From f3b29e0d760dadcc11c6147fc082eb139eed8a82 Mon Sep 17 00:00:00 2001 From: Paul Eggert Date: Sat, 15 Aug 2026 10:19:34 -0700 Subject: [PATCH] tests: remove test for buggy mbrtowc These tests should no longer be needed, as we no longer restart mbrtowc after an incomplete scan. This effectively reverts commit 77a7c6a731acd4dea226b500c21dce17b2a94ece dated 2016-07-04. * init.cfg (require_valid_ja_shiftjis_locale_) (require_valid_ja_eucjp_locale_): Remove. All uses removed. * testsuite/local.mk (check_PROGRAMS): Remove testsuite/test-mbrtowc. (testsuite_test_mbrtowc_LDADD): Remove. * testsuite/test-mbrtowc.c: Remove. --- .gitignore | 1 - init.cfg | 24 ---- testsuite/invalid-mb-seq-UMR.sh | 3 - testsuite/local.mk | 3 +- testsuite/mb-charclass-non-utf8.sh | 3 - testsuite/test-mbrtowc.c | 170 ----------------------------- 6 files changed, 1 insertion(+), 203 deletions(-) delete mode 100644 testsuite/test-mbrtowc.c diff --git a/.gitignore b/.gitignore index 654ab74..fa4a419 100644 --- a/.gitignore +++ b/.gitignore @@ -60,7 +60,6 @@ /testsuite/eval.in2 /testsuite/get-mb-cur-max /testsuite/init.sh -/testsuite/test-mbrtowc Makefile Makefile.in gt-bug* diff --git a/init.cfg b/init.cfg index 9c8bc7d..e3f6178 100644 --- a/init.cfg +++ b/init.cfg @@ -99,30 +99,6 @@ require_ja_shiftjis_locale_() test -z "$LOCALE_JA_SJIS" && skip_ 'ja_JP shift-jis locale not found' } -# Ensure the implementation of mbrtowc can detect invalid -# multibyte shiftjis sequences. Otherwise, skip the test, to avoid -# false-alarms. -# "$1" should be the name of the SHIFT-JIS locale -# (as set by 'require_ja_shiftjis_locale_' above) -require_valid_ja_shiftjis_locale_() -{ - path_prepend_ ./testsuite - local n=$(printf '\203:' | LC_ALL="$1" test-mbrtowc) - test "x$n" = "x-2,-1" || skip_ "locale '$1' is buggy" -} - -# Ensure the implementation of mbrtowc can detect invalid -# multibyte eucJP sequences. Otherwise, skip the test, to avoid -# false-alarms. -# "$1" should be the name of the ja_JP.eucJP locale -# (as set in $LOCALE_JA by m4/locale-ja.m4) -require_valid_ja_eucjp_locale_() -{ - path_prepend_ . - local n=$(printf '\262C' | LC_ALL="$1" test-mbrtowc) - test "x$n" = "x-2,-1" || skip_ "locale '$1' is buggy" -} - # When testing on a Windows machine, sed might output # line endings as CR,LF (\r\n) pairs - which will # then fail comparison with the expected output files. diff --git a/testsuite/invalid-mb-seq-UMR.sh b/testsuite/invalid-mb-seq-UMR.sh index 7f0b36a..c43c9b9 100755 --- a/testsuite/invalid-mb-seq-UMR.sh +++ b/testsuite/invalid-mb-seq-UMR.sh @@ -23,9 +23,6 @@ require_valgrind_ test "$LOCALE_JA" = none && skip_ found no Japanese EUC locale -# Ensure the implementation is not buggy (skip otherwise) -require_valid_ja_eucjp_locale_ "$LOCALE_JA" - echo a > in || framework_failure_ printf 'b\262C\n' > exp || framework_failure_ LC_ALL=$LOCALE_JA valgrind --quiet --error-exitcode=1 \ diff --git a/testsuite/local.mk b/testsuite/local.mk index c350c49..0167194 100644 --- a/testsuite/local.mk +++ b/testsuite/local.mk @@ -127,9 +127,8 @@ TESTS = $(SEDTESTS) $(T) SEDTESTS = -check_PROGRAMS = testsuite/get-mb-cur-max testsuite/test-mbrtowc +check_PROGRAMS = testsuite/get-mb-cur-max testsuite_get_mb_cur_max_LDADD = lib/libsed.a $(INTLLIBS) -testsuite_test_mbrtowc_LDADD = lib/libsed.a $(INTLLIBS) # Note that the first lines are statements. They ensure that environment # variables that can perturb tests are unset or set to expected values. diff --git a/testsuite/mb-charclass-non-utf8.sh b/testsuite/mb-charclass-non-utf8.sh index e189885..4bfeb1c 100755 --- a/testsuite/mb-charclass-non-utf8.sh +++ b/testsuite/mb-charclass-non-utf8.sh @@ -24,9 +24,6 @@ print_ver_ sed # If found, LOCALE_JA_SJIS will contain the locale name. require_ja_shiftjis_locale_ -# Ensure the implementation is not buggy (skip otherwise) -require_valid_ja_shiftjis_locale_ "$LOCALE_JA_SJIS" - # This test uses two characters: # Unicode Character 'KATAKANA LETTER ZE' (U+30BC) # Unicode Character 'KATAKANA LETTER ZO' (U+30BE) diff --git a/testsuite/test-mbrtowc.c b/testsuite/test-mbrtowc.c deleted file mode 100644 index 38e8ca7..0000000 --- a/testsuite/test-mbrtowc.c +++ /dev/null @@ -1,170 +0,0 @@ -/* Auxiliary program to test