|
Tim Waugh |
1b8d44 |
--- grep-2.5.1/src/search.c 2004-12-31 15:28:35.720391036 +0000
|
|
Tim Waugh |
1b8d44 |
+++ grep-2.5.1a/src/search.c 2005-01-07 14:53:10.308860193 +0000
|
|
Tim Waugh |
ff7a53 |
@@ -18,9 +18,13 @@
|
|
Tim Waugh |
ff7a53 |
|
|
Tim Waugh |
ff7a53 |
/* Written August 1992 by Mike Haertel. */
|
|
Tim Waugh |
ff7a53 |
|
|
Tim Waugh |
ff7a53 |
+#ifndef _GNU_SOURCE
|
|
Tim Waugh |
ff7a53 |
+# define _GNU_SOURCE 1
|
|
Tim Waugh |
ff7a53 |
+#endif
|
|
Tim Waugh |
e7e163 |
#ifdef HAVE_CONFIG_H
|
|
Tim Waugh |
e7e163 |
# include <config.h>
|
|
Tim Waugh |
e7e163 |
#endif
|
|
Tim Waugh |
e7e163 |
+#include <assert.h>
|
|
Tim Waugh |
e7e163 |
#include <sys/types.h>
|
|
Tim Waugh |
e7e163 |
#if defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H && defined HAVE_MBRTOWC
|
|
Tim Waugh |
e7e163 |
/* We can handle multibyte string. */
|
|
Tim Waugh |
ff7a53 |
@@ -39,6 +43,9 @@
|
|
Tim Waugh |
c3fd6f |
#ifdef HAVE_LIBPCRE
|
|
Tim Waugh |
c3fd6f |
# include <pcre.h>
|
|
Tim Waugh |
c3fd6f |
#endif
|
|
Tim Waugh |
c3fd6f |
+#ifdef HAVE_LANGINFO_CODESET
|
|
Tim Waugh |
c3fd6f |
+# include <langinfo.h>
|
|
Tim Waugh |
c3fd6f |
+#endif
|
|
Tim Waugh |
c3fd6f |
|
|
Tim Waugh |
c3fd6f |
#define NCHAR (UCHAR_MAX + 1)
|
|
Tim Waugh |
c3fd6f |
|
|
Tim Waugh |
ff7a53 |
@@ -70,9 +77,10 @@
|
|
Tim Waugh |
e1ca94 |
call the regexp matcher at all. */
|
|
Tim Waugh |
e1ca94 |
static int kwset_exact_matches;
|
|
Tim Waugh |
e1ca94 |
|
|
Tim Waugh |
e1ca94 |
-#if defined(MBS_SUPPORT)
|
|
Tim Waugh |
e1ca94 |
-static char* check_multibyte_string PARAMS ((char const *buf, size_t size));
|
|
Tim Waugh |
e1ca94 |
-#endif
|
|
Tim Waugh |
c3fd6f |
+/* UTF-8 encoding allows some optimizations that we can't otherwise
|
|
Tim Waugh |
c3fd6f |
+ assume in a multibyte encoding. */
|
|
Tim Waugh |
c3fd6f |
+static int using_utf8;
|
|
Tim Waugh |
c3fd6f |
+
|
|
Tim Waugh |
e1ca94 |
static void kwsinit PARAMS ((void));
|
|
Tim Waugh |
e1ca94 |
static void kwsmusts PARAMS ((void));
|
|
Tim Waugh |
e1ca94 |
static void Gcompile PARAMS ((char const *, size_t));
|
|
Tim Waugh |
ff7a53 |
@@ -84,6 +92,15 @@
|
|
Tim Waugh |
c3fd6f |
static size_t Pexecute PARAMS ((char const *, size_t, size_t *, int));
|
|
Tim Waugh |
c3fd6f |
|
|
Tim Waugh |
c3fd6f |
void
|
|
Tim Waugh |
c3fd6f |
+check_utf8 (void)
|
|
Tim Waugh |
c3fd6f |
+{
|
|
Tim Waugh |
c3fd6f |
+#ifdef HAVE_LANGINFO_CODESET
|
|
Tim Waugh |
c3fd6f |
+ if (strcmp (nl_langinfo (CODESET), "UTF-8") == 0)
|
|
Tim Waugh |
c3fd6f |
+ using_utf8 = 1;
|
|
Tim Waugh |
c3fd6f |
+#endif
|
|
Tim Waugh |
c3fd6f |
+}
|
|
Tim Waugh |
c3fd6f |
+
|
|
Tim Waugh |
c3fd6f |
+void
|
|
Tim Waugh |
c3fd6f |
dfaerror (char const *mesg)
|
|
Tim Waugh |
c3fd6f |
{
|
|
Tim Waugh |
c3fd6f |
error (2, 0, mesg);
|
|
Tim Waugh |
ff7a53 |
@@ -141,47 +158,6 @@
|
|
Tim Waugh |
e1ca94 |
}
|
|
Tim Waugh |
e1ca94 |
}
|
|
Tim Waugh |
e1ca94 |
|
|
Tim Waugh |
e1ca94 |
-#ifdef MBS_SUPPORT
|
|
Tim Waugh |
e1ca94 |
-/* This function allocate the array which correspond to "buf".
|
|
Tim Waugh |
e1ca94 |
- Then this check multibyte string and mark on the positions which
|
|
Tim Waugh |
e1ca94 |
- are not singlebyte character nor the first byte of a multibyte
|
|
Tim Waugh |
e1ca94 |
- character. Caller must free the array. */
|
|
Tim Waugh |
e1ca94 |
-static char*
|
|
Tim Waugh |
e1ca94 |
-check_multibyte_string(char const *buf, size_t size)
|
|
Tim Waugh |
e1ca94 |
-{
|
|
Tim Waugh |
e1ca94 |
- char *mb_properties = xmalloc(size);
|
|
Tim Waugh |
e1ca94 |
- mbstate_t cur_state;
|
|
Tim Waugh |
e1ca94 |
- wchar_t wc;
|
|
Tim Waugh |
e1ca94 |
- int i;
|
|
Tim Waugh |
e1ca94 |
- memset(&cur_state, 0, sizeof(mbstate_t));
|
|
Tim Waugh |
e1ca94 |
- memset(mb_properties, 0, sizeof(char)*size);
|
|
Tim Waugh |
e1ca94 |
- for (i = 0; i < size ;)
|
|
Tim Waugh |
e1ca94 |
- {
|
|
Tim Waugh |
e1ca94 |
- size_t mbclen;
|
|
Tim Waugh |
e1ca94 |
- mbclen = mbrtowc(&wc, buf + i, size - i, &cur_state);
|
|
Tim Waugh |
e1ca94 |
-
|
|
Tim Waugh |
e1ca94 |
- if (mbclen == (size_t) -1 || mbclen == (size_t) -2 || mbclen == 0)
|
|
Tim Waugh |
e1ca94 |
- {
|
|
Tim Waugh |
e1ca94 |
- /* An invalid sequence, or a truncated multibyte character.
