rpms/grep
7
0
Fork 0
Code Issues Pull Requests Releases Activity

- Fixed -Fi for multibyte input (bug #143079).

Browse Source
This commit is contained in:
Tim Waugh 2004-12-21 13:57:42 +00:00
parent e5092c306f
commit e7e16314f3
2 changed files with 249 additions and 20 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

264
grep-2.5.1-egf-speedup.patch
View File

@ -1,6 +1,14 @@
--- grep-2.5.1/src/search.c 2004-12-16 17:46:57.039678304 +0000 --- grep-2.5.1/src/search.c 2004-12-21 13:37:15.700555594 +0000
+++ grep-2.5.1/src/search.c 2004-12-17 13:03:49.300731757 +0000 +++ grep-2.5.1/src/search.c 2004-12-21 13:49:05.873811016 +0000
@@ -39,6 +39,9 @@ @@ -21,6 +21,7 @@
#ifdef HAVE_CONFIG_H
# include <config.h>
#endif
+#include <assert.h>
#include <sys/types.h>
#if defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H && defined HAVE_MBRTOWC
/* We can handle multibyte string. */
@@ -39,6 +40,9 @@
#ifdef HAVE_LIBPCRE #ifdef HAVE_LIBPCRE
# include <pcre.h> # include <pcre.h>
#endif #endif
@ -10,7 +18,7 @@
#define NCHAR (UCHAR_MAX + 1) #define NCHAR (UCHAR_MAX + 1)
@@ -70,9 +73,10 @@ @@ -70,9 +74,10 @@
call the regexp matcher at all. */ call the regexp matcher at all. */
static int kwset_exact_matches; static int kwset_exact_matches;
@ -24,7 +32,7 @@
static void kwsinit PARAMS ((void)); static void kwsinit PARAMS ((void));
static void kwsmusts PARAMS ((void)); static void kwsmusts PARAMS ((void));
static void Gcompile PARAMS ((char const *, size_t)); static void Gcompile PARAMS ((char const *, size_t));
@@ -84,6 +88,15 @@ @@ -84,6 +89,15 @@
static size_t Pexecute PARAMS ((char const *, size_t, size_t *, int)); static size_t Pexecute PARAMS ((char const *, size_t, size_t *, int));
void void
@ -40,7 +48,7 @@
dfaerror (char const *mesg) dfaerror (char const *mesg)
{ {
error (2, 0, mesg); error (2, 0, mesg);
@@ -141,47 +154,6 @@ @@ -141,47 +155,6 @@
} }
} }
@ -88,7 +96,7 @@
static void static void
Gcompile (char const *pattern, size_t size) Gcompile (char const *pattern, size_t size)
{ {
@@ -190,6 +162,7 @@ @@ -190,6 +163,7 @@
size_t total = size; size_t total = size;
char const *motif = pattern; char const *motif = pattern;
@ -96,7 +104,7 @@
re_set_syntax (RE_SYNTAX_GREP | RE_HAT_LISTS_NOT_NEWLINE | (match_icase ? RE_ICASE : 0)); re_set_syntax (RE_SYNTAX_GREP | RE_HAT_LISTS_NOT_NEWLINE | (match_icase ? RE_ICASE : 0));
dfasyntax (RE_SYNTAX_GREP | RE_HAT_LISTS_NOT_NEWLINE, match_icase, eolbyte); dfasyntax (RE_SYNTAX_GREP | RE_HAT_LISTS_NOT_NEWLINE, match_icase, eolbyte);
@@ -266,6 +239,7 @@ @@ -266,6 +240,7 @@
size_t total = size; size_t total = size;
