Home | History | Annotate | Line # | Download | only in locale
t_c8rtomb.c revision 1.2
      1 /*	$NetBSD: t_c8rtomb.c,v 1.2 2024/08/17 21:31:22 riastradh Exp $	*/
      2 
      3 /*-
      4  * Copyright (c) 2002 Tim J. Robbins
      5  * All rights reserved.
      6  *
      7  * Copyright (c) 2013 Ed Schouten <ed (at) FreeBSD.org>
      8  * All rights reserved.
      9  *
     10  * Redistribution and use in source and binary forms, with or without
     11  * modification, are permitted provided that the following conditions
     12  * are met:
     13  * 1. Redistributions of source code must retain the above copyright
     14  *    notice, this list of conditions and the following disclaimer.
     15  * 2. Redistributions in binary form must reproduce the above copyright
     16  *    notice, this list of conditions and the following disclaimer in the
     17  *    documentation and/or other materials provided with the distribution.
     18  *
     19  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
     20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
     23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     29  * SUCH DAMAGE.
     30  */
     31 /*
     32  * Test program for c8rtomb() as specified by C23.
     33  */
     34 
     35 #include <sys/cdefs.h>
     36 __RCSID("$NetBSD: t_c8rtomb.c,v 1.2 2024/08/17 21:31:22 riastradh Exp $");
     37 
     38 #include <errno.h>
     39 #include <limits.h>
     40 #include <locale.h>
     41 #include <stdio.h>
     42 #include <string.h>
     43 #include <uchar.h>
     44 
     45 #include <atf-c.h>
     46 
     47 static void
     48 require_lc_ctype(const char *locale_name)
     49 {
     50 	char *lc_ctype_set;
     51 
     52 	lc_ctype_set = setlocale(LC_CTYPE, locale_name);
     53 	if (lc_ctype_set == NULL)
     54 		atf_tc_fail("setlocale(LC_CTYPE, \"%s\") failed; errno=%d",
     55 		    locale_name, errno);
     56 
     57 	ATF_REQUIRE_EQ_MSG(strcmp(lc_ctype_set, locale_name), 0,
     58 	    "lc_ctype_set=%s locale_name=%s", lc_ctype_set, locale_name);
     59 }
     60 
     61 static mbstate_t s;
     62 static char buf[MB_LEN_MAX + 1];
     63 
     64 ATF_TC_WITHOUT_HEAD(c8rtomb_c_locale_test);
     65 ATF_TC_BODY(c8rtomb_c_locale_test, tc)
     66 {
     67 	size_t n;
     68 
     69 	require_lc_ctype("C");
     70 
     71 	/*
     72 	 * If the buffer argument is NULL, c8 is implicitly 0,
     73 	 * c8rtomb() resets its internal state.
     74 	 */
     75 	ATF_CHECK_EQ_MSG((n = c8rtomb(NULL, '\0', NULL)), 1, "n=%zu", n);
     76 	ATF_CHECK_EQ_MSG((n = c8rtomb(NULL, 0x80, NULL)), 1, "n=%zu", n);
     77 	ATF_CHECK_EQ_MSG((n = c8rtomb(NULL, 0xc0, NULL)), 1, "n=%zu", n);
     78 	ATF_CHECK_EQ_MSG((n = c8rtomb(NULL, 0xe0, NULL)), 1, "n=%zu", n);
     79 	ATF_CHECK_EQ_MSG((n = c8rtomb(NULL, 0xf0, NULL)), 1, "n=%zu", n);
     80 	ATF_CHECK_EQ_MSG((n = c8rtomb(NULL, 0xf8, NULL)), 1, "n=%zu", n);
     81 	ATF_CHECK_EQ_MSG((n = c8rtomb(NULL, 0xfc, NULL)), 1, "n=%zu", n);
     82 	ATF_CHECK_EQ_MSG((n = c8rtomb(NULL, 0xfe, NULL)), 1, "n=%zu", n);
     83 	ATF_CHECK_EQ_MSG((n = c8rtomb(NULL, 0xff, NULL)), 1, "n=%zu", n);
     84 
     85 	/* Null wide character. */
     86 	memset(&s, 0, sizeof(s));
     87 	memset(buf, 0xcc, sizeof(buf));
     88 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0, &s)), 1, "n=%zu", n);
     89 	ATF_CHECK_MSG(((unsigned char)buf[0] == 0 &&
     90 		(unsigned char)buf[1] == 0xcc),
     91 	    "buf=[%02x %02x]", buf[0], buf[1]);
     92 
     93 	/* Latin letter A, internal state. */
     94 	ATF_CHECK_EQ_MSG((n = c8rtomb(NULL, '\0', NULL)), 1, "n=%zu", n);
     95 	ATF_CHECK_EQ_MSG((n = c8rtomb(NULL, 'A', NULL)), 1, "n=%zu", n);
     96 
     97 	/* Latin letter A. */
     98 	memset(&s, 0, sizeof(s));
     99 	memset(buf, 0xcc, sizeof(buf));
    100 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 'A', &s)), 1, "n=%zu", n);
    101 	ATF_CHECK_MSG(((unsigned char)buf[0] == 'A' &&
    102 		(unsigned char)buf[1] == 0xcc),
