add new operator \J (skip search)

kkos · Jun 7, 2024 · 8860d24 · 8860d24
1 parent f2f9c69
commit 8860d24
Show file tree

Hide file tree

Showing 3 changed files with 93 additions and 5 deletions.
diff --git a/src/regexec.c b/src/regexec.c
@@ -2,7 +2,7 @@
   regexec.c -  Oniguruma (regular expression library)
 **********************************************************************/
 /*-
- * Copyright (c) 2002-2022  K.Kosako
+ * Copyright (c) 2002-2024  K.Kosako
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -177,6 +177,9 @@ typedef struct {
 #ifdef USE_CALL
   unsigned long  subexp_call_in_search_counter;
 #endif
+#ifdef USE_SKIP_SEARCH
+  UChar* skip_search;
+#endif
 } MatchArg;
 
 
@@ -1261,6 +1264,7 @@ struct OnigCalloutArgsStruct {
 #endif
 
 #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
+#ifdef USE_SKIP_SEARCH
 #define MATCH_ARG_INIT(msa, reg, arg_option, arg_region, arg_start, mpv) do { \
   (msa).stack_p  = (void* )0;\
   (msa).options  = (arg_option)|(reg)->options;\
@@ -1272,6 +1276,35 @@ struct OnigCalloutArgsStruct {
   (msa).mp = mpv;\
   (msa).best_len = ONIG_MISMATCH;\
   (msa).ptr_num  = PTR_NUM_SIZE(reg);\
+  (msa).skip_search = (UChar* )(arg_start);\
+} while(0)
+#else
+#define MATCH_ARG_INIT(msa, reg, arg_option, arg_region, arg_start, mpv) do { \
+  (msa).stack_p  = (void* )0;\
+  (msa).options  = (arg_option)|(reg)->options;\
+  (msa).region   = (arg_region);\
+  (msa).start    = (arg_start);\
+  (msa).match_stack_limit  = (mpv)->match_stack_limit;\
+  RETRY_IN_MATCH_ARG_INIT(msa,mpv)\
+  SUBEXP_CALL_IN_MATCH_ARG_INIT(msa,mpv)\
+  (msa).mp = mpv;\
+  (msa).best_len = ONIG_MISMATCH;\
+  (msa).ptr_num  = PTR_NUM_SIZE(reg);\
+} while(0)
+#endif
+#else
+#ifdef USE_SKIP_SEARCH
+#define MATCH_ARG_INIT(msa, reg, arg_option, arg_region, arg_start, mpv) do { \
+  (msa).stack_p  = (void* )0;\
+  (msa).options  = (arg_option)|(reg)->options;\
+  (msa).region   = (arg_region);\
+  (msa).start    = (arg_start);\
+  (msa).match_stack_limit  = (mpv)->match_stack_limit;\
+  RETRY_IN_MATCH_ARG_INIT(msa,mpv)\
+  SUBEXP_CALL_IN_MATCH_ARG_INIT(msa,mpv)\
+  (msa).mp = mpv;\
+  (msa).ptr_num  = PTR_NUM_SIZE(reg);\
+  (msa).skip_search = (UChar* )(arg_start);\
 } while(0)
 #else
 #define MATCH_ARG_INIT(msa, reg, arg_option, arg_region, arg_start, mpv) do { \
@@ -1286,6 +1319,7 @@ struct OnigCalloutArgsStruct {
   (msa).ptr_num  = PTR_NUM_SIZE(reg);\
 } while(0)
 #endif
+#endif
 
 #define MATCH_ARG_FREE(msa)  if ((msa).stack_p) xfree((msa).stack_p)
 
@@ -4335,6 +4369,11 @@ match_at(regex_t* reg, const UChar* str, const UChar* end,
         case UPDATE_VAR_RIGHT_RANGE_INIT:
           INIT_RIGHT_RANGE;
           break;
+#ifdef USE_SKIP_SEARCH
+        case UPDATE_VAR_SKIP_SEARCH:
+          if (s > msa->skip_search) msa->skip_search = s;
+          break;
+#endif
         }
       }
       INC_OP;
@@ -5629,6 +5668,9 @@ search_in_range(regex_t* reg, const UChar* str, const UChar* end,
           while (s <= high) {
             MATCH_AND_RETURN_CHECK(data_range);
             s += enclen(reg->enc, s);
+#ifdef USE_SKIP_SEARCH
+            if (s < msa.skip_search) s = msa.skip_search;
+#endif
           }
         } while (s < range);
         goto mismatch;
@@ -5646,10 +5688,18 @@ search_in_range(regex_t* reg, const UChar* str, const UChar* end,
             prev = s;
             s += enclen(reg->enc, s);
 
