor for a wide character case conversion. */ +#if @GNULIB_WCTRANS@ +# if !@HAVE_WCTRANS_T@ +_GL_FUNCDECL_SYS (wctrans, wctrans_t, (const char *name)); +# endif +_GL_CXXALIAS_SYS (wctrans, wctrans_t, (const char *name)); +_GL_CXXALIASWARN (wctrans); +#elif defined GNULIB_POSIXCHECK +# undef wctrans +# if HAVE_RAW_DECL_WCTRANS +_GL_WARN_ON_USE (wctrans, "wctrans is unportable - " + "use gnulib module wctrans for portability"); +# endif +#endif + +/* Perform a given case conversion on a wide character. + The argument WC must be either a wchar_t value or WEOF. + The argument DESC must have been returned by the wctrans() function. */ +#if @GNULIB_TOWCTRANS@ +# if !@HAVE_WCTRANS_T@ +_GL_FUNCDECL_SYS (towctrans, wint_t, (wint_t wc, wctrans_t desc)); +# endif +_GL_CXXALIAS_SYS (towctrans, wint_t, (wint_t wc, wctrans_t desc)); +_GL_CXXALIASWARN (towctrans); +#elif defined GNULIB_POSIXCHECK +# undef towctrans +# if HAVE_RAW_DECL_TOWCTRANS +_GL_WARN_ON_USE (towctrans, "towctrans is unportable - " + "use gnulib module towctrans for portability"); +# endif +#endif + + +#endif /* _@GUARD_PREFIX@_WCTYPE_H */ +#endif /* _@GUARD_PREFIX@_WCTYPE_H */ diff --git a/libguile/init.c b/libguile/init.c index 633f8c681..f171950fa 100644 --- a/libguile/init.c +++ b/libguile/init.c @@ -1,5 +1,6 @@ -/* Copyright (C) 1995,1996,1997,1998,1999,2000,2001, 2002, 2003, 2004, 2006, 2009, 2010, 2011 Free Software Foundation, Inc. - * +/* Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, + * 2004, 2006, 2009, 2010, 2011, 2012 Free Software Foundation, Inc. + * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public License * as published by the Free Software Foundation; either version 3 of @@ -91,7 +92,7 @@ #include "libguile/poll.h" #include "libguile/ports.h" #include "libguile/posix.h" -#ifdef HAVE_REGCOMP +#ifdef ENABLE_REGEX #include "libguile/regex-posix.h" #endif #include "libguile/print.h" @@ -453,7 +454,7 @@ scm_i_init_guile (void *base) #ifdef HAVE_POSIX scm_init_posix (); #endif -#ifdef HAVE_REGCOMP +#ifdef ENABLE_REGEX scm_init_regex_posix (); /* Requires smob_prehistory */ #endif scm_init_procs (); diff --git a/libguile/regex-posix.c b/libguile/regex-posix.c index 34230994e..bec0f89fb 100644 --- a/libguile/regex-posix.c +++ b/libguile/regex-posix.c @@ -1,4 +1,4 @@ -/* Copyright (C) 1997, 1998, 1999, 2000, 2001, 2004, 2006, 2007, 2010, 2011 Free Software Foundation, Inc. +/* Copyright (C) 1997, 1998, 1999, 2000, 2001, 2004, 2006, 2007, 2010, 2011, 2012 Free Software Foundation, Inc. * * This library is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public License @@ -35,23 +35,7 @@ #include "libguile/_scm.h" -/* Supposedly, this file is never compiled unless we know we have - POSIX regular expressions. But we still put this in an #ifdef so - the file is CPP'able (for dependency scanning) even on systems that - don't have a <regex.h> header. */ -#ifdef HAVE_REGCOMP -#ifdef HAVE_REGEX_H #include <regex.h> -#else -#ifdef HAVE_RXPOSIX_H -#include <rxposix.h> /* GNU Rx library */ -#else -#ifdef HAVE_RX_RXPOSIX_H -#include <rx/rxposix.h> /* GNU Rx library on Linux */ -#endif -#endif -#endif -#endif #ifdef HAVE_WCHAR_H #include <wchar.h> diff --git a/m4/btowc.m4 b/m4/btowc.m4 new file mode 100644 index 000000000..e565321ce --- /dev/null +++ b/m4/btowc.m4 @@ -0,0 +1,116 @@ +# btowc.m4 serial 10 +dnl Copyright (C) 2008-2012 Free Software Foundation, Inc. +dnl This file is free software; the Free Software Foundation +dnl gives unlimited permission to copy and/or distribute it, +dnl with or without modifications, as long as this notice is preserved. + +AC_DEFUN([gl_FUNC_BTOWC], +[ + AC_REQUIRE([gl_WCHAR_H_DEFAULTS]) + + dnl Check whether <wchar.h> is usable at all, first. Otherwise the test + dnl program below may lead to an endless loop. See + dnl <http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42440>. + AC_REQUIRE([gl_WCHAR_H_INLINE_OK]) + + AC_CHECK_FUNCS_ONCE([btowc]) + if test $ac_cv_func_btowc = no; then + HAVE_BTOWC=0 + else + + AC_REQUIRE([AC_PROG_CC]) + AC_REQUIRE([gt_LOCALE_FR]) + AC_REQUIRE([AC_CANONICAL_HOST]) dnl for cross-compiles + + dnl Cygwin 1.7.2 btowc('\0') is WEOF, not 0. + AC_CACHE_CHECK([whether btowc(0) is correct], + [gl_cv_func_btowc_nul], + [ + AC_RUN_IFELSE( + [AC_LANG_SOURCE([[ +#include <string.h> +/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before + <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> must be + included before <wchar.h>. */ +#include <stddef.h> +#include <stdio.h> +#include <time.h> +#include <wchar.h> +int main () +{ + if (btowc ('\0') != 0) + return 1; + return 0; +}]])], + [gl_cv_func_btowc_nul=yes], + [gl_cv_func_btowc_nul=no], + [ +changequote(,)dnl + case "$host_os" in + # Guess no on Cygwin. + cygwin*) gl_cv_func_btowc_nul="guessing no" ;; + # Guess yes otherwise. + *) gl_cv_func_btowc_nul="guessing yes" ;; + esac +changequote([,])dnl + ]) + ]) + + dnl IRIX 6.5 btowc(EOF) is 0xFF, not WEOF. + AC_CACHE_CHECK([whether btowc(EOF) is correct], + [gl_cv_func_btowc_eof], + [ + dnl Initial guess, used when cross-compiling or when no suitable locale + dnl is present. +changequote(,)dnl + case "$host_os" in + # Guess no on IRIX. + irix*) gl_cv_func_btowc_eof="guessing no" ;; + # Guess yes otherwise. + *) gl_cv_func_btowc_eof="guessing yes" ;; + esac +changequote([,])dnl + if test $LOCALE_FR != none; then + AC_RUN_IFELSE( + [AC_LANG_SOURCE([[ +#include <locale.h> +#include <string.h> +/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before + <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> must be + included before <wchar.h>. */ +#include <stddef.h> +#include <stdio.h> +#include <time.h> +#include <wchar.h> +int main () +{ + if (setlocale (LC_ALL, "$LOCALE_FR") != NULL) + { + if (btowc (EOF) != WEOF) + return 1; + } + return 0; +}]])], + [gl_cv_func_btowc_eof=yes], + [gl_cv_func_btowc_eof=no], + [:]) + fi + ]) + + case "$gl_cv_func_btowc_nul" in + *yes) ;; + *) REPLACE_BTOWC=1 ;; + esac + case "$gl_cv_func_btowc_eof" in + *yes) ;; + *) REPLACE_BTOWC=1 ;; + esac + fi +]) + +# Prerequisites of lib/btowc.c. +AC_DEFUN([gl_PREREQ_BTOWC], [ + : +]) diff --git a/m4/gnulib-cache.m4 b/m4/gnulib-cache.m4 index 5aa46aba6..4ab00ddb1 100644 --- a/m4/gnulib-cache.m4 +++ b/m4/gnulib-cache.m4 @@ -27,7 +27,7 @@ # Specification in the form of a command-line invocation: -# gnulib-tool --import --dir=. --local-dir=gnulib-local --lib=libgnu --source-base=lib --m4-base=m4 --doc-base=doc --tests-base=tests --aux-dir=build-aux --lgpl=3 --no-conditional-dependencies --libtool --macro-prefix=gl --no-vc-files accept alignof alloca-opt announce-gen autobuild bind byteswap canonicalize-lgpl ceil close connect dirfd duplocale environ extensions flock floor fpieee frexp full-read full-write func gendocs getaddrinfo getpeername getsockname getsockopt git-version-gen gitlog-to-changelog gnu-web-doc-update gnupload havelib iconv_open-utf inet_ntop inet_pton isinf isnan ldexp lib-symbol-versions lib-symbol-visibility libunistring listen localcharset locale log1p maintainer-makefile malloc-gnu malloca nproc open pipe2 putenv recv recvfrom rename send sendto setenv setsockopt shutdown socket stat-time stdlib strftime striconveh string sys_stat trunc verify vsnprintf warnings wchar +# gnulib-tool --import --dir=. --local-dir=gnulib-local --lib=libgnu --source-base=lib --m4-base=m4 --doc-base=doc --tests-base=tests --aux-dir=build-aux --lgpl=3 --no-conditional-dependencies --libtool --macro-prefix=gl --no-vc-files accept alignof alloca-opt announce-gen autobuild bind byteswap canonicalize-lgpl ceil close connect dirfd duplocale environ extensions flock floor fpieee frexp full-read full-write func gendocs getaddrinfo getpeername getsockname getsockopt git-version-gen gitlog-to-changelog gnu-web-doc-update gnupload havelib iconv_open-utf inet_ntop inet_pton isinf isnan ldexp lib-symbol-versions lib-symbol-visibility libunistring listen localcharset locale log1p maintainer-makefile malloc-gnu malloca nproc open pipe2 putenv recv recvfrom regex rename send sendto setenv setsockopt shutdown socket stat-time stdlib strftime striconveh string sys_stat trunc verify vsnprintf warnings wchar # Specification in the form of a few gnulib-tool.m4 macro invocations: gl_LOCAL_DIR([gnulib-local]) @@ -86,6 +86,7 @@ gl_MODULES([ putenv recv recvfrom + regex rename send sendto diff --git a/m4/gnulib-comp.m4 b/m4/gnulib-comp.m4 index 9ec19f03d..498c5fa04 100644 --- a/m4/gnulib-comp.m4 +++ b/m4/gnulib-comp.m4 @@ -48,6 +48,7 @@ AC_DEFUN([gl_EARLY], AB_INIT # Code from module binary-io: # Code from module bind: + # Code from module btowc: # Code from module byteswap: # Code from module c-ctype: # Code from module c-strcase: @@ -108,6 +109,7 @@ AC_DEFUN([gl_EARLY], # Code from module isnand-nolibm: # Code from module isnanf: # Code from module isnanl: + # Code from module langinfo: # Code from module largefile: AC_REQUIRE([AC_SYS_LARGEFILE]) # Code from module ldexp: @@ -124,12 +126,16 @@ AC_DEFUN([gl_EARLY], # Code from module malloc-posix: # Code from module malloca: # Code from module math: + # Code from module mbrtowc: + # Code from module mbsinit: + # Code from module mbtowc: # Code from module memchr: # Code from module msvc-inval: # Code from module msvc-nothrow: # Code from module multiarch: # Code from module netdb: # Code from module netinet_in: + # Code from module nl_langinfo: # Code from module nocrash: # Code from module nproc: # Code from module open: @@ -141,6 +147,7 @@ AC_DEFUN([gl_EARLY], # Code from module readlink: # Code from module recv: # Code from module recvfrom: + # Code from module regex: # Code from module rename: # Code from module rmdir: # Code from module safe-read: @@ -173,9 +180,12 @@ AC_DEFUN([gl_EARLY], # Code from module stdint: # Code from module stdio: # Code from module stdlib: + # Code from module strcase: + # Code from module streq: # Code from module strftime: # Code from module striconveh: # Code from module string: + # Code from module strings: # Code from module sys_file: # Code from module sys_socket: # Code from module sys_stat: @@ -200,6 +210,8 @@ AC_DEFUN([gl_EARLY], # Code from module vsnprintf: # Code from module warnings: # Code from module wchar: + # Code from module wcrtomb: + # Code from module wctype-h: # Code from module write: # Code from module xsize: ]) @@ -231,6 +243,12 @@ if test "$ac_cv_header_winsock2_h" = yes; then AC_LIBOBJ([bind]) fi gl_SYS_SOCKET_MODULE_INDICATOR([bind]) +gl_FUNC_BTOWC +if test $HAVE_BTOWC = 0 || test $REPLACE_BTOWC = 1; then + AC_LIBOBJ([btowc]) + gl_PREREQ_BTOWC +fi +gl_WCHAR_MODULE_INDICATOR([btowc]) gl_BYTESWAP gl_CANONICALIZE_LGPL if test $HAVE_CANONICALIZE_FILE_NAME = 0 || test $REPLACE_CANONICALIZE_FILE_NAME = 1; then @@ -406,6 +424,7 @@ if test $HAVE_ISNANL = 0 || test $REPLACE_ISNAN = 1; then gl_PREREQ_ISNANL fi gl_MATH_MODULE_INDICATOR([isnanl]) +gl_LANGINFO_H gl_FUNC_LDEXP gl_LD_VERSION_SCRIPT gl_VISIBILITY @@ -440,6 +459,24 @@ fi gl_STDLIB_MODULE_INDICATOR([malloc-posix]) gl_MALLOCA gl_MATH_H +gl_FUNC_MBRTOWC +if test $HAVE_MBRTOWC = 0 || test $REPLACE_MBRTOWC = 1; then + AC_LIBOBJ([mbrtowc]) + gl_PREREQ_MBRTOWC +fi +gl_WCHAR_MODULE_INDICATOR([mbrtowc]) +gl_FUNC_MBSINIT +if test $HAVE_MBSINIT = 0 || test $REPLACE_MBSINIT = 1; then + AC_LIBOBJ([mbsinit]) + gl_PREREQ_MBSINIT +fi +gl_WCHAR_MODULE_INDICATOR([mbsinit]) +gl_FUNC_MBTOWC +if test $REPLACE_MBTOWC = 1; then + AC_LIBOBJ([mbtowc]) + gl_PREREQ_MBTOWC +fi +gl_STDLIB_MODULE_INDICATOR([mbtowc]) gl_FUNC_MEMCHR if test $HAVE_MEMCHR = 0 || test $REPLACE_MEMCHR = 1; then AC_LIBOBJ([memchr]) @@ -458,6 +495,11 @@ gl_MULTIARCH gl_HEADER_NETDB gl_HEADER_NETINET_IN AC_PROG_MKDIR_P +gl_FUNC_NL_LANGINFO +if test $HAVE_NL_LANGINFO = 0 || test $REPLACE_NL_LANGINFO = 1; then + AC_LIBOBJ([nl_langinfo]) +fi +gl_LANGINFO_MODULE_INDICATOR([nl_langinfo]) gl_NPROC gl_FUNC_OPEN if test $REPLACE_OPEN = 1; then @@ -501,6 +543,11 @@ if test "$ac_cv_header_winsock2_h" = yes; then AC_LIBOBJ([recvfrom]) fi gl_SYS_SOCKET_MODULE_INDICATOR([recvfrom]) +gl_REGEX +if test $ac_use_included_regex = yes; then + AC_LIBOBJ([regex]) + gl_PREREQ_REGEX +fi gl_FUNC_RENAME if test $REPLACE_RENAME = 1; then AC_LIBOBJ([rename]) @@ -576,12 +623,22 @@ gl_STDDEF_H gl_STDINT_H gl_STDIO_H gl_STDLIB_H +gl_STRCASE +if test $HAVE_STRCASECMP = 0; then + AC_LIBOBJ([strcasecmp]) + gl_PREREQ_STRCASECMP +fi +if test $HAVE_STRNCASECMP = 0; then + AC_LIBOBJ([strncasecmp]) + gl_PREREQ_STRNCASECMP +fi gl_FUNC_GNU_STRFTIME if test $gl_cond_libtool = false; then gl_ltlibdeps="$gl_ltlibdeps $LTLIBICONV" gl_libdeps="$gl_libdeps $LIBICONV" fi gl_HEADER_STRING_H +gl_HEADER_STRINGS_H gl_HEADER_SYS_FILE_H AC_PROG_MKDIR_P gl_HEADER_SYS_SOCKET @@ -623,6 +680,13 @@ gl_FUNC_VSNPRINTF gl_STDIO_MODULE_INDICATOR([vsnprintf]) AC_SUBST([WARN_CFLAGS]) gl_WCHAR_H +gl_FUNC_WCRTOMB +if test $HAVE_WCRTOMB = 0 || test $REPLACE_WCRTOMB = 1; then + AC_LIBOBJ([wcrtomb]) + gl_PREREQ_WCRTOMB +fi +gl_WCHAR_MODULE_INDICATOR([wcrtomb]) +gl_WCTYPE_H gl_FUNC_WRITE if test $REPLACE_WRITE = 1; then AC_LIBOBJ([write]) @@ -789,6 +853,7 @@ AC_DEFUN([gl_FILE_LIST], [ lib/basename-lgpl.c lib/binary-io.h lib/bind.c + lib/btowc.c lib/byteswap.in.h lib/c-ctype.c lib/c-ctype.h @@ -847,6 +912,7 @@ AC_DEFUN([gl_FILE_LIST], [ lib/isnanf.c lib/isnanl.c lib/itold.c + lib/langinfo.in.h lib/libunistring.valgrind lib/listen.c lib/localcharset.c @@ -858,6 +924,10 @@ AC_DEFUN([gl_FILE_LIST], [ lib/malloca.h lib/malloca.valgrind lib/math.in.h + lib/mbrtowc.c + lib/mbsinit.c + lib/mbtowc-impl.h + lib/mbtowc.c lib/memchr.c lib/memchr.valgrind lib/msvc-inval.c @@ -866,6 +936,7 @@ AC_DEFUN([gl_FILE_LIST], [ lib/msvc-nothrow.h lib/netdb.in.h lib/netinet_in.in.h + lib/nl_langinfo.c lib/nproc.c lib/nproc.h lib/open.c @@ -883,6 +954,12 @@ AC_DEFUN([gl_FILE_LIST], [ lib/recvfrom.c lib/ref-add.sin lib/ref-del.sin + lib/regcomp.c + lib/regex.c + lib/regex.h + lib/regex_internal.c + lib/regex_internal.h + lib/regexec.c lib/rename.c lib/rmdir.c lib/safe-read.c @@ -909,12 +986,16 @@ AC_DEFUN([gl_FILE_LIST], [ lib/stdint.in.h lib/stdio.in.h lib/stdlib.in.h + lib/strcasecmp.c + lib/streq.h lib/strftime.c lib/strftime.h lib/striconveh.c lib/striconveh.h lib/string.in.h + lib/strings.in.h lib/stripslash.c + lib/strncasecmp.c lib/sys_file.in.h lib/sys_socket.in.h lib/sys_stat.in.h @@ -941,6 +1022,8 @@ AC_DEFUN([gl_FILE_LIST], [ lib/vsnprintf.c lib/w32sock.h lib/wchar.in.h + lib/wcrtomb.c + lib/wctype.in.h lib/write.c lib/xsize.h m4/00gnulib.m4 @@ -948,6 +1031,7 @@ AC_DEFUN([gl_FILE_LIST], [ m4/alloca.m4 m4/arpa_inet_h.m4 m4/autobuild.m4 + m4/btowc.m4 m4/byteswap.m4 m4/canonicalize.m4 m4/ceil.m4 @@ -995,6 +1079,7 @@ AC_DEFUN([gl_FILE_LIST], [ m4/isnand.m4 m4/isnanf.m4 m4/isnanl.m4 + m4/langinfo_h.m4 m4/largefile.m4 m4/ld-version-script.m4 m4/ldexp.m4 @@ -1004,6 +1089,9 @@ AC_DEFUN([gl_FILE_LIST], [ m4/libunistring-base.m4 m4/libunistring.m4 m4/localcharset.m4 + m4/locale-fr.m4 + m4/locale-ja.m4 + m4/locale-zh.m4 m4/locale_h.m4 m4/longlong.m4 m4/lstat.m4 @@ -1011,6 +1099,10 @@ AC_DEFUN([gl_FILE_LIST], [ m4/malloca.m4 m4/math_h.m4 m4/mathfunc.m4 + m4/mbrtowc.m4 + m4/mbsinit.m4 + m4/mbstate_t.m4 + m4/mbtowc.m4 m4/memchr.m4 m4/mmap-anon.m4 m4/mode_t.m4 @@ -1019,6 +1111,7 @@ AC_DEFUN([gl_FILE_LIST], [ m4/multiarch.m4 m4/netdb_h.m4 m4/netinet_in_h.m4 + m4/nl_langinfo.m4 m4/nocrash.m4 m4/nproc.m4 m4/open.m4 @@ -1029,6 +1122,7 @@ AC_DEFUN([gl_FILE_LIST], [ m4/raise.m4 m4/read.m4 m4/readlink.m4 + m4/regex.m4 m4/rename.m4 m4/rmdir.m4 m4/safe-read.m4 @@ -1052,8 +1146,10 @@ AC_DEFUN([gl_FILE_LIST], [ m4/stdint_h.m4 m4/stdio_h.m4 m4/stdlib_h.m4 + m4/strcase.m4 m4/strftime.m4 m4/string_h.m4 + m4/strings_h.m4 m4/sys_file_h.m4 m4/sys_socket_h.m4 m4/sys_stat_h.m4 @@ -1072,6 +1168,8 @@ AC_DEFUN([gl_FILE_LIST], [ m4/warnings.m4 m4/wchar_h.m4 m4/wchar_t.m4 + m4/wcrtomb.m4 + m4/wctype_h.m4 m4/wint_t.m4 m4/write.m4 m4/xsize.m4 diff --git a/m4/langinfo_h.m4 b/m4/langinfo_h.m4 new file mode 100644 index 000000000..b93fe7066 --- /dev/null +++ b/m4/langinfo_h.m4 @@ -0,0 +1,105 @@ +# langinfo_h.m4 serial 7 +dnl Copyright (C) 2009-2012 Free Software Foundation, Inc. +dnl This file is free software; the Free Software Foundation +dnl gives unlimited permission to copy and/or distribute it, +dnl with or without modifications, as long as this notice is preserved. + +AC_DEFUN([gl_LANGINFO_H], +[ + AC_REQUIRE([gl_LANGINFO_H_DEFAULTS]) + + dnl Persuade glibc-2.0.6 <langinfo.h> to define CODESET. + AC_REQUIRE([AC_USE_SYSTEM_EXTENSIONS]) + + dnl <langinfo.h> is always overridden, because of GNULIB_POSIXCHECK. + gl_CHECK_NEXT_HEADERS([langinfo.h]) + + dnl Determine whether <langinfo.h> exists. It is missing on mingw and BeOS. + HAVE_LANGINFO_CODESET=0 + HAVE_LANGINFO_T_FMT_AMPM=0 + HAVE_LANGINFO_ERA=0 + HAVE_LANGINFO_YESEXPR=0 + AC_CHECK_HEADERS_ONCE([langinfo.h]) + if test $ac_cv_header_langinfo_h = yes; then + HAVE_LANGINFO_H=1 + dnl Determine what <langinfo.h> defines. CODESET and ERA etc. are missing + dnl on OpenBSD 3.8. T_FMT_AMPM and YESEXPR, NOEXPR are missing on IRIX 5.3. + AC_CACHE_CHECK([whether langinfo.h defines CODESET], + [gl_cv_header_langinfo_codeset], + [AC_COMPILE_IFELSE( + [AC_LANG_PROGRAM([[#include <langinfo.h> +int a = CODESET; +]])], + [gl_cv_header_langinfo_codeset=yes], + [gl_cv_header_langinfo_codeset=no]) + ]) + if test $gl_cv_header_langinfo_codeset = yes; then + HAVE_LANGINFO_CODESET=1 + fi + AC_CACHE_CHECK([whether langinfo.h defines T_FMT_AMPM], + [gl_cv_header_langinfo_t_fmt_ampm], + [AC_COMPILE_IFELSE( + [AC_LANG_PROGRAM([[#include <langinfo.h> +int a = T_FMT_AMPM; +]])], + [gl_cv_header_langinfo_t_fmt_ampm=yes], + [gl_cv_header_langinfo_t_fmt_ampm=no]) + ]) + if test $gl_cv_header_langinfo_t_fmt_ampm = yes; then + HAVE_LANGINFO_T_FMT_AMPM=1 + fi + AC_CACHE_CHECK([whether langinfo.h defines ERA], + [gl_cv_header_langinfo_era], + [AC_COMPILE_IFELSE( + [AC_LANG_PROGRAM([[#include <langinfo.h> +int a = ERA; +]])], + [gl_cv_header_langinfo_era=yes], + [gl_cv_header_langinfo_era=no]) + ]) + if test $gl_cv_header_langinfo_era = yes; then + HAVE_LANGINFO_ERA=1 + fi + AC_CACHE_CHECK([whether langinfo.h defines YESEXPR], + [gl_cv_header_langinfo_yesexpr], + [AC_COMPILE_IFELSE( + [AC_LANG_PROGRAM([[#include <langinfo.h> +int a = YESEXPR; +]])], + [gl_cv_header_langinfo_yesexpr=yes], + [gl_cv_header_langinfo_yesexpr=no]) + ]) + if test $gl_cv_header_langinfo_yesexpr = yes; then + HAVE_LANGINFO_YESEXPR=1 + fi + else + HAVE_LANGINFO_H=0 + fi + AC_SUBST([HAVE_LANGINFO_H]) + AC_SUBST([HAVE_LANGINFO_CODESET]) + AC_SUBST([HAVE_LANGINFO_T_FMT_AMPM]) + AC_SUBST([HAVE_LANGINFO_ERA]) + AC_SUBST([HAVE_LANGINFO_YESEXPR]) + + dnl Check for declarations of anything we want to poison if the + dnl corresponding gnulib module is not in use. + gl_WARN_ON_USE_PREPARE([[#include <langinfo.h> + ]], [nl_langinfo]) +]) + +AC_DEFUN([gl_LANGINFO_MODULE_INDICATOR], +[ + dnl Use AC_REQUIRE here, so that the default settings are expanded once only. + AC_REQUIRE([gl_LANGINFO_H_DEFAULTS]) + gl_MODULE_INDICATOR_SET_VARIABLE([$1]) + dnl Define it also as a C macro, for the benefit of the unit tests. + gl_MODULE_INDICATOR_FOR_TESTS([$1]) +]) + +AC_DEFUN([gl_LANGINFO_H_DEFAULTS], +[ + GNULIB_NL_LANGINFO=0; AC_SUBST([GNULIB_NL_LANGINFO]) + dnl Assume proper GNU behavior unless another module says otherwise. + HAVE_NL_LANGINFO=1; AC_SUBST([HAVE_NL_LANGINFO]) + REPLACE_NL_LANGINFO=0; AC_SUBST([REPLACE_NL_LANGINFO]) +]) diff --git a/m4/locale-fr.m4 b/m4/locale-fr.m4 new file mode 100644 index 000000000..4cb14b5f5 --- /dev/null +++ b/m4/locale-fr.m4 @@ -0,0 +1,246 @@ +# locale-fr.m4 serial 14 +dnl Copyright (C) 2003, 2005-2012 Free Software Foundation, Inc. +dnl This file is free software; the Free Software Foundation +dnl gives unlimited permission to copy and/or distribute it, +dnl with or without modifications, as long as this notice is preserved. + +dnl From Bruno Haible. + +dnl Determine the name of a french locale with traditional encoding. +AC_DEFUN([gt_LOCALE_FR], +[ + AC_REQUIRE([AC_CANONICAL_HOST]) + AC_REQUIRE([AM_LANGINFO_CODESET]) + AC_CACHE_CHECK([for