    103 	    "buf=[%02x %02x]", buf[0], buf[1]);
    104 
    105 	/* Unicode character 'Pile of poo'. */
    106 	memset(&s, 0, sizeof(s));
    107 	memset(buf, 0xcc, sizeof(buf));
    108 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0xf0, &s)), 0, "n=%zu", n);
    109 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0x9f, &s)), 0, "n=%zu", n);
    110 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0x92, &s)), 0, "n=%zu", n);
    111 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0xa9, &s)), (size_t)-1,
    112 	    "n=%zu", n);
    113 	ATF_CHECK_EQ_MSG(errno, EILSEQ, "errno=%d", errno);
    114 	ATF_CHECK_EQ_MSG((unsigned char)buf[0], 0xcc, "buf=[%02x]", buf[0]);
    115 
    116 	/* Incomplete Unicode character 'Pile of poo', interrupted by NUL. */
    117 	memset(&s, 0, sizeof(s));
    118 	memset(buf, 0xcc, sizeof(buf));
    119 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0xf0, &s)), 0, "n=%zu", n);
    120 	atf_tc_expect_fail("PR lib/58615:"
    121 	    " incomplete c8rtomb, c16rtomb handles NUL termination wrong");
    122 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, '\0', &s)), 1, "n=%zu", n);
    123 	ATF_CHECK_MSG(((unsigned char)buf[0] == '\0' &&
    124 		(unsigned char)buf[1] == 0xcc),
    125 	    "buf=[%02x %02x]", buf[0], buf[1]);
    126 
    127 	memset(&s, 0, sizeof(s));
    128 	memset(buf, 0xcc, sizeof(buf));
    129 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0xf0, &s)), 0, "n=%zu", n);
    130 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0x9f, &s)), 0, "n=%zu", n);
    131 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, '\0', &s)), 1, "n=%zu", n);
    132 	ATF_CHECK_MSG(((unsigned char)buf[0] == '\0' &&
    133 		(unsigned char)buf[1] == 0xcc),
    134 	    "buf=[%02x %02x]", buf[0], buf[1]);
    135 
    136 	memset(&s, 0, sizeof(s));
    137 	memset(buf, 0xcc, sizeof(buf));
    138 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0xf0, &s)), 0, "n=%zu", n);
    139 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0x9f, &s)), 0, "n=%zu", n);
    140 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0x92, &s)), 0, "n=%zu", n);
    141 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, '\0', &s)), 1, "n=%zu", n);
    142 	ATF_CHECK_MSG(((unsigned char)buf[0] == '\0' &&
    143 		(unsigned char)buf[1] == 0xcc),
    144 	    "buf=[%02x %02x]", buf[0], buf[1]);
    145 }
    146 
    147 ATF_TC_WITHOUT_HEAD(c8rtomb_iso_8859_1_test);
    148 ATF_TC_BODY(c8rtomb_iso_8859_1_test, tc)
    149 {
    150 	size_t n;
    151 
    152 	require_lc_ctype("en_US.ISO8859-1");
    153 
    154 	/* Unicode character 'Euro sign'. */
    155 	memset(&s, 0, sizeof(s));
    156 	memset(buf, 0xcc, sizeof(buf));
    157 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0xe2, &s)), 0, "n=%zu", n);
    158 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0x82, &s)), 0, "n=%zu", n);
    159 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0xac, &s)), (size_t)-1,
    160 	    "n=%zu", n);
    161 	ATF_CHECK_EQ_MSG(errno, EILSEQ, "errno=%d", errno);
    162 	ATF_CHECK_EQ_MSG((unsigned char)buf[0], 0xcc, "buf=[%02x]", buf[0]);
    163 }
    164 
    165 ATF_TC_WITHOUT_HEAD(c8rtomb_iso_8859_15_test);
    166 ATF_TC_BODY(c8rtomb_iso_8859_15_test, tc)
    167 {
    168 	size_t n;
    169 
    170 	require_lc_ctype("en_US.ISO8859-15");
    171 
    172 	/* Unicode character 'Euro sign'. */
    173 	memset(&s, 0, sizeof(s));
    174 	memset(buf, 0xcc, sizeof(buf));
    175 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0xe2, &s)), 0, "n=%zu", n);
    176 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0x82, &s)), 0, "n=%zu", n);
    177 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0xac, &s)), 1, "n=%zu", n);
    178 	ATF_CHECK_MSG(((unsigned char)buf[0] == 0xa4 &&
    179 		(unsigned char)buf[1] == 0xcc),
    180 	    "buf=[%02x %02x]", buf[0], buf[1]);
    181 }
    182 
    183 ATF_TC_WITHOUT_HEAD(c8rtomb_utf_8_test);
