Changeset 28139 in webkit for trunk/JavaScriptCore/pcre/pcre_exec.cpp

Timestamp:

Nov 29, 2007, 3:05:07 AM (18 years ago)

Author:

[email protected]

Message:

2007-11-24 Eric Seidel <​[email protected]>

Reviewed by Maciej.

Remove register keyword and more cleanup

• pcre/pcre_compile.cpp: (find_fixedlength): (compile_branch): (is_anchored): (is_startline): (find_firstassertedchar): (calculateCompiledPatternLengthAndFlags): (jsRegExpCompile):
• pcre/pcre_exec.cpp: (MatchStack::canUseStackBufferForNextFrame): (MatchStack::allocateNextFrame): (MatchStack::pushNewFrame): (MatchStack::frameIsStackAllocated): (MatchStack::popCurrentFrame): (MatchStack::unrollAnyHeapAllocatedFrames): (getUTF8CharAndIncrementLength): (match): (jsRegExpExecute):
• pcre/pcre_internal.h: (PUT2INC): (isLeadingSurrogate): (isTrailingSurrogate): (decodeSurrogatePair): (getChar):
• pcre/pcre_ord2utf8.cpp: (_pcre_ord2utf8):
• pcre/pcre_xclass.cpp: (getUTF8CharAndAdvancePointer): (_pcre_xclass):

File:

• trunk/JavaScriptCore/pcre/pcre_exec.cpp (modified) (33 diffs)

trunk/JavaScriptCore/pcre/pcre_exec.cpp

@@ -81 +81 @@
 #endif
 
-typedef struct matchframe {
+struct MatchFrame {
   ReturnLocation returnLocation;
 
-  struct matchframe* prevframe;
+  struct MatchFrame* previousFrame;
 
   /* Function arguments that may change */
 
-  const pcre_uchar *eptr;
+  const pcre_uchar* eptr;
   const uschar* ecode;
   int offset_top;
-  eptrblock *eptrb;
+  eptrblock* eptrb;
 
   /* Function local variables */
 
-  const uschar *data;
-  const uschar *next;
-  const pcre_uchar *pp;
-  const uschar *prev;
-  const pcre_uchar *saved_eptr;
+  const uschar* data;
+  const uschar* next;
+  const pcre_uchar* pp;
+  const uschar* prev;
+  const pcre_uchar* saved_eptr;
 
   int repeat_othercase;
@@ -113 +113 @@
 
   eptrblock newptrb;
-} matchframe;
+};
 
 /* Structure for passing "static" information around between the functions
@@ -197 +197 @@
 
 static BOOL
-match_ref(int offset, register USPTR eptr, int length, match_data *md)
+match_ref(int offset, USPTR eptr, int length, match_data *md)
 {
 USPTR p = md->start_subject + md->offset_vector[offset];
@@ -327 +327 @@
      any calls to pcre_stack_malloc, yet the amount of stack used for the array is
      modest enough that we don't run out of stack. */
-    matchframe frames[16];
-    matchframe* framesEnd;
-    matchframe* currentFrame;
+    MatchFrame frames[16];
+    MatchFrame* framesEnd;
+    MatchFrame* currentFrame;
     
     inline bool canUseStackBufferForNextFrame()
@@ -336 +336 @@
     }
     
-    inline matchframe* allocateNextFrame()
+    inline MatchFrame* allocateNextFrame()
     {
         if (canUseStackBufferForNextFrame())
             return currentFrame + 1;
-        return new matchframe;
+        return new MatchFrame;
     }
     
     inline void pushNewFrame(const uschar* ecode, eptrblock* eptrb, ReturnLocation returnLocation)
     {
-        matchframe* newframe = allocateNextFrame();
-        newframe->prevframe = currentFrame;
+        MatchFrame* newframe = allocateNextFrame();
+        newframe->previousFrame = currentFrame;
 
         newframe->eptr = currentFrame->eptr;
@@ -357 +357 @@
     }
     
-    inline bool frameIsStackAllocated(matchframe* frame)
+    inline bool frameIsStackAllocated(MatchFrame* frame)
     {
         return (frame >= frames && frame < framesEnd);
@@ -364 +364 @@
     inline void popCurrentFrame()
     {
-        matchframe* oldFrame = currentFrame;
-        currentFrame = currentFrame->prevframe;
+        MatchFrame* oldFrame = currentFrame;
+        currentFrame = currentFrame->previousFrame;
         if (!frameIsStackAllocated(oldFrame))
             delete oldFrame;
@@ -373 +373 @@
     {
         while (!frameIsStackAllocated(currentFrame)) {
-            matchframe* oldFrame = currentFrame;
-            currentFrame = currentFrame->prevframe;
+            MatchFrame* oldFrame = currentFrame;
+            currentFrame = currentFrame->previousFrame;
             delete oldFrame;
         }
@@ -386 +386 @@
 }
 