char const *motif = pattern; char const *motif = pattern;
@ -104,7 +112,7 @@
if (strcmp (matcher, "awk") == 0) if (strcmp (matcher, "awk") == 0)
{ {
re_set_syntax (RE_SYNTAX_AWK | (match_icase ? RE_ICASE : 0)); re_set_syntax (RE_SYNTAX_AWK | (match_icase ? RE_ICASE : 0));
@@ -350,18 +324,8 @@ @@ -350,18 +325,8 @@
struct kwsmatch kwsm; struct kwsmatch kwsm;
size_t i, ret_val; size_t i, ret_val;
#ifdef MBS_SUPPORT #ifdef MBS_SUPPORT
@ -125,7 +133,7 @@
#endif /* MBS_SUPPORT */ #endif /* MBS_SUPPORT */
buflim = buf + size; buflim = buf + size;
@@ -373,21 +337,63 @@ @@ -373,21 +338,63 @@
if (kwset) if (kwset)
{ {
/* Find a possible match using the KWset matcher. */ /* Find a possible match using the KWset matcher. */
@ -193,7 +201,7 @@
goto success_in_beg_and_end; goto success_in_beg_and_end;
if (dfaexec (&dfa, beg, end - beg, &backref) == (size_t) -1) if (dfaexec (&dfa, beg, end - beg, &backref) == (size_t) -1)
continue; continue;
@@ -395,13 +401,47 @@ @@ -395,13 +402,47 @@
else else
{ {
/* No good fixed strings; start with DFA. */ /* No good fixed strings; start with DFA. */
@ -241,7 +249,7 @@
while (beg > buf && beg[-1] != eol) while (beg > buf && beg[-1] != eol)
--beg; --beg;
} }
@@ -469,15 +509,6 @@ @@ -469,15 +510,6 @@
} /* for (beg = end ..) */ } /* for (beg = end ..) */
failure: failure:
@ -257,7 +265,7 @@
return (size_t) -1; return (size_t) -1;
success_in_beg_and_end: success_in_beg_and_end:
@@ -486,15 +517,6 @@ @@ -486,24 +518,125 @@
/* FALLTHROUGH */ /* FALLTHROUGH */
success_in_start_and_len: success_in_start_and_len:
@ -273,15 +281,214 @@
*match_size = len; *match_size = len;
return start; return start;
} }
@@ -504,6 +526,7 @@
+static wchar_t **f_pattern;
+static char *f_initial_byte;
+static size_t f_pattern_count;
+static int f_i_multibyte; /* whether we're using the new -Fi MB method */
+
static void
Fcompile (char const *pattern, size_t size)
{ {
char const *beg, *lim, *err; char const *beg, *lim, *err;
+ check_utf8 (); + check_utf8 ();
+#ifdef MBS_SUPPORT
+ /* Support -F -i for UTF-8 input. */
+ if (match_icase && MB_CUR_MAX > 1)
+ {
+ size_t in = 0;
+
+ while (f_i_multibyte != -1 && in < size)
+ {
+ wchar_t *f_this_pattern;
+ size_t f_this_pattern_allocated = sizeof (wchar_t) * 1000;
+ mbstate_t mbs;
+ size_t out = 0;
+ f_pattern_count++;
+ f_pattern = xrealloc (f_pattern,
+ sizeof (wchar_t *) * f_pattern_count);
+ f_initial_byte = xrealloc (f_initial_byte,
+ sizeof (char) *
+ (2 * f_pattern_count + 1));
+ if (f_pattern_count == 1)
+ f_initial_byte[0] = '\0';
+
+ /* Convert pattern into wchar_t*, storing them in this_pattern.
