* oniguruma.h: merge Oniguruma 4.0.0 [ruby-dev:28290]

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@9885 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

1 files changed, 89 insertions, 72 deletions

diff --git a/regparse.c b/regparse.c
index 837ea35b30..76bce7885b 100644
--- a/regparse.c
+++ b/regparse.c
@@ -58,7 +58,21 @@ OnigSyntaxType OnigSyntaxRuby = {
 
 OnigSyntaxType*  OnigDefaultSyntax = ONIG_SYNTAX_RUBY;
 
-extern void onig_null_warn(const char* s, ...) { }
+extern void onig_null_warn(const char* s) { }
+
+#ifdef RUBY_PLATFORM
+extern void
+onig_rb_warn(const char* s)
+{
+  rb_warn(s);
+}
+
+extern void
+onig_rb_warning(const char* s)
+{
+  rb_warning(s);
+}
+#endif
 
 #ifdef DEFAULT_WARN_FUNCTION
 static OnigWarnFunc onig_warn = (OnigWarnFunc )DEFAULT_WARN_FUNCTION;
@@ -1050,12 +1064,12 @@ onig_node_free(Node* node)
 
 #ifdef USE_RECYCLE_NODE
 extern int
-onig_free_node_list(void)
+onig_free_node_list()
 {
   FreeNode* n;
 
   THREAD_ATOMIC_START;
-  while (FreeNodeList) {
+  while (IS_NOT_NULL(FreeNodeList)) {
     n = FreeNodeList;
     FreeNodeList = FreeNodeList->next;
     xfree(n);
@@ -1066,18 +1080,19 @@ onig_free_node_list(void)
 #endif
 
 static Node*
-node_new(void)
+node_new()
 {
   Node* node;
 
 #ifdef USE_RECYCLE_NODE
+  THREAD_ATOMIC_START;
   if (IS_NOT_NULL(FreeNodeList)) {
-    THREAD_ATOMIC_START;
     node = (Node* )FreeNodeList;
     FreeNodeList = FreeNodeList->next;
     THREAD_ATOMIC_END;
     return node;
   }
+  THREAD_ATOMIC_END;
 #endif
 
   node = (Node* )xmalloc(sizeof(Node));
@@ -1094,7 +1109,7 @@ initialize_cclass(CClassNode* cc)
 }
 
 static Node*
-node_new_cclass(void)
+node_new_cclass()
 {
   Node* node = node_new();
   CHECK_NULL_RETURN(node);
@@ -1106,7 +1121,7 @@ node_new_cclass(void)
 
 static Node*
 node_new_cclass_by_codepoint_range(int not,
-                   OnigCodePoint sbr[], OnigCodePoint mbr[])
+                   const OnigCodePoint sbr[], const OnigCodePoint mbr[])
 {
   CClassNode* cc;
   int n, i, j;
@@ -1163,7 +1178,7 @@ node_new_ctype(int type)
 }
 
 static Node*
-node_new_anychar(void)
+node_new_anychar()
 {
   Node* node = node_new();
   CHECK_NULL_RETURN(node);
@@ -1434,7 +1449,7 @@ node_new_str_raw(UChar* s, UChar* end)
 }
 
 static Node*
-node_new_empty(void)
+node_new_empty()
 {
   return node_new_str(NULL, NULL);
 }
@@ -2358,15 +2373,17 @@ fetch_escaped_value(UChar** src, UChar* end, ScanEnv* env)
     control:
       if (PEND) return ONIGERR_END_PATTERN_AT_CONTROL;
       PFETCH(c);
-      if (c == MC_ESC(enc)) {
-	v = fetch_escaped_value(&p, end, env);
-	if (v < 0) return v;
-        c = (OnigCodePoint )(v & 0x9f);
-      }
-      else if (c == '?')
+      if (c == '?') {
 	c = 0177;
-      else
+      }
+      else {
+        if (c == MC_ESC(enc)) {
+          v = fetch_escaped_value(&p, end, env);
+          if (v < 0) return v;
+          c = (OnigCodePoint )v;
+        }
 	c &= 0x9f;
+      }
       break;
     }
     /* fall through */
@@ -2512,11 +2529,11 @@ CC_ESC_WARN(ScanEnv* env, UChar *c)
 
