about summary refs log tree commit diff
path: root/posix
diff options
context:
space:
mode:
authorUlrich Drepper <drepper@redhat.com>2003-11-13 21:23:11 +0000
committerUlrich Drepper <drepper@redhat.com>2003-11-13 21:23:11 +0000
commita8067e8f899dc2842aba0780a6943bacb8247c8c (patch)
tree10a2eae6158fe4e31fe5dc9ae61b833688436e8e /posix
parent78c81ab7b4a25563697ce988ecff73c9937cef16 (diff)
downloadglibc-a8067e8f899dc2842aba0780a6943bacb8247c8c.tar.gz
glibc-a8067e8f899dc2842aba0780a6943bacb8247c8c.tar.xz
glibc-a8067e8f899dc2842aba0780a6943bacb8247c8c.zip
Update.
2003-11-13  Jakub Jelinek  <jakub@redhat.com>

	* posix/regcomp.c (optimize_utf8): Optimize even if SIMPLE_BRACKET
	with no bits set for >= 0x80 chars is seen.
	* posix/bug-regex20.c (tests): Add new tests.  Expect [ABC] to be
	optimized.
	(main): Run all tests with RE_ICASE as well.
Diffstat (limited to 'posix')
-rw-r--r--posix/bug-regex20.c44
-rw-r--r--posix/regcomp.c8
2 files changed, 47 insertions, 5 deletions
diff --git a/posix/bug-regex20.c b/posix/bug-regex20.c
index 69f3fbe572..11b9484faf 100644
--- a/posix/bug-regex20.c
+++ b/posix/bug-regex20.c
@@ -47,8 +47,8 @@ static struct
   {RE_SYNTAX_POSIX_BASIC, "^x\\\\y\\{6\\}z\\+", "x\\yyyyyyzz\xc3\xb6", 0, 1},
   {RE_SYNTAX_POSIX_BASIC, "^x\\\\y\\{2,36\\}z\\+", "x\\yzz\xc3\xb6", -1, 1},
   {RE_SYNTAX_POSIX_BASIC, "^x\\\\y\\{,3\\}z\\+", "x\\yyyzz\xc3\xb6", 0, 1},
-  /* FIXME.  This one should be optimizable, but is not ATM.  */
-  {RE_SYNTAX_POSIX_BASIC, "x[ABC]y", "axCy", 1, 0},
+  {RE_SYNTAX_POSIX_BASIC, "x[C]y", "axCy", 1, 1},
+  {RE_SYNTAX_POSIX_BASIC, "x[ABC]y", "axCy", 1, 1},
   {RE_SYNTAX_POSIX_BASIC, "\\`x\\|z\\'", "x\xe2\x80\x94", 0, 1},
   {RE_SYNTAX_POSIX_BASIC, "\\(xy\\)z\\1a\\1", "\xe2\x80\x94xyzxyaxy\xc3\x84", 3, 1},
   {RE_SYNTAX_POSIX_BASIC, "xy\\?z", "\xc3\x84xz\xc3\xb6", 2, 1},
@@ -57,8 +57,8 @@ static struct
   {RE_SYNTAX_POSIX_EXTENDED, "^x\\\\y{6}z+", "x\\yyyyyyzz\xc3\xb6", 0, 1},
   {RE_SYNTAX_POSIX_EXTENDED, "^x\\\\y{2,36}z+", "x\\yzz\xc3\xb6", -1, 1},
   {RE_SYNTAX_POSIX_EXTENDED, "^x\\\\y{,3}z+", "x\\yyyzz\xc3\xb6", 0, 1},
-  /* FIXME.  This one should be optimizable, but is not ATM.  */
-  {RE_SYNTAX_POSIX_EXTENDED, "x[ABC]y", "axCy", 1, 0},
+  {RE_SYNTAX_POSIX_EXTENDED, "x[C]y", "axCy", 1, 1},
+  {RE_SYNTAX_POSIX_EXTENDED, "x[ABC]y", "axCy", 1, 1},
   {RE_SYNTAX_POSIX_EXTENDED, "\\`x|z\\'", "x\xe2\x80\x94", 0, 1},
   {RE_SYNTAX_POSIX_EXTENDED, "(xy)z\\1a\\1", "\xe2\x80\x94xyzxyaxy\xc3\x84", 3, 1},
   {RE_SYNTAX_POSIX_EXTENDED, "xy?z", "\xc3\x84xz\xc3\xb6", 2, 1},
@@ -69,6 +69,8 @@ static struct
   {RE_SYNTAX_POSIX_BASIC, "x[A-Z,]y", "axCy", 1, 0},
   {RE_SYNTAX_POSIX_BASIC, "x[^y]z", "ax\xe2\x80\x94z", 1, 0},
   {RE_SYNTAX_POSIX_BASIC, "x[[:alnum:]]z", "ax\xc3\x96z", 1, 0},
+  {RE_SYNTAX_POSIX_BASIC, "x[[=A=]]z", "axAz", 1, 0},
