1
0
mirror of https://git.savannah.gnu.org/git/gnulib.git synced 2026-10-24 00:48:56 +03:00
Files
gnulib/tests/test-mbscspn.c
Bruno Haible 1e7cbc30fd mbuiterf: Implement multi-byte per encoding error (MEE) consistently.
* lib/mbuiterf.h: Include mbiter-aux.h.
(struct mbuif_state): Add field is_utf8.
(mbuiterf_next): Invoke mbiter_is_utf8, mbiter_utf8_maximal_subpart.
(mbuif_init): Initialize the field is_utf8.
* modules/mbuiterf (Depends-on): Add mbiter-aux.
* tests/test-mbslen.c (OR): New macro, copied from
tests/test-mbsnlen.c.
(main): Add more test cases with incomplete characters.
* tests/test-mbschr2.sh: Renamed from tests/test-mbschr.sh.
* tests/test-mbschr2.c: Renamed from tests/test-mbschr.c.
* tests/test-mbschr1.sh: New file, based on
tests/test-mbmemcasecmp-3.sh.
* tests/test-mbschr1.c: New file.
* modules/mbschr-tests (Files): Update accordingly. Add locale-en.m4,
locale-fr.m4.
(configure.ac): Invoke gt_LOCALE_EN_UTF8, gt_LOCALE_FR_UTF8.
(Makefile.am): Arrange to compile test-mbschr1 and test-mbschr2 and to
run test-mbschr1.sh, test-mbschr2.sh.
* tests/test-mbsrchr2.sh: Renamed from tests/test-mbsrchr.sh.
* tests/test-mbsrchr2.c: Renamed from tests/test-mbsrchr.c.
* tests/test-mbsrchr1.sh: New file, based on
tests/test-mbmemcasecmp-3.sh.
* tests/test-mbsrchr1.c: New file.
* modules/mbsrchr-tests (Files): Update accordingly. Add locale-en.m4,
locale-fr.m4.
(configure.ac): Invoke gt_LOCALE_EN_UTF8, gt_LOCALE_FR_UTF8.
(Makefile.am): Arrange to compile test-mbsrchr1 and test-mbsrchr2 and to
run test-mbsrchr1.sh, test-mbsrchr2.sh.
* tests/test-mbscspn.c (OR): New macro, copied from
tests/test-mbsnlen.c.
(main): Add test cases with incomplete characters.
* tests/test-mbspbrk.c (OR): New macro, copied from
tests/test-mbsnlen.c.
(main): Add test cases with incomplete characters.
* tests/test-mbsspn.c (OR): New macro, copied from
tests/test-mbsnlen.c.
(main): Add test cases with incomplete characters.
2026-05-26 01:31:23 +02:00

140 lines
4.2 KiB
C

/* Test of searching a string for a character among a given set of characters.
Copyright (C) 2007-2026 Free Software Foundation, Inc.
This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see <https://www.gnu.org/licenses/>. */
/* Written by Bruno Haible <bruno@clisp.org>, 2007. */
#include <config.h>
#include <string.h>
#include <locale.h>
#include "macros.h"
/* The mcel-based implementation of mbsnlen behaves differently than the
original one. Namely, for invalid/incomplete byte sequences:
Where we ideally should have multi-byte-per-encoding-error (MEE) behaviour
everywhere, mcel implements single-byte-per-encoding-error (SEE) behaviour.
See <https://lists.gnu.org/archive/html/bug-gnulib/2023-07/msg00131.html>,
<https://lists.gnu.org/archive/html/bug-gnulib/2023-07/msg00145.html>.
Therefore, here we have different expected results, depending on the
implementation. */
#if GNULIB_MCEL_PREFER
# define OR(a,b) b
#else
# define OR(a,b) a
#endif
int
main ()
{
/* configure should already have checked that the locale is supported. */
if (setlocale (LC_ALL, "") == NULL)
return 1;
ASSERT (mbscspn ("Some text", "") == 9);
{
const char input[] = "A long sentence";
ASSERT (mbscspn (input, "aeiou") == 3);
ASSERT (mbscspn (input, "iI") == 15);
}
/* The following tests shows how mbscspn() is different from strcspn(). */
{
const char input[] = "B\303\266se B\303\274bchen"; /* "Böse Bübchen" */
ASSERT (mbscspn (input, "\303\244\303\274") == 7); /* "äü" */
}
#if 0 /* This would be for a GB18030 locale. */
{
const char input[] = "B\201\060\213\062se B\250\271bchen"; /* "Böse Bübchen" */
ASSERT (mbscspn (input, "\201\060\212\061\250\271") == 9); /* "äü" */
}
#endif
{
const char input[] = "B\303\266se B\303\274bchen"; /* "Böse Bübchen" */
ASSERT (mbscspn (input, "\303") == 14); /* invalid multibyte sequence */
}
/* Incomplete characters. See
https://www.unicode.org/versions/Unicode15.0.0/ch03.pdf
page 128 table 3-11. */
/* "\341\200\240" = 0xE1 0x80 0xA0 = U+1020. */
{
const char input[] = "\341\200\240x\341\200y";
ASSERT (mbscspn (input, "\341\200") == 4);
}
{
const char input[] = "\341\200\240x\341\200";
ASSERT (mbscspn (input, "\341\200") == 4);
}
{
const char input[] = "\341\200\240x\341\200";
ASSERT (mbscspn (input, "\341") == OR(6,4));
}
{
const char input[] = "\341\200\240x\341y";
ASSERT (mbscspn (input, "\341") == 4);
}
{
const char input[] = "\341\200\240x\341";
ASSERT (mbscspn (input, "\341") == 4);
}
/* "\360\221\222\240" = 0xF0 0x91 0x92 0xA0 = U+114A0. */
{
const char input[] = "\360\221\222\240x\360\221\222y";
ASSERT (mbscspn (input, "\360\221\222") == 5);
}
{
const char input[] = "\360\221\222\240x\360\221\222";
ASSERT (mbscspn (input, "\360\221\222") == 5);
}
{
const char input[] = "\360\221\222\240x\360\221\222";
ASSERT (mbscspn (input, "\360\221") == OR(8,5));
}
{
const char input[] = "\360\221\222\240x\360\221y";
ASSERT (mbscspn (input, "\360\221") == 5);
}
{
const char input[] = "\360\221\222\240x\360\221";
ASSERT (mbscspn (input, "\360\221") == 5);
}
{
const char input[] = "\360\221\222\240x\360\221\222";
ASSERT (mbscspn (input, "\360") == OR(8,5));
}
{
const char input[] = "\360\221\222\240x\360\221";
ASSERT (mbscspn (input, "\360") == OR(7,5));
}
{
const char input[] = "\360\221\222\240x\360y";
ASSERT (mbscspn (input, "\360") == 5);
}
{
const char input[] = "\360\221\222\240x\360";
ASSERT (mbscspn (input, "\360") == 5);
}
return test_exit_status;
}