lh | 9ed821d | 2023-04-07 01:36:19 -0700 | [diff] [blame] | 1 | #define _GNU_SOURCE 1 |
| 2 | |
| 3 | #include <fcntl.h> |
| 4 | #include <locale.h> |
| 5 | #include <regex.h> |
| 6 | #include <stdio.h> |
| 7 | #include <stdlib.h> |
| 8 | #include <string.h> |
| 9 | #include <sys/stat.h> |
| 10 | #include <sys/time.h> |
| 11 | #include <time.h> |
| 12 | #include <unistd.h> |
| 13 | #include <errno.h> |
| 14 | |
| 15 | static int |
| 16 | do_test(void) |
| 17 | { |
| 18 | static const char *pat[] = { |
| 19 | ".?.?.?.?.?.?.?Log\\.13", |
| 20 | "(.?)(.?)(.?)(.?)(.?)(.?)(.?)Log\\.13", |
| 21 | "((((((((((.?))))))))))((((((((((.?))))))))))((((((((((.?))))))))))" |
| 22 | "((((((((((.?))))))))))((((((((((.?))))))))))((((((((((.?))))))))))" |
| 23 | "((((((((((.?))))))))))Log\\.13" |
| 24 | }; |
| 25 | char *buf, *string; |
| 26 | const char *fname = "tst-regex2.dat"; |
| 27 | struct stat st; |
| 28 | unsigned len; |
| 29 | int testno; |
| 30 | int exitcode = 0; |
| 31 | |
| 32 | int fd = open(fname, O_RDONLY); |
| 33 | if (fd < 0) { |
| 34 | printf("Couldn't open %s: %s\n", fname, strerror(errno)); |
| 35 | return 1; |
| 36 | } |
| 37 | if (fstat(fd, &st) < 0) { |
| 38 | printf("Couldn't fstat %s: %s\n", fname, strerror(errno)); |
| 39 | return 1; |
| 40 | } |
| 41 | len = st.st_size; |
| 42 | string = buf = malloc(len + 1); |
| 43 | if (buf == NULL) { |
| 44 | printf("Couldn't allocate %u bytes\n", len + 1); |
| 45 | return 1; |
| 46 | } |
| 47 | if (read(fd, buf, st.st_size) != (ssize_t) st.st_size) { |
| 48 | printf("Couldn't read %s\n", fname); |
| 49 | return 1; |
| 50 | } |
| 51 | |
| 52 | close(fd); |
| 53 | buf[len] = '\0'; |
| 54 | |
| 55 | #if defined __UCLIBC_HAS_XLOCALE__ || !defined __UCLIBC__ |
| 56 | setlocale(LC_ALL, "de_DE.UTF-8"); |
| 57 | #endif |
| 58 | |
| 59 | for (testno = 0; testno < 2; ++testno) { |
| 60 | int i; |
| 61 | for (i = 0; i < sizeof(pat) / sizeof(pat[0]); ++i) { |
| 62 | struct timeval start, stop; |
| 63 | regex_t rbuf; |
| 64 | int err; |
| 65 | |
| 66 | printf("test %d pattern %d '%s'\n", testno, i, pat[i]); |
| 67 | gettimeofday(&start, NULL); |
| 68 | |
| 69 | err = regcomp(&rbuf, pat[i], |
| 70 | REG_EXTENDED | (testno ? REG_NOSUB : 0)); |
| 71 | if (err != 0) { |
| 72 | char errstr[300]; |
| 73 | regerror(err, &rbuf, errstr, sizeof(errstr)); |
| 74 | puts(errstr); |
| 75 | exitcode = 1; |
| 76 | goto contin1; |
| 77 | } |
| 78 | |
| 79 | regmatch_t pmatch[71]; |
| 80 | err = regexec(&rbuf, string, 71, pmatch, 0); |
| 81 | if (err == REG_NOMATCH) { |
| 82 | puts("regexec failed"); |
| 83 | exitcode = 1; |
| 84 | goto contin1; |
| 85 | } |
| 86 | |
| 87 | if (testno == 0) { |
| 88 | if (pmatch[0].rm_eo != pmatch[0].rm_so + 13 |
| 89 | || pmatch[0].rm_eo > len |
| 90 | || pmatch[0].rm_so < len - 100 |
| 91 | || strncmp(string + pmatch[0].rm_so, |
| 92 | " ChangeLog.13 for earlier changes", |
| 93 | sizeof " ChangeLog.13 for earlier changes" - 1 |
| 94 | ) != 0 |
| 95 | ) { |
| 96 | puts("regexec without REG_NOSUB did not find the correct match"); |
| 97 | exitcode = 1; |
| 98 | goto contin1; |
| 99 | } |
| 100 | |
| 101 | if (i > 0) { |
| 102 | int j, k, l; |
| 103 | for (j = 0, l = 1; j < 7; ++j) { |
| 104 | for (k = 0; k < (i == 1 ? 1 : 10); ++k, ++l) { |
| 105 | if (pmatch[l].rm_so != pmatch[0].rm_so + j |
| 106 | || pmatch[l].rm_eo != pmatch[l].rm_so + 1 |
| 107 | ) { |
| 108 | printf("pmatch[%d] incorrect\n", l); |
| 109 | exitcode = 1; |
| 110 | goto contin1; |
| 111 | } |
| 112 | } |
| 113 | } |
| 114 | } |
| 115 | } |
| 116 | |
| 117 | gettimeofday(&stop, NULL); |
| 118 | stop.tv_sec -= start.tv_sec; |
| 119 | if (stop.tv_usec < start.tv_usec) { |
| 120 | stop.tv_sec--; |
| 121 | stop.tv_usec += 1000000; |
| 122 | } |
| 123 | stop.tv_usec -= start.tv_usec; |
| 124 | printf(" %lu.%06lus\n", (unsigned long) stop.tv_sec, |
| 125 | (unsigned long) stop.tv_usec); |
| 126 | contin1: |
