Fix for bug 81753, do not read past the end of unicode strings

source/i18n/rematch.cpp

Index: source/i18n/rematch.cpp
===================================================================
--- source/i18n/rematch.cpp	(revision 98343)
+++ source/i18n/rematch.cpp	(working copy)
@@ -5598,6 +5598,7 @@
                     const UChar *foldChars = NULL;
                     int32_t foldOffset, foldLength;
                     UChar32 c;
+                    UBool c_is_valid = FALSE;
                     
                     #ifdef REGEX_SMART_BACKTRACKING
                     int32_t originalInputIdx = fp->fInputIdx;
@@ -5607,10 +5608,16 @@
                     foldOffset = foldLength = 0;
 
                     while (patternChars < patternEnd && success) {
-                        if(foldOffset < foldLength) {
+                        if (fp->fInputIdx >= fActiveLimit) {
+                          // don't read past end of string
+                        } else if(foldOffset < foldLength) {
                             U16_NEXT_UNSAFE(foldChars, foldOffset, c);
+                            c_is_valid = TRUE;
                         } else {

[jungshik at Google, 2011/09/14 18:35:29]

How about rewritting the above if - elseif - else as following?

if (fp->fInputIdx < fActiveLimit)
    if (foldOffset < foldLength) 
          .....
     else 
          .....
}

[Brad Chen, 2011/09/14 19:10:06]

Done. Note this makes the patch quite a bit longer due to
indentation changes.

On 2011/09/14 18:35:29, Jungshik Shin wrote:

+                            // test pre-condition of U16_NEXT: i < length
+                            U_ASSERT(fp->fInputIdx < fActiveLimit);
                             U16_NEXT(inputBuf, fp->fInputIdx, fActiveLimit, c);
+                            c_is_valid = TRUE;
                             foldLength = ucase_toFullFolding(csp, c, &foldChars, U_FOLD_CASE_DEFAULT);
                             if(foldLength >= 0) {
                                 if(foldLength <= UCASE_MAX_STRING_LENGTH) {   // !!!: Does not correctly handle chars that fold to 0-length strings
@@ -5623,7 +5630,7 @@
                             }
                         }
                         
-                        if (fp->fInputIdx <= fActiveLimit) {
+                        if (fp->fInputIdx <= fActiveLimit && c_is_valid) {
                             if (U_IS_BMP(c)) {
                                 success = (*patternChars == c);
                                 patternChars += 1;
@@ -6070,4 +6077,3 @@
 U_NAMESPACE_END
 
 #endif  // !UCONFIG_NO_REGULAR_EXPRESSIONS
-