   if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_WARN_CC_OP_NOT_ESCAPED) &&
       IS_SYNTAX_BV(env->syntax, ONIG_SYN_BACKSLASH_ESCAPE_IN_CC)) {
-    char buf[WARN_BUFSIZE];
+    UChar buf[WARN_BUFSIZE];
     onig_snprintf_with_pattern(buf, WARN_BUFSIZE, env->enc,
 		env->pattern, env->pattern_end,
-		"character class has '%s' without escape", c);
-    (*onig_warn)(buf);
+                (UChar* )"character class has '%s' without escape", c);
+    (*onig_warn)((char* )buf);
   }
 }
 
@@ -2526,11 +2543,11 @@ CCEND_ESC_WARN(ScanEnv* env, UChar* c)
   if (onig_warn == onig_null_warn) return ;
 
   if (IS_SYNTAX_BV((env)->syntax, ONIG_SYN_WARN_CC_OP_NOT_ESCAPED)) {
-    char buf[WARN_BUFSIZE];
+    UChar buf[WARN_BUFSIZE];
     onig_snprintf_with_pattern(buf, WARN_BUFSIZE, (env)->enc,
 		(env)->pattern, (env)->pattern_end,
-		"regular expression has '%s' without escape", c);
-    (*onig_warn)(buf);
+		(UChar* )"regular expression has '%s' without escape", c);
+    (*onig_warn)((char* )buf);
   }
 }
 
@@ -2794,7 +2811,7 @@ fetch_token_in_cc(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env)
 	tok->type = TK_CC_CC_OPEN;
       }
       else {
-	CC_ESC_WARN(env, "[");
+	CC_ESC_WARN(env, (UChar* )"[");
       }
     }
   }
@@ -2833,7 +2850,7 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env)
   tok->backp = p;
 
   PFETCH(c);
-  if (c == MC_ESC(enc)) {
+  if (IS_MC_ESC_CODE(c, enc, syn)) {
     if (PEND) return ONIGERR_END_PATTERN_AT_ESCAPE;
 
     tok->backp = p;
@@ -3365,7 +3382,7 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env)
 
     case ']':
       if (*src > env->pattern)   /* /].../ is allowed. */
-	CCEND_ESC_WARN(env, "]");
+	CCEND_ESC_WARN(env, (UChar* )"]");
       break;
 
     case '#':
@@ -3400,7 +3417,7 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env)
 
 static int
 add_ctype_to_cc_by_range(CClassNode* cc, int ctype, int not, OnigEncoding enc,
-                         OnigCodePoint sbr[], OnigCodePoint mbr[])
+                         const OnigCodePoint sbr[], const OnigCodePoint mbr[])
 {
   int i, r;
   OnigCodePoint j;
@@ -3464,7 +3481,7 @@ static int
 add_ctype_to_cc(CClassNode* cc, int ctype, int not, ScanEnv* env)
 {
   int c, r;
-  OnigCodePoint *sbr, *mbr;
+  const OnigCodePoint *sbr, *mbr;
   OnigEncoding enc = env->enc;
 
   r = ONIGENC_GET_CTYPE_CODE_RANGE(enc, ctype, &sbr, &mbr);
@@ -3602,19 +3619,19 @@ parse_posix_bracket(CClassNode* cc, UChar** src, UChar* end, ScanEnv* env)
 #define POSIX_BRACKET_NAME_MAX_LEN         6
 