+/* Get the next UTF-8 character, not advancing the pointer, incrementing length
+ if there are extra bytes. This is called when we know we are in UTF-8 mode. */
+
+static inline void getUTF8CharAndIncrementLength(int& c, const uschar* eptr, int& len)
+{
+    c = *eptr;
+    if ((c & 0xc0) == 0xc0) {
+        int gcaa = _pcre_utf8_table4[c & 0x3f];  /* Number of additional bytes */
+        int gcss = 6 * gcaa;
+        c = (c & _pcre_utf8_table3[gcaa]) << gcss;
+        for (int gcii = 1; gcii <= gcaa; gcii++) {
+            gcss -= 6;
+            c |= (eptr[gcii] & 0x3f) << gcss;
+        }
+        len += gcaa;
+    }
+}
+
 static int match(USPTR eptr, const uschar* ecode, int offset_top, match_data* md)
 {
-    register int is_match = false;
-    register int i;
-    register int c;
+    int is_match = false;
+    int i;
+    int c;
     
     unsigned rdepth = 0;
@@ -759 +777 @@
                 else {
                     const pcre_uchar *lastptr = stack.currentFrame->eptr - 1;
-                    while(ISMIDCHAR(*lastptr))
+                    while(isTrailingSurrogate(*lastptr))
                         lastptr--;
-                    GETCHAR(c, lastptr);
+                    getChar(c, lastptr);
                     prev_is_word = c < 128 && (md->ctypes[c] & ctype_word) != 0;
                 }
@@ -767 +785 @@
                     cur_is_word = false;
                 else {
-                    GETCHAR(c, stack.currentFrame->eptr);
+                    getChar(c, stack.currentFrame->eptr);
                     cur_is_word = c < 128 && (md->ctypes[c] & ctype_word) != 0;
                 }
@@ -784 +802 @@
                 if (stack.currentFrame->eptr++ >= md->end_subject)
                     RRETURN_NO_MATCH;
-                while (stack.currentFrame->eptr < md->end_subject && ISMIDCHAR(*stack.currentFrame->eptr))
+                while (stack.currentFrame->eptr < md->end_subject && isTrailingSurrogate(*stack.currentFrame->eptr))
                     stack.currentFrame->eptr++;
                 stack.currentFrame->ecode++;
@@ -1172 +1190 @@
                 stack.currentFrame->length = 1;
                 stack.currentFrame->ecode++;
-                GETUTF8CHARLEN(stack.currentFrame->fc, stack.currentFrame->ecode, stack.currentFrame->length);
+                getUTF8CharAndIncrementLength(stack.currentFrame->fc, stack.currentFrame->ecode, stack.currentFrame->length);
             {
                 int dc;
@@ -1181 +1199 @@
                 case 1:
                     dc = *stack.currentFrame->eptr++;
-                    if (IS_LEADING_SURROGATE(dc))
+                    if (isLeadingSurrogate(dc))
                         RRETURN_NO_MATCH;
                     break;
@@ -1197 +1215 @@
                 stack.currentFrame->length = 1;
                 stack.currentFrame->ecode++;
-                GETUTF8CHARLEN(stack.currentFrame->fc, stack.currentFrame->ecode, stack.currentFrame->length);
+                getUTF8CharAndIncrementLength(stack.currentFrame->fc, stack.currentFrame->ecode, stack.currentFrame->length);
                 
                 if (md->end_subject - stack.currentFrame->eptr == 0)
@@ -1206 +1224 @@
                 if (md->end_subject - stack.currentFrame->eptr == 1) {
                     dc = *stack.currentFrame->eptr++;
-                    if (IS_LEADING_SURROGATE(dc))
+                    if (isLeadingSurrogate(dc))
                         RRETURN_NO_MATCH;
                 } else
@@ -1280 +1298 @@
                 
