1 #include <sys/cdefs.h> 2 __FBSDID("$FreeBSD$"); 3 4 #include <sys/types.h> 5 #include <assert.h> 6 #include <regex.h> 7 #include <stdio.h> 8 #include <stdlib.h> 9 #include <string.h> 10 #include <unistd.h> 11 12 #include "debug.ih" 13 #include "main.ih" 14 #include "split.ih" 15 16 char *progname; 17 int debug = 0; 18 int line = 0; 19 int status = 0; 20 21 int copts = REG_EXTENDED; 22 int eopts = 0; 23 regoff_t startoff = 0; 24 regoff_t endoff = 0; 25 26 27 /* 28 - main - do the simple case, hand off to regress() for regression 29 */ 30 int 31 main(int argc, char **argv) 32 { 33 regex_t re; 34 # define NS 10 35 regmatch_t subs[NS]; 36 char erbuf[100]; 37 int err; 38 size_t len; 39 int c; 40 int errflg = 0; 41 int i; 42 extern int optind; 43 extern char *optarg; 44 45 progname = argv[0]; 46 47 while ((c = getopt(argc, argv, "c:e:S:E:x")) != -1) 48 switch (c) { 49 case 'c': /* compile options */ 50 copts = options('c', optarg); 51 break; 52 case 'e': /* execute options */ 53 eopts = options('e', optarg); 54 break; 55 case 'S': /* start offset */ 56 startoff = (regoff_t)atoi(optarg); 57 break; 58 case 'E': /* end offset */ 59 endoff = (regoff_t)atoi(optarg); 60 break; 61 case 'x': /* Debugging. */ 62 debug++; 63 break; 64 case '?': 65 default: 66 errflg++; 67 break; 68 } 69 if (errflg) { 70 fprintf(stderr, "usage: %s ", progname); 71 fprintf(stderr, "[-c copt][-C][-d] [re]\n"); 72 exit(2); 73 } 74 75 if (optind >= argc) { 76 regress(stdin); 77 exit(status); 78 } 79 80 err = regcomp(&re, argv[optind++], copts); 81 if (err) { 82 len = regerror(err, &re, erbuf, sizeof(erbuf)); 83 fprintf(stderr, "error %s, %zu/%zu `%s'\n", 84 eprint(err), len, sizeof(erbuf), erbuf); 85 exit(status); 86 } 87 regprint(&re, stdout); 88 89 if (optind >= argc) { 90 regfree(&re); 91 exit(status); 92 } 93 94 if ((eopts & REG_STARTEND) != 0) { 95 subs[0].rm_so = startoff; 96 subs[0].rm_eo = strlen(argv[optind]) - endoff; 97 } 98 err = regexec(&re, argv[optind], (size_t)NS, subs, eopts); 99 if (err) { 100 len = regerror(err, &re, erbuf, sizeof(erbuf)); 101 fprintf(stderr, "error %s, %zu/%zu `%s'\n", 102 eprint(err), len, sizeof(erbuf), erbuf); 103 exit(status); 104 } 105 if ((copts & REG_NOSUB) == 0) { 106 len = (int)(subs[0].rm_eo - subs[0].rm_so); 107 if (subs[0].rm_so != -1) { 108 if (len != 0) 109 printf("match `%.*s'\n", (int)len, 110 argv[optind] + subs[0].rm_so); 111 else 112 printf("match `'@%.1s\n", 113 argv[optind] + subs[0].rm_so); 114 } 115 for (i = 1; i < NS; i++) 116 if (subs[i].rm_so != -1) 117 printf("(%d) `%.