   static PosixBracketEntryType PBS[] = {
-    { "alnum",  ONIGENC_CTYPE_ALNUM,  5 },
-    { "alpha",  ONIGENC_CTYPE_ALPHA,  5 },
-    { "blank",  ONIGENC_CTYPE_BLANK,  5 },
-    { "cntrl",  ONIGENC_CTYPE_CNTRL,  5 },
-    { "digit",  ONIGENC_CTYPE_DIGIT,  5 },
-    { "graph",  ONIGENC_CTYPE_GRAPH,  5 },
-    { "lower",  ONIGENC_CTYPE_LOWER,  5 },
-    { "print",  ONIGENC_CTYPE_PRINT,  5 },
-    { "punct",  ONIGENC_CTYPE_PUNCT,  5 },
-    { "space",  ONIGENC_CTYPE_SPACE,  5 },
-    { "upper",  ONIGENC_CTYPE_UPPER,  5 },
-    { "xdigit", ONIGENC_CTYPE_XDIGIT, 6 },
-    { "ascii",  ONIGENC_CTYPE_ASCII,  5 }, /* I don't know origin. Perl? */
+    { (UChar* )"alnum",  ONIGENC_CTYPE_ALNUM,  5 },
+    { (UChar* )"alpha",  ONIGENC_CTYPE_ALPHA,  5 },
+    { (UChar* )"blank",  ONIGENC_CTYPE_BLANK,  5 },
+    { (UChar* )"cntrl",  ONIGENC_CTYPE_CNTRL,  5 },
+    { (UChar* )"digit",  ONIGENC_CTYPE_DIGIT,  5 },
+    { (UChar* )"graph",  ONIGENC_CTYPE_GRAPH,  5 },
+    { (UChar* )"lower",  ONIGENC_CTYPE_LOWER,  5 },
+    { (UChar* )"print",  ONIGENC_CTYPE_PRINT,  5 },
+    { (UChar* )"punct",  ONIGENC_CTYPE_PUNCT,  5 },
+    { (UChar* )"space",  ONIGENC_CTYPE_SPACE,  5 },
+    { (UChar* )"upper",  ONIGENC_CTYPE_UPPER,  5 },
+    { (UChar* )"xdigit", ONIGENC_CTYPE_XDIGIT, 6 },
+    { (UChar* )"ascii",  ONIGENC_CTYPE_ASCII,  5 },
     { (UChar* )NULL, -1, 0 }
   };
 
@@ -3638,7 +3655,7 @@ parse_posix_bracket(CClassNode* cc, UChar** src, UChar* end, ScanEnv* env)
   for (pb = PBS; IS_NOT_NULL(pb->name); pb++) {
     if (onigenc_with_ascii_strncmp(enc, p, end, pb->name, pb->len) == 0) {
       p = (UChar* )onigenc_step(enc, p, end, pb->len);
-      if (onigenc_with_ascii_strncmp(enc, p, end, ":]", 2) != 0)
+      if (onigenc_with_ascii_strncmp(enc, p, end, (UChar* )":]", 2) != 0)
 	return ONIGERR_INVALID_POSIX_BRACKET_TYPE;
 
       r = add_ctype_to_cc(cc, pb->ctype, not, env);
@@ -3673,19 +3690,19 @@ static int
 property_name_to_ctype(UChar* p, UChar* end, OnigEncoding enc)
 {
   static PosixBracketEntryType PBS[] = {
-    { "Alnum",  ONIGENC_CTYPE_ALNUM,  5 },
-    { "Alpha",  ONIGENC_CTYPE_ALPHA,  5 },
-    { "Blank",  ONIGENC_CTYPE_BLANK,  5 },
-    { "Cntrl",  ONIGENC_CTYPE_CNTRL,  5 },
-    { "Digit",  ONIGENC_CTYPE_DIGIT,  5 },
-    { "Graph",  ONIGENC_CTYPE_GRAPH,  5 },
-    { "Lower",  ONIGENC_CTYPE_LOWER,  5 },
-    { "Print",  ONIGENC_CTYPE_PRINT,  5 },
-    { "Punct",  ONIGENC_CTYPE_PUNCT,  5 },
-    { "Space",  ONIGENC_CTYPE_SPACE,  5 },
-    { "Upper",  ONIGENC_CTYPE_UPPER,  5 },
-    { "XDigit", ONIGENC_CTYPE_XDIGIT, 6 },
-    { "ASCII",  ONIGENC_CTYPE_ASCII,  5 },
+    { (UChar* )"Alnum",  ONIGENC_CTYPE_ALNUM,  5 },
+    { (UChar* )"Alpha",  ONIGENC_CTYPE_ALPHA,  5 },
+    { (UChar* )"Blank",  ONIGENC_CTYPE_BLANK,  5 },
+    { (UChar* )"Cntrl",  ONIGENC_CTYPE_CNTRL,  5 },
+    { (UChar* )"Digit",  ONIGENC_CTYPE_DIGIT,  5 },
+    { (UChar* )"Graph",  ONIGENC_CTYPE_GRAPH,  5 },
+    { (UChar* )"Lower",  ONIGENC_CTYPE_LOWER,  5 },
+    { (UChar* )"Print",  ONIGENC_CTYPE_PRINT,  5 },
+    { (UChar* )"Punct",  ONIGENC_CTYPE_PUNCT,  5 },
+    { (UChar* )"Space",  ONIGENC_CTYPE_SPACE,  5 },
+    { (UChar* )"Upper",  ONIGENC_CTYPE_UPPER,  5 },
+    { (UChar* )"XDigit", ONIGENC_CTYPE_XDIGIT, 6 },
+    { (UChar* )"ASCII",  ONIGENC_CTYPE_ASCII,  5 },
     { (UChar* )NULL, -1, 0 }
   };
 
