Hello,
On Tue, Jan 31, 2006 at 08:12:40PM +0200, Wladimir Mutel wrote:
>
> Here is the testcase. May be fgrep in some way distinguishes
> between end-of-word and end-of-line, as my search samples were
> words at the end of lines :
The attached patch should fix this bug in "fgrep -w".
In the first chunk, we need to check the next char because the previous
char is the beginning of a multibyte char.
In the second chunk, we match a word if the last matched char is not
the beginning of a multibyte char, OR if we are out of the buffer.
Kind Regards,
--
Nekral
diff -rauN ../orig/grep-2.5.1.ds2/debian/patches/64-egf-speedup.patch
./grep-2.5.1.ds2/debian/patches/64-egf-speedup.patch
--- ../orig/grep-2.5.1.ds2/debian/patches/64-egf-speedup.patch 2006-05-16
11:21:45.000000000 -0500
+++ ./grep-2.5.1.ds2/debian/patches/64-egf-speedup.patch 2006-05-16
18:39:25.000000000 -0500
@@ -658,7 +658,7 @@
+ }
+ else
+#endif /* MBS_SUPPORT */
-+ if (!WCHAR ((unsigned char) beg[-1]))
++ if (WCHAR ((unsigned char) beg[-1]))
+ goto next_char;
+ }
#ifdef MBS_SUPPORT
@@ -685,7 +685,7 @@
+ else
#endif /* MBS_SUPPORT */
- return offset;
-+ if (beg + len >= buf + size && !WCHAR ((unsigned char)
beg[len]))
++ if (beg + len >= buf + size || !WCHAR ((unsigned char)
beg[len]))
+ word_match = 1;
+ if (word_match)
+ {
