[bug-grep] Testing for UTF-8 bugs [was: Applying outstanding patches]

[Julian Foad]

Charles Levert wrote:
> * On Thursday 2005-04-28 at 17:27:06 +0100, Julian Foad wrote:
> > when the script is run from within "make check" the 
> > LANG environment variable does not get through from the calling line in my 
> > script to the function defined in the same script.
>
> Isn't using LANG instead of the higher priority
> LC_ALL risky in any case?

Good point - in fact, that was the problem (the test suite sets LC_ALL=C).  
Thanks.

Does this attached patch look right?

- Julian

Index: ChangeLog
===================================================================
RCS file: /cvsroot/grep/grep/ChangeLog,v
retrieving revision 1.237
diff -u -3 -p -d -r1.237 ChangeLog
--- ChangeLog   28 Apr 2005 15:17:13 -0000      1.237
+++ ChangeLog   28 Apr 2005 19:05:19 -0000
@@ -2,6 +2,10 @@
 
        * tests/foad1.sh: Remove Bash-specific syntax.
 
+       * src/dfa.c: Fix a bug whereby a bracket "[" was matched by the
+         pattern "[[:alpha:]]" in UTF-8 locales.  Patch #3800, by Tim Waugh.
+       * tests/foad1.sh: Add a regression test for that.
+
 2005-04-27  Julian Foad  <address@hidden>
 
        Fix a bug reported by Elliott Hughes in patch #1834 whereby "grep -Fw"
Index: src/dfa.c
===================================================================
RCS file: /cvsroot/grep/grep/src/dfa.c,v
retrieving revision 1.34
diff -u -3 -p -d -r1.34 dfa.c
--- src/dfa.c   11 Apr 2005 22:36:32 -0000      1.34
+++ src/dfa.c   28 Apr 2005 19:05:20 -0000
@@ -632,7 +632,7 @@ parse_bracket_exp_mb ()
                      work_mbc->coll_elems[work_mbc->ncoll_elems++] = elem;
                    }
                }
-             wc = WEOF;
+             wc1 = wc = WEOF;
            }
          else
            /* We treat '[' as a normal character here.  */
Index: tests/foad1.sh
===================================================================
RCS file: /cvsroot/grep/grep/tests/foad1.sh,v
retrieving revision 1.5
diff -u -3 -p -d -r1.5 foad1.sh
--- tests/foad1.sh      28 Apr 2005 15:17:14 -0000      1.5
+++ tests/foad1.sh      28 Apr 2005 19:05:21 -0000
@@ -80,4 +80,13 @@ grep_test "A/CX/B/C/" "A/B/C/" -wF -e A 
 grep_test "LIN7C 55327/" "" -wF -e 5327 -e 5532
 
 
+# Test character class erroneously matching a '[' character.
+# If the UTF-8 locale doesn't work, skip this test silently.
+if LC_ALL=cs_CZ.UTF-8 locale -k LC_CTYPE 2>/dev/null |
+  "${GREP}" -q "charmap.*UTF-8"
+then
+  LC_ALL=cs_CZ.UTF-8 grep_test "[/" "" "[[:alpha:]]" -E
+fi
+
+
 exit $failures