@@ -3935,7 +3952,7 @@ parse_char_class(Node** np, OnigToken* tok, UChar** src, UChar* end,
                            *src, env->pattern_end, 1, env->enc))
       return ONIGERR_EMPTY_CHAR_CLASS;
 
-    CC_ESC_WARN(env, "]");
+    CC_ESC_WARN(env, (UChar* )"]");
     r = tok->type = TK_CHAR;  /* allow []...] */
   }
 
@@ -4038,7 +4055,7 @@ parse_char_class(Node** np, OnigToken* tok, UChar** src, UChar* end,
       r = parse_posix_bracket(cc, &p, end, env);
       if (r < 0) goto err;
       if (r == 1) {  /* is not POSIX bracket */
-	CC_ESC_WARN(env, "[");
+	CC_ESC_WARN(env, (UChar* )"[");
 	p = tok->backp;
 	v = (OnigCodePoint )tok->u.c;
 	in_israw = 0;
@@ -4084,7 +4101,7 @@ parse_char_class(Node** np, OnigToken* tok, UChar** src, UChar* end,
 	  goto val_entry;
 	}
 	else if (r == TK_CC_AND) {
-	  CC_ESC_WARN(env, "-");
+	  CC_ESC_WARN(env, (UChar* )"-");
 	  goto range_end_val;
 	}
 	state = CCS_RANGE;
@@ -4099,12 +4116,12 @@ parse_char_class(Node** np, OnigToken* tok, UChar** src, UChar* end,
 	fetched = 1;
 	/* [--x] or [a&&-x] is warned. */
 	if (r == TK_CC_RANGE || and_start != 0)
-	  CC_ESC_WARN(env, "-");
+	  CC_ESC_WARN(env, (UChar* )"-");
 
 	goto val_entry;
       }
       else if (state == CCS_RANGE) {
-	CC_ESC_WARN(env, "-");
+	CC_ESC_WARN(env, (UChar* )"-");
 	goto sb_char;  /* [!--x] is allowed */
       }
       else { /* CCS_COMPLETE */
@@ -4113,12 +4130,12 @@ parse_char_class(Node** np, OnigToken* tok, UChar** src, UChar* end,
 	fetched = 1;
 	if (r == TK_CC_CLOSE) goto range_end_val; /* allow [a-b-] */
 	else if (r == TK_CC_AND) {
-	  CC_ESC_WARN(env, "-");
+	  CC_ESC_WARN(env, (UChar* )"-");
 	  goto range_end_val;
 	}
 	
