ICU-62123.0.1.tar.gz

[apple/icu.git] / icuSources / i18n / rematch.cpp

diff --git a/icuSources/i18n/rematch.cpp b/icuSources/i18n/rematch.cpp
index 3350820b2afe551980a21c8719d4c2ad1b156a8f..38e5fb29503e0fd23912dd69b45fa63263b43ee8 100644 (file)
--- a/icuSources/i18n/rematch.cpp
+++ b/icuSources/i18n/rematch.cpp
@@ -1,3 +1,5 @@
+// © 2016 and later: Unicode, Inc. and others.
+// License & terms of use: http://www.unicode.org/copyright.html
 /*
 **************************************************************************
 *   Copyright (C) 2002-2016 International Business Machines Corporation
@@ -436,7 +438,7 @@ RegexMatcher &RegexMatcher::appendReplacement(UText *dest,
                         status = U_REGEX_INVALID_CAPTURE_GROUP_NAME;
                     }
                 }
-                        
+
             } else if (u_isdigit(nextChar)) {
                 // $n    Scan for a capture group number
                 int32_t numCaptureGroups = fPattern->fGroupMap->size();
@@ -457,7 +459,7 @@ RegexMatcher &RegexMatcher::appendReplacement(UText *dest,
                         break;
                     }
                     (void)UTEXT_NEXT32(replacement);
-                    groupNum=groupNum*10 + nextDigitVal; 
+                    groupNum=groupNum*10 + nextDigitVal;
                     ++numDigits;
                 }
             } else {
@@ -1074,6 +1076,15 @@ UBool RegexMatcher::findUsingChunk(UErrorCode &status) {
             if (fMatch) {
                 return TRUE;
             }
+            // In bug 31063104 which has a zero-length text buffer we get here with
+            // inputBuf=NULL, startPos=fActiveLimit=0 (and fMatch F) which violates the
+            // requirement for U16_FWD_1 (utf16.h) that startPos < fActiveLimit. Having
+            // inputBuf=NULL (chunkContexts NULL) is probably due to an error in the
+            // CFStringUText functions. Nevertheless, to be defensive, add test below.
+            if (startPos >= testLen) {
+                fHitEnd = TRUE;
+                return FALSE;
+            }
             U16_FWD_1(inputBuf, startPos, fActiveLimit);
         }
 
@@ -2185,7 +2196,7 @@ int32_t  RegexMatcher::split(UText *input,
                     break;
                 }
                 i++;
-                dest[i] = utext_extract_replace(fInputText, dest[i], 
+                dest[i] = utext_extract_replace(fInputText, dest[i],
                                                start64(groupNum, status), end64(groupNum, status), &status);
             }
 
@@ -2198,7 +2209,7 @@ int32_t  RegexMatcher::split(UText *input,
                     if (dest[i] == NULL) {
                         dest[i] = utext_openUChars(NULL, NULL, 0, &status);
                     } else {
-                        static UChar emptyString[] = {(UChar)0};
+                        static const UChar emptyString[] = {(UChar)0};
                         utext_replace(dest[i], 0, utext_nativeLength(dest[i]), emptyString, 0, &status);
                     }
                 }
@@ -3564,7 +3575,14 @@ GC_Done:
                         }
                     }
                     fp = StateSave(fp, fp->fPatIdx, status);
+                } else {
+                    // Increment time-out counter. (StateSave() does it if count >= minCount)
+                    fTickCounter--;
+                    if (fTickCounter <= 0) {
+                        IncrementTime(status);    // Re-initializes fTickCounter
+                    }
                 }
+
                 fp->fPatIdx = opValue + 4;    // Loop back.
             }
             break;
@@ -3621,6 +3639,11 @@ GC_Done:
                     // We haven't met the minimum number of matches yet.
                     //   Loop back for another one.
                     fp->fPatIdx = opValue + 4;    // Loop back.
+                    // Increment time-out counter. (StateSave() does it if count >= minCount)
+                    fTickCounter--;
+                    if (fTickCounter <= 0) {
+                        IncrementTime(status);    // Re-initializes fTickCounter
+                    }
                 } else {
                     // We do have the minimum number of matches.
 
@@ -5097,6 +5120,12 @@ GC_Done:
                         }
                     }
                     fp = StateSave(fp, fp->fPatIdx, status);
+                } else {
+                    // Increment time-out counter. (StateSave() does it if count >= minCount)
+                    fTickCounter--;
+                    if (fTickCounter <= 0) {
+                        IncrementTime(status);    // Re-initializes fTickCounter
+                    }
                 }
                 fp->fPatIdx = opValue + 4;    // Loop back.
             }
@@ -5154,6 +5183,10 @@ GC_Done:
                     // We haven't met the minimum number of matches yet.
                     //   Loop back for another one.
                     fp->fPatIdx = opValue + 4;    // Loop back.
+                    fTickCounter--;
+                    if (fTickCounter <= 0) {
+                        IncrementTime(status);    // Re-initializes fTickCounter
+                    }
                 } else {
                     // We do have the minimum number of matches.
 
@@ -5445,7 +5478,7 @@ GC_Done:
                 if (lbStartIdx < 0) {
                     // First time through loop.
                     lbStartIdx = fp->fInputIdx - minML;
-                    if (lbStartIdx > 0) {
+                    if (lbStartIdx > 0 && lbStartIdx < fInputLength) {
                         U16_SET_CP_START(inputBuf, 0, lbStartIdx);
                     }
                 } else {
@@ -5522,7 +5555,7 @@ GC_Done:
                 if (lbStartIdx < 0) {
                     // First time through loop.
                     lbStartIdx = fp->fInputIdx - minML;
-                    if (lbStartIdx > 0) {
+                    if (lbStartIdx > 0 && lbStartIdx < fInputLength) {
                         U16_SET_CP_START(inputBuf, 0, lbStartIdx);
                     }
                 } else {
@@ -5794,3 +5827,4 @@ UOBJECT_DEFINE_RTTI_IMPLEMENTATION(RegexMatcher)
 U_NAMESPACE_END
 
 #endif  // !UCONFIG_NO_REGULAR_EXPRESSIONS
+