a traditional french locale], [gt_cv_locale_fr], [ + AC_LANG_CONFTEST([AC_LANG_SOURCE([ +changequote(,)dnl +#include <locale.h> +#include <time.h> +#if HAVE_LANGINFO_CODESET +# include <langinfo.h> +#endif +#include <stdlib.h> +#include <string.h> +struct tm t; +char buf[16]; +int main () { + /* Check whether the given locale name is recognized by the system. */ +#if (defined _WIN32 || defined __WIN32__) && !defined __CYGWIN__ + /* On native Windows, setlocale(category, "") looks at the system settings, + not at the environment variables. Also, when an encoding suffix such + as ".65001" or ".54936" is speficied, it succeeds but sets the LC_CTYPE + category of the locale to "C". */ + if (setlocale (LC_ALL, getenv ("LC_ALL")) == NULL + || strcmp (setlocale (LC_CTYPE, NULL), "C") == 0) + return 1; +#else + if (setlocale (LC_ALL, "") == NULL) return 1; +#endif + /* Check whether nl_langinfo(CODESET) is nonempty and not "ASCII" or "646". + On MacOS X 10.3.5 (Darwin 7.5) in the fr_FR locale, nl_langinfo(CODESET) + is empty, and the behaviour of Tcl 8.4 in this locale is not useful. + On OpenBSD 4.0, when an unsupported locale is specified, setlocale() + succeeds but then nl_langinfo(CODESET) is "646". In this situation, + some unit tests fail. + On MirBSD 10, when an unsupported locale is specified, setlocale() + succeeds but then nl_langinfo(CODESET) is "UTF-8". */ +#if HAVE_LANGINFO_CODESET + { + const char *cs = nl_langinfo (CODESET); + if (cs[0] == '\0' || strcmp (cs, "ASCII") == 0 || strcmp (cs, "646") == 0 + || strcmp (cs, "UTF-8") == 0) + return 1; + } +#endif +#ifdef __CYGWIN__ + /* On Cygwin, avoid locale names without encoding suffix, because the + locale_charset() function relies on the encoding suffix. Note that + LC_ALL is set on the command line. */ + if (strchr (getenv ("LC_ALL"), '.') == NULL) return 1; +#endif + /* Check whether in the abbreviation of the second month, the second + character (should be U+00E9: LATIN SMALL LETTER E WITH ACUTE) is only + one byte long. This excludes the UTF-8 encoding. */ + t.tm_year = 1975 - 1900; t.tm_mon = 2 - 1; t.tm_mday = 4; + if (strftime (buf, sizeof (buf), "%b", &t) < 3 || buf[2] != 'v') return 1; + /* Check whether the decimal separator is a comma. + On NetBSD 3.0 in the fr_FR.ISO8859-1 locale, localeconv()->decimal_point + are nl_langinfo(RADIXCHAR) are both ".". */ + if (localeconv () ->decimal_point[0] != ',') return 1; + return 0; +} +changequote([,])dnl + ])]) + if AC_TRY_EVAL([ac_link]) && test -s conftest$ac_exeext; then + case "$host_os" in + # Handle native Windows specially, because there setlocale() interprets + # "ar" as "Arabic" or "Arabic_Saudi Arabia.1256", + # "fr" or "fra" as "French" or "French_France.1252", + # "ge"(!) or "deu"(!) as "German" or "German_Germany.1252", + # "ja" as "Japanese" or "Japanese_Japan.932", + # and similar. + mingw*) + # Test for the native Windows locale name. + if (LC_ALL=French_France.1252 LC_TIME= LC_CTYPE= ./conftest; exit) 2>/dev/null; then + gt_cv_locale_fr=French_France.1252 + else + # None found. + gt_cv_locale_fr=none + fi + ;; + *) + # Setting LC_ALL is not enough. Need to set LC_TIME to empty, because + # otherwise on MacOS X 10.3.5 the LC_TIME=C from the beginning of the + # configure script would override the LC_ALL setting. Likewise for + # LC_CTYPE, which is also set at the beginning of the configure script. + # Test for the usual locale name. + if (LC_ALL=fr_FR LC_TIME= LC_CTYPE= ./conftest; exit) 2>/dev/null; then + gt_cv_locale_fr=fr_FR + else + # Test for the locale name with explicit encoding suffix. + if (LC_ALL=fr_FR.ISO-8859-1 LC_TIME= LC_CTYPE= ./conftest; exit) 2>/dev/null; then + gt_cv_locale_fr=fr_FR.ISO-8859-1 + else + # Test for the AIX, OSF/1, FreeBSD, NetBSD, OpenBSD locale name. + if (LC_ALL=fr_FR.ISO8859-1 LC_TIME= LC_CTYPE= ./conftest; exit) 2>/dev/null; then + gt_cv_locale_fr=fr_FR.ISO8859-1 + else + # Test for the HP-UX locale name. + if (LC_ALL=fr_FR.iso88591 LC_TIME= LC_CTYPE= ./conftest; exit) 2>/dev/null; then + gt_cv_locale_fr=fr_FR.iso88591 + else + # Test for the Solaris 7 locale name. + if (LC_ALL=fr LC_TIME= LC_CTYPE= ./conftest; exit) 2>/dev/null; then + gt_cv_locale_fr=fr + else + # None found. + gt_cv_locale_fr=none + fi + fi + fi + fi + fi + ;; + esac + fi + rm -fr conftest* + ]) + LOCALE_FR=$gt_cv_locale_fr + AC_SUBST([LOCALE_FR]) +]) + +dnl Determine the name of a french locale with UTF-8 encoding. +AC_DEFUN([gt_LOCALE_FR_UTF8], +[ + AC_REQUIRE([AM_LANGINFO_CODESET]) + AC_CACHE_CHECK([for a french Unicode locale], [gt_cv_locale_fr_utf8], [ + AC_LANG_CONFTEST([AC_LANG_SOURCE([ +changequote(,)dnl +#include <locale.h> +#include <time.h> +#if HAVE_LANGINFO_CODESET +# include <langinfo.h> +#endif +#include <stdlib.h> +#include <string.h> +struct tm t; +char buf[16]; +int main () { + /* On BeOS and Haiku, locales are not implemented in libc. Rather, libintl + imitates locale dependent behaviour by looking at the environment + variables, and all locales use the UTF-8 encoding. */ +#if !(defined __BEOS__ || defined __HAIKU__) + /* Check whether the given locale name is recognized by the system. */ +# if (defined _WIN32 || defined __WIN32__) && !defined __CYGWIN__ + /* On native Windows, setlocale(category, "") looks at the system settings, + not at the environment variables. Also, when an encoding suffix such + as ".65001" or ".54936" is speficied, it succeeds but sets the LC_CTYPE + category of the locale to "C". */ + if (setlocale (LC_ALL, getenv ("LC_ALL")) == NULL + || strcmp (setlocale (LC_CTYPE, NULL), "C") == 0) + return 1; +# else + if (setlocale (LC_ALL, "") == NULL) return 1; +# endif + /* Check whether nl_langinfo(CODESET) is nonempty and not "ASCII" or "646". + On MacOS X 10.3.5 (Darwin 7.5) in the fr_FR locale, nl_langinfo(CODESET) + is empty, and the behaviour of Tcl 8.4 in this locale is not useful. + On OpenBSD 4.0, when an unsupported locale is specified, setlocale() + succeeds but then nl_langinfo(CODESET) is "646". In this situation, + some unit tests fail. */ +# if HAVE_LANGINFO_CODESET + { + const char *cs = nl_langinfo (CODESET); + if (cs[0] == '\0' || strcmp (cs, "ASCII") == 0 || strcmp (cs, "646") == 0) + return 1; + } +# endif +# ifdef __CYGWIN__ + /* On Cygwin, avoid locale names without encoding suffix, because the + locale_charset() function relies on the encoding suffix. Note that + LC_ALL is set on the command line. */ + if (strchr (getenv ("LC_ALL"), '.') == NULL) return 1; +# endif + /* Check whether in the abbreviation of the second month, the second + character (should be U+00E9: LATIN SMALL LETTER E WITH ACUTE) is + two bytes long, with UTF-8 encoding. */ + t.tm_year = 1975 - 1900; t.tm_mon = 2 - 1; t.tm_mday = 4; + if (strftime (buf, sizeof (buf), "%b", &t) < 4 + || buf[1] != (char) 0xc3 || buf[2] != (char) 0xa9 || buf[3] != 'v') + return 1; +#endif + /* Check whether the decimal separator is a comma. + On NetBSD 3.0 in the fr_FR.ISO8859-1 locale, localeconv()->decimal_point + are nl_langinfo(RADIXCHAR) are both ".". */ + if (localeconv () ->decimal_point[0] != ',') return 1; + return 0; +} +changequote([,])dnl + ])]) + if AC_TRY_EVAL([ac_link]) && test -s conftest$ac_exeext; then + case "$host_os" in + # Handle native Windows specially, because there setlocale() interprets + # "ar" as "Arabic" or "Arabic_Saudi Arabia.1256", + # "fr" or "fra" as "French" or "French_France.1252", + # "ge"(!) or "deu"(!) as "German" or "German_Germany.1252", + # "ja" as "Japanese" or "Japanese_Japan.932", + # and similar. + mingw*) + # Test for the hypothetical native Windows locale name. + if (LC_ALL=French_France.65001 LC_TIME= LC_CTYPE= ./conftest; exit) 2>/dev/null; then + gt_cv_locale_fr_utf8=French_France.65001 + else + # None found. + gt_cv_locale_fr_utf8=none + fi + ;; + *) + # Setting LC_ALL is not enough. Need to set LC_TIME to empty, because + # otherwise on MacOS X 10.3.5 the LC_TIME=C from the beginning of the + # configure script would override the LC_ALL setting. Likewise for + # LC_CTYPE, which is also set at the beginning of the configure script. + # Test for the usual locale name. + if (LC_ALL=fr_FR LC_TIME= LC_CTYPE= ./conftest; exit) 2>/dev/null; then + gt_cv_locale_fr_utf8=fr_FR + else + # Test for the locale name with explicit encoding suffix. + if (LC_ALL=fr_FR.UTF-8 LC_TIME= LC_CTYPE= ./conftest; exit) 2>/dev/null; then + gt_cv_locale_fr_utf8=fr_FR.UTF-8 + else + # Test for the Solaris 7 locale name. + if (LC_ALL=fr.UTF-8 LC_TIME= LC_CTYPE= ./conftest; exit) 2>/dev/null; then + gt_cv_locale_fr_utf8=fr.UTF-8 + else + # None found. + gt_cv_locale_fr_utf8=none + fi + fi + fi + ;; + esac + fi + rm -fr conftest* + ]) + LOCALE_FR_UTF8=$gt_cv_locale_fr_utf8 + AC_SUBST([LOCALE_FR_UTF8]) +]) diff --git a/m4/locale-ja.m4 b/m4/locale-ja.m4 new file mode 100644 index 000000000..b427f0992 --- /dev/null +++ b/m4/locale-ja.m4 @@ -0,0 +1,136 @@ +# locale-ja.m4 serial 10 +dnl Copyright (C) 2003, 2005-2012 Free Software Foundation, Inc. +dnl This file is free software; the Free Software Foundation +dnl gives unlimited permission to copy and/or distribute it, +dnl with or without modifications, as long as this notice is preserved. + +dnl From Bruno Haible. + +dnl Determine the name of a japanese locale with EUC-JP encoding. +AC_DEFUN([gt_LOCALE_JA], +[ + AC_REQUIRE([AC_CANONICAL_HOST]) + AC_REQUIRE([AM_LANGINFO_CODESET]) + AC_CACHE_CHECK([for a traditional japanese locale], [gt_cv_locale_ja], [ + AC_LANG_CONFTEST([AC_LANG_SOURCE([ +changequote(,)dnl +#include <locale.h> +#include <time.h> +#if HAVE_LANGINFO_CODESET +# include <langinfo.h> +#endif +#include <stdlib.h> +#include <string.h> +struct tm t; +char buf[16]; +int main () +{ + const char *p; + /* Check whether the given locale name is recognized by the system. */ +#if (defined _WIN32 || defined __WIN32__) && !defined __CYGWIN__ + /* On native Windows, setlocale(category, "") looks at the system settings, + not at the environment variables. Also, when an encoding suffix such + as ".65001" or ".54936" is speficied, it succeeds but sets the LC_CTYPE + category of the locale to "C". */ + if (setlocale (LC_ALL, getenv ("LC_ALL")) == NULL + || strcmp (setlocale (LC_CTYPE, NULL), "C") == 0) + return 1; +#else + if (setlocale (LC_ALL, "") == NULL) return 1; +#endif + /* Check whether nl_langinfo(CODESET) is nonempty and not "ASCII" or "646". + On MacOS X 10.3.5 (Darwin 7.5) in the fr_FR locale, nl_langinfo(CODESET) + is empty, and the behaviour of Tcl 8.4 in this locale is not useful. + On OpenBSD 4.0, when an unsupported locale is specified, setlocale() + succeeds but then nl_langinfo(CODESET) is "646". In this situation, + some unit tests fail. + On MirBSD 10, when an unsupported locale is specified, setlocale() + succeeds but then nl_langinfo(CODESET) is "UTF-8". */ +#if HAVE_LANGINFO_CODESET + { + const char *cs = nl_langinfo (CODESET); + if (cs[0] == '\0' || strcmp (cs, "ASCII") == 0 || strcmp (cs, "646") == 0 + || strcmp (cs, "UTF-8") == 0) + return 1; + } +#endif +#ifdef __CYGWIN__ + /* On Cygwin, avoid locale names without encoding suffix, because the + locale_charset() function relies on the encoding suffix. Note that + LC_ALL is set on the command line. */ + if (strchr (getenv ("LC_ALL"), '.') == NULL) return 1; +#endif + /* Check whether MB_CUR_MAX is > 1. This excludes the dysfunctional locales + on Cygwin 1.5.x. */ + if (MB_CUR_MAX == 1) + return 1; + /* Check whether in a month name, no byte in the range 0x80..0x9F occurs. + This excludes the UTF-8 encoding (except on MirBSD). */ + t.tm_year = 1975 - 1900; t.tm_mon = 2 - 1; t.tm_mday = 4; + if (strftime (buf, sizeof (buf), "%B", &t) < 2) return 1; + for (p = buf; *p != '\0'; p++) + if ((unsigned char) *p >= 0x80 && (unsigned char) *p < 0xa0) + return 1; + return 0; +} +changequote([,])dnl + ])]) + if AC_TRY_EVAL([ac_link]) && test -s conftest$ac_exeext; then + case "$host_os" in + # Handle native Windows specially, because there setlocale() interprets + # "ar" as "Arabic" or "Arabic_Saudi Arabia.1256", + # "fr" or "fra" as "French" or "French_France.1252", + # "ge"(!) or "deu"(!) as "German" or "German_Germany.1252", + # "ja" as "Japanese" or "Japanese_Japan.932", + # and similar. + mingw*) + # Note that on native Windows, the Japanese locale is + # Japanese_Japan.932, and CP932 is very different from EUC-JP, so we + # cannot use it here. + gt_cv_locale_ja=none + ;; + *) + # Setting LC_ALL is not enough. Need to set LC_TIME to empty, because + # otherwise on MacOS X 10.3.5 the LC_TIME=C from the beginning of the + # configure script would override the LC_ALL setting. Likewise for + # LC_CTYPE, which is also set at the beginning of the configure script. + # Test for the AIX locale name. + if (LC_ALL=ja_JP LC_TIME= LC_CTYPE= ./conftest; exit) 2>/dev/null; then + gt_cv_locale_ja=ja_JP + else + # Test for the locale name with explicit encoding suffix. + if (LC_ALL=ja_JP.EUC-JP LC_TIME= LC_CTYPE= ./conftest; exit) 2>/dev/null; then + gt_cv_locale_ja=ja_JP.EUC-JP + else + # Test for the HP-UX, OSF/1, NetBSD locale name. + if (LC_ALL=ja_JP.eucJP LC_TIME= LC_CTYPE= ./conftest; exit) 2>/dev/null; then + gt_cv_locale_ja=ja_JP.eucJP + else + # Test for the IRIX, FreeBSD locale name. + if (LC_ALL=ja_JP.EUC LC_TIME= LC_CTYPE= ./conftest; exit) 2>/dev/null; then + gt_cv_locale_ja=ja_JP.EUC + else + # Test for the Solaris 7 locale name. + if (LC_ALL=ja LC_TIME= LC_CTYPE= ./conftest; exit) 2>/dev/null; then + gt_cv_locale_ja=ja + else + # Special test for NetBSD 1.6. + if test -f /usr/share/locale/ja_JP.eucJP/LC_CTYPE; then + gt_cv_locale_ja=ja_JP.eucJP + else + # None found. + gt_cv_locale_ja=none + fi + fi + fi + fi + fi + fi + ;; + esac + fi + rm -fr conftest* + ]) + LOCALE_JA=$gt_cv_locale_ja + AC_SUBST([LOCALE_JA]) +]) diff --git a/m4/locale-zh.m4 b/m4/locale-zh.m4 new file mode 100644 index 000000000..de1a43b30 --- /dev/null +++ b/m4/locale-zh.m4 @@ -0,0 +1,130 @@ +# locale-zh.m4 serial 10 +dnl Copyright (C) 2003, 2005-2012 Free Software Foundation, Inc. +dnl This file is free software; the Free Software Foundation +dnl gives unlimited permission to copy and/or distribute it, +dnl with or without modifications, as long as this notice is preserved. + +dnl From Bruno Haible. + +dnl Determine the name of a chinese locale with GB18030 encoding. +AC_DEFUN([gt_LOCALE_ZH_CN], +[ + AC_REQUIRE([AC_CANONICAL_HOST]) + AC_REQUIRE([AM_LANGINFO_CODESET]) + AC_CACHE_CHECK([for a transitional chinese locale], [gt_cv_locale_zh_CN], [ + AC_LANG_CONFTEST([AC_LANG_SOURCE([ +changequote(,)dnl +#include <locale.h> +#include <stdlib.h> +#include <time.h> +#if HAVE_LANGINFO_CODESET +# include <langinfo.h> +#endif +#include <stdlib.h> +#include <string.h> +struct tm t; +char buf[16]; +int main () +{ + const char *p; + /* Check whether the given locale name is recognized by the system. */ +#if (defined _WIN32 || defined __WIN32__) && !defined __CYGWIN__ + /* On native Windows, setlocale(category, "") looks at the system settings, + not at the environment variables. Also, when an encoding suffix such + as ".65001" or ".54936" is speficied, it succeeds but sets the LC_CTYPE + category of the locale to "C". */ + if (setlocale (LC_ALL, getenv ("LC_ALL")) == NULL + || strcmp (setlocale (LC_CTYPE, NULL), "C") == 0) + return 1; +#else + if (setlocale (LC_ALL, "") == NULL) return 1; +#endif + /* Check whether nl_langinfo(CODESET) is nonempty and not "ASCII" or "646". + On MacOS X 10.3.5 (Darwin 7.5) in the fr_FR locale, nl_langinfo(CODESET) + is empty, and the behaviour of Tcl 8.4 in this locale is not useful. + On OpenBSD 4.0, when an unsupported locale is specified, setlocale() + succeeds but then nl_langinfo(CODESET) is "646". In this situation, + some unit tests fail. + On MirBSD 10, when an unsupported locale is specified, setlocale() + succeeds but then nl_langinfo(CODESET) is "UTF-8". */ +#if HAVE_LANGINFO_CODESET + { + const char *cs = nl_langinfo (CODESET); + if (cs[0] == '\0' || strcmp (cs, "ASCII") == 0 || strcmp (cs, "646") == 0 + || strcmp (cs, "UTF-8") == 0) + return 1; + } +#endif +#ifdef __CYGWIN__ + /* On Cygwin, avoid locale names without encoding suffix, because the + locale_charset() function relies on the encoding suffix. Note that + LC_ALL is set on the command line. */ + if (strchr (getenv ("LC_ALL"), '.') == NULL) return 1; +#endif + /* Check whether in a month name, no byte in the range 0x80..0x9F occurs. + This excludes the UTF-8 encoding (except on MirBSD). */ + t.tm_year = 1975 - 1900; t.tm_mon = 2 - 1; t.tm_mday = 4; + if (strftime (buf, sizeof (buf), "%B", &t) < 2) return 1; + for (p = buf; *p != '\0'; p++) + if ((unsigned char) *p >= 0x80 && (unsigned char) *p < 0xa0) + return 1; + /* Check whether a typical GB18030 multibyte sequence is recognized as a + single wide character. This excludes the GB2312 and GBK encodings. */ + if (mblen ("\203\062\332\066", 5) != 4) + return 1; + return 0; +} +changequote([,])dnl + ])]) + if AC_TRY_EVAL([ac_link]) && test -s conftest$ac_exeext; then + case "$host_os" in + # Handle native Windows specially, because there setlocale() interprets + # "ar" as "Arabic" or "Arabic_Saudi Arabia.1256", + # "fr" or "fra" as "French" or "French_France.1252", + # "ge"(!) or "deu"(!) as "German" or "German_Germany.1252", + # "ja" as "Japanese" or "Japanese_Japan.932", + # and similar. + mingw*) + # Test for the hypothetical native Windows locale name. + if (LC_ALL=Chinese_China.54936 LC_TIME= LC_CTYPE= ./conftest; exit) 2>/dev/null; then + gt_cv_locale_zh_CN=Chinese_China.54936 + else + # None found. + gt_cv_locale_zh_CN=none + fi + ;; + solaris2.8) + # On Solaris 8, the locales zh_CN.GB18030, zh_CN.GBK, zh.GBK are + # broken. One witness is the test case in gl_MBRTOWC_SANITYCHECK. + # Another witness is that "LC_ALL=zh_CN.GB18030 bash -c true" dumps core. + gt_cv_locale_zh_CN=none + ;; + *) + # Setting LC_ALL is not enough. Need to set LC_TIME to empty, because + # otherwise on MacOS X 10.3.5 the LC_TIME=C from the beginning of the + # configure script would override the LC_ALL setting. Likewise for + # LC_CTYPE, which is also set at the beginning of the configure script. + # Test for the locale name without encoding suffix. + if (LC_ALL=zh_CN LC_TIME= LC_CTYPE= ./conftest; exit) 2>/dev/null; then + gt_cv_locale_zh_CN=zh_CN + else + # Test for the locale name with explicit encoding suffix. + if (LC_ALL=zh_CN.GB18030 LC_TIME= LC_CTYPE= ./conftest; exit) 2>/dev/null; then + gt_cv_locale_zh_CN=zh_CN.GB18030 + else + # None found. + gt_cv_locale_zh_CN=none + fi + fi + ;; + esac + else + # If there was a link error, due to mblen(), the system is so old that + # it certainly doesn't have a chinese locale. + gt_cv_locale_zh_CN=none + fi + rm -fr conftest* + ]) + LOCALE_ZH_CN=$gt_cv_locale_zh_CN + AC_SUBST([LOCALE_ZH_CN]) +]) diff --git a/m4/mbrtowc.m4 b/m4/mbrtowc.m4 new file mode 100644 index 000000000..8f829c8ea --- /dev/null +++ b/m4/mbrtowc.m4 @@ -0,0 +1,572 @@ +# mbrtowc.m4 serial 25 +dnl Copyright (C) 2001-2002, 2004-2005, 2008-2012 Free Software Foundation, +dnl Inc. +dnl This file is free software; the Free Software Foundation +dnl gives unlimited permission to copy and/or distribute it, +dnl with or without modifications, as long as this notice is preserved. + +AC_DEFUN([gl_FUNC_MBRTOWC], +[ + AC_REQUIRE([gl_WCHAR_H_DEFAULTS]) + + AC_REQUIRE([AC_TYPE_MBSTATE_T]) + gl_MBSTATE_T_BROKEN + + AC_CHECK_FUNCS_ONCE([mbrtowc]) + if test $ac_cv_func_mbrtowc = no; then + HAVE_MBRTOWC=0 + AC_CHECK_DECLS([mbrtowc],,, [[ +/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before + <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> must be + included before <wchar.h>. */ +#include <stddef.h> +#include <stdio.h> +#include <time.h> +#include <wchar.h> +]]) + if test $ac_cv_have_decl_mbrtowc = yes; then + dnl On Minix 3.1.8, the system's <wchar.h> declares mbrtowc() although + dnl it does not have the function. Avoid a collision with gnulib's + dnl replacement. + REPLACE_MBRTOWC=1 + fi + else + if test $REPLACE_MBSTATE_T = 1; then + REPLACE_MBRTOWC=1 + else + gl_MBRTOWC_NULL_ARG1 + gl_MBRTOWC_NULL_ARG2 + gl_MBRTOWC_RETVAL + gl_MBRTOWC_NUL_RETVAL + case "$gl_cv_func_mbrtowc_null_arg1" in + *yes) ;; + *) AC_DEFINE([MBRTOWC_NULL_ARG1_BUG], [1], + [Define if the mbrtowc function has the NULL pwc argument bug.]) + REPLACE_MBRTOWC=1 + ;; + esac + case "$gl_cv_func_mbrtowc_null_arg2" in + *yes) ;; + *) AC_DEFINE([MBRTOWC_NULL_ARG2_BUG], [1], + [Define if the mbrtowc function has the NULL string argument bug.]) + REPLACE_MBRTOWC=1 + ;; + esac + case "$gl_cv_func_mbrtowc_retval" in + *yes) ;; + *) AC_DEFINE([MBRTOWC_RETVAL_BUG], [1], + [Define if the mbrtowc function returns a wrong return value.]) + REPLACE_MBRTOWC=1 + ;; + esac + case "$gl_cv_func_mbrtowc_nul_retval" in + *yes) ;; + *) AC_DEFINE([MBRTOWC_NUL_RETVAL_BUG], [1], + [Define if the mbrtowc function does not return 0 for a NUL character.]) + REPLACE_MBRTOWC=1 + ;; + esac + fi + fi +]) + +dnl Test whether mbsinit() and mbrtowc() need to be overridden in a way that +dnl redefines the semantics of the given mbstate_t type. +dnl Result is REPLACE_MBSTATE_T. +dnl When this is set to 1, we replace both mbsinit() and mbrtowc(), in order to +dnl avoid inconsistencies. + +AC_DEFUN([gl_MBSTATE_T_BROKEN], +[ + AC_REQUIRE([gl_WCHAR_H_DEFAULTS]) + + AC_REQUIRE([AC_TYPE_MBSTATE_T]) + AC_CHECK_FUNCS_ONCE([mbsinit]) + AC_CHECK_FUNCS_ONCE([mbrtowc]) + if test $ac_cv_func_mbsinit = yes && test $ac_cv_func_mbrtowc = yes; then + gl_MBRTOWC_INCOMPLETE_STATE + gl_MBRTOWC_SANITYCHECK + REPLACE_MBSTATE_T=0 + case "$gl_cv_func_mbrtowc_incomplete_state" in + *yes) ;; + *) REPLACE_MBSTATE_T=1 ;; + esac + case "$gl_cv_func_mbrtowc_sanitycheck" in + *yes) ;; + *) REPLACE_MBSTATE_T=1 ;; + esac + else + REPLACE_MBSTATE_T=1 + fi +]) + +dnl Test whether mbrtowc puts the state into non-initial state when parsing an +dnl incomplete multibyte character. +dnl Result is gl_cv_func_mbrtowc_incomplete_state. + +AC_DEFUN([gl_MBRTOWC_INCOMPLETE_STATE], +[ + AC_REQUIRE([AC_PROG_CC]) + AC_REQUIRE([gt_LOCALE_JA]) + AC_REQUIRE([AC_CANONICAL_HOST]) dnl for cross-compiles + AC_CACHE_CHECK([whether mbrtowc handles incomplete characters], + [gl_cv_func_mbrtowc_incomplete_state], + [ + dnl Initial guess, used when cross-compiling or when no suitable locale + dnl is present. +changequote(,)dnl + case "$host_os" in + # Guess no on AIX and OSF/1. + aix* | osf*) gl_cv_func_mbrtowc_incomplete_state="guessing no" ;; + # Guess yes otherwise. + *) gl_cv_func_mbrtowc_incomplete_state="guessing yes" ;; + esac +changequote([,])dnl + if test $LOCALE_JA != none; then + AC_RUN_IFELSE( + [AC_LANG_SOURCE([[ +#include <locale.h> +#include <string.h> +/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before + <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> must be + included before <wchar.h>. */ +#include <stddef.h> +#include <stdio.h> +#include <time.h> +#include <wchar.h> +int main () +{ + if (setlocale (LC_ALL, "$LOCALE_JA") != NULL) + { + const char input[] = "B\217\253\344\217\251\316er"; /* "Büßer" */ + mbstate_t state; + wchar_t wc; + + memset (&state, '\0', sizeof (mbstate_t)); + if (mbrtowc (&wc, input + 1, 1, &state) == (size_t)(-2)) + if (mbsinit (&state)) + return 1; + } + return 0; +}]])], + [gl_cv_func_mbrtowc_incomplete_state=yes], + [gl_cv_func_mbrtowc_incomplete_state=no], + [:]) + fi + ]) +]) + +dnl Test whether mbrtowc works not worse than mbtowc. +dnl Result is gl_cv_func_mbrtowc_sanitycheck. + +AC_DEFUN([gl_MBRTOWC_SANITYCHECK], +[ + AC_REQUIRE([AC_PROG_CC]) + AC_REQUIRE([gt_LOCALE_ZH_CN]) + AC_REQUIRE([AC_CANONICAL_HOST]) dnl for cross-compiles + AC_CACHE_CHECK([whether mbrtowc works as well as mbtowc], + [gl_cv_func_mbrtowc_sanitycheck], + [ + dnl Initial guess, used when cross-compiling or when no suitable locale + dnl is present. +changequote(,)dnl + case "$host_os" in + # Guess no on Solaris 8. + solaris2.8) gl_cv_func_mbrtowc_sanitycheck="guessing no" ;; + # Guess yes otherwise. + *) gl_cv_func_mbrtowc_sanitycheck="guessing yes" ;; + esac +changequote([,])dnl + if test $LOCALE_ZH_CN != none; then + AC_RUN_IFELSE( + [AC_LANG_SOURCE([[ +#include <locale.h> +#include <stdlib.h> +#include <string.h> +/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before + <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> must be + included before <wchar.h>. */ +#include <stddef.h> +#include <stdio.h> +#include <time.h> +#include <wchar.h> +int main () +{ + /* This fails on Solaris 8: + mbrtowc returns 2, and sets wc to 0x00F0. + mbtowc returns 4 (correct) and sets wc to 0x5EDC. */ + if (setlocale (LC_ALL, "$LOCALE_ZH_CN") != NULL) + { + char input[] = "B\250\271\201\060\211\070er"; /* "Büßer" */ + mbstate_t state; + wchar_t wc; + + memset (&state, '\0', sizeof (mbstate_t)); + if (mbrtowc (&wc, input + 3, 6, &state) != 4 + && mbtowc (&wc, input + 3, 6) == 4) + return 1; + } + return 0; +}]])], + [gl_cv_func_mbrtowc_sanitycheck=yes], + [gl_cv_func_mbrtowc_sanitycheck=no], + [:]) + fi + ]) +]) + +dnl Test whether mbrtowc supports a NULL pwc argument correctly. +dnl Result is gl_cv_func_mbrtowc_null_arg1. + +AC_DEFUN([gl_MBRTOWC_NULL_ARG1], +[ + AC_REQUIRE([AC_PROG_CC]) + AC_REQUIRE([gt_LOCALE_FR_UTF8]) + AC_REQUIRE([AC_CANONICAL_HOST]) dnl for cross-compiles + AC_CACHE_CHECK([whether mbrtowc handles a NULL pwc argument], + [gl_cv_func_mbrtowc_null_arg1], + [ + dnl Initial guess, used when cross-compiling or when no suitable locale + dnl is present. +changequote(,)dnl + case "$host_os" in + # Guess no on Solaris. + solaris*) gl_cv_func_mbrtowc_null_arg1="guessing no" ;; + # Guess yes otherwise. + *) gl_cv_func_mbrtowc_null_arg1="guessing yes" ;; + esac +changequote([,])dnl + if test $LOCALE_FR_UTF8 != none; then + AC_RUN_IFELSE( + [AC_LANG_SOURCE([[ +#include <locale.h> +#include <stdlib.h> +#include <string.h> +/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before + <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> must be + included before <wchar.h>. */ +#include <stddef.h> +#include <stdio.h> +#include <time.h> +#include <wchar.h> +int main () +{ + int result = 0; + + if (setlocale (LC_ALL, "$LOCALE_FR_UTF8") != NULL) + { + char input[] = "\303\237er"; + mbstate_t state; + wchar_t wc; + size_t ret; + + memset (&state, '\0', sizeof (mbstate_t)); + wc = (wchar_t) 0xBADFACE; + ret = mbrtowc (&wc, input, 5, &state); + if (ret != 2) + result |= 1; + if (!mbsinit (&state)) + result |= 2; + + memset (&state, '\0', sizeof (mbstate_t)); + ret = mbrtowc (NULL, input, 5, &state); + if (ret != 2) /* Solaris 7 fails here: ret is -1. */ + result |= 4; + if (!mbsinit (&state)) + result |= 8; + } + return result; +}]])], + [gl_cv_func_mbrtowc_null_arg1=yes], + [gl_cv_func_mbrtowc_null_arg1=no], + [:]) + fi + ]) +]) + +dnl Test whether mbrtowc supports a NULL string argument correctly. +dnl Result is gl_cv_func_mbrtowc_null_arg2. + +AC_DEFUN([gl_MBRTOWC_NULL_ARG2], +[ + AC_REQUIRE([AC_PROG_CC]) + AC_REQUIRE([gt_LOCALE_FR_UTF8]) + AC_REQUIRE([AC_CANONICAL_HOST]) dnl for cross-compiles + AC_CACHE_CHECK([whether mbrtowc handles a NULL string argument], + [gl_cv_func_mbrtowc_null_arg2], + [ + dnl Initial guess, used when cross-compiling or when no suitable locale + dnl is present. +changequote(,)dnl + case "$host_os" in + # Guess no on OSF/1. + osf*) gl_cv_func_mbrtowc_null_arg2="guessing no" ;; + # Guess yes otherwise. + *) gl_cv_func_mbrtowc_null_arg2="guessing yes" ;; + esac +changequote([,])dnl + if test $LOCALE_FR_UTF8 != none; then + AC_RUN_IFELSE( + [AC_LANG_SOURCE([[ +#include <locale.h> +#include <string.h> +/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before + <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> must be + included before <wchar.h>. */ +#include <stddef.h> +#include <stdio.h> +#include <time.h> +#include <wchar.h> +int main () +{ + if (setlocale (LC_ALL, "$LOCALE_FR_UTF8") != NULL) + { + mbstate_t state; + wchar_t wc; + int ret; + + memset (&state, '\0', sizeof (mbstate_t)); + wc = (wchar_t) 0xBADFACE; + mbrtowc (&wc, NULL, 5, &state); + /* Check that wc was not modified. */ + if (wc != (wchar_t) 0xBADFACE) + return 1; + } + return 0; +}]])], + [gl_cv_func_mbrtowc_null_arg2=yes], + [gl_cv_func_mbrtowc_null_arg2=no], + [:]) + fi + ]) +]) + +dnl Test whether mbrtowc, when parsing the end of a multibyte character, +dnl correctly returns the number of bytes that were needed to complete the +dnl character (not the total number of bytes of the multibyte character). +dnl Result is gl_cv_func_mbrtowc_retval. + +AC_DEFUN([gl_MBRTOWC_RETVAL], +[ + AC_REQUIRE([AC_PROG_CC]) + AC_REQUIRE([gt_LOCALE_FR_UTF8]) + AC_REQUIRE([gt_LOCALE_JA]) + AC_REQUIRE([AC_CANONICAL_HOST]) + AC_CACHE_CHECK([whether mbrtowc has a correct return value], + [gl_cv_func_mbrtowc_retval], + [ + dnl Initial guess, used when cross-compiling or when no suitable locale + dnl is present. +changequote(,)dnl + case "$host_os" in + # Guess no on HP-UX, Solaris, native Windows. + hpux* | solaris* | mingw*) gl_cv_func_mbrtowc_retval="guessing no" ;; + # Guess yes otherwise. + *) gl_cv_func_mbrtowc_retval="guessing yes" ;; + esac +changequote([,])dnl + if test $LOCALE_FR_UTF8 != none || test $LOCALE_JA != none \ + || { case "$host_os" in mingw*) true;; *) false;; esac; }; then + AC_RUN_IFELSE( + [AC_LANG_SOURCE([[ +#include <locale.h> +#include <string.h> +/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before + <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> must be + included before <wchar.h>. */ +#include <stddef.h> +#include <stdio.h> +#include <time.h> +#include <wchar.h> +int main () +{ + int result = 0; + int found_some_locale = 0; + /* This fails on Solaris. */ + if (setlocale (LC_ALL, "$LOCALE_FR_UTF8") != NULL) + { + char input[] = "B\303\274\303\237er"; /* "Büßer" */ + mbstate_t state; + wchar_t wc; + + memset (&state, '\0', sizeof (mbstate_t)); + if (mbrtowc (&wc, input + 1, 1, &state) == (size_t)(-2)) + { + input[1] = '\0'; + if (mbrtowc (&wc, input + 2, 5, &state) != 1) + result |= 