\d, \s and \w are now non Unicode class. [ruby-dev:39026]

* include/ruby/oniguruma.h
  (ONIGENC_CTYPE_SPECIAL_MASK): added.
  (ONIGENC_CTYPE_D): ditto.
  (ONIGENC_CTYPE_S): ditto.
  (ONIGENC_CTYPE_W): ditto.

* regparse.c: \d, \s and \w are now non Unicode class.
  [ruby-dev:39026]
  (fetch_token_in_cc): use ONIGENC_CTYPE_[DSW] for \d/\s/\w.
  (fetch_token): ditto.
  (add_ctype_to_cc): add routines for ONIGENC_CTYPE_[DSW].
  (parse_exp): ditto.

* test/ruby/test_regexp.rb (TestRegexp#test_char_class):
  add tests for above.

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@24544 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

4 files changed, 80 insertions, 12 deletions

diff --git a/ChangeLog b/ChangeLog
index add764aa3f..697e60f81c 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,21 @@
+Sun Aug 16 00:30:33 2009  NARUSE, Yui  <[email protected]>
+
+	* include/ruby/oniguruma.h
+	  (ONIGENC_CTYPE_SPECIAL_MASK): added.
+	  (ONIGENC_CTYPE_D): ditto.
+	  (ONIGENC_CTYPE_S): ditto.
+	  (ONIGENC_CTYPE_W): ditto.
+
+	* regparse.c: \d, \s and \w are now non Unicode class.
+	  [ruby-dev:39026]
+	  (fetch_token_in_cc): use ONIGENC_CTYPE_[DSW] for \d/\s/\w.
+	  (fetch_token): ditto.
+	  (add_ctype_to_cc): add routines for ONIGENC_CTYPE_[DSW].
+	  (parse_exp): ditto.
+
+	* test/ruby/test_regexp.rb (TestRegexp#test_char_class):
+	  add tests for above.
+
 Sat Aug 15 10:39:53 2009  Nobuyoshi Nakada  <[email protected]>
 
 	* parse.y (fname, string_dvar, sym, dsym, f_arglist): removed
diff --git a/include/ruby/oniguruma.h b/include/ruby/oniguruma.h
index 00b2999157..52f0c687ef 100644
--- a/include/ruby/oniguruma.h
+++ b/include/ruby/oniguruma.h
@@ -200,6 +200,14 @@ ONIG_EXTERN OnigEncodingType OnigEncodingASCII;
 #define ONIGENC_CTYPE_ALNUM    13  /* alpha || digit */
 #define ONIGENC_CTYPE_ASCII    14
 #define ONIGENC_MAX_STD_CTYPE  ONIGENC_CTYPE_ASCII
+#define ONIGENC_CTYPE_SPECIAL_MASK        128
+#define ONIGENC_CTYPE_S            /* [\t\n\v\f\r\s] */ \
+    ONIGENC_CTYPE_SPECIAL_MASK | ONIGENC_CTYPE_SPACE
+#define ONIGENC_CTYPE_D            /* [0-9] */ \
+    ONIGENC_CTYPE_SPECIAL_MASK | ONIGENC_CTYPE_DIGIT
+#define ONIGENC_CTYPE_W            /* [0-9A-Za-z_] */ \
+    ONIGENC_CTYPE_SPECIAL_MASK | ONIGENC_CTYPE_WORD
+#define ONIGENC_CTYPE_SPECIAL_P(ctype) ((ctype) & ONIGENC_CTYPE_SPECIAL_MASK)
 
 
 #define onig_enc_len(enc,p,e)                ONIGENC_MBC_ENC_LEN(enc, p, e)
diff --git a/regparse.c b/regparse.c
index 435e4de59d..876ad546ae 100644
--- a/regparse.c
+++ b/regparse.c
@@ -2974,32 +2974,32 @@ fetch_token_in_cc(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env)
     switch (c) {
     case 'w':
       tok->type = TK_CHAR_TYPE;
-      tok->u.prop.ctype = ONIGENC_CTYPE_WORD;
+      tok->u.prop.ctype = ONIGENC_CTYPE_W;
       tok->u.prop.not   = 0;
       break;
     case 'W':
       tok->type = TK_CHAR_TYPE;
-      tok->u.prop.ctype = ONIGENC_CTYPE_WORD;
+      tok->u.prop.ctype = ONIGENC_CTYPE_W;
       tok->u.prop.not   = 1;
       break;
     case 'd':
       tok->type = TK_CHAR_TYPE;
-      tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;
+      tok->u.prop.ctype = ONIGENC_CTYPE_D;
       tok->u.prop.not   = 0;
       break;
     case 'D':
       tok->type = TK_CHAR_TYPE;
-      tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;
+      tok->u.prop.ctype = ONIGENC_CTYPE_D;
       tok->u.prop.not   = 1;
       break;
     case 's':
       tok->type = TK_CHAR_TYPE;
-      tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;
+      tok->u.prop.ctype = ONIGENC_CTYPE_S;
       tok->u.prop.not   = 0;
       break;
     case 'S':
       tok->type = TK_CHAR_TYPE;
-      tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;
+      tok->u.prop.ctype = ONIGENC_CTYPE_S;
       tok->u.prop.not   = 1;
       break;
     case 'h':
@@ -3261,14 +3261,14 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env)
     case 'w':
       if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_W_WORD)) break;
       tok->type = TK_CHAR_TYPE;
-      tok->u.prop.ctype = ONIGENC_CTYPE_WORD;
+      tok->u.prop.ctype = ONIGENC_CTYPE_W;
       tok->u.prop.not   = 0;
       break;
 
     case 'W':
       if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_W_WORD)) break;
       tok->type = TK_CHAR_TYPE;
-      tok->u.prop.ctype = ONIGENC_CTYPE_WORD;
+      tok->u.prop.ctype = ONIGENC_CTYPE_W;
       tok->u.prop.not   = 1;
       break;
 