|
|
Tim Waugh |
e1ca94 |
- We treat it as a singlebyte character. */
|
|
Tim Waugh |
e1ca94 |
- mbclen = 1;
|
|
Tim Waugh |
e1ca94 |
- }
|
|
Tim Waugh |
e1ca94 |
- else if (match_icase)
|
|
Tim Waugh |
e1ca94 |
- {
|
|
Tim Waugh |
e1ca94 |
- if (iswupper((wint_t)wc))
|
|
Tim Waugh |
e1ca94 |
- {
|
|
Tim Waugh |
e1ca94 |
- wc = towlower((wint_t)wc);
|
|
Tim Waugh |
e1ca94 |
- wcrtomb(buf + i, wc, &cur_state);
|
|
Tim Waugh |
e1ca94 |
- }
|
|
Tim Waugh |
e1ca94 |
- }
|
|
Tim Waugh |
e1ca94 |
- mb_properties[i] = mbclen;
|
|
Tim Waugh |
e1ca94 |
- i += mbclen;
|
|
Tim Waugh |
e1ca94 |
- }
|
|
Tim Waugh |
e1ca94 |
-
|
|
Tim Waugh |
e1ca94 |
- return mb_properties;
|
|
Tim Waugh |
e1ca94 |
-}
|
|
Tim Waugh |
e1ca94 |
-#endif
|
|
Tim Waugh |
e1ca94 |
-
|
|
Tim Waugh |
e1ca94 |
static void
|
|
Tim Waugh |
e1ca94 |
Gcompile (char const *pattern, size_t size)
|
|
Tim Waugh |
e1ca94 |
{
|
|
Tim Waugh |
ff7a53 |
@@ -190,6 +166,7 @@
|
|
Tim Waugh |
c3fd6f |
size_t total = size;
|
|
Tim Waugh |
c3fd6f |
char const *motif = pattern;
|
|
Tim Waugh |
c3fd6f |
|
|
Tim Waugh |
c3fd6f |
+ check_utf8 ();
|
|
Tim Waugh |
c3fd6f |
re_set_syntax (RE_SYNTAX_GREP | RE_HAT_LISTS_NOT_NEWLINE | (match_icase ? RE_ICASE : 0));
|
|
Tim Waugh |
c3fd6f |
dfasyntax (RE_SYNTAX_GREP | RE_HAT_LISTS_NOT_NEWLINE, match_icase, eolbyte);
|
|
Tim Waugh |
c3fd6f |
|
|
Tim Waugh |
ff7a53 |
@@ -266,6 +243,7 @@
|
|
Tim Waugh |
c3fd6f |
size_t total = size;
|
|
Tim Waugh |
c3fd6f |
char const *motif = pattern;
|
|
Tim Waugh |
c3fd6f |
|
|
Tim Waugh |
c3fd6f |
+ check_utf8 ();
|
|
Tim Waugh |
c3fd6f |
if (strcmp (matcher, "awk") == 0)
|
|
Tim Waugh |
c3fd6f |
{
|
|
Tim Waugh |
c3fd6f |
re_set_syntax (RE_SYNTAX_AWK | (match_icase ? RE_ICASE : 0));
|
|
Tim Waugh |
1b8d44 |
@@ -350,18 +328,9 @@
|
|
Tim Waugh |
e1ca94 |
struct kwsmatch kwsm;
|
|
Tim Waugh |
e1ca94 |
size_t i, ret_val;
|
|
Tim Waugh |
e1ca94 |
#ifdef MBS_SUPPORT
|
|
Tim Waugh |
e1ca94 |
- char *mb_properties = NULL;
|
|
Tim Waugh |
e1ca94 |
- if (MB_CUR_MAX > 1)
|
|
Tim Waugh |
e1ca94 |
- {
|
|
Tim Waugh |
e1ca94 |
- if (match_icase)
|
|
Tim Waugh |
e1ca94 |
- {
|
|
Tim Waugh |
e1ca94 |
- char *case_buf = xmalloc(size);
|
|
Tim Waugh |
e1ca94 |
- memcpy(case_buf, buf, size);
|
|
Tim Waugh |
e1ca94 |
- buf = case_buf;
|
|
Tim Waugh |
e1ca94 |
- }
|
|
Tim Waugh |
e1ca94 |
- if (kwset)
|
|
Tim Waugh |
e1ca94 |
- mb_properties = check_multibyte_string(buf, size);
|
|
Tim Waugh |
e1ca94 |
- }
|
|
Tim Waugh |
1b8d44 |
+ int mb_cur_max = MB_CUR_MAX;
|
|
Tim Waugh |
e1ca94 |
+ mbstate_t mbs;
|
|
Tim Waugh |
e1ca94 |
+ memset (&mbs, '\0', sizeof (mbstate_t));
|
|
Tim Waugh |
e1ca94 |
#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
e1ca94 |
|
|
Tim Waugh |
e1ca94 |
buflim = buf + size;
|
|
Tim Waugh |
1b8d44 |
@@ -373,21 +342,63 @@
|
|
Tim Waugh |
9fa0ec |
if (kwset)
|
|
Tim Waugh |
9fa0ec |
{
|
|
Tim Waugh |
9fa0ec |
/* Find a possible match using the KWset matcher. */
|
|
Tim Waugh |
8443e1 |
- size_t offset = kwsexec (kwset, beg, buflim - beg, &kwsm);
|
|
Tim Waugh |
9fa0ec |
+#ifdef MBS_SUPPORT
|
|
Tim Waugh |
9fa0ec |
+ size_t bytes_left = 0;
|
|
Tim Waugh |
9fa0ec |
+#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
8443e1 |
+ size_t offset;
|
|
Tim Waugh |
8443e1 |
+#ifdef MBS_SUPPORT
|
|
Tim Waugh |
a5ca95 |
+ /* kwsexec doesn't work with match_icase and multibyte input. */
|
|
Tim Waugh |
1b8d44 |
+ if (match_icase && mb_cur_max > 1)
|
|
Tim Waugh |
8443e1 |
+ /* Avoid kwset */
|
|
Tim Waugh |
8443e1 |
+ offset = 0;
|
|
Tim Waugh |
8443e1 |
+ else
|
|
Tim Waugh |
8443e1 |
+#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
8443e1 |
+ offset = kwsexec (kwset, beg, buflim - beg, &kwsm);
|
|
Tim Waugh |
e1ca94 |
if (offset == (size_t) -1)
|
|
Tim Waugh |
e1ca94 |
goto failure;
|
|
Tim Waugh |
e1ca94 |
+#ifdef MBS_SUPPORT
|
|
Tim Waugh |
1b8d44 |
+ if (mb_cur_max > 1 && !using_utf8)
|
|
Tim Waugh |
e1ca94 |
+ {
|
|
Tim Waugh |
9fa0ec |
+ bytes_left = offset;
|
|
Tim Waugh |
9fa0ec |
+ while (bytes_left)
|
|
Tim Waugh |
e1ca94 |
+ {
|
|
Tim Waugh |
b077d6 |
+ size_t mlen = mbrlen (beg, bytes_left, &mbs);
|
|
Tim Waugh |
b077d6 |
+ if (mlen == (size_t) -1 || mlen == 0)
|
|
Tim Waugh |
6f2e16 |
+ {
|
|
Tim Waugh |
2bb286 |
+ /* Incomplete character: treat as single-byte. */
|
|
Tim Waugh |
6f2e16 |
+ memset (&mbs, '\0', sizeof (mbstate_t));
|
|
Tim Waugh |
2bb286 |
+ beg++;
|
|
Tim Waugh |
2bb286 |
+ bytes_left--;
|
|
Tim Waugh |
2bb286 |
+ continue;
|
|
Tim Waugh |
6f2e16 |
+ }
|
|
Tim Waugh |
e1ca94 |
+
|
|
Tim Waugh |