+ Don't read more than we're given. */
+ f_this_pattern = xmalloc (f_this_pattern_allocated);
+ memset (&mbs, '\0', sizeof (mbs));
+ while (in < size)
+ {
+ size_t c;
+ wchar_t this_wc;
+ if (out == f_this_pattern_allocated)
+ {
+ f_this_pattern_allocated *= 2;
+ f_this_pattern = xrealloc (f_this_pattern,
+ f_this_pattern_allocated);
+ }
+
+ c = mbrtowc (&this_wc, pattern + in, size - in, &mbs);
+ if (c < 1)
+ {
+ /* Fall back to old method. */
+ f_i_multibyte = -1;
+ while (f_pattern_count--)
+ free (f_pattern[f_pattern_count]);
+ free (f_pattern);
+ f_pattern = NULL;
+ break;
+ }
+
+ f_this_pattern[out] = towlower (this_wc);
+ if (out == 0)
+ {
+ /* First character. Work out the first byte of upper and
+ lower case multibyte strings for the first character. */
+ wchar_t wc;
+ char mbs[MB_CUR_MAX];
+ mbstate_t ps;
+
+ if (iswupper (this_wc))
+ {
+ wc = towlower (this_wc);
+ }
+ else
+ {
+ wc = towupper (this_wc);
+ }
+
+ memset (&ps, '\0', sizeof (ps));
+ wcrtomb (mbs, this_wc, &ps);
+ mbs[1] = '\0';
+ strcat (f_initial_byte, mbs);
+
+ memset (&ps, '\0', sizeof (ps));
+ wcrtomb (mbs, wc, &ps);
+ mbs[1] = '\0';
+ strcat (f_initial_byte, mbs);
+ }
+
+ in += c;
+
+ if (this_wc == L'\n')
+ break;
+
+ out++;
+ }
+
+ if (f_i_multibyte == -1)
+ break;
+
+ /* Nul-terminate it. */
+ if (out == f_this_pattern_allocated)
+ {
+ f_this_pattern_allocated++;
+ f_this_pattern = xrealloc (f_this_pattern,
+ f_this_pattern_allocated);
+ }
+
+ f_this_pattern[out] = L'\0';
+ f_pattern[f_pattern_count - 1] = f_this_pattern;
+ f_i_multibyte = 1;
+ }
+ }
+#endif /* MBS_SUPPORT */
+
+
kwsinit (); kwsinit ();
beg = pattern; beg = pattern;
do do
@@ -531,17 +554,8 @@ @@ -523,6 +656,87 @@
}
static size_t
+Fimbexec (const char *buf, size_t size, size_t *plen)
+{
+ char const *beg;
+ size_t len;
+ mbstate_t mbs;
+
+ assert (match_icase && f_i_multibyte == 1);
+ assert (MB_CUR_MAX > 1);
+
+ memset (&mbs, '\0', sizeof (mbs));
+ beg = buf;
+ len = 0;
+ while (beg < buf + size)
+ {
+ wchar_t wc;
+ char const *p;
+ char const *next_char;
+ unsigned char match[f_pattern_count];
+ size_t i, letter;
+ int patterns_left;
+
+ for (p = beg;
+ (p < buf + size) && !strchr (f_initial_byte, *p);
+ p++)
+ ;
+
+ if (p == NULL || p == buf + size)
+ break;
+
+ /* First byte matches, now check the rest */
+ beg = p;
+ letter = len = 0;
+ memset (match, '\1', f_pattern_count);
+ patterns_left = 1;
+ while (patterns_left)
+ {
+ size_t c;
+
+ patterns_left = 0;
+
+ c = mbrtowc (&wc, beg + len, size - (beg - buf) - len, &mbs);
+ if (c < 1)
+ {
+ memset (&mbs, '\0', sizeof (mbs));
+ next_char = beg + 1;
+ break;
+ }
+
+ if (!len)
+ next_char = beg + c;
+
+ wc = towlower (wc);
+ for (i = 0; i < f_pattern_count; i++)
+ {
+ if (match[i])
+ {
+ if (f_pattern[i][letter] == L'\0')
+ {
+ /* Found a match. */
+ *plen = len;
+ return beg - buf;
+ }
+
+ if (f_pattern[i][letter] == wc)
