[pypy-svn] r61082 - in pypy/trunk/pypy/module/_codecs: . test

[fijal at codespeak.net]

Author: fijal
Date: Sun Jan 18 17:57:56 2009
New Revision: 61082

Modified:
   pypy/trunk/pypy/module/_codecs/app_codecs.py
   pypy/trunk/pypy/module/_codecs/test/test_codecs.py
Log:
a bit of fixes for _codecs module. This is not over yet thouhg.


Modified: pypy/trunk/pypy/module/_codecs/app_codecs.py
==============================================================================
--- pypy/trunk/pypy/module/_codecs/app_codecs.py	(original)
+++ pypy/trunk/pypy/module/_codecs/app_codecs.py	Sun Jan 18 17:57:56 2009
@@ -1,3 +1,4 @@
+
 # Note:
 # This *is* now explicitly RPython.
 # Please make sure not to break this.
@@ -48,12 +49,6 @@
     """
     pass
 
-def readbuffer_encode( obj, errors='strict'):
-    """None
-    """
-    res = str(obj)
-    return res, len(res)
-
 def escape_encode( obj, errors='strict'):
     """None
     """
@@ -142,10 +137,17 @@
             t = 0
             h = 0
             for j in range(start, stop, step):
-                t += ord(unistr[i+j])<<(h*8)
+                t += ord(unistr[i+j])<<(h*8)    
                 h += 1
             i += unicode_bytes
-            p += unichr(t)
+            try:
+                p += unichr(t)
+            except ValueError:
+                startpos = i - unicode_bytes
+                endpos = i
+                raise UnicodeDecodeError('unicode_internal', unistr, startpos,
+                                         endpos,
+                                         "unichr(%s) not in range" % (t,))
         res = u''.join(p)
         return res, len(unistr)
 
@@ -416,8 +418,8 @@
                         inShift = 1
                     
                 elif SPECIAL(ch, 0, 0) :
-                    raise  UnicodeDecodeError, "unexpected special character"
-                        
+                    msg = "unexpected special character"
+                    raise UnicodeDecodeError('utf-7', s, i-1, i, msg)
                 else:  
                     p +=  ch 
             else:
@@ -437,7 +439,8 @@
                 
         elif (SPECIAL(ch, 0, 0)):
             i += 1
-            raise UnicodeDecodeError, "unexpected special character"
+            msg = "unexpected special character"
+            raise UnicodeDecodeError('utf-7', s, i-1, i, msg)
         else:
             p +=  ch 
             i += 1
@@ -445,7 +448,8 @@
     if (inShift) :
         #XXX This aint right
         endinpos = size
-        raise UnicodeDecodeError, "unterminated shift sequence"
+        msg = "unterminated shift sequence"
+        raise UnicodeDecodeError('utf-7', s, i-1, i, msg)
         
     return p
 
@@ -857,7 +861,7 @@
                 raise KeyError
             else:
                 raise TypeError
-        except KeyError:
+        except (KeyError, IndexError):
             x = unicode_call_errorhandler(errors, "charmap",
                 "character maps to <undefined>", s, inpos, inpos+1)
             p += x[0]

Modified: pypy/trunk/pypy/module/_codecs/test/test_codecs.py
==============================================================================
--- pypy/trunk/pypy/module/_codecs/test/test_codecs.py	(original)
+++ pypy/trunk/pypy/module/_codecs/test/test_codecs.py	Sun Jan 18 17:57:56 2009
@@ -352,3 +352,21 @@
         s = '\\\n'
         decoded = _codecs.unicode_escape_decode(s)[0]
         assert decoded == ''
+
+    def test_charmap_decode(self):
+        import codecs
+        res = codecs.charmap_decode("\x00\x01\x02", "replace", u"ab")
+        assert res == (u"ab\ufffd", 3)
+
+    def test_decode_errors(self):
+        import sys
+        if sys.maxunicode > 0xffff:
+            try:
+                "\x00\x00\x00\x00\x00\x11\x11\x00".decode("unicode_internal")
+            except UnicodeDecodeError, ex:
+                assert "unicode_internal" == ex.encoding
+                assert "\x00\x00\x00\x00\x00\x11\x11\x00" == ex.object
+                assert ex.start == 4
+                assert ex.end == 8
+            else:
+                raise Exception("DID NOT RAISE")