1; + } + found_some_locale = 1; + } + /* This fails on HP-UX 11.11. */ + if (setlocale (LC_ALL, "$LOCALE_JA") != NULL) + { + char input[] = "B\217\253\344\217\251\316er"; /* "Büßer" */ + mbstate_t state; + wchar_t wc; + + memset (&state, '\0', sizeof (mbstate_t)); + if (mbrtowc (&wc, input + 1, 1, &state) == (size_t)(-2)) + { + input[1] = '\0'; + if (mbrtowc (&wc, input + 2, 5, &state) != 2) + result |= 2; + } + found_some_locale = 1; + } + /* This fails on native Windows. */ + if (setlocale (LC_ALL, "Japanese_Japan.932") != NULL) + { + char input[] = "<\223\372\226\173\214\352>"; /* "<日本語>" */ + mbstate_t state; + wchar_t wc; + + memset (&state, '\0', sizeof (mbstate_t)); + if (mbrtowc (&wc, input + 3, 1, &state) == (size_t)(-2)) + { + input[3] = '\0'; + if (mbrtowc (&wc, input + 4, 4, &state) != 1) + result |= 4; + } + found_some_locale = 1; + } + if (setlocale (LC_ALL, "Chinese_Taiwan.950") != NULL) + { + char input[] = "<\244\351\245\273\273\171>"; /* "<日本語>" */ + mbstate_t state; + wchar_t wc; + + memset (&state, '\0', sizeof (mbstate_t)); + if (mbrtowc (&wc, input + 3, 1, &state) == (size_t)(-2)) + { + input[3] = '\0'; + if (mbrtowc (&wc, input + 4, 4, &state) != 1) + result |= 8; + } + found_some_locale = 1; + } + if (setlocale (LC_ALL, "Chinese_China.936") != NULL) + { + char input[] = "<\310\325\261\276\325\132>"; /* "<日本語>" */ + mbstate_t state; + wchar_t wc; + + memset (&state, '\0', sizeof (mbstate_t)); + if (mbrtowc (&wc, input + 3, 1, &state) == (size_t)(-2)) + { + input[3] = '\0'; + if (mbrtowc (&wc, input + 4, 4, &state) != 1) + result |= 16; + } + found_some_locale = 1; + } + return (found_some_locale ? result : 77); +}]])], + [gl_cv_func_mbrtowc_retval=yes], + [if test $? != 77; then + gl_cv_func_mbrtowc_retval=no + fi + ], + [:]) + fi + ]) +]) + +dnl Test whether mbrtowc, when parsing a NUL character, correctly returns 0. +dnl Result is gl_cv_func_mbrtowc_nul_retval. + +AC_DEFUN([gl_MBRTOWC_NUL_RETVAL], +[ + AC_REQUIRE([AC_PROG_CC]) + AC_REQUIRE([gt_LOCALE_ZH_CN]) + AC_REQUIRE([AC_CANONICAL_HOST]) dnl for cross-compiles + AC_CACHE_CHECK([whether mbrtowc returns 0 when parsing a NUL character], + [gl_cv_func_mbrtowc_nul_retval], + [ + dnl Initial guess, used when cross-compiling or when no suitable locale + dnl is present. +changequote(,)dnl + case "$host_os" in + # Guess no on Solaris 8 and 9. + solaris2.[89]) gl_cv_func_mbrtowc_nul_retval="guessing no" ;; + # Guess yes otherwise. + *) gl_cv_func_mbrtowc_nul_retval="guessing yes" ;; + esac +changequote([,])dnl + if test $LOCALE_ZH_CN != none; then + AC_RUN_IFELSE( + [AC_LANG_SOURCE([[ +#include <locale.h> +#include <string.h> +/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before + <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> must be + included before <wchar.h>. */ +#include <stddef.h> +#include <stdio.h> +#include <time.h> +#include <wchar.h> +int main () +{ + /* This fails on Solaris 8 and 9. */ + if (setlocale (LC_ALL, "$LOCALE_ZH_CN") != NULL) + { + mbstate_t state; + wchar_t wc; + + memset (&state, '\0', sizeof (mbstate_t)); + if (mbrtowc (&wc, "", 1, &state) != 0) + return 1; + } + return 0; +}]])], + [gl_cv_func_mbrtowc_nul_retval=yes], + [gl_cv_func_mbrtowc_nul_retval=no], + [:]) + fi + ]) +]) + +# Prerequisites of lib/mbrtowc.c. +AC_DEFUN([gl_PREREQ_MBRTOWC], [ + : +]) + + +dnl From Paul Eggert + +dnl This is an override of an autoconf macro. + +AC_DEFUN([AC_FUNC_MBRTOWC], +[ + dnl Same as AC_FUNC_MBRTOWC in autoconf-2.60. + AC_CACHE_CHECK([whether mbrtowc and mbstate_t are properly declared], + gl_cv_func_mbrtowc, + [AC_LINK_IFELSE( + [AC_LANG_PROGRAM( + [[/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be + included before <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> + must be included before <wchar.h>. */ + #include <stddef.h> + #include <stdio.h> + #include <time.h> + #include <wchar.h>]], + [[wchar_t wc; + char const s[] = ""; + size_t n = 1; + mbstate_t state; + return ! (sizeof state && (mbrtowc) (&wc, s, n, &state));]])], + gl_cv_func_mbrtowc=yes, + gl_cv_func_mbrtowc=no)]) + if test $gl_cv_func_mbrtowc = yes; then + AC_DEFINE([HAVE_MBRTOWC], [1], + [Define to 1 if mbrtowc and mbstate_t are properly declared.]) + fi +]) diff --git a/m4/mbsinit.m4 b/m4/mbsinit.m4 new file mode 100644 index 000000000..da56c3d11 --- /dev/null +++ b/m4/mbsinit.m4 @@ -0,0 +1,51 @@ +# mbsinit.m4 serial 8 +dnl Copyright (C) 2008, 2010-2012 Free Software Foundation, Inc. +dnl This file is free software; the Free Software Foundation +dnl gives unlimited permission to copy and/or distribute it, +dnl with or without modifications, as long as this notice is preserved. + +AC_DEFUN([gl_FUNC_MBSINIT], +[ + AC_REQUIRE([gl_WCHAR_H_DEFAULTS]) + AC_REQUIRE([AC_CANONICAL_HOST]) + + AC_REQUIRE([AC_TYPE_MBSTATE_T]) + gl_MBSTATE_T_BROKEN + + AC_CHECK_FUNCS_ONCE([mbsinit]) + if test $ac_cv_func_mbsinit = no; then + HAVE_MBSINIT=0 + AC_CHECK_DECLS([mbsinit],,, [[ +/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before + <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> must be + included before <wchar.h>. */ +#include <stddef.h> +#include <stdio.h> +#include <time.h> +#include <wchar.h> +]]) + if test $ac_cv_have_decl_mbsinit = yes; then + dnl On Minix 3.1.8, the system's <wchar.h> declares mbsinit() although + dnl it does not have the function. Avoid a collision with gnulib's + dnl replacement. + REPLACE_MBSINIT=1 + fi + else + if test $REPLACE_MBSTATE_T = 1; then + REPLACE_MBSINIT=1 + else + dnl On mingw, mbsinit() always returns 1, which is inappropriate for + dnl states produced by mbrtowc() for an incomplete multibyte character + dnl in multibyte locales. + case "$host_os" in + mingw*) REPLACE_MBSINIT=1 ;; + esac + fi + fi +]) + +# Prerequisites of lib/mbsinit.c. +AC_DEFUN([gl_PREREQ_MBSINIT], [ + : +]) diff --git a/m4/mbstate_t.m4 b/m4/mbstate_t.m4 new file mode 100644 index 000000000..61a8190ce --- /dev/null +++ b/m4/mbstate_t.m4 @@ -0,0 +1,41 @@ +# mbstate_t.m4 serial 13 +dnl Copyright (C) 2000-2002, 2008-2012 Free Software Foundation, Inc. +dnl This file is free software; the Free Software Foundation +dnl gives unlimited permission to copy and/or distribute it, +dnl with or without modifications, as long as this notice is preserved. + +# From Paul Eggert. + +# BeOS 5 has <wchar.h> but does not define mbstate_t, +# so you can't declare an object of that type. +# Check for this incompatibility with Standard C. + +# AC_TYPE_MBSTATE_T +# ----------------- +AC_DEFUN([AC_TYPE_MBSTATE_T], +[ + AC_REQUIRE([AC_USE_SYSTEM_EXTENSIONS]) dnl for HP-UX 11.11 + + AC_CACHE_CHECK([for mbstate_t], [ac_cv_type_mbstate_t], + [AC_COMPILE_IFELSE( + [AC_LANG_PROGRAM( + [AC_INCLUDES_DEFAULT[ +/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before + <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> must be + included before <wchar.h>. */ +#include <stddef.h> +#include <stdio.h> +#include <time.h> +#include <wchar.h>]], + [[mbstate_t x; return sizeof x;]])], + [ac_cv_type_mbstate_t=yes], + [ac_cv_type_mbstate_t=no])]) + if test $ac_cv_type_mbstate_t = yes; then + AC_DEFINE([HAVE_MBSTATE_T], [1], + [Define to 1 if <wchar.h> declares mbstate_t.]) + else + AC_DEFINE([mbstate_t], [int], + [Define to a type if <wchar.h> does not define.]) + fi +]) diff --git a/m4/mbtowc.m4 b/m4/mbtowc.m4 new file mode 100644 index 000000000..fec0d2582 --- /dev/null +++ b/m4/mbtowc.m4 @@ -0,0 +1,19 @@ +# mbtowc.m4 serial 2 +dnl Copyright (C) 2011-2012 Free Software Foundation, Inc. +dnl This file is free software; the Free Software Foundation +dnl gives unlimited permission to copy and/or distribute it, +dnl with or without modifications, as long as this notice is preserved. + +AC_DEFUN([gl_FUNC_MBTOWC], +[ + AC_REQUIRE([gl_STDLIB_H_DEFAULTS]) + + if false; then + REPLACE_MBTOWC=1 + fi +]) + +# Prerequisites of lib/mbtowc.c. +AC_DEFUN([gl_PREREQ_MBTOWC], [ + : +]) diff --git a/m4/nl_langinfo.m4 b/m4/nl_langinfo.m4 new file mode 100644 index 000000000..80fe60d57 --- /dev/null +++ b/m4/nl_langinfo.m4 @@ -0,0 +1,50 @@ +# nl_langinfo.m4 serial 5 +dnl Copyright (C) 2009-2012 Free Software Foundation, Inc. +dnl This file is free software; the Free Software Foundation +dnl gives unlimited permission to copy and/or distribute it, +dnl with or without modifications, as long as this notice is preserved. + +AC_DEFUN([gl_FUNC_NL_LANGINFO], +[ + AC_REQUIRE([gl_LANGINFO_H_DEFAULTS]) + AC_REQUIRE([gl_LANGINFO_H]) + AC_CHECK_FUNCS_ONCE([nl_langinfo]) + AC_REQUIRE([AC_CANONICAL_HOST]) dnl for cross-compiles + if test $ac_cv_func_nl_langinfo = yes; then + # On Irix 6.5, YESEXPR is defined, but nl_langinfo(YESEXPR) is broken. + AC_CACHE_CHECK([whether YESEXPR works], + [gl_cv_func_nl_langinfo_yesexpr_works], + [AC_RUN_IFELSE( + [AC_LANG_PROGRAM([[#include <langinfo.h> +]], [[return !