b077d6 |
+ if (mlen == (size_t) -2)
|
|
Tim Waugh |
2bb286 |
+ /* Offset points inside multibyte character:
|
|
Tim Waugh |
2bb286 |
+ * no good. */
|
|
Tim Waugh |
2bb286 |
+ break;
|
|
Tim Waugh |
2bb286 |
+
|
|
Tim Waugh |
b077d6 |
+ beg += mlen;
|
|
Tim Waugh |
b077d6 |
+ bytes_left -= mlen;
|
|
Tim Waugh |
e1ca94 |
+ }
|
|
Tim Waugh |
e1ca94 |
+ }
|
|
Tim Waugh |
6f2e16 |
+ else
|
|
Tim Waugh |
6f2e16 |
+#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
e1ca94 |
beg += offset;
|
|
Tim Waugh |
e1ca94 |
/* Narrow down to the line containing the candidate, and
|
|
Tim Waugh |
e1ca94 |
run it through DFA. */
|
|
Tim Waugh |
e1ca94 |
end = memchr(beg, eol, buflim - beg);
|
|
Tim Waugh |
e1ca94 |
end++;
|
|
Tim Waugh |
9fa0ec |
#ifdef MBS_SUPPORT
|
|
Tim Waugh |
e1ca94 |
- if (MB_CUR_MAX > 1 && mb_properties[beg - buf] == 0)
|
|
Tim Waugh |
1b8d44 |
+ if (mb_cur_max > 1 && bytes_left)
|
|
Tim Waugh |
9fa0ec |
continue;
|
|
Tim Waugh |
e1ca94 |
-#endif
|
|
Tim Waugh |
9fa0ec |
+#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
e1ca94 |
while (beg > buf && beg[-1] != eol)
|
|
Tim Waugh |
e1ca94 |
--beg;
|
|
Tim Waugh |
a5ca95 |
- if (kwsm.index < kwset_exact_matches)
|
|
Tim Waugh |
e5092c |
+ if (
|
|
Tim Waugh |
e5092c |
+#ifdef MBS_SUPPORT
|
|
Tim Waugh |
1b8d44 |
+ !(match_icase && mb_cur_max > 1) &&
|
|
Tim Waugh |
e5092c |
+#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
a5ca95 |
+ (kwsm.index < kwset_exact_matches))
|
|
Tim Waugh |
a5ca95 |
goto success_in_beg_and_end;
|
|
Tim Waugh |
a5ca95 |
if (dfaexec (&dfa, beg, end - beg, &backref) == (size_t) -1)
|
|
Tim Waugh |
a5ca95 |
continue;
|
|
Tim Waugh |
1b8d44 |
@@ -395,13 +406,47 @@
|
|
Tim Waugh |
9fa0ec |
else
|
|
Tim Waugh |
9fa0ec |
{
|
|
Tim Waugh |
9fa0ec |
/* No good fixed strings; start with DFA. */
|
|
Tim Waugh |
9fa0ec |
+#ifdef MBS_SUPPORT
|
|
Tim Waugh |
9fa0ec |
+ size_t bytes_left = 0;
|
|
Tim Waugh |
9fa0ec |
+#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
9fa0ec |
size_t offset = dfaexec (&dfa, beg, buflim - beg, &backref);
|
|
Tim Waugh |
e1ca94 |
if (offset == (size_t) -1)
|
|
Tim Waugh |
e1ca94 |
break;
|
|
Tim Waugh |
e1ca94 |
/* Narrow down to the line we've found. */
|
|
Tim Waugh |
e1ca94 |
+#ifdef MBS_SUPPORT
|
|
Tim Waugh |
1b8d44 |
+ if (mb_cur_max > 1 && !using_utf8)
|
|
Tim Waugh |
e1ca94 |
+ {
|
|
Tim Waugh |
9fa0ec |
+ bytes_left = offset;
|
|
Tim Waugh |
9fa0ec |
+ while (bytes_left)
|
|
Tim Waugh |
e1ca94 |
+ {
|
|
Tim Waugh |
b077d6 |
+ size_t mlen = mbrlen (beg, bytes_left, &mbs);
|
|
Tim Waugh |
b077d6 |
+ if (mlen == (size_t) -1 || mlen == 0)
|
|
Tim Waugh |
6f2e16 |
+ {
|
|
Tim Waugh |
2bb286 |
+ /* Incomplete character: treat as single-byte. */
|
|
Tim Waugh |
6f2e16 |
+ memset (&mbs, '\0', sizeof (mbstate_t));
|
|
Tim Waugh |
2bb286 |
+ beg++;
|
|
Tim Waugh |
2bb286 |
+ bytes_left--;
|
|
Tim Waugh |
2bb286 |
+ continue;
|
|
Tim Waugh |
6f2e16 |
+ }
|
|
Tim Waugh |
e1ca94 |
+
|
|
Tim Waugh |
b077d6 |
+ if (mlen == (size_t) -2)
|
|
Tim Waugh |
2bb286 |
+ /* Offset points inside multibyte character:
|
|
Tim Waugh |
2bb286 |
+ * no good. */
|
|
Tim Waugh |
2bb286 |
+ break;
|
|
Tim Waugh |
2bb286 |
+
|
|
Tim Waugh |
b077d6 |
+ beg += mlen;
|
|
Tim Waugh |
b077d6 |
+ bytes_left -= mlen;
|
|
Tim Waugh |
e1ca94 |
+ }
|
|
Tim Waugh |
e1ca94 |
+ }
|
|
Tim Waugh |
6f2e16 |
+ else
|
|
Tim Waugh |
6f2e16 |
+#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
e1ca94 |
beg += offset;
|
|
Tim Waugh |
e1ca94 |
end = memchr (beg, eol, buflim - beg);
|
|
Tim Waugh |
e1ca94 |
end++;
|
|
Tim Waugh |
9fa0ec |
+#ifdef MBS_SUPPORT
|
|
Tim Waugh |
1b8d44 |
+ if (mb_cur_max > 1 && bytes_left)
|
|
Tim Waugh |
9fa0ec |
+ continue;
|
|
Tim Waugh |
9fa0ec |
+#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
9fa0ec |
while (beg > buf && beg[-1] != eol)
|
|
Tim Waugh |
9fa0ec |
--beg;
|
|
Tim Waugh |
9fa0ec |
}
|
|
Tim Waugh |
1b8d44 |
@@ -469,15 +514,6 @@
|
|
Tim Waugh |
e1ca94 |
} /* for (beg = end ..) */
|
|
Tim Waugh |
e1ca94 |
|
|
Tim Waugh |
e1ca94 |
failure:
|
|
Tim Waugh |
e1ca94 |
-#ifdef MBS_SUPPORT
|
|
Tim Waugh |
e1ca94 |
- if (MB_CUR_MAX > 1)
|
|
Tim Waugh |
e1ca94 |
- {
|
|
Tim Waugh |
e1ca94 |
- if (mb_properties)
|
|
Tim Waugh |
e1ca94 |
- free (mb_properties);
|
|
Tim Waugh |
e1ca94 |
- if (match_icase)
|
|
Tim Waugh |
e1ca94 |
- free ((char *) buf);
|
|
Tim Waugh |
e1ca94 |
- }
|
|
Tim Waugh |
e1ca94 |
-#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
e1ca94 |
return (size_t) -1;
|
|
Tim Waugh |
e1ca94 |
|
|
Tim Waugh |
