better encoding and decoding

3 files changed, 76 insertions, 13 deletions

diff --git a/modules/language/python/bytes.scm b/modules/language/python/bytes.scm
index daf7adc..4ba7d2b 100644
--- a/modules/language/python/bytes.scm
+++ b/modules/language/python/bytes.scm
@@ -2,10 +2,12 @@
   #:use-module (oop goops)
   #:use-module (oop pf-objects)
   #:use-module (ice-9 match)
+  #:use-module (ice-9 iconv)
   #:use-module (rnrs bytevectors)
   #:use-module (system foreign)
   #:use-module (language python string)
   #:use-module (language python for)
+  #:use-module (language python def)
   #:use-module (language python try)
   #:use-module (language python exceptions)
   #:use-module (language python list)
@@ -596,12 +598,46 @@
             (lp (+ i 1) (+ j 1)))))
     (bytes s)))
 
-(define-py** decode (py-decode bytes o n . l)
-  (let lp ((i 0) (r '()))
-    (if (< i n)
-        (lp (+ i 1) (cons (b-ref o i) r))
-        (utf8->string (list->b (reverse r))))))
+(define-python-class UnicodeDecodeError (Exception))
 
+(define-py** decode (py-decode bytes o n . l)
+  (apply
+   (lam ((= encoding "UTF-8") (= errors "strict"))	
+	(set! errors   (py-lower (scm-str errors)))
+	(set! errors   (cond
+			((equal? errors "strict")
+			 'error)
+			((equal? errors "replace")
+			 'substitute)
+			((equal? errors "ignore")
+			 (warn
+			  (string-append
+			   "not possible to use ignore "
+			   "encodong error strategy "
+			   "using replace in stead"))
+			 'substitute)
+			(else
+			 (warn
+			  "not a correct encodong error strategy")
+			 'error)))
+	(set! encoding (py-upper (scm-str encoding)))
+	
+	(let lp ((i 0) (r '()))
+	  (if (< i n)
+	      (lp (+ i 1) (cons (b-ref o i) r))
+	      (catch #t
+		(lambda ()
+		  (bytevector->string
+		   (list->b (reverse r))
+		   encoding
+		   errors))
+		(lambda x
+		  (raise (UnicodeDecodeError
+			  (+
+			   "failed to decode "
+			   encoding))))))))
+   l))
+			     
 ;;;py-encode
 
 (define-py* endswith (py-endswith bytes o n suff . l)
@@ -1330,3 +1366,5 @@
 			    zfill)))))
     (pylist-sort! l)
     l))
+
+(set! (@@ (language python string) bytes) bytes)
diff --git a/modules/language/python/compile.scm b/modules/language/python/compile.scm
index 36182a4..42da128 100644
--- a/modules/language/python/compile.scm
+++ b/modules/language/python/compile.scm
@@ -419,9 +419,11 @@
    ((swapcase)   (S 'py-swapcase))
    ((translate)  (S 'py-translate))
    ((zfill)      (S 'py-zfill))
-
-   ;;bytevectors
+   ((encode)     (S 'py-encode))
+  
+   ;;Nytevectors   
    ((decode)     (B 'py-decode))
+  
    ;;DICTS
    ((copy)       (Di 'py-copy))
    ((fromkeys)   (Di 'py-fromkeys))
diff --git a/modules/language/python/string.scm b/modules/language/python/string.scm
index a560801..c1d7d1b 100644
--- a/modules/language/python/string.scm
+++ b/modules/language/python/string.scm
@@ -4,16 +4,19 @@
   #:use-module (oop pf-objects)
   #:use-module (language python hash)
   #:use-module (ice-9 match)
+  #:use-module (ice-9 iconv)
   #:use-module (language python list)
   #:use-module (language python exceptions)
   #:use-module (language python for)
+  #:use-module (language python def)
   #:use-module (language python bool)
   #:use-module (language python persist)
+  #:use-module (rnrs bytevectors)
   #:export (py-format py-capitalize py-center py-endswith
                       py-expandtabs py-find py-rfind
                       py-isalnum py-isalpha py-isdigit py-islower
                       py-isspace py-isupper py-istitle py-join py-ljust
-                      py-rjust py-format-map
+                      py-rjust py-format-map py-encode
                       py-rljust py-lower py-upper py-lstrip py-rstrip
                       py-partition py-replace py-strip py-title
                       py-rpartitio py-rindex py-split py-rsplit py-splitlines
@@ -100,10 +103,6 @@
     
     
 
-;;;py-decode
-;;;py-encode
-
-
 (define-py (py-endswith endswith o (suff <string>) . l)
   (let* ((n   (string-length o))
          (ns  (string-length suff))
@@ -376,7 +375,31 @@
   (let ((n (len s)))
     (- n (apply pylist-index (string-reverse s) l) 1)))
 
-
+(define bytes #f)
+
+(define-py (py-encode encode s . l) 
+  (apply (lam ((= encoding "UTF-8") (= errors "strict"))	
+	      (set! errors   (py-lower (scm-str errors)))
+	      (set! errors   (cond
+			      ((equal? errors "strict")
+			       'error)
+			      ((equal? errors "replace")
+			       'substitute)
+			      ((equal? errors "ignore")
+			       (warn
+				(string-append
+				 "not possible to use ignore "
+				 "encodong error strategy "
+				 "using replace in stead"))
+			       'substitute)
+			      (else
+			       (warn
+				"not a correct encodong error strategy")
+			       'error)))
+	      (set! encoding (py-upper (scm-str encoding)))
+
+	      (bytes (string->bytevector (scm-str s) encoding errors)))
+	 l))
 
 (define-py (py-split split s . l)
   (define ws (f+ (f-reg "[ \t\n]")))