autoupdate
[gnulib.git] / lib / regcomp.c
index 398df66..0fdc3bc 100644 (file)
@@ -1,5 +1,5 @@
 /* Extended regular expression matching and search library.
-   Copyright (C) 2002-2012 Free Software Foundation, Inc.
+   Copyright (C) 2002-2013 Free Software Foundation, Inc.
    This file is part of the GNU C Library.
    Contributed by Isamu Hasegawa <isamu@yamato.ibm.com>.
 
@@ -94,20 +94,20 @@ static reg_errcode_t build_charclass (RE_TRANSLATE_TYPE trans,
                                      bitset_t sbcset,
                                      re_charset_t *mbcset,
                                      Idx *char_class_alloc,
-                                     const unsigned char *class_name,
+                                     const char *class_name,
                                      reg_syntax_t syntax);
 #else  /* not RE_ENABLE_I18N */
 static reg_errcode_t build_equiv_class (bitset_t sbcset,
                                        const unsigned char *name);
 static reg_errcode_t build_charclass (RE_TRANSLATE_TYPE trans,
                                      bitset_t sbcset,
-                                     const unsigned char *class_name,
+                                     const char *class_name,
                                      reg_syntax_t syntax);
 #endif /* not RE_ENABLE_I18N */
 static bin_tree_t *build_charclass_op (re_dfa_t *dfa,
                                       RE_TRANSLATE_TYPE trans,
-                                      const unsigned char *class_name,
-                                      const unsigned char *extra,
+                                      const char *class_name,
+                                      const char *extra,
                                       bool non_match, reg_errcode_t *err);
 static bin_tree_t *create_tree (re_dfa_t *dfa,
                                bin_tree_t *left, bin_tree_t *right,
@@ -2422,8 +2422,8 @@ parse_expression (re_string_t *regexp, regex_t *preg, re_token_t *token,
     case OP_WORD:
     case OP_NOTWORD:
       tree = build_charclass_op (dfa, regexp->trans,
-                                (const unsigned char *) "alnum",
-                                (const unsigned char *) "_",
+                                "alnum",
+                                "_",
                                 token->type == OP_NOTWORD, err);
       if (BE (*err != REG_NOERROR && tree == NULL, 0))
        return NULL;
@@ -2431,8 +2431,8 @@ parse_expression (re_string_t *regexp, regex_t *preg, re_token_t *token,
     case OP_SPACE:
     case OP_NOTSPACE:
       tree = build_charclass_op (dfa, regexp->trans,
-                                (const unsigned char *) "space",
-                                (const unsigned char *) "",
+                                "space",
+                                "",
                                 token->type == OP_NOTSPACE, err);
       if (BE (*err != REG_NOERROR && tree == NULL, 0))
        return NULL;
@@ -2712,7 +2712,6 @@ build_range_exp (const reg_syntax_t syntax,
     wchar_t wc;
     wint_t start_wc;
     wint_t end_wc;
-    wchar_t cmp_buf[6] = {L'\0', L'\0', L'\0', L'\0', L'\0', L'\0'};
 
     start_ch = ((start_elem->type == SB_CHAR) ? start_elem->opr.ch
                : ((start_elem->type == COLL_SYM) ? start_elem->opr.name[0]
@@ -2726,11 +2725,7 @@ build_range_exp (const reg_syntax_t syntax,
              ? __btowc (end_ch) : end_elem->opr.wch);
     if (start_wc == WEOF || end_wc == WEOF)
       return REG_ECOLLATE;
-    cmp_buf[0] = start_wc;
-    cmp_buf[4] = end_wc;
-
-    if (BE ((syntax & RE_NO_EMPTY_RANGES)
-            && wcscoll (cmp_buf, cmp_buf + 4) > 0, 0))
+    else if (BE ((syntax & RE_NO_EMPTY_RANGES) && start_wc > end_wc, 0))
       return REG_ERANGE;
 
     /* Got valid collation sequence values, add them as a new entry.
@@ -2771,9 +2766,7 @@ build_range_exp (const reg_syntax_t syntax,
     /* Build the table for single byte characters.  */
     for (wc = 0; wc < SBC_MAX; ++wc)
       {
-       cmp_buf[2] = wc;
-       if (wcscoll (cmp_buf, cmp_buf + 2) <= 0
-           && wcscoll (cmp_buf + 2, cmp_buf + 4) <= 0)
+       if (start_wc <= wc && wc <= end_wc)
          bitset_set (sbcset, wc);
       }
   }
@@ -2970,6 +2963,7 @@ parse_bracket_exp (re_string_t *regexp, re_dfa_t *dfa, re_token_t *token,
              0))
        return REG_ERANGE;
 
+      /* FIXME: Implement rational ranges here, too.  */
       start_collseq = lookup_collation_sequence_value (start_elem);
       end_collseq = lookup_collation_sequence_value (end_elem);
       /* Check start/end collation sequence values.  */
@@ -3577,14 +3571,14 @@ static reg_errcode_t
 #ifdef RE_ENABLE_I18N
 build_charclass (RE_TRANSLATE_TYPE trans, bitset_t sbcset,
                 re_charset_t *mbcset, Idx *char_class_alloc,
-                const unsigned char *class_name, reg_syntax_t syntax)
+                const char *class_name, reg_syntax_t syntax)
 #else /* not RE_ENABLE_I18N */
 build_charclass (RE_TRANSLATE_TYPE trans, bitset_t sbcset,
-                const unsigned char *class_name, reg_syntax_t syntax)
+                const char *class_name, reg_syntax_t syntax)
 #endif /* not RE_ENABLE_I18N */
 {
   int i;
-  const char *name = (const char *) class_name;
+  const char *name = class_name;
 
   /* In case of REG_ICASE "upper" and "lower" match the both of
      upper and lower cases.  */
@@ -3658,8 +3652,8 @@ build_charclass (RE_TRANSLATE_TYPE trans, bitset_t sbcset,
 
 static bin_tree_t *
 build_charclass_op (re_dfa_t *dfa, RE_TRANSLATE_TYPE trans,
-                   const unsigned char *class_name,
-                   const unsigned char *extra, bool non_match,
+                   const char *class_name,
+                   const char *extra, bool non_match,
                    reg_errcode_t *err)
 {
   re_bitset_ptr_t sbcset;