+ patterns_left = 1;
+ else
+ match[i] = '\0';
+ }
+ }
+
+ len += c;
+ letter++;
+ }
+
+ beg = next_char;
+ }
+
+ return -1;
+}
+
+static size_t
Fexecute (char const *buf, size_t size, size_t *match_size, int exact)
{
register char const *beg, *try, *end;
@@ -531,27 +745,50 @@
struct kwsmatch kwsmatch; struct kwsmatch kwsmatch;
size_t ret_val; size_t ret_val;
#ifdef MBS_SUPPORT #ifdef MBS_SUPPORT
@ -301,7 +508,16 @@
#endif /* MBS_SUPPORT */ #endif /* MBS_SUPPORT */
for (beg = buf; beg <= buf + size; ++beg) for (beg = buf; beg <= buf + size; ++beg)
@@ -550,8 +564,33 @@ {
- size_t offset = kwsexec (kwset, beg, buf + size - beg, &kwsmatch);
+ size_t offset;
+#ifdef MBS_SUPPORT
+ if (match_icase && f_i_multibyte == 1)
+ offset = Fimbexec (beg, buf + size - beg, &kwsmatch.size[0]);
+ else
+#endif /* MBS_SUPPORT */
+ offset = kwsexec (kwset, beg, buf + size - beg, &kwsmatch);
+
if (offset == (size_t) -1) if (offset == (size_t) -1)
goto failure; goto failure;
#ifdef MBS_SUPPORT #ifdef MBS_SUPPORT
@ -337,7 +553,17 @@
#endif /* MBS_SUPPORT */ #endif /* MBS_SUPPORT */
beg += offset; beg += offset;
len = kwsmatch.size[0]; len = kwsmatch.size[0];
@@ -587,6 +626,36 @@ @@ -583,10 +820,46 @@
{
/* Try a shorter length anchored at the same place. */
--len;
+#ifdef MBS_SUPPORT
+ if (match_icase && f_i_multibyte == 1)
+ offset = Fimbexec (beg, len, &kwsmatch.size[0]);
+ else
+#endif /* MBS_SUPPORT */
offset = kwsexec (kwset, beg, len, &kwsmatch);
+
if (offset == -1) { if (offset == -1) {
break; /* Try a different anchor. */ break; /* Try a different anchor. */
} }
@ -374,7 +600,7 @@
beg += offset; beg += offset;
len = kwsmatch.size[0]; len = kwsmatch.size[0];
} }
@@ -597,19 +666,31 @@ @@ -597,19 +870,31 @@
} }
failure: failure:
@ -414,7 +640,7 @@
end++; end++;
while (buf < beg && beg[-1] != eol) while (buf < beg && beg[-1] != eol)
--beg; --beg;
@@ -618,15 +699,6 @@ @@ -618,15 +903,6 @@
success_in_beg_and_len: success_in_beg_and_len:
*match_size = len; *match_size = len;

5
grep.spec
View File

@ -1,7 +1,7 @@
Summary: The GNU versions of grep pattern matching utilities. Summary: The GNU versions of grep pattern matching utilities.
Name: grep Name: grep
Version: 2.5.1 Version: 2.5.1
Release: 43 Release: 44
License: GPL License: GPL
Group: Applications/Text Group: Applications/Text
Source: ftp://ftp.gnu.org/pub/gnu/grep/grep-%{version}.tar.bz2 Source: ftp://ftp.gnu.org/pub/gnu/grep/grep-%{version}.tar.bz2
@ -85,6 +85,9 @@ fi
%{_mandir}/*/* %{_mandir}/*/*
%changelog %changelog
* Tue Dec 21 2004 Tim Waugh <twaugh@redhat.com> 2.5.1-44
- Fixed -Fi for multibyte input (bug #143079).
* Thu Dec 16 2004 Tim Waugh <twaugh@redhat.com> 2.5.1-43 * Thu Dec 16 2004 Tim Waugh <twaugh@redhat.com> 2.5.1-43
- Bypass kwset matching when ignoring case and processing multibyte input - Bypass kwset matching when ignoring case and processing multibyte input
(bug #143079). (bug #143079).