e1ca94 |
success_in_beg_and_end:
|
|
Tim Waugh |
1b8d44 |
@@ -486,24 +522,144 @@
|
|
Tim Waugh |
e1ca94 |
/* FALLTHROUGH */
|
|
Tim Waugh |
e1ca94 |
|
|
Tim Waugh |
e1ca94 |
success_in_start_and_len:
|
|
Tim Waugh |
e1ca94 |
-#ifdef MBS_SUPPORT
|
|
Tim Waugh |
e1ca94 |
- if (MB_CUR_MAX > 1)
|
|
Tim Waugh |
e1ca94 |
- {
|
|
Tim Waugh |
e1ca94 |
- if (mb_properties)
|
|
Tim Waugh |
e1ca94 |
- free (mb_properties);
|
|
Tim Waugh |
e1ca94 |
- if (match_icase)
|
|
Tim Waugh |
e1ca94 |
- free ((char *) buf);
|
|
Tim Waugh |
e1ca94 |
- }
|
|
Tim Waugh |
e1ca94 |
-#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
e1ca94 |
*match_size = len;
|
|
Tim Waugh |
e1ca94 |
return start;
|
|
Tim Waugh |
e1ca94 |
}
|
|
Tim Waugh |
e7e163 |
|
|
Tim Waugh |
ff7a53 |
+#ifdef MBS_SUPPORT
|
|
Tim Waugh |
e7e163 |
+static int f_i_multibyte; /* whether we're using the new -Fi MB method */
|
|
Tim Waugh |
ff7a53 |
+static struct
|
|
Tim Waugh |
ff7a53 |
+{
|
|
Tim Waugh |
ff7a53 |
+ wchar_t **patterns;
|
|
Tim Waugh |
ff7a53 |
+ size_t count, maxlen;
|
|
Tim Waugh |
ff7a53 |
+ unsigned char *match;
|
|
Tim Waugh |
ff7a53 |
+} Fimb;
|
|
Tim Waugh |
ff7a53 |
+#endif
|
|
Tim Waugh |
e7e163 |
+
|
|
Tim Waugh |
e7e163 |
static void
|
|
Tim Waugh |
e7e163 |
Fcompile (char const *pattern, size_t size)
|
|
Tim Waugh |
c3fd6f |
{
|
|
Tim Waugh |
1b8d44 |
+ int mb_cur_max = MB_CUR_MAX;
|
|
Tim Waugh |
c3fd6f |
char const *beg, *lim, *err;
|
|
Tim Waugh |
c3fd6f |
|
|
Tim Waugh |
c3fd6f |
+ check_utf8 ();
|
|
Tim Waugh |
e7e163 |
+#ifdef MBS_SUPPORT
|
|
Tim Waugh |
e7e163 |
+ /* Support -F -i for UTF-8 input. */
|
|
Tim Waugh |
1b8d44 |
+ if (match_icase && mb_cur_max > 1)
|
|
Tim Waugh |
e7e163 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ mbstate_t mbs;
|
|
Tim Waugh |
ff7a53 |
+ wchar_t *wcpattern = xmalloc ((size + 1) * sizeof (wchar_t));
|
|
Tim Waugh |
ff7a53 |
+ const char *patternend = pattern;
|
|
Tim Waugh |
ff7a53 |
+ size_t wcsize;
|
|
Tim Waugh |
ff7a53 |
+ kwset_t fimb_kwset = NULL;
|
|
Tim Waugh |
ff7a53 |
+ char *starts = NULL;
|
|
Tim Waugh |
ff7a53 |
+ wchar_t *wcbeg, *wclim;
|
|
Tim Waugh |
ff7a53 |
+ size_t allocated = 0;
|
|
Tim Waugh |
ff7a53 |
+
|
|
Tim Waugh |
ff7a53 |
+ memset (&mbs, '\0', sizeof (mbs));
|
|
Tim Waugh |
ff7a53 |
+# ifdef __GNU_LIBRARY__
|
|
Tim Waugh |
ff7a53 |
+ wcsize = mbsnrtowcs (wcpattern, &patternend, size, size, &mbs);
|
|
Tim Waugh |
ff7a53 |
+ if (patternend != pattern + size)
|
|
Tim Waugh |
ff7a53 |
+ wcsize = (size_t) -1;
|
|
Tim Waugh |
ff7a53 |
+# else
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ char *patterncopy = xmalloc (size + 1);
|
|
Tim Waugh |
ff7a53 |
+
|
|
Tim Waugh |
ff7a53 |
+ memcpy (patterncopy, pattern, size);
|
|
Tim Waugh |
ff7a53 |
+ patterncopy[size] = '\0';
|
|
Tim Waugh |
ff7a53 |
+ patternend = patterncopy;
|
|
Tim Waugh |
ff7a53 |
+ wcsize = mbsrtowcs (wcpattern, &patternend, size, &mbs);
|
|
Tim Waugh |
ff7a53 |
+ if (patternend != patterncopy + size)
|
|
Tim Waugh |
ff7a53 |
+ wcsize = (size_t) -1;
|
|
Tim Waugh |
ff7a53 |
+ free (patterncopy);
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
+# endif
|
|
Tim Waugh |
ff7a53 |
+ if (wcsize + 2 <= 2)
|
|
Tim Waugh |
e7e163 |
+ {
|
|
Tim Waugh |
ff7a53 |
+fimb_fail:
|
|
Tim Waugh |
ff7a53 |
+ free (wcpattern);
|
|
Tim Waugh |
ff7a53 |
+ free (starts);
|
|
Tim Waugh |
ff7a53 |
+ if (fimb_kwset)
|
|
Tim Waugh |
ff7a53 |
+ kwsfree (fimb_kwset);
|
|
Tim Waugh |
ff7a53 |
+ free (Fimb.patterns);
|
|
Tim Waugh |
ff7a53 |
+ Fimb.patterns = NULL;
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
+ else
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ if (!(fimb_kwset = kwsalloc (NULL)))
|
|
Tim Waugh |
ff7a53 |
+ error (2, 0, _("memory exhausted"));
|
|
Tim Waugh |
ff7a53 |
+
|
|
Tim Waugh |
1b8d44 |
+ starts = xmalloc (mb_cur_max * 3);
|
|
Tim Waugh |
ff7a53 |
+ wcbeg = wcpattern;
|
|
Tim Waugh |
ff7a53 |
+ do
|
|
Tim Waugh |
e7e163 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ int i;
|
|
Tim Waugh |
ff7a53 |
+ size_t wclen;
|
|
Tim Waugh |
e7e163 |
+
|
|
Tim Waugh |
ff7a53 |
+ if (Fimb.count >= allocated)
|
|
Tim Waugh |
e7e163 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ if (allocated == 0)
|
|
Tim Waugh |
ff7a53 |
+ allocated = 128;
|
|
Tim Waugh |
ff7a53 |
+ else
|
|
Tim Waugh |
ff7a53 |
+ allocated *= 2;
|
|
Tim Waugh |
ff7a53 |
+ Fimb.patterns = xrealloc (Fimb.patterns,
|
|
Tim Waugh |
ff7a53 |
+ sizeof (wchar_t *) * allocated);
|
|
Tim Waugh |
e7e163 |
+ }
|
|
Tim Waugh |
ff7a53 |
