[Substrate-commits] SF.net SVN: substrate: [302] trunk/Foundation/NSUnicodeString.m

[<la...@us...>]

Revision: 302
          http://svn.sourceforge.net/substrate/?rev=302&view=rev
Author:   landonf
Date:     2006-09-01 22:51:51 -0700 (Fri, 01 Sep 2006)

Log Message:
-----------
Fix a heap stomper -- ucnv_toUChars() wants the buffer length in characters, not bytes.

Modified Paths:
--------------
    trunk/Foundation/NSUnicodeString.m

Modified: trunk/Foundation/NSUnicodeString.m
===================================================================
--- trunk/Foundation/NSUnicodeString.m	2006-09-02 01:33:04 UTC (rev 301)
+++ trunk/Foundation/NSUnicodeString.m	2006-09-02 05:51:51 UTC (rev 302)
@@ -99,54 +99,49 @@
 - (id) initWithBytes:(const void *)bytes length:(unsigned int)length encoding:(NSStringEncoding)encoding {
     UConverter *conv;
     UErrorCode err = U_ZERO_ERROR;
-    size_t allocSize;
     
     self = [super init];
     if (!self)
         return nil;
 
+    /* Default to same number of characters as source bytes */
+    _length = length;
+
     /* Acquire a converter, and make an educated guess as to the correct allocation */
     switch (encoding) {
         /* 7-bit ASCII encoding, using 8-bit characters. */
         case NSASCIIStringEncoding:
             conv = ucnv_open("US-ASCII", &err);
-            allocSize = length * sizeof(UChar);
             break;
 
         /* 8-bit EUC encoding for Japanese. */
         case NSJapaneseEUCStringEncoding:
             conv = ucnv_open("EUC-JP", &err);
-            allocSize = length * sizeof(UChar);
             break;
 
         /* 8-bit Unicode encoding (UTF-8). */
         case NSUTF8StringEncoding:
             conv = ucnv_open("UTF-8", &err);
-            allocSize = length * sizeof(UChar);
             break;
 
         /* 8-bit ISO Latin 1 Encoding. */
         case NSISOLatin1StringEncoding:
             conv = ucnv_open("latin1", &err);
-            allocSize = length * sizeof(UChar);
             break;
 
         /* 7-bit non-lossy Unicode encoding (UTF-7). */
         case NSNonLossyASCIIStringEncoding:
             conv = ucnv_open("UTF-7", &err);
-            allocSize = length * sizeof(UChar);
             break;
 
         /* 8-bit Japanese Shift_JIS Encoding. */
         case NSShiftJISStringEncoding:
             conv = ucnv_open("Shift_JIS", &err);
-            allocSize = length * sizeof(UChar);
             break;
 
         /* 8-bit ISO Latin 2 Encoding. */
         case NSISOLatin2StringEncoding:
             conv = ucnv_open("latin2", &err);
-            allocSize = length * sizeof(UChar);
             break;
 
         case NSProprietaryStringEncoding: /* Fall through to UTF-16 */
@@ -169,49 +164,41 @@
                     conv = ucnv_open("UTF-16", &err);
                     break;
             }
-            allocSize = length;
             break;
 
         /* Microsoft Windows codepage 1251 for Cryillic characters. */
         case NSWindowsCP1251StringEncoding:
             conv = ucnv_open("windows-1251", &err);
-            allocSize = length * sizeof(UChar);
             break;
 
         /* Microsoft Windows codepage 1252 (WinLatin1). */
         case NSWindowsCP1252StringEncoding:
             conv = ucnv_open("windows-1252", &err);
-            allocSize = length * sizeof(UChar);
             break;
 
         /* Microsoft Windows codepage 1253, for Greek characters. */
         case NSWindowsCP1253StringEncoding:
             conv = ucnv_open("windows-1253", &err);
-            allocSize = length * sizeof(UChar);
             break;
 
         /* Microsoft Windows codepage 1254, for Turkish characters. */
         case NSWindowsCP1254StringEncoding:
             conv = ucnv_open("windows-1254", &err);
-            allocSize = length * sizeof(UChar);
             break;
 
         /* Microsoft Windows codepage 1250 (WinLatin2). */
         case NSWindowsCP1250StringEncoding:
             conv = ucnv_open("windows-1250", &err);
-            allocSize = length * sizeof(UChar);
             break;
 
         /* ISO 2022 Japanese for e-mail. */
         case NSISO2022JPStringEncoding:
             conv = ucnv_open("ISO-2022-JP", &err);
-            allocSize = length * sizeof(UChar);
             break;
 
         /* Legacy Mac Roman Encoding. */
         case NSMacOSRomanStringEncoding:
             conv = ucnv_open("macintosh", &err);
-            allocSize = length * sizeof(UChar);
             break;
 
         /* Unhandled encodings */
@@ -241,25 +228,24 @@
     }
 
     /* Do the conversion */
-    _string = NSZoneMalloc(NULL, allocSize);
+    _string = NSZoneMalloc(NULL, _length * sizeof(UChar));
     while (1) {
-        size_t newAllocSize;
+        unsigned int newLength;
 
         err = U_ZERO_ERROR;
-        _length = ucnv_toUChars(conv, _string, allocSize, bytes, length, &err);
-        newAllocSize = _length * sizeof(UChar);
+        newLength = ucnv_toUChars(conv, _string, _length, bytes, length, &err);
 
         if (U_SUCCESS(err)) {
             /* Don't bother calling realloc for very small differences */
-            if (newAllocSize != allocSize && newAllocSize - allocSize >= 5)
-                _string = NSZoneRealloc(NULL, _string, newAllocSize);
-            allocSize = newAllocSize;
+            if (newLength != _length && newLength - _length >= 5)
+                _string = NSZoneRealloc(NULL, _string, newLength * sizeof(UChar));
+            _length = newLength;
             break;
         }
 
         if (err == U_BUFFER_OVERFLOW_ERROR) {
-            allocSize = newAllocSize;
-            _string = NSZoneRealloc(NULL, _string, allocSize);
+            _length = newLength;
+            _string = NSZoneRealloc(NULL, _string, _length * sizeof(UChar));
             continue;
         } else {
             [NSException raise: NSCharacterConversionException format: @"Failure in IBM ICU's ucnv_toUChars()"];


This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.