    184 ATF_TC_BODY(c8rtomb_utf_8_test, tc)
    185 {
    186 	size_t n;
    187 
    188 	require_lc_ctype("en_US.UTF-8");
    189 
    190 	/* Unicode character 'Pile of poo'. */
    191 	memset(&s, 0, sizeof(s));
    192 	memset(buf, 0xcc, sizeof(buf));
    193 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0xf0, &s)), 0, "n=%zu", n);
    194 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0x9f, &s)), 0, "n=%zu", n);
    195 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0x92, &s)), 0, "n=%zu", n);
    196 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0xa9, &s)), 4, "n=%zu", n);
    197 	ATF_CHECK_MSG(((unsigned char)buf[0] == 0xf0 &&
    198 		(unsigned char)buf[1] == 0x9f &&
    199 		(unsigned char)buf[2] == 0x92 &&
    200 		(unsigned char)buf[3] == 0xa9 &&
    201 		(unsigned char)buf[4] == 0xcc),
    202 	    "buf=[%02x %02x %02x %02x %02x]",
    203 	    buf[0], buf[1], buf[2], buf[3], buf[4]);
    204 
    205 	/* Invalid code; 'Pile of poo' without the last byte. */
    206 	memset(&s, 0, sizeof(s));
    207 	memset(buf, 0xcc, sizeof(buf));
    208 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0xf0, &s)), 0, "n=%zu", n);
    209 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0x9f, &s)), 0, "n=%zu", n);
    210 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0x92, &s)), 0, "n=%zu", n);
    211 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 'A', &s)), (size_t)-1,
    212 	    "n=%zu", n);
    213 	ATF_CHECK_EQ_MSG(errno, EILSEQ, "errno=%d", errno);
    214 	ATF_CHECK_EQ_MSG((unsigned char)buf[0], 0xcc, "buf=[%02x]", buf[0]);
    215 
    216 	/* Invalid code; 'Pile of poo' without the first byte. */
    217 	memset(&s, 0, sizeof(s));
    218 	memset(buf, 0xcc, sizeof(buf));
    219 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0x9f, &s)), (size_t)-1,
    220 	    "n=%zu", n);
    221 	ATF_CHECK_EQ_MSG(errno, EILSEQ, "errno=%d", errno);
    222 	ATF_CHECK_EQ_MSG((unsigned char)buf[0], 0xcc, "buf=[%02x]", buf[0]);
    223 
    224 	/* Incomplete Unicode character 'Pile of poo', interrupted by NUL. */
    225 	memset(&s, 0, sizeof(s));
    226 	memset(buf, 0xcc, sizeof(buf));
    227 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0xf0, &s)), 0, "n=%zu", n);
    228 	atf_tc_expect_fail("PR lib/58615:"
    229 	    " incomplete c8rtomb, c16rtomb handles NUL termination wrong");
    230 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, '\0', &s)), 1, "n=%zu", n);
    231 	ATF_CHECK_MSG(((unsigned char)buf[0] == '\0' &&
    232 		(unsigned char)buf[1] == 0xcc),
    233 	    "buf=[%02x %02x]", buf[0], buf[1]);
    234 
    235 	memset(&s, 0, sizeof(s));
    236 	memset(buf, 0xcc, sizeof(buf));
    237 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0xf0, &s)), 0, "n=%zu", n);
    238 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0x9f, &s)), 0, "n=%zu", n);
    239 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, '\0', &s)), 1, "n=%zu", n);
    240 	ATF_CHECK_MSG(((unsigned char)buf[0] == '\0' &&
    241 		(unsigned char)buf[1] == 0xcc),
    242 	    "buf=[%02x %02x]", buf[0], buf[1]);
    243 
    244 	memset(&s, 0, sizeof(s));
    245 	memset(buf, 0xcc, sizeof(buf));
    246 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0xf0, &s)), 0, "n=%zu", n);
    247 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0x9f, &s)), 0, "n=%zu", n);
    248 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, 0x92, &s)), 0, "n=%zu", n);
    249 	ATF_CHECK_EQ_MSG((n = c8rtomb(buf, '\0', &s)), 1, "n=%zu", n);
    250 	ATF_CHECK_MSG(((unsigned char)buf[0] == '\0' &&
    251 		(unsigned char)buf[1] == 0xcc),
    252 	    "buf=[%02x %02x]", buf[0], buf[1]);
    253 }
    254 
    255 ATF_TP_ADD_TCS(tp)
    256 {
    257 
    258 	ATF_TP_ADD_TC(tp, c8rtomb_c_locale_test);
    259 	ATF_TP_ADD_TC(tp, c8rtomb_iso_8859_1_test);
    260 	ATF_TP_ADD_TC(tp, c8rtomb_iso_8859_15_test);
    261 	ATF_TP_ADD_TC(tp, c8rtomb_utf_8_test);
    262 
    263 	return (atf_no_error());
    264 }
    265