remove bruno's 11-03 entry for unicodeio.c

[gnulib.git] / lib / regex.c

diff --git a/lib/regex.c b/lib/regex.c
index 8a20da1..25a219e 100644 (file)
--- a/lib/regex.c
+++ b/lib/regex.c
@@ -89,6 +89,10 @@
 /* This is for other GNU distributions with internationalized messages.  */
 #if HAVE_LIBINTL_H || defined _LIBC
 # include <libintl.h>
+# ifdef _LIBC
+#  undef gettext
+#  define gettext(msgid) __dcgettext ("libc", msgid, LC_MESSAGES)
+# endif
 #else
 # define gettext(msgid) (msgid)
 #endif
@@ -197,35 +201,34 @@ char *realloc ();
    eliminate the && through constant folding."
    Solaris defines some of these symbols so we must undefine them first.  */
 
-#undef ISASCII
 #if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
-# define ISASCII(c) 1
+# define IN_CTYPE_DOMAIN(c) 1
 #else
-# define ISASCII(c) isascii(c)
+# define IN_CTYPE_DOMAIN(c) isascii(c)
 #endif
 
 #ifdef isblank
-# define ISBLANK(c) (ISASCII (c) && isblank (c))
+# define ISBLANK(c) (IN_CTYPE_DOMAIN (c) && isblank (c))
 #else
 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
 #endif
 #ifdef isgraph
-# define ISGRAPH(c) (ISASCII (c) && isgraph (c))
+# define ISGRAPH(c) (IN_CTYPE_DOMAIN (c) && isgraph (c))
 #else
-# define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
+# define ISGRAPH(c) (IN_CTYPE_DOMAIN (c) && isprint (c) && !isspace (c))
 #endif
 
 #undef ISPRINT
-#define ISPRINT(c) (ISASCII (c) && isprint (c))
-#define ISDIGIT(c) (ISASCII (c) && isdigit (c))
-#define ISALNUM(c) (ISASCII (c) && isalnum (c))
-#define ISALPHA(c) (ISASCII (c) && isalpha (c))
-#define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
-#define ISLOWER(c) (ISASCII (c) && islower (c))
-#define ISPUNCT(c) (ISASCII (c) && ispunct (c))
-#define ISSPACE(c) (ISASCII (c) && isspace (c))
-#define ISUPPER(c) (ISASCII (c) && isupper (c))
-#define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
+#define ISPRINT(c) (IN_CTYPE_DOMAIN (c) && isprint (c))
+#define ISDIGIT(c) (IN_CTYPE_DOMAIN (c) && isdigit (c))
+#define ISALNUM(c) (IN_CTYPE_DOMAIN (c) && isalnum (c))
+#define ISALPHA(c) (IN_CTYPE_DOMAIN (c) && isalpha (c))
+#define ISCNTRL(c) (IN_CTYPE_DOMAIN (c) && iscntrl (c))
+#define ISLOWER(c) (IN_CTYPE_DOMAIN (c) && islower (c))
+#define ISPUNCT(c) (IN_CTYPE_DOMAIN (c) && ispunct (c))
+#define ISSPACE(c) (IN_CTYPE_DOMAIN (c) && isspace (c))
+#define ISUPPER(c) (IN_CTYPE_DOMAIN (c) && isupper (c))
+#define ISXDIGIT(c) (IN_CTYPE_DOMAIN (c) && isxdigit (c))
 
 #ifdef _tolower
 # define TOLOWER(c) _tolower(c)
@@ -1234,7 +1237,7 @@ typedef struct
 # if defined MATCH_MAY_ALLOCATE
 /* 4400 was enough to cause a crash on Alpha OSF/1,
    whose default stack limit is 2mb.  */
-int re_max_failures = 20000;
+int re_max_failures = 4000;
 # else
 int re_max_failures = 2000;
 # endif
@@ -2712,7 +2715,7 @@ regex_compile (pattern, size, syntax, bufp)
                            PATFETCH (c);
 
                            /* Now add the multibyte character(s) we found
-                              to the acceptabed list.
+                              to the accept list.
 
                               XXX Note that this is not entirely correct.
                               we would have to match multibyte sequences
@@ -2725,7 +2728,10 @@ regex_compile (pattern, size, syntax, bufp)
                            if (c1 == 1)
                              range_start = extra[idx];
                            while (c1-- > 0)
-                             SET_LIST_BIT (extra[idx++]);
+                             {
+                               SET_LIST_BIT (extra[idx]);
+                               ++idx;
+                             }
                          }
 #endif
                        had_char_class = false;
@@ -3507,23 +3513,19 @@ compile_range (range_start_char, p_ptr, pend, translate, syntax, b)
      unsigned char *b;
 {
   unsigned this_char;
-
   const char *p = *p_ptr;
   reg_errcode_t ret;
-  char range_start[2];
-  char range_end[2];
-  char ch[2];
+#if _LIBC
+  const unsigned char *collseq;
+  unsigned int start_colseq;
+  unsigned int end_colseq;
+#else
+  unsigned end_char;
+#endif
 
   if (p == pend)
     return REG_ERANGE;
 
-  /* Fetch the endpoints without translating them; the
-     appropriate translation is done in the bit-setting loop below.  */
-  range_start[0] = range_start_char;
-  range_start[1] = '\0';
-  range_end[0] = p[0];
-  range_end[1] = '\0';
-
   /* Have to increment the pointer into the pattern string, so the
      caller isn't still at the ending character.  */
   (*p_ptr)++;
@@ -3531,19 +3533,34 @@ compile_range (range_start_char, p_ptr, pend, translate, syntax, b)
   /* Report an error if the range is empty and the syntax prohibits this.  */
   ret = syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
 
-  /* Here we see why `this_char' has to be larger than an `unsigned
-     char' -- we would otherwise go into an infinite loop, since all
-     characters <= 0xff.  */
-  ch[1] = '\0';
+#if _LIBC
+  collseq = (const unsigned char *) _NL_CURRENT (LC_COLLATE,
+                                                _NL_COLLATE_COLLSEQMB);
+
+  start_colseq = collseq[(unsigned char) TRANSLATE (range_start_char)];
+  end_colseq = collseq[(unsigned char) TRANSLATE (p[0])];
   for (this_char = 0; this_char <= (unsigned char) -1; ++this_char)
     {
-      ch[0] = this_char;
-      if (strcoll (range_start, ch) <= 0 && strcoll (ch, range_end) <= 0)
+      unsigned int this_colseq = collseq[(unsigned char) TRANSLATE (this_char)];
+
+      if (start_colseq <= this_colseq && this_colseq <= end_colseq)
        {
          SET_LIST_BIT (TRANSLATE (this_char));
          ret = REG_NOERROR;
        }
     }
+#else
+  /* Here we see why `this_char' has to be larger than an `unsigned
+     char' -- we would otherwise go into an infinite loop, since all
+     characters <= 0xff.  */
+  range_start_char = TRANSLATE (range_start_char);
+  end_char = TRANSLATE (p[0]);
+  for (this_char = range_start_char; this_char <= end_char; ++this_char)
+    {
+      SET_LIST_BIT (TRANSLATE (this_char));
+      ret = REG_NOERROR;
+    }
+#endif
 
   return ret;
 }