+  {RE_SYNTAX_POSIX_BASIC, "x[[=\xc3\x84=]]z", "ax\xc3\x84z", 1, 0},
   {RE_SYNTAX_POSIX_BASIC, "\\<g", "\xe2\x80\x94g", 3, 0},
   {RE_SYNTAX_POSIX_BASIC, "\\bg\\b", "\xe2\x80\x94g", 3, 0},
   {RE_SYNTAX_POSIX_BASIC, "\\Bg\\B", "\xc3\xa4g\xc3\xa4", 2, 0},
@@ -80,6 +82,8 @@ static struct
   {RE_SYNTAX_POSIX_EXTENDED, "x[A-Z,]y", "axCy", 1, 0},
   {RE_SYNTAX_POSIX_EXTENDED, "x[^y]z", "ax\xe2\x80\x94z", 1, 0},
   {RE_SYNTAX_POSIX_EXTENDED, "x[[:alnum:]]z", "ax\xc3\x96z", 1, 0},
+  {RE_SYNTAX_POSIX_EXTENDED, "x[[=A=]]z", "axAz", 1, 0},
+  {RE_SYNTAX_POSIX_EXTENDED, "x[[=\xc3\x84=]]z", "ax\xc3\x84z", 1, 0},
   {RE_SYNTAX_POSIX_EXTENDED, "\\<g", "\xe2\x80\x94g", 3, 0},
   {RE_SYNTAX_POSIX_EXTENDED, "\\bg\\b", "\xe2\x80\x94g", 3, 0},
   {RE_SYNTAX_POSIX_EXTENDED, "\\Bg\\B", "\xc3\xa4g\xc3\xa4", 2, 0},
@@ -101,6 +105,7 @@ main (void)
   for (i = 0; i < sizeof (tests) / sizeof (tests[0]); ++i)
     {
       int res, optimized;
+
       re_set_syntax (tests[i].syntax);
       memset (&regbuf, '\0', sizeof (regbuf));
       err = re_compile_pattern (tests[i].pattern, strlen (tests[i].pattern),
@@ -131,6 +136,37 @@ main (void)
 	  continue;
 	}
       regfree (&regbuf);
+
+      re_set_syntax (tests[i].syntax | RE_ICASE);
+      memset (&regbuf, '\0', sizeof (regbuf));
+      err = re_compile_pattern (tests[i].pattern, strlen (tests[i].pattern),
+                                &regbuf);
+      if (err != NULL)
+	{
+	  printf ("re_compile_pattern failed: %s\n", err);
+	  ret = 1;
+	  continue;
+	}
+
+      /* Check if re_search will be done as multi-byte or single-byte.  */
+      optimized = ((re_dfa_t *) regbuf.buffer)->mb_cur_max == 1;
+      if (optimized)
+        {
+          printf ("pattern %zd optimized while it should not be when case insensitive\n",
+		  i);
+	  ret = 1;
+        }
+
+      res = re_search (&regbuf, tests[i].string, strlen (tests[i].string), 0,
+		       strlen (tests[i].string), NULL);
+      if (res != tests[i].res)
+	{
+	  printf ("re_search %zd failed: %d\n", i, res);
+	  ret = 1;
+	  regfree (&regbuf);
+	  continue;
+	}
+      regfree (&regbuf);
     }
 
   return ret;
diff --git a/posix/regcomp.c b/posix/regcomp.c
index a762859f8c..9f56b389df 100644
--- a/posix/regcomp.c
+++ b/posix/regcomp.c
@@ -964,7 +964,7 @@ static void
 optimize_utf8 (dfa)
      re_dfa_t *dfa;
 {
-  int node;
+  int node, i;
 
   for (node = 0; node < dfa->nodes_len; ++node)
     switch (dfa->nodes[node].type)
@@ -999,6 +999,12 @@ optimize_utf8 (dfa)
       case OP_OPEN_SUBEXP:
       case OP_CLOSE_SUBEXP:
 	break;
+      case SIMPLE_BRACKET:
+	/* Just double check.  */
+        for (i = 0x80 / UINT_BITS; i < BITSET_UINTS; ++i)
+	  if (dfa->nodes[node].opr.sbcset[i])
+	    return;
+	break;
       default:
 	return;
       }