+ Fimb.patterns[Fimb.count++] = wcbeg;
|
|
Tim Waugh |
ff7a53 |
+ for (wclim = wcbeg;
|
|
Tim Waugh |
ff7a53 |
+ wclim < wcpattern + wcsize && *wclim != L'\n'; ++wclim)
|
|
Tim Waugh |
ff7a53 |
+ *wclim = towlower (*wclim);
|
|
Tim Waugh |
ff7a53 |
+ *wclim = L'\0';
|
|
Tim Waugh |
ff7a53 |
+ wclen = wclim - wcbeg;
|
|
Tim Waugh |
ff7a53 |
+ if (wclen > Fimb.maxlen)
|
|
Tim Waugh |
ff7a53 |
+ Fimb.maxlen = wclen;
|
|
Tim Waugh |
ff7a53 |
+ if (wclen > 3)
|
|
Tim Waugh |
ff7a53 |
+ wclen = 3;
|
|
Tim Waugh |
ff7a53 |
+ if (wclen == 0)
|
|
Tim Waugh |
e7e163 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ if ((err = kwsincr (fimb_kwset, "", 0)) != 0)
|
|
Tim Waugh |
ff7a53 |
+ error (2, 0, err);
|
|
Tim Waugh |
e7e163 |
+ }
|
|
Tim Waugh |
ff7a53 |
+ else
|
|
Tim Waugh |
ff7a53 |
+ for (i = 0; i < (1 << wclen); i++)
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ char *p = starts;
|
|
Tim Waugh |
ff7a53 |
+ int j, k;
|
|
Tim Waugh |
ff7a53 |
+
|
|
Tim Waugh |
ff7a53 |
+ for (j = 0; j < wclen; ++j)
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ wchar_t wc = wcbeg[j];
|
|
Tim Waugh |
ff7a53 |
+ if (i & (1 << j))
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ wc = towupper (wc);
|
|
Tim Waugh |
ff7a53 |
+ if (wc == wcbeg[j])
|
|
Tim Waugh |
ff7a53 |
+ continue;
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
+ k = wctomb (p, wc);
|
|
Tim Waugh |
ff7a53 |
+ if (k <= 0)
|
|
Tim Waugh |
ff7a53 |
+ goto fimb_fail;
|
|
Tim Waugh |
ff7a53 |
+ p += k;
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
+ if ((err = kwsincr (fimb_kwset, starts, p - starts)) != 0)
|
|
Tim Waugh |
ff7a53 |
+ error (2, 0, err);
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
+ if (wclim < wcpattern + wcsize)
|
|
Tim Waugh |
ff7a53 |
+ ++wclim;
|
|
Tim Waugh |
ff7a53 |
+ wcbeg = wclim;
|
|
Tim Waugh |
e7e163 |
+ }
|
|
Tim Waugh |
ff7a53 |
+ while (wcbeg < wcpattern + wcsize);
|
|
Tim Waugh |
e7e163 |
+ f_i_multibyte = 1;
|
|
Tim Waugh |
ff7a53 |
+ kwset = fimb_kwset;
|
|
Tim Waugh |
ff7a53 |
+ free (starts);
|
|
Tim Waugh |
ff7a53 |
+ Fimb.match = xmalloc (Fimb.count);
|
|
Tim Waugh |
ff7a53 |
+ if ((err = kwsprep (kwset)) != 0)
|
|
Tim Waugh |
ff7a53 |
+ error (2, 0, err);
|
|
Tim Waugh |
ff7a53 |
+ return;
|
|
Tim Waugh |
e7e163 |
+ }
|
|
Tim Waugh |
e7e163 |
+ }
|
|
Tim Waugh |
e7e163 |
+#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
e7e163 |
+
|
|
Tim Waugh |
e7e163 |
+
|
|
Tim Waugh |
c3fd6f |
kwsinit ();
|
|
Tim Waugh |
c3fd6f |
beg = pattern;
|
|
Tim Waugh |
c3fd6f |
do
|
|
Tim Waugh |
1b8d44 |
@@ -522,6 +678,76 @@
|
|
Tim Waugh |
ff7a53 |
error (2, 0, err);
|
|
Tim Waugh |
e7e163 |
}
|
|
Tim Waugh |
e7e163 |
|
|
Tim Waugh |
ff7a53 |
+#ifdef MBS_SUPPORT
|
|
Tim Waugh |
ff7a53 |
+static int
|
|
Tim Waugh |
ff7a53 |
+Fimbexec (const char *buf, size_t size, size_t *plen, int exact)
|
|
Tim Waugh |
e7e163 |
+{
|
|
Tim Waugh |
ff7a53 |
+ size_t len, letter, i;
|
|
Tim Waugh |
ff7a53 |
+ int ret = -1;
|
|
Tim Waugh |
e7e163 |
+ mbstate_t mbs;
|
|
Tim Waugh |
ff7a53 |
+ wchar_t wc;
|
|
Tim Waugh |
ff7a53 |
+ int patterns_left;
|
|
Tim Waugh |
e7e163 |
+
|
|
Tim Waugh |
e7e163 |
+ assert (match_icase && f_i_multibyte == 1);
|
|
Tim Waugh |
e7e163 |
+ assert (MB_CUR_MAX > 1);
|
|
Tim Waugh |
e7e163 |
+
|
|
Tim Waugh |
e7e163 |
+ memset (&mbs, '\0', sizeof (mbs));
|
|
Tim Waugh |
ff7a53 |
+ memset (Fimb.match, '\1', Fimb.count);
|
|
Tim Waugh |
ff7a53 |
+ letter = len = 0;
|
|
Tim Waugh |
ff7a53 |
+ patterns_left = 1;
|
|
Tim Waugh |
ff7a53 |
+ while (patterns_left && len <= size)
|
|
Tim Waugh |
e7e163 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ size_t c;
|
|
Tim Waugh |
e7e163 |
+
|
|
Tim Waugh |
ff7a53 |
+ patterns_left = 0;
|
|
Tim Waugh |
ff7a53 |
+ if (len < size)
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ c = mbrtowc (&wc, buf + len, size - len, &mbs);
|
|
Tim Waugh |
ff7a53 |
+ if (c + 2 <= 2)
|
|
Tim Waugh |
ff7a53 |
+ return ret;
|
|
Tim Waugh |
e7e163 |
+
|
|
Tim Waugh |
e7e163 |
+ wc = towlower (wc);
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
+ else
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ c = 1;
|
|
Tim Waugh |
ff7a53 |
+ wc = L'\0';
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
+
|
|
Tim Waugh |
ff7a53 |
+ for (i = 0; i < Fimb.count; i++)
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ if (Fimb.match[i])
|
|
Tim Waugh |
e7e163 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ if (Fimb.patterns[i][letter] == L'\0')
|
|
Tim Waugh |
