Paul Eggert wrote:
> >> * What happens when strings contain encoding errors? It's not clear from
> >> the spec. I hope behavior isn't simply undefined.
> >
> > When the str_* functions are used, the byte-wise encoding will matter.
>
> I thought that str_* functions didn't care about locale, which means the
> character encoding does not matter for them.
Yes, that's what I meant. Sorry for the misunderstanding.
> > When the mbiter primitives are used, recall that they cope with
> > encoding errors (via the 'bool cur.wc_valid'); thus I expect that
> > encoding errors in the range of the suffix will match if it's the
> > same encoding error in both argument strings.
>
> Don't we have problems with mbs_startswith, though? If the prefix ends
> in an incomplete multibyte character (an encoding error), ...
Oh, I now see what you mean.
mbs_endswith needs a tweak to support that case; patch below.
mbs_startswith needs code with mbiter as well, for this case. Still working
on that...
2025-01-04 Bruno Haible <br...@clisp.org>
mbs_endswith: Fix abort in the case of incomplete characters.
Reported by Paul Eggert.
* lib/mbs_endswith.c: Don't include <stdlib.h>.
(mbs_endswith): Instead of aborting, return false.
* tests/test-mbs_endswith2.c (main): Test invalid and incomplete
characters.
diff --git a/lib/mbs_endswith.c b/lib/mbs_endswith.c
index 8f163e1aa5..00d5128d43 100644
--- a/lib/mbs_endswith.c
+++ b/lib/mbs_endswith.c
@@ -23,8 +23,6 @@
#include "mbiter.h"
-#include <stdlib.h>
-
bool
mbs_endswith (const char *string, const char *suffix)
{
@@ -62,13 +60,15 @@ mbs_endswith (const char *string, const char *suffix)
for (; len > n; len--)
{
if (!mbi_avail (iter))
- abort ();
+ /* We can get here due to incomplete multibyte characters. */
+ return false;
mbi_advance (iter);
}
if (!mbi_avail (iter))
- abort ();
+ /* We can get here due to incomplete multibyte characters. */
+ return false;
return strcmp (mbi_cur_ptr (iter), suffix) == 0;
}
}
- return 0;
+ return false;
}
diff --git a/tests/test-mbs_endswith2.c b/tests/test-mbs_endswith2.c
index f76f9b04e4..35c3b8652d 100644
--- a/tests/test-mbs_endswith2.c
+++ b/tests/test-mbs_endswith2.c
@@ -62,5 +62,29 @@ main ()
ASSERT (mbs_endswith ("\341\272\213\303\277\341\272\221",
"\341\272\213\303\277\341\272\221")); /* "ẋÿẑ" "ẋÿẑ" */
ASSERT (mbs_endswith
("\303\277\341\272\213\341\272\213\303\277\341\272\221",
"\341\272\213\303\277\341\272\221")); /* "ÿẋẋÿẑ" "ẋÿẑ" */
+ /* Test cases with invalid or incomplete characters. */
+
+ /* A valid character should not match an invalid character. */
+ ASSERT (!mbs_endswith ("\303\247", "\301\247"));
+ ASSERT (!mbs_endswith ("\301\247", "\303\247"));
+
+ /* A valid character should not match an incomplete character. */
+ ASSERT (!mbs_endswith ("\303\247", "\343\247"));
+ ASSERT (!mbs_endswith ("\343\247", "\303\247"));
+
+ /* An invalid character should not match an incomplete character. */
+ ASSERT (!mbs_endswith ("\301\247", "\343\247"));
+ ASSERT (!mbs_endswith ("\343\247", "\301\247"));
+
+ /* Two invalid characters should match only if they are identical. */
+ ASSERT (!mbs_endswith ("\301\246", "\301\247"));
+ ASSERT (!mbs_endswith ("\301\247", "\301\246"));
+ ASSERT (mbs_endswith ("\301\247", "\301\247"));
+
+ /* Two incomplete characters should match only if they are identical. */
+ ASSERT (!mbs_endswith ("\343\246", "\343\247"));
+ ASSERT (!mbs_endswith ("\343\247", "\343\246"));
+ ASSERT (mbs_endswith ("\343\247", "\343\247"));
+
return test_exit_status;
}