mbrtowc(3) behaviour. - Copyright 2016-2026 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, 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 . */ - -/* Test the operating-system's native mbrtowc(3) function, - by feeding it multibyte sequences one byte at a time, - and reporting the result. - - The program prints the following values after each mbrtowc invocation, - separated by commas: - - -2 the octet is contributes to a valid yet incomplete multibyte sequence - in the current locale. - - -1 the octet causes an encoding error. - - 0 the octet represents a NUL byte - - 1 the octet is a valid single-byte character, OR - completes a valid multibyte sequence. - - Because the program invokes mbrtowc(3) byte-by-byte, the reported - result should never be larger than 1. - - Example of typical output with UTF-8 encoding - --------------------------------------------- - - The unicode character 'N-ARY SUMMATION' (U+2211), encoded in UTF-8 as: - hex: 0xE2 0x88 0x91 - oct: 342 210 211 - - Decoding the valid sequence byte-by-byte gives: - $ printf '\342\210\221' | LC_ALL=en_US.UTF-8 test-mbrtowc - -2,-2,1 - - '\210' is not a valid leading byte in UTF-8, - thus the first byte gives -1, and the 'X' is treated - as a valid single-byte character: - - $ printf '\210X' | LC_ALL=en_US.UTF-8 test-mbrtowc - -1,1 - - '\342' is a valid yet incomplete multibyte sequence. - Passing it to mbrtowc results in value '-2'. - The following value 'X' gives an encoding error '-1' - (as 'X' is not a valid trailing byte in a multibyte UTF-8 sequence): - - $ printf '\342X' | LC_ALL=en_US.UTF-8 test-mbrtowc - -2,-1 - - - Detecting implementation bugs in mbrtowc - ---------------------------------------- - - UTF-8 implementation is correct on most operating systems. - Other multibyte locales might present more difficulties. - An example is the Japanese SHIFT-JIS locale under Mac OS X. - NOTE: The locale is 'ja_JP.SJIS' under Mac OS X, 'ja_JP.shiftjis' - under Ubuntu. 'ja_JP.sjis' was also found on some systems. - - Using unicode character 'KATAKANA LETTER ZE' (U+30BC) - UTF-8: hex: 0xE3 0x82 0xBC - Shift-jis hex: 0x83 0x5B - oct: 203 133 - - The following is a valid multibyte sequence in SHIFT-JIS, - the first byte should result in '-2' (valid yet incomplete), - and the second byte should result in '1' (a valid multibyte sequence - completed): - - $ printf '\203\133' | LC_ALL=ja_JP.SJIS test-mbrtowc - -2,1 - - The following is an INVALID multibyte sequence in SHIFT-JIS - (The byte ':' is not valid as a second octet). - Buggy implementations will accept this as a valid multibyte sequence: - - # NOTE: this result indicates a buggy mbrtowc - $ printf '\203:' | LC_ALL=ja_JP.SJIS test-mbrtowc - -2,1 - - A correct implementations should report '-1' for the second byte (i.e. - an encoding error): - - $ printf '\203:' | LC_ALL=ja_JP.SJIS test-mbrtowc - -2,-1 - - - Expected results with correct implementations - --------------------------------------------- - - In GNU Sed some tests purposely use invalid multibyte sequences - to test sed's behaviour. A buggy implementation of mbrtowc - would result in false-alarm failures. - - The following are expected results in correct implementations: - (locale names are from Mac OS X): - - $ printf '\203\133' | LC_ALL=ja_JP.SJIS test-mbrtowc - -2,1 - $ printf '\203:' | LC_ALL=ja_JP.SJIS test-mbrtowc - -2,-1 - $ printf '\262C' | LC_ALL=ja_JP.eucJP test-mbrtowc - -2,-1 -*/ - -#include -#include -#include -#include -#include -#include - -#include "progname.h" - -/* stub replacement for non-standard err(3) */ -static int -die (const char *msg) -{ - error (0, 0, "%s: error: %s\n", program_name, msg); - exit (EXIT_FAILURE); -} - -int -main (int argc, char **argv) -{ - int c; - int first = 1; - - set_program_name (argv[0]); - if (!setlocale (LC_ALL, "")) - die ("failed to set locale"); - - while ((c = getchar ()) != EOF) - { - wchar_t wc; - char ch = (unsigned char) c; - int i = (int) mbrtowc (&wc, &ch, 1, NULL); - - if (!first) - putchar (','); - first = 0; - - printf ("%d", i); - } - - if (first) - die ("empty input"); - - putchar ('\n'); - - if (ferror (stdin)) - die ("read error"); - if (ferror (stdout) || fclose (stdout) != 0) - die ("write error"); - - exit (EXIT_SUCCESS); -}