*nl_langinfo(YESEXPR); +]])], + [gl_cv_func_nl_langinfo_yesexpr_works=yes], + [gl_cv_func_nl_langinfo_yesexpr_works=no], + [ + case "$host_os" in + # Guess no on irix systems. + irix*) gl_cv_func_nl_langinfo_yesexpr_works="guessing no";; + # Guess yes elsewhere. + *) gl_cv_func_nl_langinfo_yesexpr_works="guessing yes";; + esac + ]) + ]) + case $gl_cv_func_nl_langinfo_yesexpr_works in + *yes) FUNC_NL_LANGINFO_YESEXPR_WORKS=1 ;; + *) FUNC_NL_LANGINFO_YESEXPR_WORKS=0 ;; + esac + AC_DEFINE_UNQUOTED([FUNC_NL_LANGINFO_YESEXPR_WORKS], + [$FUNC_NL_LANGINFO_YESEXPR_WORKS], + [Define to 1 if nl_langinfo (YESEXPR) returns a non-empty string.]) + if test $HAVE_LANGINFO_CODESET = 1 && test $HAVE_LANGINFO_ERA = 1 \ + && test $FUNC_NL_LANGINFO_YESEXPR_WORKS = 1; then + : + else + REPLACE_NL_LANGINFO=1 + AC_DEFINE([REPLACE_NL_LANGINFO], [1], + [Define if nl_langinfo exists but is overridden by gnulib.]) + fi + else + HAVE_NL_LANGINFO=0 + fi +]) diff --git a/m4/regex.m4 b/m4/regex.m4 new file mode 100644 index 000000000..08f1352e2 --- /dev/null +++ b/m4/regex.m4 @@ -0,0 +1,223 @@ +# serial 60 + +# Copyright (C) 1996-2001, 2003-2012 Free Software Foundation, Inc. +# +# This file is free software; the Free Software Foundation +# gives unlimited permission to copy and/or distribute it, +# with or without modifications, as long as this notice is preserved. + +dnl Initially derived from code in GNU grep. +dnl Mostly written by Jim Meyering. + +AC_PREREQ([2.50]) + +AC_DEFUN([gl_REGEX], +[ + AC_ARG_WITH([included-regex], + [AS_HELP_STRING([--without-included-regex], + [don't compile regex; this is the default on systems + with recent-enough versions of the GNU C Library + (use with caution on other systems).])]) + + case $with_included_regex in #( + yes|no) ac_use_included_regex=$with_included_regex + ;; + '') + # If the system regex support is good enough that it passes the + # following run test, then default to *not* using the included regex.c. + # If cross compiling, assume the test would fail and use the included + # regex.c. + AC_CACHE_CHECK([for working re_compile_pattern], + [gl_cv_func_re_compile_pattern_working], + [AC_RUN_IFELSE( + [AC_LANG_PROGRAM( + [AC_INCLUDES_DEFAULT[ + #include <locale.h> + #include <limits.h> + #include <regex.h> + ]], + [[int result = 0; + static struct re_pattern_buffer regex; + unsigned char folded_chars[UCHAR_MAX + 1]; + int i; + const char *s; + struct re_registers regs; + + /* http://sourceware.org/ml/libc-hacker/2006-09/msg00008.html + This test needs valgrind to catch the bug on Debian + GNU/Linux 3.1 x86, but it might catch the bug better + on other platforms and it shouldn't hurt to try the + test here. */ + if (setlocale (LC_ALL, "en_US.UTF-8")) + { + static char const pat[] = "insert into"; + static char const data[] = + "\xFF\0\x12\xA2\xAA\xC4\xB1,K\x12\xC4\xB1*\xACK"; + re_set_syntax (RE_SYNTAX_GREP | RE_HAT_LISTS_NOT_NEWLINE + | RE_ICASE); + memset (®ex, 0, sizeof regex); + s = re_compile_pattern (pat, sizeof pat - 1, ®ex); + if (s) + result |= 1; + else if (re_search (®ex, data, sizeof data - 1, + 0, sizeof data - 1, ®s) + != -1) + result |= 1; + if (! setlocale (LC_ALL, "C")) + return 1; + } + + /* This test is from glibc bug 3957, reported by Andrew Mackey. */ + re_set_syntax (RE_SYNTAX_EGREP | RE_HAT_LISTS_NOT_NEWLINE); + memset (®ex, 0, sizeof regex); + s = re_compile_pattern ("a[^x]b", 6, ®ex); + if (s) + result |= 2; + /* This should fail, but succeeds for glibc-2.5. */ + else if (re_search (®ex, "a\nb", 3, 0, 3, ®s) != -1) + result |= 2; + + /* This regular expression is from Spencer ere test number 75 + in grep-2.3. */ + re_set_syntax (RE_SYNTAX_POSIX_EGREP); + memset (®ex, 0, sizeof regex); + for (i = 0; i <= UCHAR_MAX; i++) + folded_chars[i] = i; + regex.translate = folded_chars; + s = re_compile_pattern ("a[[:@:>@:]]b\n", 11, ®ex); + /* This should fail with _Invalid character class name_ error. */ + if (!s) + result |= 4; + + /* Ensure that [b-a] is diagnosed as invalid, when + using RE_NO_EMPTY_RANGES. */ + re_set_syntax (RE_SYNTAX_POSIX_EGREP | RE_NO_EMPTY_RANGES); + memset (®ex, 0, sizeof regex); + s = re_compile_pattern ("a[b-a]", 6, ®ex); + if (s == 0) + result |= 8; + + /* This should succeed, but does not for glibc-2.1.3. */ + memset (®ex, 0, sizeof regex); + s = re_compile_pattern ("{1", 2, ®ex); + if (s) + result |= 8; + + /* The following example is derived from a problem report + against gawk from Jorge Stolfi <stolfi@ic.unicamp.br>. */ + memset (®ex, 0, sizeof regex); + s = re_compile_pattern ("[an\371]*n", 7, ®ex); + if (s) + result |= 8; + /* This should match, but does not for glibc-2.2.1. */ + else if (re_match (®ex, "an", 2, 0, ®s) != 2) + result |= 8; + + memset (®ex, 0, sizeof regex); + s = re_compile_pattern ("x", 1, ®ex); + if (s) + result |= 8; + /* glibc-2.2.93 does not work with a negative RANGE argument. */ + else if (re_search (®ex, "wxy", 3, 2, -2, ®s) != 1) + result |= 8; + + /* The version of regex.c in older versions of gnulib + ignored RE_ICASE. Detect that problem too. */ + re_set_syntax (RE_SYNTAX_EMACS | RE_ICASE); + memset (®ex, 0, sizeof regex); + s = re_compile_pattern ("x", 1, ®ex); + if (s) + result |= 16; + else if (re_search (®ex, "WXY", 3, 0, 3, ®s) < 0) + result |= 16; + + /* Catch a bug reported by Vin Shelton in + http://lists.gnu.org/archive/html/bug-coreutils/2007-06/msg00089.html + */ + re_set_syntax (RE_SYNTAX_POSIX_BASIC + & ~RE_CONTEXT_INVALID_DUP + & ~RE_NO_EMPTY_RANGES); + memset (®ex, 0, sizeof regex); + s = re_compile_pattern ("[[:alnum:]_-]\\\\+$", 16, ®ex); + if (s) + result |= 32; + + /* REG_STARTEND was added to glibc on 2004-01-15. + Reject older versions. */ + if (! REG_STARTEND) + result |= 64; + +#if 0 + /* It would be nice to reject hosts whose regoff_t values are too + narrow (including glibc on hosts with 64-bit ptrdiff_t and + 32-bit int), but we should wait until glibc implements this + feature. Otherwise, support for equivalence classes and + multibyte collation symbols would always be broken except + when compiling --without-included-regex. */ + if (sizeof (regoff_t) < sizeof (ptrdiff_t) + || sizeof (regoff_t) < sizeof (ssize_t)) + result |= 64; +#endif + + return result; + ]])], + [gl_cv_func_re_compile_pattern_working=yes], + [gl_cv_func_re_compile_pattern_working=no], + dnl When crosscompiling, assume it is not working. + [gl_cv_func_re_compile_pattern_working=no])]) + case $gl_cv_func_re_compile_pattern_working in #( + yes) ac_use_included_regex=no;; #( + no) ac_use_included_regex=yes;; + esac + ;; + *) AC_MSG_ERROR([Invalid value for --with-included-regex: $with_included_regex]) + ;; + esac + + if test $ac_use_included_regex = yes; then + AC_DEFINE([_REGEX_LARGE_OFFSETS], [1], + [Define if you want regoff_t to be at least as wide POSIX requires.]) + AC_DEFINE([re_syntax_options], [rpl_re_syntax_options], + [Define to rpl_re_syntax_options if the replacement should be used.]) + AC_DEFINE([re_set_syntax], [rpl_re_set_syntax], + [Define to rpl_re_set_syntax if the replacement should be used.]) + AC_DEFINE([re_compile_pattern], [rpl_re_compile_pattern], + [Define to rpl_re_compile_pattern if the replacement should be used.]) + AC_DEFINE([re_compile_fastmap], [rpl_re_compile_fastmap], + [Define to rpl_re_compile_fastmap if the replacement should be used.]) + AC_DEFINE([re_search], [rpl_re_search], + [Define to rpl_re_search if the replacement should be used.]) + AC_DEFINE([re_search_2], [rpl_re_search_2], + [Define to rpl_re_search_2 if the replacement should be used.]) + AC_DEFINE([re_match], [rpl_re_match], + [Define to rpl_re_match if the replacement should be used.]) + AC_DEFINE([re_match_2], [rpl_re_match_2], + [Define to rpl_re_match_2 if the replacement should be used.]) + AC_DEFINE([re_set_registers], [rpl_re_set_registers], + [Define to rpl_re_set_registers if the replacement should be used.]) + AC_DEFINE([re_comp], [rpl_re_comp], + [Define to rpl_re_comp if the replacement should be used.]) + AC_DEFINE([re_exec], [rpl_re_exec], + [Define to rpl_re_exec if the replacement should be used.]) + AC_DEFINE([regcomp], [rpl_regcomp], + [Define to rpl_regcomp if the replacement should be used.]) + AC_DEFINE([regexec], [rpl_regexec], + [Define to rpl_regexec if the replacement should be used.]) + AC_DEFINE([regerror], [rpl_regerror], + [Define to rpl_regerror if the replacement should be used.]) + AC_DEFINE([regfree], [rpl_regfree], + [Define to rpl_regfree if the replacement should be used.]) + fi +]) + +# Prerequisites of lib/regex.c and lib/regex_internal.c. +AC_DEFUN([gl_PREREQ_REGEX], +[ + AC_REQUIRE([AC_USE_SYSTEM_EXTENSIONS]) + AC_REQUIRE([AC_C_INLINE]) + AC_REQUIRE([AC_C_RESTRICT]) + AC_REQUIRE([AC_TYPE_MBSTATE_T]) + AC_CHECK_HEADERS([libintl.h]) + AC_CHECK_FUNCS_ONCE([isblank iswctype wcscoll]) + AC_CHECK_DECLS([isblank], [], [], [[#include <ctype.h>]]) +]) diff --git a/m4/strcase.m4 b/m4/strcase.m4 new file mode 100644 index 000000000..717fa9c32 --- /dev/null +++ b/m4/strcase.m4 @@ -0,0 +1,45 @@ +# strcase.m4 serial 11 +dnl Copyright (C) 2002, 2005-2012 Free Software Foundation, Inc. +dnl This file is free software; the Free Software Foundation +dnl gives unlimited permission to copy and/or distribute it, +dnl with or without modifications, as long as this notice is preserved. + +AC_DEFUN([gl_STRCASE], +[ + gl_FUNC_STRCASECMP + gl_FUNC_STRNCASECMP +]) + +AC_DEFUN([gl_FUNC_STRCASECMP], +[ + AC_REQUIRE([gl_HEADER_STRINGS_H_DEFAULTS]) + AC_CHECK_FUNCS([strcasecmp]) + if test $ac_cv_func_strcasecmp = no; then + HAVE_STRCASECMP=0 + fi +]) + +AC_DEFUN([gl_FUNC_STRNCASECMP], +[ + AC_REQUIRE([gl_HEADER_STRINGS_H_DEFAULTS]) + AC_CHECK_FUNCS([strncasecmp]) + if test $ac_cv_func_strncasecmp = yes; then + HAVE_STRNCASECMP=1 + else + HAVE_STRNCASECMP=0 + fi + AC_CHECK_DECLS([strncasecmp]) + if test $ac_cv_have_decl_strncasecmp = no; then + HAVE_DECL_STRNCASECMP=0 + fi +]) + +# Prerequisites of lib/strcasecmp.c. +AC_DEFUN([gl_PREREQ_STRCASECMP], [ + : +]) + +# Prerequisites of lib/strncasecmp.c. +AC_DEFUN([gl_PREREQ_STRNCASECMP], [ + : +]) diff --git a/m4/strings_h.m4 b/m4/strings_h.m4 new file mode 100644 index 000000000..a057e1c8b --- /dev/null +++ b/m4/strings_h.m4 @@ -0,0 +1,52 @@ +# Configure a replacement for <strings.h>. +# serial 6 + +# Copyright (C) 2007, 2009-2012 Free Software Foundation, Inc. +# This file is free software; the Free Software Foundation +# gives unlimited permission to copy and/or distribute it, +# with or without modifications, as long as this notice is preserved. + +AC_DEFUN([gl_HEADER_STRINGS_H], +[ + dnl Use AC_REQUIRE here, so that the default behavior below is expanded + dnl once only, before all statements that occur in other macros. + AC_REQUIRE([gl_HEADER_STRINGS_H_BODY]) +]) + +AC_DEFUN([gl_HEADER_STRINGS_H_BODY], +[ + AC_REQUIRE([gl_HEADER_STRINGS_H_DEFAULTS]) + + gl_CHECK_NEXT_HEADERS([strings.h]) + if test $ac_cv_header_strings_h = yes; then + HAVE_STRINGS_H=1 + else + HAVE_STRINGS_H=0 + fi + AC_SUBST([HAVE_STRINGS_H]) + + dnl Check for declarations of anything we want to poison if the + dnl corresponding gnulib module is not in use. + gl_WARN_ON_USE_PREPARE([[ + /* Minix 3.1.8 has a bug: <sys/types.h> must be included before + <strings.h>. */ + #include <sys/types.h> + #include <strings.h> + ]], [ffs strcasecmp strncasecmp]) +]) + +AC_DEFUN([gl_STRINGS_MODULE_INDICATOR], +[ + dnl Use AC_REQUIRE here, so that the default settings are expanded once only. + AC_REQUIRE([gl_HEADER_STRINGS_H_DEFAULTS]) + gl_MODULE_INDICATOR_SET_VARIABLE([$1]) +]) + +AC_DEFUN([gl_HEADER_STRINGS_H_DEFAULTS], +[ + GNULIB_FFS=0; AC_SUBST([GNULIB_FFS]) + dnl Assume proper GNU behavior unless another module says otherwise. + HAVE_FFS=1; AC_SUBST([HAVE_FFS]) + HAVE_STRCASECMP=1; AC_SUBST([HAVE_STRCASECMP]) + HAVE_DECL_STRNCASECMP=1; AC_SUBST([HAVE_DECL_STRNCASECMP]) +]) diff --git a/m4/wcrtomb.m4 b/m4/wcrtomb.m4 new file mode 100644 index 000000000..00d7302e8 --- /dev/null +++ b/m4/wcrtomb.m4 @@ -0,0 +1,112 @@ +# wcrtomb.m4 serial 11 +dnl Copyright (C) 2008-2012 Free Software Foundation, Inc. +dnl This file is free software; the Free Software Foundation +dnl gives unlimited permission to copy and/or distribute it, +dnl with or without modifications, as long as this notice is preserved. + +AC_DEFUN([gl_FUNC_WCRTOMB], +[ + AC_REQUIRE([gl_WCHAR_H_DEFAULTS]) + + AC_REQUIRE([AC_TYPE_MBSTATE_T]) + gl_MBSTATE_T_BROKEN + + AC_CHECK_FUNCS_ONCE([wcrtomb]) + if test $ac_cv_func_wcrtomb = no; then + HAVE_WCRTOMB=0 + AC_CHECK_DECLS([wcrtomb],,, [[ +/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before + <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> must be + included before <wchar.h>. */ +#include <stddef.h> +#include <stdio.h> +#include <time.h> +#include <wchar.h> +]]) + if test $ac_cv_have_decl_wcrtomb = yes; then + dnl On Minix 3.1.8, the system's <wchar.h> declares wcrtomb() although + dnl it does not have the function. Avoid a collision with gnulib's + dnl replacement. + REPLACE_WCRTOMB=1 + fi + else + if test $REPLACE_MBSTATE_T = 1; then + REPLACE_WCRTOMB=1 + else + dnl On AIX 4.3, OSF/1 5.1 and Solaris 10, wcrtomb (NULL, 0, NULL) sometimes + dnl returns 0 instead of 1. + AC_REQUIRE([AC_PROG_CC]) + AC_REQUIRE([gt_LOCALE_FR]) + AC_REQUIRE([gt_LOCALE_FR_UTF8]) + AC_REQUIRE([gt_LOCALE_JA]) + AC_REQUIRE([gt_LOCALE_ZH_CN]) + AC_REQUIRE([AC_CANONICAL_HOST]) dnl for cross-compiles + AC_CACHE_CHECK([whether wcrtomb return value is correct], + [gl_cv_func_wcrtomb_retval], + [ + dnl Initial guess, used when cross-compiling or when no suitable locale + dnl is present. +changequote(,)dnl + case "$host_os" in + # Guess no on AIX 4, OSF/1 and Solaris. + aix4* | osf* | solaris*) gl_cv_func_wcrtomb_retval="guessing no" ;; + # Guess yes otherwise. + *) gl_cv_func_wcrtomb_retval="guessing yes" ;; + esac +changequote([,])dnl + if test $LOCALE_FR != none || test $LOCALE_FR_UTF8 != none || test $LOCALE_JA != none || test $LOCALE_ZH_CN != none; then + AC_RUN_IFELSE( + [AC_LANG_SOURCE([[ +#include <locale.h> +#include <string.h> +/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before + <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> must be + included before <wchar.h>. */ +#include <stddef.h> +#include <stdio.h> +#include <time.h> +#include <wchar.h> +int main () +{ + int result = 0; + if (setlocale (LC_ALL, "$LOCALE_FR") != NULL) + { + if (wcrtomb (NULL, 0, NULL) != 1) + result |= 1; + } + if (setlocale (LC_ALL, "$LOCALE_FR_UTF8") != NULL) + { + if (wcrtomb (NULL, 0, NULL) != 1) + result |= 2; + } + if (setlocale (LC_ALL, "$LOCALE_JA") != NULL) + { + if (wcrtomb (NULL, 0, NULL) != 1) + result |= 4; + } + if (setlocale (LC_ALL, "$LOCALE_ZH_CN") != NULL) + { + if (wcrtomb (NULL, 0, NULL) != 1) + result |= 8; + } + return result; +}]])], + [gl_cv_func_wcrtomb_retval=yes], + [gl_cv_func_wcrtomb_retval=no], + [:]) + fi + ]) + case "$gl_cv_func_wcrtomb_retval" in + *yes) ;; + *) REPLACE_WCRTOMB=1 ;; + esac + fi + fi +]) + +# Prerequisites of lib/wcrtomb.c. +AC_DEFUN([gl_PREREQ_WCRTOMB], [ + : +]) diff --git a/m4/wctype_h.m4 b/m4/wctype_h.m4 new file mode 100644 index 000000000..a10938307 --- /dev/null +++ b/m4/wctype_h.m4 @@ -0,0 +1,211 @@ +# wctype_h.m4 serial 16 + +dnl A placeholder for ISO C99 <wctype.h>, for platforms that lack it. + +dnl Copyright (C) 2006-2012 Free Software Foundation, Inc. +dnl This file is free software; the Free Software Foundation +dnl gives unlimited permission to copy and/or distribute it, +dnl with or without modifications, as long as this notice is preserved. + +dnl Written by Paul Eggert. + +AC_DEFUN([gl_WCTYPE_H], +[ + AC_REQUIRE([gl_WCTYPE_H_DEFAULTS]) + AC_REQUIRE([AC_PROG_CC]) + AC_REQUIRE([AC_CANONICAL_HOST]) + AC_CHECK_FUNCS_ONCE([iswcntrl]) + if test $ac_cv_func_iswcntrl = yes; then + HAVE_ISWCNTRL=1 + else + HAVE_ISWCNTRL=0 + fi + AC_SUBST([HAVE_ISWCNTRL]) + + AC_REQUIRE([AC_C_INLINE]) + + AC_REQUIRE([gt_TYPE_WINT_T]) + if test $gt_cv_c_wint_t = yes; then + HAVE_WINT_T=1 + else + HAVE_WINT_T=0 + fi + AC_SUBST([HAVE_WINT_T]) + + gl_CHECK_NEXT_HEADERS([wctype.h]) + if test $ac_cv_header_wctype_h = yes; then + if test $ac_cv_func_iswcntrl = yes; then + dnl Linux libc5 has an iswprint function that returns 0 for all arguments. + dnl The other functions are likely broken in the same way. + AC_CACHE_CHECK([whether iswcntrl works], [gl_cv_func_iswcntrl_works], + [ + AC_RUN_IFELSE( + [AC_LANG_SOURCE([[ + /* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be + included before <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> + must be included before <wchar.h>. */ + #include <stddef.h> + #include <stdio.h> + #include <time.h> + #include <wchar.h> + #include <wctype.h> + int main () { return iswprint ('x') == 0; } + ]])], + [gl_cv_func_iswcntrl_works=yes], [gl_cv_func_iswcntrl_works=no], + [AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[#include <stdlib.h> + #if __GNU_LIBRARY__ == 1 + Linux libc5 i18n is broken. + #endif]], [])], + [gl_cv_func_iswcntrl_works=yes], [gl_cv_func_iswcntrl_works=no]) + ]) + ]) + fi + HAVE_WCTYPE_H=1 + else + HAVE_WCTYPE_H=0 + fi + AC_SUBST([HAVE_WCTYPE_H]) + + if test "$gl_cv_func_iswcntrl_works" = no; then + REPLACE_ISWCNTRL=1 + else + REPLACE_ISWCNTRL=0 + fi + AC_SUBST([REPLACE_ISWCNTRL]) + + if test $HAVE_ISWCNTRL = 0 || test $REPLACE_ISWCNTRL = 1; then + dnl Redefine all of iswcntrl, ..., iswxdigit in <wctype.h>. + : + fi + + if test $REPLACE_ISWCNTRL = 1; then + REPLACE_TOWLOWER=1 + else + AC_CHECK_FUNCS([towlower]) + if test $ac_cv_func_towlower = yes; then + REPLACE_TOWLOWER=0 + else + AC_CHECK_DECLS([towlower],,, + [[/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be + included before <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> + must be included before <wchar.h>. */ + #include <stddef.h> + #include <stdio.h> + #include <time.h> + #include <wchar.h> + #if HAVE_WCTYPE_H + # include <wctype.h> + #endif + ]]) + if test $ac_cv_have_decl_towlower = yes; then + dnl On Minix 3.1.8, the system's <wctype.h> declares towlower() and + dnl towupper() although it does not have the functions. Avoid a + dnl collision with gnulib's replacement. + REPLACE_TOWLOWER=1 + else + REPLACE_TOWLOWER=0 + fi + fi + fi + AC_SUBST([REPLACE_TOWLOWER]) + + if test $HAVE_ISWCNTRL = 0 || test $REPLACE_TOWLOWER = 1; then + dnl Redefine towlower, towupper in <wctype.h>. + : + fi + + dnl We assume that the wctype() and iswctype() functions exist if and only + dnl if the type wctype_t is defined in <wchar.h> or in <wctype.h> if that + dnl exists. + dnl HP-UX 11.00 declares all these in <wchar.h> and lacks <wctype.h>. + AC_CACHE_CHECK([for wctype_t], [gl_cv_type_wctype_t], + [AC_COMPILE_IFELSE( + [AC_LANG_PROGRAM( + [[/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be + included before <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> + must be included before <wchar.h>. */ + #include <stddef.h> + #include <stdio.h> + #include <time.h> + #include <wchar.h> + #if HAVE_WCTYPE_H + # include <wctype.h> + #endif + wctype_t a; + ]], + [[]])], + [gl_cv_type_wctype_t=yes], + [gl_cv_type_wctype_t=no]) + ]) + if test $gl_cv_type_wctype_t = no; then + HAVE_WCTYPE_T=0 + fi + + dnl We assume that the wctrans() and towctrans() functions exist if and only + dnl if the type wctrans_t is defined in <wctype.h>. + AC_CACHE_CHECK([for wctrans_t], [gl_cv_type_wctrans_t], + [AC_COMPILE_IFELSE( + [AC_LANG_PROGRAM( + [[/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be + included before <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> + must be included before <wchar.h>. */ + #include <stddef.h> + #include <stdio.h> + #include <time.h> + #include <wchar.h> + #include <wctype.h> + wctrans_t a; + ]], + [[]])], + [gl_cv_type_wctrans_t=yes], + [gl_cv_type_wctrans_t=no]) + ]) + if test $gl_cv_type_wctrans_t = no; then + HAVE_WCTRANS_T=0 + fi + + dnl Check for declarations of anything we want to poison if the + dnl corresponding gnulib module is not in use. + gl_WARN_ON_USE_PREPARE([[ +/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included before + <wchar.h>. + BSD/OS 4.0.1 has a bug: <stddef.h>, <stdio.h> and <time.h> must be + included before <wchar.h>. */ +#if !(defined __GLIBC__ && !defined __UCLIBC__) +# include <stddef.h> +# include <stdio.h> +# include <time.h> +# include <wchar.h> +#endif +#include <wctype.h> + ]], + [wctype iswctype wctrans towctrans + ]) +]) + +AC_DEFUN([gl_WCTYPE_MODULE_INDICATOR], +[ + dnl Use AC_REQUIRE here, so that the default settings are expanded once only. + AC_REQUIRE([gl_WCTYPE_H_DEFAULTS]) + gl_MODULE_INDICATOR_SET_VARIABLE([$1]) + dnl Define it also as a C macro, for the benefit of the unit tests. + gl_MODULE_INDICATOR_FOR_TESTS([$1]) +]) + +AC_DEFUN([gl_WCTYPE_H_DEFAULTS], +[ + GNULIB_ISWBLANK=0; AC_SUBST([GNULIB_ISWBLANK]) + GNULIB_WCTYPE=0; AC_SUBST([GNULIB_WCTYPE]) + GNULIB_ISWCTYPE=0; AC_SUBST([GNULIB_ISWCTYPE]) + GNULIB_WCTRANS=0; AC_SUBST([GNULIB_WCTRANS]) + GNULIB_TOWCTRANS=0; AC_SUBST([GNULIB_TOWCTRANS]) + dnl Assume proper GNU behavior unless another module says otherwise. + HAVE_ISWBLANK=1; AC_SUBST([HAVE_ISWBLANK]) + HAVE_WCTYPE_T=1; AC_SUBST([HAVE_WCTYPE_T]) + HAVE_WCTRANS_T=1; AC_SUBST([HAVE_WCTRANS_T]) + REPLACE_ISWBLANK=0; AC_SUBST([REPLACE_ISWBLANK]) +]) |