e7e163 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ /* Found a match. */
|
|
Tim Waugh |
ff7a53 |
+ *plen = len;
|
|
Tim Waugh |
ff7a53 |
+ if (!exact && !match_words)
|
|
Tim Waugh |
ff7a53 |
+ return 0;
|
|
Tim Waugh |
ff7a53 |
+ else
|
|
Tim Waugh |
e7e163 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ /* For -w or exact look for longest match. */
|
|
Tim Waugh |
ff7a53 |
+ ret = 0;
|
|
Tim Waugh |
ff7a53 |
+ Fimb.match[i] = '\0';
|
|
Tim Waugh |
ff7a53 |
+ continue;
|
|
Tim Waugh |
e7e163 |
+ }
|
|
Tim Waugh |
e7e163 |
+ }
|
|
Tim Waugh |
e7e163 |
+
|
|
Tim Waugh |
ff7a53 |
+ if (Fimb.patterns[i][letter] == wc)
|
|
Tim Waugh |
ff7a53 |
+ patterns_left = 1;
|
|
Tim Waugh |
ff7a53 |
+ else
|
|
Tim Waugh |
ff7a53 |
+ Fimb.match[i] = '\0';
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
e7e163 |
+ }
|
|
Tim Waugh |
e7e163 |
+
|
|
Tim Waugh |
ff7a53 |
+ len += c;
|
|
Tim Waugh |
ff7a53 |
+ letter++;
|
|
Tim Waugh |
e7e163 |
+ }
|
|
Tim Waugh |
ff7a53 |
+
|
|
Tim Waugh |
ff7a53 |
+ return ret;
|
|
Tim Waugh |
e7e163 |
+}
|
|
Tim Waugh |
ff7a53 |
+#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
e7e163 |
+
|
|
Tim Waugh |
ff7a53 |
static size_t
|
|
Tim Waugh |
e7e163 |
Fexecute (char const *buf, size_t size, size_t *match_size, int exact)
|
|
Tim Waugh |
e7e163 |
{
|
|
Tim Waugh |
1b8d44 |
@@ -531,80 +757,258 @@
|
|
Tim Waugh |
e1ca94 |
struct kwsmatch kwsmatch;
|
|
Tim Waugh |
e1ca94 |
size_t ret_val;
|
|
Tim Waugh |
e1ca94 |
#ifdef MBS_SUPPORT
|
|
Tim Waugh |
e1ca94 |
- char *mb_properties = NULL;
|
|
Tim Waugh |
e1ca94 |
- if (MB_CUR_MAX > 1)
|
|
Tim Waugh |
e1ca94 |
- {
|
|
Tim Waugh |
e1ca94 |
- if (match_icase)
|
|
Tim Waugh |
e1ca94 |
- {
|
|
Tim Waugh |
e1ca94 |
- char *case_buf = xmalloc(size);
|
|
Tim Waugh |
e1ca94 |
- memcpy(case_buf, buf, size);
|
|
Tim Waugh |
e1ca94 |
- buf = case_buf;
|
|
Tim Waugh |
e1ca94 |
- }
|
|
Tim Waugh |
e1ca94 |
- mb_properties = check_multibyte_string(buf, size);
|
|
Tim Waugh |
e1ca94 |
- }
|
|
Tim Waugh |
ff7a53 |
+ int mb_cur_max = MB_CUR_MAX;
|
|
Tim Waugh |
e1ca94 |
+ mbstate_t mbs;
|
|
Tim Waugh |
e1ca94 |
+ memset (&mbs, '\0', sizeof (mbstate_t));
|
|
Tim Waugh |
ff7a53 |
+ const char *last_char = NULL;
|
|
Tim Waugh |
e1ca94 |
#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
e1ca94 |
|
|
Tim Waugh |
e1ca94 |
for (beg = buf; beg <= buf + size; ++beg)
|
|
Tim Waugh |
e7e163 |
{
|
|
Tim Waugh |
e7e163 |
- size_t offset = kwsexec (kwset, beg, buf + size - beg, &kwsmatch);
|
|
Tim Waugh |
e7e163 |
+ size_t offset;
|
|
Tim Waugh |
e7e163 |
+ offset = kwsexec (kwset, beg, buf + size - beg, &kwsmatch);
|
|
Tim Waugh |
e7e163 |
+
|
|
Tim Waugh |
e1ca94 |
if (offset == (size_t) -1)
|
|
Tim Waugh |
e1ca94 |
goto failure;
|
|
Tim Waugh |
e1ca94 |
#ifdef MBS_SUPPORT
|
|
Tim Waugh |
e1ca94 |
- if (MB_CUR_MAX > 1 && mb_properties[offset+beg-buf] == 0)
|
|
Tim Waugh |
e1ca94 |
- continue; /* It is a part of multibyte character. */
|
|
Tim Waugh |
ff7a53 |
+ if (mb_cur_max > 1 && !using_utf8)
|
|
Tim Waugh |
e1ca94 |
+ {
|
|
Tim Waugh |
9fa0ec |
+ size_t bytes_left = offset;
|
|
Tim Waugh |
9fa0ec |
+ while (bytes_left)
|
|
Tim Waugh |
e1ca94 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ size_t mlen = mbrlen (beg, bytes_left, &mbs);
|
|
Tim Waugh |
ff7a53 |
+
|
|
Tim Waugh |
ff7a53 |
+ last_char = beg;
|
|
Tim Waugh |
ff7a53 |
+ if (mlen == (size_t) -1 || mlen == 0)
|
|
Tim Waugh |
6f2e16 |
+ {
|
|
Tim Waugh |
2bb286 |
+ /* Incomplete character: treat as single-byte. */
|
|
Tim Waugh |
6f2e16 |
+ memset (&mbs, '\0', sizeof (mbstate_t));
|
|
Tim Waugh |
2bb286 |
+ beg++;
|
|
Tim Waugh |
2bb286 |
+ bytes_left--;
|
|
Tim Waugh |
2bb286 |
+ continue;
|
|
Tim Waugh |
6f2e16 |
+ }
|
|
Tim Waugh |
6f2e16 |
+
|
|
Tim Waugh |
ff7a53 |
+ if (mlen == (size_t) -2)
|
|
Tim Waugh |
2bb286 |
+ /* Offset points inside multibyte character: no good. */
|
|
Tim Waugh |
2bb286 |
+ break;
|
|
Tim Waugh |
2bb286 |
+
|
|
Tim Waugh |
ff7a53 |
+ beg += mlen;
|
|
Tim Waugh |
ff7a53 |
+ bytes_left -= mlen;
|
|
Tim Waugh |
e1ca94 |
+ }
|
|
Tim Waugh |
e1ca94 |
+
|
|
Tim Waugh |
9fa0ec |
+ if (bytes_left)
|
|
Tim Waugh |
6f2e16 |
+ continue;
|
|
Tim Waugh |
e1ca94 |
+ }
|
|
Tim Waugh |
6f2e16 |
+ else
|
|
Tim Waugh |
6f2e16 |
#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
e1ca94 |
beg += offset;
|
|
Tim Waugh |
e7e163 |
+#ifdef MBS_SUPPORT
|
|
Tim Waugh |
ff7a53 |
+ /* For f_i_multibyte, the string at beg now matches first 3 chars of
|
|
Tim Waugh |
ff7a53 |
+ one of the search strings (less if there are shorter search strings).
