From: Paul Eggert <eggert@cs.ucla.edu>
Date: Sun, 30 Dec 2012 07:31:08 +0000 (-0800)
Subject: regex: implement rational ranges
X-Git-Tag: v0.1~292
X-Git-Url: http://erislabs.org.uk/gitweb/?a=commitdiff_plain;h=6410c7a6ea65cceb588aadb497715279d4ec5daa;p=gnulib.git

regex: implement rational ranges

Reported by Aharon Robbins in
<http://sourceware.org/ml/libc-alpha/2012-12/msg00456.html>.
* lib/regcomp.c (build_range_exp) [!_LIBC]:
* lib/regexec.c (check_node_accept_bytes) [!_LIBC]:
Implement rational ranges.
---

diff --git a/ChangeLog b/ChangeLog
index 0824b0835..8c61adb68 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,12 @@
 2012-12-29  Paul Eggert  <eggert@cs.ucla.edu>
 
+	regex: implement rational ranges
+	Reported by Aharon Robbins in
+	<http://sourceware.org/ml/libc-alpha/2012-12/msg00456.html>.
+	* lib/regcomp.c (build_range_exp) [!_LIBC]:
+	* lib/regexec.c (check_node_accept_bytes) [!_LIBC]:
+	Implement rational ranges.
+
 	regex: avoid redefining __wctype
 	Reported by Aharon Robbins in
 	<http://sourceware.org/ml/libc-alpha/2012-12/msg00456.html>.
diff --git a/lib/regcomp.c b/lib/regcomp.c
index 398df66ce..5cc2c67b3 100644
--- a/lib/regcomp.c
+++ b/lib/regcomp.c
@@ -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.  */
diff --git a/lib/regexec.c b/lib/regexec.c
index 4e4b295b7..22e8dd67c 100644
--- a/lib/regexec.c
+++ b/lib/regexec.c
@@ -3936,6 +3936,7 @@ check_node_accept_bytes (const re_dfa_t *dfa, Idx node_idx,
 		in_collseq = find_collation_sequence_value (pin, elem_len);
 	    }
 	  /* match with range expression?  */
+	  /* FIXME: Implement rational ranges here, too.  */
 	  for (i = 0; i < cset->nranges; ++i)
 	    if (cset->range_starts[i] <= in_collseq
 		&& in_collseq <= cset->range_ends[i])
@@ -3987,18 +3988,9 @@ check_node_accept_bytes (const re_dfa_t *dfa, Idx node_idx,
 # endif /* _LIBC */
 	{
 	  /* match with range expression?  */
-#if __GNUC__ >= 2 && ! (__STDC_VERSION__ < 199901L && defined __STRICT_ANSI__)
-	  wchar_t cmp_buf[] = {L'\0', L'\0', wc, L'\0', L'\0', L'\0'};
-#else
-	  wchar_t cmp_buf[] = {L'\0', L'\0', L'\0', L'\0', L'\0', L'\0'};
-	  cmp_buf[2] = wc;
-#endif
 	  for (i = 0; i < cset->nranges; ++i)
 	    {
-	      cmp_buf[0] = cset->range_starts[i];
-	      cmp_buf[4] = cset->range_ends[i];
-	      if (wcscoll (cmp_buf, cmp_buf + 2) <= 0
-		  && wcscoll (cmp_buf + 2, cmp_buf + 4) <= 0)
+	      if (cset->range_starts[i] <= wc && wc <= cset->range_ends[i])
 		{
 		  match_len = char_len;
 		  goto check_node_accept_bytes_match;