Fix for the UTF-8 memory allocation bug and the UTF-8 encoding

bug related to lone high surrogates.

diff --git a/Lib/test/test_unicode.py b/Lib/test/test_unicode.py
index 8d4bed56474136464c6e1d15c9630628b4a4e1ab..5368f6ef6064a3855fb691798c2154976e82470c 100644 (file)
--- a/Lib/test/test_unicode.py
+++ b/Lib/test/test_unicode.py
@@ -495,19 +495,18 @@ else:
 verify(unicode('+3ADYAA-', 'utf-7', 'replace') == u'\ufffd')
 
 # UTF-8 specific encoding tests:
-verify(u'\u20ac'.encode('utf-8') == \
-       ''.join((chr(0xe2), chr(0x82), chr(0xac))) )
-verify(u'\ud800\udc02'.encode('utf-8') == \
-       ''.join((chr(0xf0), chr(0x90), chr(0x80), chr(0x82))) )
-verify(u'\ud84d\udc56'.encode('utf-8') == \
-       ''.join((chr(0xf0), chr(0xa3), chr(0x91), chr(0x96))) )
+verify(u'\u20ac'.encode('utf-8') == '\xe2\x82\xac')
+verify(u'\ud800\udc02'.encode('utf-8') == '\xf0\x90\x80\x82')
+verify(u'\ud84d\udc56'.encode('utf-8') == '\xf0\xa3\x91\x96')
+verify(u'\ud800'.encode('utf-8') == '\xed\xa0\x80')
+verify(u'\udc00'.encode('utf-8') == '\xed\xb0\x80')
+verify((u'\ud800\udc02'*1000).encode('utf-8') ==
+       '\xf0\x90\x80\x82'*1000)
+
 # UTF-8 specific decoding tests
-verify(unicode(''.join((chr(0xf0), chr(0xa3), chr(0x91), chr(0x96))),
-               'utf-8') == u'\U00023456' )
-verify(unicode(''.join((chr(0xf0), chr(0x90), chr(0x80), chr(0x82))),
-               'utf-8') == u'\U00010002' )
-verify(unicode(''.join((chr(0xe2), chr(0x82), chr(0xac))),
-               'utf-8') == u'\u20ac' )
+verify(unicode('\xf0\xa3\x91\x96', 'utf-8') == u'\U00023456' )
+verify(unicode('\xf0\x90\x80\x82', 'utf-8') == u'\U00010002' )
+verify(unicode('\xe2\x82\xac', 'utf-8') == u'\u20ac' )
 
 # Other possible utf-8 test cases:
 # * strict decoding testing for all of the

diff --git a/Objects/unicodeobject.c b/Objects/unicodeobject.c
index 68afaa05c85fb25c5e3c1a7932b0878d36f2ee9f..c7e5c8af2ebaf29467ab0f8ee57d6bca5d879cf1 100644 (file)
--- a/Objects/unicodeobject.c
+++ b/Objects/unicodeobject.c
@@ -1171,61 +1171,64 @@ PyObject *PyUnicode_EncodeUTF8(const Py_UNICODE *s,
 {
     PyObject *v;
     char *p;
-    char *q;
-    Py_UCS4 ch2;
-    unsigned int cbAllocated = 3 * size;
+    unsigned int cbAllocated = 2 * size;
     unsigned int cbWritten = 0;
     int i = 0;
 
-    v = PyString_FromStringAndSize(NULL, cbAllocated);
+    v = PyString_FromStringAndSize(NULL, cbAllocated + 4);
     if (v == NULL)
         return NULL;
     if (size == 0)
         return v;
 
-    p = q = PyString_AS_STRING(v);
+    p = PyString_AS_STRING(v);
     while (i < size) {
         Py_UCS4 ch = s[i++];
+
         if (ch < 0x80) {
             *p++ = (char) ch;
             cbWritten++;
         }
+
         else if (ch < 0x0800) {
             *p++ = 0xc0 | (ch >> 6);
             *p++ = 0x80 | (ch & 0x3f);
             cbWritten += 2;
         }
-        else if (ch < 0x10000) {
-            /* Check for high surrogate */
-            if (0xD800 <= ch && ch <= 0xDBFF) {
-                if (i != size) {
-                    ch2 = s[i];
-                    if (0xDC00 <= ch2 && ch2 <= 0xDFFF) {
                         
-                        if (cbWritten >= (cbAllocated - 4)) {
-                           /* Provide enough room for some more
-                              surrogates */
-                           cbAllocated += 4*10;
-                            if (_PyString_Resize(&v, cbAllocated))
+        else {
+           
+           /* Assure that we have enough room for high order Unicode
+              ordinals */
+           if (cbWritten >= cbAllocated) {
+               cbAllocated += 4 * 10;
+               if (_PyString_Resize(&v, cbAllocated + 4))
                                goto onError;
+               p = PyString_AS_STRING(v) + cbWritten;
                         }
 
-                        /* combine the two values */
+           if (ch < 0x10000) {
+               /* Check for high surrogate */
+               if (0xD800 <= ch && ch <= 0xDBFF && i != size) {
+                   Py_UCS4 ch2 = s[i];
+                   /* Check for low surrogate */
+                   if (0xDC00 <= ch2 && ch2 <= 0xDFFF) {
                         ch = ((ch - 0xD800)<<10 | (ch2-0xDC00))+0x10000;
-                    
                         *p++ = (char)((ch >> 18) | 0xf0);
                         *p++ = (char)(0x80 | ((ch >> 12) & 0x3f));
+                       *p++ = (char)(0x80 | ((ch >> 6) & 0x3f));
+                       *p++ = (char)(0x80 | (ch & 0x3f));
                         i++;
                         cbWritten += 4;
+                       continue;
                     }
+                   /* Fall through: handles isolated high surrogates */
                 }
-            }
-            else {
                 *p++ = (char)(0xe0 | (ch >> 12));
-                cbWritten += 3;
-            }
             *p++ = (char)(0x80 | ((ch >> 6) & 0x3f));
             *p++ = (char)(0x80 | (ch & 0x3f));
+               cbWritten += 3;
+    
         } else {
             *p++ = 0xf0 | (ch>>18);
             *p++ = 0x80 | ((ch>>12) & 0x3f);
@@ -1234,8 +1237,9 @@ PyObject *PyUnicode_EncodeUTF8(const Py_UNICODE *s,
             cbWritten += 4;
        }
     }
+    }
     *p = '\0';
-    if (_PyString_Resize(&v, p - q))
+    if (_PyString_Resize(&v, cbWritten))
        goto onError;
     return v;