From 1b091165d1ed2d1ed9e575bfab4f1b1808a85f04 Mon Sep 17 00:00:00 2001 From: Jim Meyering Date: Sat, 18 Mar 2023 23:25:03 -0700 Subject: [PATCH] grep: -P (--perl-regexp) \D once again works lie [^0-9] * NEWS: Mention \D, too. * doc/grep.texi: Likewise * src/grep.c (pcre_pattern_expand_backslash_d): Handle \D. * tests/pcre-ascii-digits: Test \D, too. Add comments. Tighten one test by using returns_ 1. Reported by Paul Eggert in https://bugs.gnu.org/62267#8 --- NEWS | 2 +- doc/grep.texi | 1 + src/grep.c | 7 +++++-- tests/pcre-ascii-digits | 9 ++++++++- 4 files changed, 15 insertions(+), 4 deletions(-) diff --git a/NEWS b/NEWS index a24cebd..6f77d16 100644 --- a/NEWS +++ b/NEWS @@ -9,7 +9,7 @@ GNU grep NEWS -*- outline -*- properly had the undesirable side effect of making \d also match e.g., the Arabic digits: ٠١٢٣٤٥٦٧٨٩. With grep-3.9, -P '\d+' would match that ten-digit (20-byte) string. Now, to match such - a digit, you would use \p{Nd}. + a digit, you would use \p{Nd}. Similarly, \D is now mapped to [^0-9]. [bug introduced in grep 3.9] diff --git a/doc/grep.texi b/doc/grep.texi index eaad6e1..ad034f1 100644 --- a/doc/grep.texi +++ b/doc/grep.texi @@ -1149,6 +1149,7 @@ in-regexp directives like @samp{(?aD)}. Use @samp{\p@{Nd@}} if you require to match non-ASCII digits. Once pcre2 support for @samp{(?aD)} is widespread enough, we expect to make that the default, so it will be overridable. +Similarly, @samp{\D} matches anything but those ten ASCII digits. @c Using pcre2 git commit pcre2-10.40-112-g6277357, this demonstrates how @c we'll prefix with (?aD) to make \d's ASCII-only behavior the default: @c $ LC_ALL=en_US.UTF-8 ./pcre2grep -u '(?aD)^\d+' <<< '٠١٢٣٤٥٦٧٨٩' diff --git a/src/grep.c b/src/grep.c index 6ba881e..79459f3 100644 --- a/src/grep.c +++ b/src/grep.c @@ -2382,14 +2382,14 @@ fgrep_to_grep_pattern (char **keys_p, idx_t *len_p) /* Replace each \d in *KEYS_P with [0-9], to ensure that \d matches only ASCII digits. Now that we enable PCRE2_UCP for pcre regexps, \d would otherwise match non-ASCII digits in some locales. Use \p{Nd} if you require to match - those. */ + those. Similarly, replace each \D with [^0-9]. */ static void pcre_pattern_expand_backslash_d (char **keys_p, idx_t *len_p) { idx_t len = *len_p; char *keys = *keys_p; mbstate_t mb_state = { 0 }; - char *new_keys = xnmalloc (len / 2 + 1, 5); + char *new_keys = xnmalloc (len / 2 + 1, 6); char *p = new_keys; bool prev_backslash = false; @@ -2428,6 +2428,9 @@ pcre_pattern_expand_backslash_d (char **keys_p, idx_t *len_p) case 'd': p = mempcpy (p, "[0-9]", 5); break; + case 'D': + p = mempcpy (p, "[^0-9]", 6); + break; default: *p++ = '\\'; *p++ = *keys; diff --git a/tests/pcre-ascii-digits b/tests/pcre-ascii-digits index ae713f7..0159286 100755 --- a/tests/pcre-ascii-digits +++ b/tests/pcre-ascii-digits @@ -1,6 +1,7 @@ #!/bin/sh # Ensure that grep -P's \d matches only the 10 ASCII digits. # With, grep-3.9, \d would match e.g., the multibyte Arabic digits. +# The same applied to \D. # # Copyright (C) 2023 Free Software Foundation, Inc. # @@ -24,8 +25,14 @@ fail=0 # \331\245\331\246\331\247\331\250\331\251 printf '\331\240\331\241\331\242\331\243\331\244' > in || framework_failure_ printf '\331\245\331\246\331\247\331\250\331\251' >> in || framework_failure_ +printf '\n' >> in || framework_failure_ -grep -P '\d+' in > out && fail=1 +# Ensure that \d matches no character. +returns_ 1 grep -P '\d' in > out || fail=1 compare /dev/null out || fail=1 +# Ensure that ^\D+$ matches the entire line. +grep -P '^\D+$' in > out || fail=1 +compare in out || fail=1 + Exit $fail -- 2.40.0.rc2