@@ -3301,28 +3301,28 @@ fetch_token(OnigToken* tok, UChar** src, UChar* end, ScanEnv* env)
     case 's':
       if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_S_WHITE_SPACE)) break;
       tok->type = TK_CHAR_TYPE;
-      tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;
+      tok->u.prop.ctype = ONIGENC_CTYPE_S;
       tok->u.prop.not   = 0;
       break;
 
     case 'S':
       if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_S_WHITE_SPACE)) break;
       tok->type = TK_CHAR_TYPE;
-      tok->u.prop.ctype = ONIGENC_CTYPE_SPACE;
+      tok->u.prop.ctype = ONIGENC_CTYPE_S;
       tok->u.prop.not   = 1;
       break;
 
     case 'd':
       if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_D_DIGIT)) break;
       tok->type = TK_CHAR_TYPE;
-      tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;
+      tok->u.prop.ctype = ONIGENC_CTYPE_D;
       tok->u.prop.not   = 0;
       break;
 
     case 'D':
       if (! IS_SYNTAX_OP(syn, ONIG_SYN_OP_ESC_D_DIGIT)) break;
       tok->type = TK_CHAR_TYPE;
-      tok->u.prop.ctype = ONIGENC_CTYPE_DIGIT;
+      tok->u.prop.ctype = ONIGENC_CTYPE_D;
       tok->u.prop.not   = 1;
       break;
 
@@ -3864,6 +3864,28 @@ add_ctype_to_cc(CClassNode* cc, int ctype, int not, ScanEnv* env)
   OnigCodePoint sb_out;
   OnigEncoding enc = env->enc;
 
+  switch (ctype) {
+  case ONIGENC_CTYPE_D:
+  case ONIGENC_CTYPE_S:
+  case ONIGENC_CTYPE_W:
+    ctype ^= ONIGENC_CTYPE_SPECIAL_MASK;
+    if (not != 0) {
+      for (c = 0; c < SINGLE_BYTE_SIZE; c++) {
+	if (! ONIGENC_IS_ASCII_CODE_CTYPE((OnigCodePoint )c, ctype))
+	  BITSET_SET_BIT_CHKDUP(cc->bs, c);
+      }
+      ADD_ALL_MULTI_BYTE_RANGE(enc, cc->mbuf);
+    }
+    else {
+      for (c = 0; c < SINGLE_BYTE_SIZE; c++) {
+	if (ONIGENC_IS_ASCII_CODE_CTYPE((OnigCodePoint )c, ctype))
+	  BITSET_SET_BIT_CHKDUP(cc->bs, c);
+      }
+    }
+    return 0;
+    break;
+  }
+
   r = ONIGENC_GET_CTYPE_CODE_RANGE(enc, ctype, &sb_out, &ranges);
   if (r == 0) {
     return add_ctype_to_cc_by_range(cc, ctype, not, env, sb_out, ranges);
@@ -5212,6 +5234,19 @@ parse_exp(Node** np, OnigToken* tok, int term,
   case TK_CHAR_TYPE:
     {
       switch (tok->u.prop.ctype) {
+      case ONIGENC_CTYPE_D:
+      case ONIGENC_CTYPE_S:
+      case ONIGENC_CTYPE_W:
+	{
+	    CClassNode* cc;
+	    *np = node_new_cclass();
+	    CHECK_NULL_RETURN_MEMERR(*np);
+	    cc = NCCLASS(*np);
+	    add_ctype_to_cc(cc, tok->u.prop.ctype, 0, env);
+	    if (tok->u.prop.not != 0) NCCLASS_SET_NOT(cc);
+	}
+	break;
+
       case ONIGENC_CTYPE_WORD:
 	*np = node_new_ctype(tok->u.prop.ctype, tok->u.prop.not);
 	CHECK_NULL_RETURN_MEMERR(*np);
diff --git a/test/ruby/test_regexp.rb b/test/ruby/test_regexp.rb
index 7d52012c00..30651c4c91 100644
--- a/test/ruby/test_regexp.rb
+++ b/test/ruby/test_regexp.rb
@@ -665,6 +665,13 @@ class TestRegexp < Test::Unit::TestCase
     check(/\A[[^b-c]&&[^e]&&a-f]\z/, %w(a d f), %w(b c e g 0))
     check(/\A[\n\r\t]\z/, ["\n", "\r", "\t"])
     failcheck('[9-1]')
+
+    assert_match(/\A\d+\z/, "0123456789")
+    assert_no_match(/\d/, "\uff10\uff11\uff12\uff13\uff14\uff15\uff16\uff17\uff18\uff19")
+    assert_match(/\A\w+\z/, "09azAZ_")
+    assert_no_match(/\w/, "\uff10\uff19\uff41\uff5a\uff21\uff3a")
+    assert_match(/\A\s+\z/, "\r\n\v\f\r\s")
+    assert_no_match(/\s/, "\u0085")
   end
 
   def test_posix_bracket