From: Paul Eggert <eggert@cs.ucla.edu>
Date: Tue, 26 Jan 2021 17:23:54 +0000 (-0800)
Subject: expr: fix bug with unmatched \(...\)
X-Git-Tag: v9.0~153
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=735083ba24878075235007b4417982ad5700436d;p=thirdparty%2Fcoreutils.git

expr: fix bug with unmatched \(...\)

Problem reported by Qiuhao Li.
* NEWS: Mention this.
* doc/coreutils.texi (String expressions):
Document the correct behavior, which POSIX requires.
* src/expr.c (docolon): Treat unmatched \(...\) as empty.
* tests/misc/expr.pl: New test.
---

diff --git a/NEWS b/NEWS
index 0929f0ae2f..351a2983aa 100644
--- a/NEWS
+++ b/NEWS
@@ -17,6 +17,9 @@ GNU coreutils NEWS                                    -*- outline -*-
   heavily changed during the run.
   [bug introduced in coreutils-8.25]
 
+  expr no longer mishandles unmatched \(...\) in regular expressions.
+  [bug introduced in coreutils-6.0]
+
   ls no longer crashes when printing the SELinux context for unstatable files.
   [bug introduced in coreutils-6.9.91]
 
diff --git a/doc/coreutils.texi b/doc/coreutils.texi
index 94c9fbfa56..c90c4d5128 100644
--- a/doc/coreutils.texi
+++ b/doc/coreutils.texi
@@ -13559,12 +13559,14 @@ second is considered to be a (basic, a la GNU @code{grep}) regular
 expression, with a @code{^} implicitly prepended.  The first argument is
 then matched against this regular expression.
 
-If the match succeeds and @var{regex} uses @samp{\(} and @samp{\)}, the
-@code{:} expression returns the part of @var{string} that matched the
-subexpression; otherwise, it returns the number of characters matched.
-
-If the match fails, the @code{:} operator returns the null string if
-@samp{\(} and @samp{\)} are used in @var{regex}, otherwise 0.
+If @var{regex} does not use @samp{\(} and @samp{\)}, the @code{:}
+expression returns the number of characters matched, or 0 if the match
+fails.
+
+If @var{regex} uses @samp{\(} and @samp{\)}, the @code{:} expression
+returns the part of @var{string} that matched the subexpression, or
+the null string if the match failed or the subexpression did not
+contribute to the match.
 
 @kindex \( @r{regexp operator}
 Only the first @samp{\( @dots{} \)} pair is relevant to the return
diff --git a/src/expr.c b/src/expr.c
index afd8b9ca56..4893d948f3 100644
--- a/src/expr.c
+++ b/src/expr.c
@@ -614,8 +614,13 @@ docolon (VALUE *sv, VALUE *pv)
       /* Were \(...\) used? */
       if (re_buffer.re_nsub > 0)
         {
-          sv->u.s[re_regs.end[1]] = '\0';
-          v = str_value (sv->u.s + re_regs.start[1]);
+          if (re_regs.end[1] < 0)
+            v = str_value ("");
+          else
+            {
+              sv->u.s[re_regs.end[1]] = '\0';
+              v = str_value (sv->u.s + re_regs.start[1]);
+            }
         }
       else
         {
diff --git a/tests/misc/expr.pl b/tests/misc/expr.pl
index 31f903aff1..0e547af9fd 100755
--- a/tests/misc/expr.pl
+++ b/tests/misc/expr.pl
@@ -84,6 +84,9 @@ my @Tests =
      # In 5.94 and earlier, anchors incorrectly matched newlines.
      ['anchor', "'a\nb' : 'a\$'", {OUT => '0'}, {EXIT => 1}],
 
+     # In 8.32, \( ... \) that did not match caused memory errors.
+     ['emptysub', '"a" : "\\(b\\)*"', {OUT => ''}, {EXIT => 1}],
+
      # These tests are taken from grep/tests/bre.tests.
      ['bre1', '"abc" : "a\\(b\\)c"', {OUT => 'b'}],
      ['bre2', '"a(" : "a("', {OUT => '2'}],