 	if (IS_SYNTAX_BV(env->syntax, ONIG_SYN_ALLOW_DOUBLE_RANGE_OP_IN_CC)) {
-	  CC_ESC_WARN(env, "-");
+	  CC_ESC_WARN(env, (UChar* )"-");
 	  goto sb_char;   /* [0-9-a] is allowed as [0-9\-a] */
 	}
 	r = ONIGERR_UNMATCHED_RANGE_SPECIFIER_IN_CHAR_CLASS;
@@ -4495,7 +4512,7 @@ set_qualifier(Node* qnode, Node* target, int group, ScanEnv* env)
       if (qn->by_number == 0 && qnt->by_number == 0 &&
 	  IS_SYNTAX_BV(env->syntax, ONIG_SYN_WARN_REDUNDANT_NESTED_REPEAT)) {
         int nestq_num, targetq_num;
-        char buf[WARN_BUFSIZE];
+        UChar buf[WARN_BUFSIZE];
 
         nestq_num   = popular_qualifier_num(qn);
         targetq_num = popular_qualifier_num(qnt);
@@ -4507,9 +4524,9 @@ set_qualifier(Node* qnode, Node* target, int group, ScanEnv* env)
         case RQ_DEL:
           if (onig_verb_warn != onig_null_warn) {
             onig_snprintf_with_pattern(buf, WARN_BUFSIZE, env->enc,
-                                       env->pattern, env->pattern_end,
-                                       "redundant nested repeat operator");
-            (*onig_verb_warn)(buf);
+                                 env->pattern, env->pattern_end,
+                                 (UChar* )"redundant nested repeat operator");
+            (*onig_verb_warn)((char* )buf);
           }
           goto warn_exit;
           break;
@@ -4518,10 +4535,10 @@ set_qualifier(Node* qnode, Node* target, int group, ScanEnv* env)
           if (onig_verb_warn != onig_null_warn) {
             onig_snprintf_with_pattern(buf, WARN_BUFSIZE, env->enc,
                                        env->pattern, env->pattern_end,
-            "nested repeat operator %s and %s was replaced with '%s'",
+            (UChar* )"nested repeat operator %s and %s was replaced with '%s'",
             PopularQStr[targetq_num], PopularQStr[nestq_num],
             ReduceQStr[ReduceTypeTable[targetq_num][nestq_num]]);
-            (*onig_verb_warn)(buf);
+            (*onig_verb_warn)((char* )buf);
           }
           goto warn_exit;
           break;
@@ -4553,8 +4570,8 @@ make_compound_alt_node_from_cc(OnigAmbigType ambig_flag, OnigEncoding enc,
   int r, i, j, k, clen, len, ncode, n;
   UChar buf[ONIGENC_CODE_TO_MBC_MAXLEN];
   Node **ptail, *snode = NULL_NODE;
-  OnigCompAmbigCodes* ccs;
-  OnigCompAmbigCodeItem* ci;
+  const OnigCompAmbigCodes* ccs;
+  const OnigCompAmbigCodeItem* ci;
   OnigAmbigType amb;
 
   n = 0;
@@ -4662,7 +4679,7 @@ i_free_shared_class(type_cclass_key* key, Node* node, void* arg)
 }
 
 extern int
-onig_free_shared_cclass_table(void)
+onig_free_shared_cclass_table()
 {
   if (IS_NOT_NULL(OnigTypeCClassTable)) {
     onig_st_foreach(OnigTypeCClassTable, i_free_shared_class, 0);
@@ -4819,7 +4836,7 @@ parse_exp(Node** np, OnigToken* tok, int term,
 	  int ctype, not;
 
 #ifdef USE_SHARED_CCLASS_TABLE
-          OnigCodePoint *sbr, *mbr;
+          const OnigCodePoint *sbr, *mbr;
 
 	  ctype = parse_ctype_to_enc_ctype(tok->u.subtype, &not);
           r = ONIGENC_GET_CTYPE_CODE_RANGE(env->enc, ctype, &sbr, &mbr);
@@ -4901,7 +4918,7 @@ parse_exp(Node** np, OnigToken* tok, int term,
 
       if (IS_IGNORECASE(env->option)) {
         int i, n, in_cc;
-        OnigPairAmbigCodes* ccs;
+        const OnigPairAmbigCodes* ccs;
         BitSetRef bs = cc->bs;
         OnigAmbigType amb;