                 stack.currentFrame->length = 1;
-                GETUTF8CHARLEN(stack.currentFrame->fc, stack.currentFrame->ecode, stack.currentFrame->length);
+                getUTF8CharAndIncrementLength(stack.currentFrame->fc, stack.currentFrame->ecode, stack.currentFrame->length);
                 if (min * (stack.currentFrame->fc > 0xFFFF ? 2 : 1) > md->end_subject - stack.currentFrame->eptr)
                     RRETURN_NO_MATCH;
@@ -1333 +1351 @@
                     for (i = 1; i <= min; i++) {
                         int nc;
-                        GETCHAR(nc, stack.currentFrame->eptr);
+                        getChar(nc, stack.currentFrame->eptr);
                         if (nc != stack.currentFrame->fc)
                             RRETURN_NO_MATCH;
@@ -1350 +1368 @@
                             if (stack.currentFrame->fi >= stack.currentFrame->max || stack.currentFrame->eptr >= md->end_subject)
                                 RRETURN;
-                            GETCHAR(nc, stack.currentFrame->eptr);
+                            getChar(nc, stack.currentFrame->eptr);
                             if (*stack.currentFrame->eptr != stack.currentFrame->fc)
                                 RRETURN;
@@ -1362 +1380 @@
                             if (stack.currentFrame->eptr > md->end_subject - 2)
                                 break;
-                            GETCHAR(nc, stack.currentFrame->eptr);
+                            getChar(nc, stack.currentFrame->eptr);
                             if (*stack.currentFrame->eptr != stack.currentFrame->fc)
                                 break;
@@ -1455 +1473 @@
                     
                     {
-                        register int d;
+                        int d;
                         for (i = 1; i <= min; i++) {
                             GETCHARINC(d, stack.currentFrame->eptr);
@@ -1469 +1487 @@
                     
                     if (minimize) {
-                        register int d;
+                        int d;
                         for (stack.currentFrame->fi = min;; stack.currentFrame->fi++) {
                             RMATCH(38, stack.currentFrame->ecode, stack.currentFrame->eptrb, 0);
@@ -1489 +1507 @@
                         
                         {
-                            register int d;
+                            int d;
                             for (i = min; i < stack.currentFrame->max; i++) {
                                 int len = 1;
@@ -1520 +1538 @@
                 else {
                     {
-                        register int d;
+                        int d;
                         for (i = 1; i <= min; i++) {
                             GETCHARINC(d, stack.currentFrame->eptr);
@@ -1532 +1550 @@
                     
                     if (minimize) {
-                        register int d;
+                        int d;
                         for (stack.currentFrame->fi = min;; stack.currentFrame->fi++) {
                             RMATCH(42, stack.currentFrame->ecode, stack.currentFrame->eptrb, 0);
@@ -1550 +1568 @@
                         
                         {
-                            register int d;
+                            int d;
                             for (i = min; i < stack.currentFrame->max; i++) {
                                 int len = 1;
@@ -1630 +1648 @@
                                     RRETURN_NO_MATCH;
                                 ++stack.currentFrame->eptr;
-                                while (stack.currentFrame->eptr < md->end_subject && ISMIDCHAR(*stack.currentFrame->eptr))
+                                while (stack.currentFrame->eptr < md->end_subject && isTrailingSurrogate(*stack.currentFrame->eptr))
                                     stack.currentFrame->eptr++;
                             }
@@ -1658 +1676 @@
                                     (*stack.currentFrame->eptr < 128 && (md->ctypes[*stack.currentFrame->eptr] & ctype_space) != 0))
                                     RRETURN_NO_MATCH;
-                                while (++stack.currentFrame->eptr < md->end_subject && ISMIDCHAR(*stack.currentFrame->eptr)) { }
+                                while (++stack.currentFrame->eptr < md->end_subject && isTrailingSurrogate(*stack.currentFrame->eptr)) { }
                             }
                             break;
@@ -1676 +1694 @@
                                     (*stack.currentFrame->eptr < 128 && (md->ctypes[*stack.currentFrame->eptr] & ctype_word) != 0))
                                     RRETURN_NO_MATCH;
-                                while (++stack.currentFrame->eptr < md->end_subject && ISMIDCHAR(*stack.currentFrame->eptr)) { }
+                                while (++stack.currentFrame->eptr < md->end_subject && isTrailingSurrogate(*stack.currentFrame->eptr)) { }
                             }
                             break;
@@ -2132 +2150 @@
     USPTR start_match = (USPTR)subject + start_offset;
     USPTR req_byte_ptr = start_match - 1;
+    bool startline = re->options & PCRE_STARTLINE;
     
     do {
@@ -2138 +2157 @@
         /* Reset the maximum number of extractions we might see. */
         
-        if (match_block.offset_vector != NULL) {
+        if (match_block.offset_vector) {
             int* iptr = match_block.offset_vector;
             int* iend = iptr + resetcount;
@@ -2171 +2190 @@
         /* Or to just after \n for a multiline match if possible */
         
-        else if (re->options & PCRE_STARTLINE) {
+        else if (startline) {
             if (start_match > match_block.start_subject + start_offset) {
                 while (start_match < end_subject && !isNewline(start_match[-1]))
@@ -2265 +2284 @@
         if (returnCode == MATCH_NOMATCH) {
             start_match++;
-            while(start_match < end_subject && ISMIDCHAR(*start_match))
+            while(start_match < end_subject && isTrailingSurrogate(*start_match))
                 start_match++;
             continue;