|
|
Tim Waugh |
ff7a53 |
+ See if this is a real match. */
|
|
Tim Waugh |
ff7a53 |
+ if (f_i_multibyte
|
|
Tim Waugh |
ff7a53 |
+ && Fimbexec (beg, buf + size - beg, &kwsmatch.size[0], exact))
|
|
Tim Waugh |
ff7a53 |
+ goto next_char;
|
|
Tim Waugh |
ff7a53 |
+#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
ff7a53 |
len = kwsmatch.size[0];
|
|
Tim Waugh |
ff7a53 |
if (exact && !match_words)
|
|
Tim Waugh |
ff7a53 |
goto success_in_beg_and_len;
|
|
Tim Waugh |
ff7a53 |
if (match_lines)
|
|
Tim Waugh |
ff7a53 |
{
|
|
Tim Waugh |
ff7a53 |
if (beg > buf && beg[-1] != eol)
|
|
Tim Waugh |
ff7a53 |
- continue;
|
|
Tim Waugh |
ff7a53 |
+ goto next_char;
|
|
Tim Waugh |
ff7a53 |
if (beg + len < buf + size && beg[len] != eol)
|
|
Tim Waugh |
ff7a53 |
- continue;
|
|
Tim Waugh |
ff7a53 |
+ goto next_char;
|
|
Tim Waugh |
ff7a53 |
goto success;
|
|
Tim Waugh |
ff7a53 |
}
|
|
Tim Waugh |
ff7a53 |
else if (match_words)
|
|
Tim Waugh |
ff7a53 |
- for (try = beg; len; )
|
|
Tim Waugh |
ff7a53 |
- {
|
|
Tim Waugh |
ff7a53 |
- if (try > buf && WCHAR((unsigned char) try[-1]))
|
|
Tim Waugh |
ff7a53 |
- break;
|
|
Tim Waugh |
ff7a53 |
- if (try + len < buf + size && WCHAR((unsigned char) try[len]))
|
|
Tim Waugh |
ff7a53 |
- {
|
|
Tim Waugh |
ff7a53 |
- offset = kwsexec (kwset, beg, --len, &kwsmatch);
|
|
Tim Waugh |
ff7a53 |
- if (offset == (size_t) -1)
|
|
Tim Waugh |
ff7a53 |
- {
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ while (1)
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ int word_match = 0;
|
|
Tim Waugh |
ff7a53 |
+ if (beg > buf)
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
#ifdef MBS_SUPPORT
|
|
Tim Waugh |
ff7a53 |
- if (MB_CUR_MAX > 1)
|
|
Tim Waugh |
ff7a53 |
- free (mb_properties);
|
|
Tim Waugh |
ff7a53 |
+ if (mb_cur_max > 1)
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ const char *s;
|
|
Tim Waugh |
ff7a53 |
+ int mr;
|
|
Tim Waugh |
ff7a53 |
+ wchar_t pwc;
|
|
Tim Waugh |
ff7a53 |
+
|
|
Tim Waugh |
ff7a53 |
+ if (using_utf8)
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ s = beg - 1;
|
|
Tim Waugh |
ff7a53 |
+ while (s > buf
|
|
Tim Waugh |
ff7a53 |
+ && (unsigned char) *s >= 0x80
|
|
Tim Waugh |
ff7a53 |
+ && (unsigned char) *s <= 0xbf)
|
|
Tim Waugh |
ff7a53 |
+ --s;
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
+ else
|
|
Tim Waugh |
ff7a53 |
+ s = last_char;
|
|
Tim Waugh |
ff7a53 |
+ mr = mbtowc (&pwc, s, beg - s);
|
|
Tim Waugh |
ff7a53 |
+ if (mr <= 0)
|
|
Tim Waugh |
ff7a53 |
+ memset (&mbs, '\0', sizeof (mbstate_t));
|
|
Tim Waugh |
ff7a53 |
+ else if ((iswalnum (pwc) || pwc == L'_')
|
|
Tim Waugh |
ff7a53 |
+ && mr == (int) (beg - s))
|
|
Tim Waugh |
ff7a53 |
+ goto next_char;
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
e7e163 |
+ else
|
|
Tim Waugh |
ff7a53 |
#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
ff7a53 |
- return offset;
|
|
Tim Waugh |
ff7a53 |
- }
|
|
Tim Waugh |
ff7a53 |
- try = beg + offset;
|
|
Tim Waugh |
ff7a53 |
- len = kwsmatch.size[0];
|
|
Tim Waugh |
ff7a53 |
- }
|
|
Tim Waugh |
ff7a53 |
- else
|
|
Tim Waugh |
ff7a53 |
- goto success;
|
|
Tim Waugh |
ff7a53 |
- }
|
|
Tim Waugh |
b566e4 |
+ if (WCHAR ((unsigned char) beg[-1]))
|
|
Tim Waugh |
ff7a53 |
+ goto next_char;
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
+#ifdef MBS_SUPPORT
|
|
Tim Waugh |
ff7a53 |
+ if (mb_cur_max > 1)
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ wchar_t nwc;
|
|
Tim Waugh |
ff7a53 |
+ int mr;
|
|
Tim Waugh |
ff7a53 |
+
|
|
Tim Waugh |
ff7a53 |
+ mr = mbtowc (&nwc, beg + len, buf + size - beg - len);
|
|
Tim Waugh |
ff7a53 |
+ if (mr <= 0)
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ memset (&mbs, '\0', sizeof (mbstate_t));
|
|
Tim Waugh |
ff7a53 |
+ word_match = 1;
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
+ else if (!iswalnum (nwc) && nwc != L'_')
|
|
Tim Waugh |
ff7a53 |
+ word_match = 1;
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
+ else
|
|
Tim Waugh |
e7e163 |
+#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
23c991 |
+ if (beg + len >= buf + size || !WCHAR ((unsigned char) beg[len]))
|
|
Tim Waugh |
ff7a53 |
+ word_match = 1;
|
|
Tim Waugh |
ff7a53 |
+ if (word_match)
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ if (!exact)
|
|
Tim Waugh |
ff7a53 |
+ /* Returns the whole line now we know there's a word match. */
|
|
Tim Waugh |
ff7a53 |
+ goto success;
|
|
Tim Waugh |
ff7a53 |
+ else
|
|
Tim Waugh |
ff7a53 |
+ /* Returns just this word match. */
|
|
Tim Waugh |
ff7a53 |
+ goto success_in_beg_and_len;
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
+ if (len > 0)
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ /* Try a shorter length anchored at the same place. */
|
|
Tim Waugh |
ff7a53 |
+ --len;
|
|
Tim Waugh |
ff7a53 |
+ offset = kwsexec (kwset, beg, len, &kwsmatch);
|
|
Tim Waugh |
e7e163 |
+
|
|
Tim Waugh |
ff7a53 |
+ if (offset == -1)
|
|
Tim Waugh |
ff7a53 |
+ goto next_char; /* Try a different anchor. */
|
|
Tim Waugh |
e1ca94 |
+#ifdef MBS_SUPPORT
|
|
Tim Waugh |
ff7a53 |
+ if (mb_cur_max > 1 && !using_utf8)
|
|
Tim Waugh |
e1ca94 |
+ {
|
|
Tim Waugh |
9fa0ec |
+ size_t bytes_left = offset;
|
|
Tim Waugh |
9fa0ec |
+ while (bytes_left)
|
|
Tim Waugh |
e1ca94 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ size_t mlen = mbrlen (beg, bytes_left, &mbs);
|
|
Tim Waugh |
ff7a53 |
+
|
|
Tim Waugh |
ff7a53 |
+ last_char = beg;
|
|
Tim Waugh |
ff7a53 |
+ if (mlen == (size_t) -1 || mlen == 0)
|
|
Tim Waugh |
6f2e16 |
+ {
|
|
Tim Waugh |
2bb286 |
+ /* Incomplete character: treat as single-byte. */