| 127 | regfree(&rbuf); |
| 128 | } |
| 129 | } |
| 130 | |
| 131 | for (testno = 2; testno < 4; ++testno) { |
| 132 | int i; |
| 133 | for (i = 0; i < sizeof(pat) / sizeof(pat[0]); ++i) { |
| 134 | struct timeval start, stop; |
| 135 | struct re_pattern_buffer rpbuf; |
| 136 | struct re_registers regs; |
| 137 | const char *s; |
| 138 | int match; |
| 139 | |
| 140 | printf("test %d pattern %d '%s'\n", testno, i, pat[i]); |
| 141 | gettimeofday(&start, NULL); |
| 142 | |
| 143 | re_set_syntax(RE_SYNTAX_POSIX_EGREP |
| 144 | | (testno == 3 ? RE_NO_SUB : 0)); |
| 145 | memset(&rpbuf, 0, sizeof(rpbuf)); |
| 146 | s = re_compile_pattern(pat[i], strlen(pat[i]), &rpbuf); |
| 147 | if (s != NULL) { |
| 148 | printf("%s\n", s); |
| 149 | exitcode = 1; |
| 150 | goto contin2; |
| 151 | } |
| 152 | |
| 153 | memset(®s, 0, sizeof(regs)); |
| 154 | match = re_search(&rpbuf, string, len, 0, len, ®s); |
| 155 | if (match < 0) { |
| 156 | printf("re_search failed (err:%d)\n", match); |
| 157 | exitcode = 1; |
| 158 | goto contin2; |
| 159 | } |
| 160 | if (match + 13 > len) { |
| 161 | printf("re_search: match+13 > len (%d > %d)\n", match + 13, len); |
| 162 | exitcode = 1; |
| 163 | goto contin2; |
| 164 | } |
| 165 | if (match < len - 100) { |
| 166 | printf("re_search: match < len-100 (%d < %d)\n", match, len - 100); |
| 167 | exitcode = 1; |
| 168 | goto contin2; |
| 169 | } |
| 170 | if (strncmp(string + match, " ChangeLog.13 for earlier changes", |
| 171 | sizeof(" ChangeLog.13 for earlier changes") - 1 |
| 172 | ) != 0 |
| 173 | ) { |
| 174 | printf("re_search did not find the correct match" |
| 175 | "(found '%s' instead)\n", string + match); |
| 176 | exitcode = 1; |
| 177 | goto contin2; |
| 178 | } |
| 179 | |
| 180 | if (testno == 2) { |
| 181 | int expected = 72; |
| 182 | if (i == 0) |
| 183 | expected = 2; |
| 184 | if (i == 1) |
| 185 | expected = 9; |
| 186 | if (regs.num_regs != expected) { |
| 187 | printf("incorrect num_regs %d, expected %d\n", regs.num_regs, expected); |
| 188 | exitcode = 1; |
| 189 | goto contin2; |
| 190 | } |
| 191 | if (regs.start[0] != match || regs.end[0] != match + 13) { |
| 192 | printf("incorrect regs.{start,end}[0] = { %d, %d }," |
| 193 | " expected { %d, %d }\n", |
| 194 | regs.start[0], regs.end[0], |
| 195 | match, match + 13 |
| 196 | ); |
| 197 | exitcode = 1; |
| 198 | goto contin2; |
| 199 | } |
| 200 | if (regs.start[regs.num_regs - 1] != -1 |
| 201 | || regs.end[regs.num_regs - 1] != -1 |
| 202 | ) { |
| 203 | printf("incorrect regs.{start,end}[num_regs - 1] = { %d, %d }," |
| 204 | " expected { -1, -1 }\n", |
| 205 | regs.start[regs.num_regs - 1], regs.end[regs.num_regs - 1] |
| 206 | ); |
| 207 | exitcode = 1; |
| 208 | goto contin2; |
| 209 | } |
| 210 | |
| 211 | if (i > 0) { |
| 212 | int j, k, l; |
| 213 | for (j = 0, l = 1; j < 7; ++j) { |
| 214 | for (k = 0; k < (i == 1 ? 1 : 10); ++k, ++l) { |
| 215 | if (regs.start[l] != match + j |
| 216 | || regs.end[l] != match + j + 1 |
| 217 | ) { |
| 218 | printf("incorrect regs.{start,end}[%d] = { %d, %d }," |
| 219 | " expected { %d, %d }\n", |
| 220 | l, |
| 221 | regs.start[l], regs.end[l], |
| 222 | match + j, match + j + 1 |
| 223 | ); |
| 224 | exitcode = 1; |
| 225 | goto contin2; |
| 226 | } |
| 227 | } |
| 228 | } |
| 229 | } |
| 230 | } |
| 231 | |
| 232 | gettimeofday(&stop, NULL); |
| 233 | stop.tv_sec -= start.tv_sec; |
| 234 | if (stop.tv_usec < start.tv_usec) { |
| 235 | stop.tv_sec--; |
| 236 | stop.tv_usec += 1000000; |
| 237 | } |
| 238 | stop.tv_usec -= start.tv_usec; |
| 239 | printf(" %lu.%06lus\n", (unsigned long) stop.tv_sec, |
| 240 | (unsigned long) stop.tv_usec); |
| 241 | contin2: |
| 242 | regfree(&rpbuf); |
| 243 | } |
| 244 | } |
| 245 | return exitcode; |
| 246 | } |
| 247 | |
| 248 | #define TIMEOUT 20 |
| 249 | #define TEST_FUNCTION do_test() |
| 250 | #include "../test-skeleton.c" |