add map_offset member into re_pattern_buffer

diff --git a/src/regcomp.c b/src/regcomp.c
index fc4b3d50ffab61e321993423ad4f5e3124fa39b9..9b8150b020d8dc877018be44afa7cd130ab2378c 100644 (file)
--- a/src/regcomp.c
+++ b/src/regcomp.c
@@ -4555,28 +4555,48 @@ setup_tree(Node* node, regex_t* reg, int state, ScanEnv* env)
 #ifdef USE_SUNDAY_QUICK_SEARCH_ALGORITHM
 static int
 set_sunday_quick_search_skip_table(UChar* s, UChar* end, OnigEncoding enc,
-                                   UChar skip[])
+                                   UChar skip[], int* roffset)
 {
-  int i, len, emin;
+  int i, len, offset;
 
-  emin = ONIGENC_MBC_MINLEN(enc);
+  offset = 1;
+  if (ONIGENC_MBC_MINLEN(enc) > 1) {
+    UChar* p = s;
+    while (1) {
+      len = enclen(enc, p);
+      if (p + len >= end) {
+        UChar* q = p + (ONIGENC_MBC_MINLEN(enc) - 1);
+        while (q > p) {
+          if (*q != '\0') {
+            offset = q - p + 1;
+            break;
+          }
+          q--;
+        }
+        break;
+      }
+      p += len;
+    }
+  }
 
   len = (int )(end - s);
-  if (len + emin >= ONIG_CHAR_TABLE_SIZE)
+  if (len + offset >= ONIG_CHAR_TABLE_SIZE)
     return ONIGERR_PARSER_BUG;
 
-  for (i = 0; i < ONIG_CHAR_TABLE_SIZE; i++) skip[i] = (UChar )(len + emin);
+  *roffset = offset;
+
+  for (i = 0; i < ONIG_CHAR_TABLE_SIZE; i++) skip[i] = (UChar )(len + offset);
 
   for (i = 0; i < len; i++)
-    skip[s[i]] = len - i + (emin - 1);
+    skip[s[i]] = len - i + (offset - 1);
 
   return 0;
 }
 #else
 /* set skip map for Boyer-Moore-Horspool search */
 static int
-set_bmh_search_skip_table(UChar* s, UChar* end,
-                          OnigEncoding enc ARG_UNUSED, UChar skip[])
+set_bmh_search_skip_table(UChar* s, UChar* end, OnigEncoding enc ARG_UNUSED,
+                          UChar skip[])
 {
   int i, len;
 
@@ -5585,7 +5605,8 @@ set_optimize_exact(regex_t* reg, OptExact* e)
     if (e->len >= 3 || (e->len >= 2 && allow_reverse)) {
 #ifdef USE_SUNDAY_QUICK_SEARCH_ALGORITHM
       r = set_sunday_quick_search_skip_table(reg->exact, reg->exact_end,
-                                             reg->enc, reg->map);
+                                             reg->enc, reg->map,
+                                             &(reg->map_offset));
 #else
       r = set_bmh_search_skip_table(reg->exact, reg->exact_end,
                                     reg->enc, reg->map);
@@ -5706,6 +5727,7 @@ clear_optimize_info(regex_t* reg)
   reg->anchor_dmax   = 0;
   reg->sub_anchor    = 0;
   reg->exact_end     = (UChar* )NULL;
+  reg->map_offset    = 0;
   reg->threshold_len = 0;
   if (IS_NOT_NULL(reg->exact)) {
     xfree(reg->exact);

diff --git a/src/regexec.c b/src/regexec.c
index 2ab410efea5064e9c79b105ea47c250b0533239c..61ca4134c1e67441227d0099a24f7364179b5cb8 100644 (file)
--- a/src/regexec.c
+++ b/src/regexec.c
@@ -4123,7 +4123,7 @@ sunday_quick_search_step_forward(regex_t* reg,
   const UChar *s, *se, *t, *p, *end;
   const UChar *tail;
   int skip, tlen1;
-  int enc_minlen;
+  int map_offset;
 
 #ifdef ONIG_DEBUG_SEARCH
   fprintf(stderr,
@@ -4136,7 +4136,7 @@ sunday_quick_search_step_forward(regex_t* reg,
   if (end + tlen1 > text_end)
     end = text_end - tlen1;
 
-  enc_minlen = ONIGENC_MBC_MINLEN(reg->enc);
+  map_offset = reg->map_offset;
   s = text;
 
   while (s < end) {
@@ -4146,8 +4146,8 @@ sunday_quick_search_step_forward(regex_t* reg,
       if (t == target) return (UChar* )s;
       p--; t--;
     }
-    if (se + enc_minlen >= text_end) break;
-    skip = reg->map[*(se + enc_minlen)];
+    if (se + map_offset >= text_end) break;
+    skip = reg->map[*(se + map_offset)];
     t = s;
     do {
       s += enclen(reg->enc, s);
@@ -4164,13 +4164,13 @@ sunday_quick_search(regex_t* reg, const UChar* target, const UChar* target_end,
 {
   const UChar *s, *t, *p, *end;
   const UChar *tail;
-  int enc_minlen;
+  int map_offset;
 
   end = text_range + (target_end - target);
   if (end > text_end)
     end = text_end;
 
-  enc_minlen = ONIGENC_MBC_MINLEN(reg->enc);
+  map_offset = reg->map_offset;
   tail = target_end - 1;
   s = text + (tail - target);
 
@@ -4181,8 +4181,8 @@ sunday_quick_search(regex_t* reg, const UChar* target, const UChar* target_end,
       if (t == target) return (UChar* )p;
       p--; t--;
     }
-    if (s + enc_minlen >= text_end) break;
-    s += reg->map[*(s + enc_minlen)];
+    if (s + map_offset >= text_end) break;
+    s += reg->map[*(s + map_offset)];
   }
 
   return (UChar* )NULL;

diff --git a/src/regint.h b/src/regint.h
index 826e1652140c94b06b42c671d8e10f7d8e0d3c3c..844de3c76d8b9f3e33c8711e3c4c0c6c25102cc9 100644 (file)
--- a/src/regint.h
+++ b/src/regint.h
@@ -294,14 +294,15 @@ struct re_pattern_buffer {
   int            optimize;          /* optimize flag */
   int            threshold_len;     /* search str-length for apply optimize */
   int            anchor;            /* BEGIN_BUF, BEGIN_POS, (SEMI_)END_BUF */
-  OnigLen   anchor_dmin;       /* (SEMI_)END_BUF anchor distance */
-  OnigLen   anchor_dmax;       /* (SEMI_)END_BUF anchor distance */
+  OnigLen        anchor_dmin;       /* (SEMI_)END_BUF anchor distance */
+  OnigLen        anchor_dmax;       /* (SEMI_)END_BUF anchor distance */
   int            sub_anchor;        /* start-anchor for exact or map */
   unsigned char *exact;
   unsigned char *exact_end;
-  unsigned char  map[ONIG_CHAR_TABLE_SIZE]; /* used as BM skip or char-map */
-  OnigLen   dmin;                      /* min-distance of exact or map */
-  OnigLen   dmax;                      /* max-distance of exact or map */
+  unsigned char  map[ONIG_CHAR_TABLE_SIZE]; /* used as BMH skip or char-map */
+  int            map_offset;
+  OnigLen        dmin;                      /* min-distance of exact or map */
+  OnigLen        dmax;                      /* max-distance of exact or map */
 
   /* regex_t link chain */
   struct re_pattern_buffer* chain;  /* escape compile-conflict */