From f9df2006dfd1840e4e793b0a44d3318fd4fb0a09 Mon Sep 17 00:00:00 2001 From: Jim Meyering Date: Sat, 16 Nov 2019 09:24:00 -0800 Subject: [PATCH] grep: avoid false -Fw match in non-UTF8 multibyte locales For example, this command would erroneously print its input line: echo ab | LC_CTYPE=ja_JP.eucjp grep -Fw b * src/kwsearch.c (Fexecute): Set MB_START also when there is no preceding newline. Reported by NIDE, Naoyuki in https://bugs.gnu.org/38223. --- src/kwsearch.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/src/kwsearch.c b/src/kwsearch.c index 42567e9..2ad8a87 100644 --- a/src/kwsearch.c +++ b/src/kwsearch.c @@ -223,12 +223,14 @@ Fexecute (void *vcp, char const *buf, size_t size, size_t *match_size, if (! match_words) goto success; + /* We need a preceding mb_start pointer. Use the beginning of line + if there is a preceding newline, else BUF. */ + char const *bol = memrchr (mb_start, eol, beg - mb_start); + mb_start = bol ? bol + 1 : buf; + /* Succeed if the preceding and following characters are word constituents. If the following character is not a word constituent, keep trying with shorter matches. */ - char const *bol = memrchr (mb_start, eol, beg - mb_start); - if (bol) - mb_start = bol + 1; if (! wordchar_prev (mb_start, beg, buf + size)) for (;;) { -- 2.24.0.155.gd9f6f3b619