Issue #19327: Fixed the working of regular expressions with too big charset.

serhiy-storchaka · serhiy-storchaka · commit be80fc9a843e · 2013-10-24T22:02:58.000+03:00
diff --git a/Lib/sre_compile.py b/Lib/sre_compile.py
@@ -345,7 +345,7 @@ def _optimize_unicode(charset, fixup):
     else:
         code = 'I'
     # Convert block indices to byte array of 256 bytes
-    mapping = array.array('b', mapping).tobytes()
+    mapping = array.array('B', mapping).tobytes()
     # Convert byte array to word array
     mapping = array.array(code, mapping)
     assert mapping.itemsize == _sre.CODESIZE
diff --git a/Lib/test/test_re.py b/Lib/test/test_re.py
@@ -428,6 +428,9 @@ def test_bigcharset(self):
                                   "\u2222").group(1), "\u2222")
         self.assertEqual(re.match("([\u2222\u2223])",
                                   "\u2222", re.UNICODE).group(1), "\u2222")
+        r = '[%s]' % ''.join(map(chr, range(256, 2**16, 255)))
+        self.assertEqual(re.match(r,
+                                  "\uff01", re.UNICODE).group(), "\uff01")
 
     def test_big_codesize(self):
         # Issue #1160
diff --git a/Misc/NEWS b/Misc/NEWS
@@ -81,6 +81,8 @@ Core and Builtins
 Library
 -------
 
+- Issue #19327: Fixed the working of regular expressions with too big charset.
+
 - Issue #19350: Increasing the test coverage of macurl2path. Patch by Colin
   Williams.
 
diff --git a/Modules/_sre.c b/Modules/_sre.c
@@ -451,7 +451,7 @@ SRE_CHARSET(SRE_CODE* set, SRE_CODE ch)
             count = *(set++);
 
             if (sizeof(SRE_CODE) == 2) {
-                block = ((char*)set)[ch >> 8];
+                block = ((unsigned char*)set)[ch >> 8];
                 set += 128;
                 if (set[block*16 + ((ch & 255)>>4)] & (1 << (ch & 15)))
                     return ok;
@@ -461,7 +461,7 @@ SRE_CHARSET(SRE_CODE* set, SRE_CODE ch)
                 /* !(c & ~N) == (c < N+1) for any unsigned c, this avoids
                  * warnings when c's type supports only numbers < N+1 */
                 if (!(ch & ~65535))
-                    block = ((char*)set)[ch >> 8];
+                    block = ((unsigned char*)set)[ch >> 8];
                 else
                     block = -1;
                 set += 64;
