Changeset 10818 in webkit for trunk/JavaScriptCore/kjs/string_object.cpp

Timestamp:

Oct 11, 2005, 1:43:49 PM (20 years ago)

Author:

ggaren

Message:

• Implemented caching of match state inside the global RegExp object (lastParen, leftContext, rightContext, lastMatch, input).

exec(), test(), match(), search(), and replace() now dipatch regular
expression matching through the RegExp object's performMatch function,
to facilitate caching. This replaces registerRegexp and
setSubPatterns.

• Implemented the special '$' aliases (e.g. RegExp.input aliases to RegExp.$_).

• Moved support for backreferences into the new static hash table used for other special RegExp properties. Truncated backreferences at $9 to match IE, FF, and the "What's New in Netscape 1.2?" doc. (String.replace still supports double-digit backreferences.)

• Tweaked RegExp.prototype.exec to handle ginormous values in lastIndex.

Fixes 11 -- count em, 11 -- JavaScriptCore tests.

Reviewed by NOBODY (OOPS!).

• JavaScriptCore.xcodeproj/project.pbxproj: Added regexp_object.lut.h
• kjs/create_hash_table: Tweaked to allow for more exotic characters.

We now rely on the compiler to catch illegal
identifiers.

• kjs/regexp.cpp: (KJS::RegExp::RegExp):
• kjs/regexp_object.cpp: (RegExpProtoFuncImp::callAsFunction): (RegExpObjectImp::RegExpObjectImp): (RegExpObjectImp::performMatch): (RegExpObjectImp::arrayOfMatches): (RegExpObjectImp::backrefGetter): (RegExpObjectImp::getLastMatch): (RegExpObjectImp::getLastParen): (RegExpObjectImp::getLeftContext): (RegExpObjectImp::getRightContext): (RegExpObjectImp::getOwnPropertySlot): (RegExpObjectImp::getValueProperty): (RegExpObjectImp::put): (RegExpObjectImp::putValueProperty):
• kjs/regexp_object.h: (KJS::RegExpObjectImp::):
• kjs/string_object.cpp: (substituteBackreferences): (replace): (StringProtoFuncImp::callAsFunction):

File:

• trunk/JavaScriptCore/kjs/string_object.cpp (modified) (10 diffs)

trunk/JavaScriptCore/kjs/string_object.cpp

@@ -233 +233 @@
 }
 
-static inline UString substituteBackreferences(const UString &replacement, const UString &source, int **ovector, RegExp *reg)
+static inline UString substituteBackreferences(const UString &replacement, const UString &source, int *ovector, RegExp *reg)
 {
   UString substitutedReplacement = replacement;
@@ -247 +247 @@
     unsigned backrefIndex = substitutedReplacement.substr(i+1,1).toUInt32(&converted, false /* tolerate empty string */);
     if (converted && backrefIndex <= (unsigned)reg->subPatterns()) {
-      int backrefStart = (*ovector)[2*backrefIndex];
-      int backrefLength = (*ovector)[2*backrefIndex+1] - backrefStart;
+      int backrefStart = ovector[2*backrefIndex];
+      int backrefLength = ovector[2*backrefIndex+1] - backrefStart;
       substitutedReplacement = substitutedReplacement.substr(0,i)
         + source.substr(backrefStart, backrefLength)
@@ -289 +289 @@
     // This is either a loop (if global is set) or a one-way (if not).
     do {
-      int **ovector = regExpObj->registerRegexp( reg, source );
-      UString matchString = reg->match(source, startPosition, &matchIndex, ovector);
-      regExpObj->setSubPatterns(reg->subPatterns());
+      int *ovector;
+      UString matchString = regExpObj->performMatch(reg, source, startPosition, &matchIndex, &ovector);
       if (matchIndex == -1)
         break;
@@ -299 +298 @@
 
       if (replacementFunction) {
-          int completeMatchStart = (*ovector)[0];
+          int completeMatchStart = ovector[0];
           List args;
 
@@ -305 +304 @@
           
           for (unsigned i = 0; i < reg->subPatterns(); i++) {
-              int matchStart = (*ovector)[(i + 1) * 2];
-              int matchLen = (*ovector)[(i + 1) * 2 + 1] - matchStart;
+              int matchStart = ovector[(i + 1) * 2];
+              int matchLen = ovector[(i + 1) * 2 + 1] - matchStart;
               
               args.append(jsString(source.substr(matchStart, matchLen)));
@@ -455 +454 @@
     RegExp *reg, *tmpReg = 0;
     RegExpImp *imp = 0;
-    if (a0->isObject() && a0->getObject()->inherits(&RegExpImp::info))
-    {
+    if (a0->isObject() && a0->getObject()->inherits(&RegExpImp::info)) {
       imp = static_cast<RegExpImp *>(a0);
       reg = imp->regExp();
-    }
-    else
-    { /*
+    } else { 
+      /*
        *  ECMA 15.5.4.12 String.prototype.search (regexp)
        *  If regexp is not an object whose [[Class]] property is "RegExp", it is
@@ -469 +466 @@
     }
     RegExpObjectImp* regExpObj = static_cast<RegExpObjectImp*>(exec->lexicalInterpreter()->builtinRegExp());
-    int **ovector = regExpObj->registerRegexp(reg, u);
-    UString mstr = reg->match(u, -1, &pos, ovector);
+    UString mstr = regExpObj->performMatch(reg, u, 0, &pos);
     if (id == Search) {
       result = Number(pos);
@@ -480 +476 @@
           result = Null();
         } else {
-          regExpObj->setSubPatterns(reg->subPatterns());
           result = regExpObj->arrayOfMatches(exec,mstr);
         }
@@ -494 +489 @@
           lastIndex = pos;
           pos += mstr.isEmpty() ? 1 : mstr.size();
-          delete [] *ovector;
-          mstr = reg->match(u, pos, &pos, ovector);
+          mstr = regExpObj->performMatch(reg, u, pos, &pos);
         }
         if (imp)
@@ -515 +509 @@
     result = replace(exec, s, a0, a1);
     break;
-  case Slice: // http://developer.netscape.com/docs/manuals/js/client/jsref/string.htm#1194366
+  case Slice:
     {
         // The arg processing is very much like ArrayProtoFunc::Slice