-            while (!ONIGENC_IS_MBC_NEWLINE(reg->enc, prev, end) && s < range) {
-              prev = s;
-              s += enclen(reg->enc, s);
+#ifdef USE_SKIP_SEARCH
+            if (s < msa.skip_search) s = msa.skip_search;
+            else {
+#endif
+              while (!ONIGENC_IS_MBC_NEWLINE(reg->enc, prev, end) &&
+                     s < range) {
+                prev = s;
+                s += enclen(reg->enc, s);
+              }
+#ifdef USE_SKIP_SEARCH
             }
+#endif
           }
           goto mismatch;
         }
@@ -5660,6 +5710,13 @@ search_in_range(regex_t* reg, const UChar* str, const UChar* end,
       MATCH_AND_RETURN_CHECK(data_range);
       if (s >= range) break;
       s += enclen(reg->enc, s);
+
+#ifdef USE_SKIP_SEARCH
+      if (s < msa.skip_search) {
+        s = msa.skip_search;
+        if (s > range) break;
+      }
+#endif
     }
   }
   else {  /* backward search */

diff --git a/src/regint.h b/src/regint.h
@@ -4,7 +4,7 @@
   regint.h -  Oniguruma (regular expression library)
 **********************************************************************/
 /*-
- * Copyright (c) 2002-2023  K.Kosako
+ * Copyright (c) 2002-2024  K.Kosako
  * All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
@@ -62,6 +62,7 @@
 #define USE_REGSET
 #define USE_CALL
 #define USE_CALLOUT
+#define USE_SKIP_SEARCH
 #define USE_BACKREF_WITH_LEVEL        /* \k<name+n>, \k<name-n> */
 #define USE_WHOLE_OPTIONS
 #define USE_RIGID_CHECK_CAPTURES_IN_EMPTY_REPEAT        /* /(?:()|())*\2/ */
@@ -584,6 +585,9 @@ enum UpdateVarType {
   UPDATE_VAR_RIGHT_RANGE_FROM_S_STACK = 3,
   UPDATE_VAR_RIGHT_RANGE_TO_S         = 4,
   UPDATE_VAR_RIGHT_RANGE_INIT         = 5,
+#ifdef USE_SKIP_SEARCH
+  UPDATE_VAR_SKIP_SEARCH              = 6,
+#endif
 };
 
 enum CheckPositionType {

diff --git a/src/regparse.c b/src/regparse.c
@@ -2781,6 +2781,16 @@ node_new_keep(Node** node, ParseEnv* env)
   return ONIG_NORMAL;
 }
 
+#ifdef USE_SKIP_SEARCH
+static int
+node_new_skip_search(Node** node, ParseEnv* env)
+{
+  int r;
+  r = node_new_update_var_gimmick(node, UPDATE_VAR_SKIP_SEARCH, 0, env);
+  return r;
+}
+#endif
+
 #ifdef USE_CALLOUT
 
 extern void
@@ -4526,6 +4536,9 @@ enum TokenSyms {
   TK_QUOTE_OPEN,
   TK_CHAR_PROPERTY,    /* \p{...}, \P{...} */
   TK_KEEP,             /* \K */
+#ifdef USE_SKIP_SEARCH
+  TK_SKIP_SEARCH,      /* \J */
+#endif
   TK_GENERAL_NEWLINE,  /* \R */
   TK_NO_NEWLINE,       /* \N */
   TK_TRUE_ANYCHAR,     /* \O */
@@ -5743,6 +5756,13 @@ fetch_token(PToken* tok, UChar** src, UChar* end, ParseEnv* env)
       tok->type = TK_KEEP;
       break;
 
+#ifdef USE_SKIP_SEARCH
+    case 'J':
+      if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_CAPITAL_K_KEEP)) break;
+      tok->type = TK_SKIP_SEARCH;
+      break;
+#endif
+
     case 'R':
       if (! IS_SYNTAX_OP2(syn, ONIG_SYN_OP2_ESC_CAPITAL_R_GENERAL_NEWLINE)) break;
       tok->type = TK_GENERAL_NEWLINE;
@@ -9093,6 +9113,13 @@ prs_exp(Node** np, PToken* tok, int term, UChar** src, UChar* end,
     if (r < 0) return r;
     break;
 
+#ifdef USE_SKIP_SEARCH
+  case TK_SKIP_SEARCH:
+    r = node_new_skip_search(np, env);
+    if (r < 0) return r;
+    break;
+#endif
+
   case TK_GENERAL_NEWLINE:
     r = node_new_general_newline(np, env);
     if (r < 0) return r;