|
|
Tim Waugh |
6f2e16 |
+ memset (&mbs, '\0', sizeof (mbstate_t));
|
|
Tim Waugh |
2bb286 |
+ beg++;
|
|
Tim Waugh |
2bb286 |
+ bytes_left--;
|
|
Tim Waugh |
2bb286 |
+ continue;
|
|
Tim Waugh |
6f2e16 |
+ }
|
|
Tim Waugh |
6f2e16 |
+
|
|
Tim Waugh |
ff7a53 |
+ if (mlen == (size_t) -2)
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ /* Offset points inside multibyte character:
|
|
Tim Waugh |
ff7a53 |
+ * no good. */
|
|
Tim Waugh |
ff7a53 |
+ break;
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
2bb286 |
+
|
|
Tim Waugh |
ff7a53 |
+ beg += mlen;
|
|
Tim Waugh |
ff7a53 |
+ bytes_left -= mlen;
|
|
Tim Waugh |
e1ca94 |
+ }
|
|
Tim Waugh |
e1ca94 |
+
|
|
Tim Waugh |
9fa0ec |
+ if (bytes_left)
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ memset (&mbs, '\0', sizeof (mbstate_t));
|
|
Tim Waugh |
ff7a53 |
+ goto next_char; /* Try a different anchor. */
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
e1ca94 |
+ }
|
|
Tim Waugh |
6f2e16 |
+ else
|
|
Tim Waugh |
e1ca94 |
+#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
ff7a53 |
+ beg += offset;
|
|
Tim Waugh |
ff7a53 |
+#ifdef MBS_SUPPORT
|
|
Tim Waugh |
ff7a53 |
+ /* The string at beg now matches first 3 chars of one of
|
|
Tim Waugh |
ff7a53 |
+ the search strings (less if there are shorter search
|
|
Tim Waugh |
ff7a53 |
+ strings). See if this is a real match. */
|
|
Tim Waugh |
ff7a53 |
+ if (f_i_multibyte
|
|
Tim Waugh |
ff7a53 |
+ && Fimbexec (beg, len - offset, &kwsmatch.size[0],
|
|
Tim Waugh |
ff7a53 |
+ exact))
|
|
Tim Waugh |
ff7a53 |
+ goto next_char;
|
|
Tim Waugh |
ff7a53 |
+#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
ff7a53 |
+ len = kwsmatch.size[0];
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
else
|
|
Tim Waugh |
ff7a53 |
goto success;
|
|
Tim Waugh |
ff7a53 |
+next_char:;
|
|
Tim Waugh |
ff7a53 |
+#ifdef MBS_SUPPORT
|
|
Tim Waugh |
ff7a53 |
+ /* Advance to next character. For MB_CUR_MAX == 1 case this is handled
|
|
Tim Waugh |
ff7a53 |
+ by ++beg above. */
|
|
Tim Waugh |
ff7a53 |
+ if (mb_cur_max > 1)
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ if (using_utf8)
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ unsigned char c = *beg;
|
|
Tim Waugh |
ff7a53 |
+ if (c >= 0xc2)
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ if (c < 0xe0)
|
|
Tim Waugh |
ff7a53 |
+ ++beg;
|
|
Tim Waugh |
ff7a53 |
+ else if (c < 0xf0)
|
|
Tim Waugh |
ff7a53 |
+ beg += 2;
|
|
Tim Waugh |
ff7a53 |
+ else if (c < 0xf8)
|
|
Tim Waugh |
ff7a53 |
+ beg += 3;
|
|
Tim Waugh |
ff7a53 |
+ else if (c < 0xfc)
|
|
Tim Waugh |
ff7a53 |
+ beg += 4;
|
|
Tim Waugh |
ff7a53 |
+ else if (c < 0xfe)
|
|
Tim Waugh |
ff7a53 |
+ beg += 5;
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
+ else
|
|
Tim Waugh |
ff7a53 |
+ {
|
|
Tim Waugh |
ff7a53 |
+ size_t l = mbrlen (beg, buf + size - beg, &mbs);
|
|
Tim Waugh |
ff7a53 |
+
|
|
Tim Waugh |
ff7a53 |
+ last_char = beg;
|
|
Tim Waugh |
ff7a53 |
+ if (l + 2 >= 2)
|
|
Tim Waugh |
ff7a53 |
+ beg += l - 1;
|
|
Tim Waugh |
ff7a53 |
+ else
|
|
Tim Waugh |
ff7a53 |
+ memset (&mbs, '\0', sizeof (mbstate_t));
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
+ }
|
|
Tim Waugh |
ff7a53 |
+#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
e1ca94 |
}
|
|
Tim Waugh |
e1ca94 |
|
|
Tim Waugh |
e1ca94 |
failure:
|
|
Tim Waugh |
c3fd6f |
+ return -1;
|
|
Tim Waugh |
c3fd6f |
+
|
|
Tim Waugh |
c3fd6f |
+ success:
|
|
Tim Waugh |
c3fd6f |
#ifdef MBS_SUPPORT
|
|
Tim Waugh |
e1ca94 |
- if (MB_CUR_MAX > 1)
|
|
Tim Waugh |
ff7a53 |
+ if (mb_cur_max > 1 && !using_utf8)
|
|
Tim Waugh |
c3fd6f |
{
|
|
Tim Waugh |
e1ca94 |
- if (match_icase)
|
|
Tim Waugh |
e1ca94 |
- free((char *) buf);
|
|
Tim Waugh |
e1ca94 |
- if (mb_properties)
|
|
Tim Waugh |
e1ca94 |
- free(mb_properties);
|
|
Tim Waugh |
c3fd6f |
+ end = beg + len;
|
|
Tim Waugh |
c3fd6f |
+ while (end < buf + size)
|
|
Tim Waugh |
e1ca94 |
+ {
|
|
Tim Waugh |
b077d6 |
+ size_t mlen = mbrlen (end, buf + size - end, &mbs);
|
|
Tim Waugh |
b077d6 |
+ if (mlen == (size_t) -1 || mlen == (size_t) -2 || mlen == 0)
|
|
Tim Waugh |
c3fd6f |
+ {
|
|
Tim Waugh |
c3fd6f |
+ memset (&mbs, '\0', sizeof (mbstate_t));
|
|
Tim Waugh |
b077d6 |
+ mlen = 1;
|
|
Tim Waugh |
c3fd6f |
+ }
|
|
Tim Waugh |
b077d6 |
+ if (mlen == 1 && *end == eol)
|
|
Tim Waugh |
c3fd6f |
+ break;
|
|
Tim Waugh |
e1ca94 |
+
|
|
Tim Waugh |
b077d6 |
+ end += mlen;
|
|
Tim Waugh |
c3fd6f |
+ }
|
|
Tim Waugh |
c3fd6f |
}
|
|
Tim Waugh |
c3fd6f |
+ else
|
|
Tim Waugh |
c3fd6f |
#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
c3fd6f |
- return -1;
|
|
Tim Waugh |
c3fd6f |
-
|
|
Tim Waugh |
c3fd6f |
- success:
|
|
Tim Waugh |
e1ca94 |
end = memchr (beg + len, eol, (buf + size) - (beg + len));
|
|
Tim Waugh |
c3fd6f |
+
|
|
Tim Waugh |
e1ca94 |
end++;
|
|
Tim Waugh |
e1ca94 |
while (buf < beg && beg[-1] != eol)
|
|
Tim Waugh |
e1ca94 |
--beg;
|
|
Tim Waugh |
1b8d44 |
@@ -613,15 +1017,6 @@
|
|
Tim Waugh |
e1ca94 |
|
|
Tim Waugh |
e1ca94 |
success_in_beg_and_len:
|
|
Tim Waugh |
e1ca94 |
*match_size = len;
|
|
Tim Waugh |
e1ca94 |
-#ifdef MBS_SUPPORT
|
|
Tim Waugh |
e1ca94 |
- if (MB_CUR_MAX > 1)
|
|
Tim Waugh |
e1ca94 |
- {
|
|
Tim Waugh |
e1ca94 |
- if (mb_properties)
|
|
Tim Waugh |
e1ca94 |
- free (mb_properties);
|
|
Tim Waugh |
e1ca94 |
- if (match_icase)
|
|
Tim Waugh |
e1ca94 |
- free ((char *) buf);
|
|
Tim Waugh |
e1ca94 |
- }
|
|
Tim Waugh |
e1ca94 |
-#endif /* MBS_SUPPORT */
|
|
Tim Waugh |
e1ca94 |
return beg - buf;
|
|
Tim Waugh |
e1ca94 |
}
|
|
Tim Waugh |
e1ca94 |
|