*s'\n", i, 118 (int)(subs[i].rm_eo - subs[i].rm_so), 119 argv[optind] + subs[i].rm_so); 120 } 121 exit(status); 122 } 123 124 /* 125 - regress - main loop of regression test 126 == void regress(FILE *in); 127 */ 128 void 129 regress(FILE *in) 130 { 131 char inbuf[1000]; 132 # define MAXF 10 133 char *f[MAXF]; 134 int nf; 135 int i; 136 char erbuf[100]; 137 size_t ne; 138 char *badpat = "invalid regular expression"; 139 # define SHORT 10 140 char *bpname = "REG_BADPAT"; 141 regex_t re; 142 143 while (fgets(inbuf, sizeof(inbuf), in) != NULL) { 144 line++; 145 if (inbuf[0] == '#' || inbuf[0] == '\n') 146 continue; /* NOTE CONTINUE */ 147 inbuf[strlen(inbuf)-1] = '\0'; /* get rid of stupid \n */ 148 if (debug) 149 fprintf(stdout, "%d:\n", line); 150 nf = split(inbuf, f, MAXF, "\t\t"); 151 if (nf < 3) { 152 fprintf(stderr, "bad input, line %d\n", line); 153 exit(1); 154 } 155 for (i = 0; i < nf; i++) 156 if (strcmp(f[i], "\"\"") == 0) 157 f[i] = ""; 158 if (nf <= 3) 159 f[3] = NULL; 160 if (nf <= 4) 161 f[4] = NULL; 162 try(f[0], f[1], f[2], f[3], f[4], options('c', f[1])); 163 if (opt('&', f[1])) /* try with either type of RE */ 164 try(f[0], f[1], f[2], f[3], f[4], 165 options('c', f[1]) &~ REG_EXTENDED); 166 } 167 168 ne = regerror(REG_BADPAT, (regex_t *)NULL, erbuf, sizeof(erbuf)); 169 if (strcmp(erbuf, badpat) != 0 || ne != strlen(badpat)+1) { 170 fprintf(stderr, "end: regerror() test gave `%s' not `%s'\n", 171 erbuf, badpat); 172 status = 1; 173 } 174 ne = regerror(REG_BADPAT, (regex_t *)NULL, erbuf, (size_t)SHORT); 175 if (strncmp(erbuf, badpat, SHORT-1) != 0 || erbuf[SHORT-1] != '\0' || 176 ne != strlen(badpat)+1) { 177 fprintf(stderr, "end: regerror() short test gave `%s' not `%.*s'\n", 178 erbuf, SHORT-1, badpat); 179 status = 1; 180 } 181 ne = regerror(REG_ITOA|REG_BADPAT, (regex_t *)NULL, erbuf, sizeof(erbuf)); 182 if (strcmp(erbuf, bpname) != 0 || ne != strlen(bpname) + 1) { 183 fprintf(stderr, "end: regerror() ITOA test gave `%s' not `%s'\n", 184 erbuf, bpname); 185 status = 1; 186 } 187 re.re_endp = bpname; 188 ne = regerror(REG_ATOI, &re, erbuf, sizeof(erbuf)); 189 if (atoi(erbuf) != (int)REG_BADPAT) { 190 fprintf(stderr, "end: regerror() ATOI test gave `%s' not `%ld'\n", 191 erbuf, (long)REG_BADPAT); 192 status = 1; 193 } else if (ne != strlen(erbuf) + 1) { 194 fprintf(stderr, "end: regerror() ATOI test len(`%s') = %ld\n", 195 erbuf, (long)REG_BADPAT); 196 status = 1; 197 } 198 } 199 200 /* 201 - try - try it, and report on problems 202 == void try(char *f0, char *f1, char *f2, char *f3, char *f4, int opts); 203 - opts: may not match f1 204 */ 205 void 206 try(char *f0, char *f1, char *f2, char *f3, char *f4, int opts) 207 { 208 regex_t re; 209 # define NSUBS 10 210 regmatch_t subs[NSUBS]; 211 # define NSHOULD 15 212 char *should[NSHOULD]; 213 char erbuf[100]; 214 size_t len; 215 int err, i, nshould; 216 char *grump; 217 char *type = (opts & REG_EXTENDED) ? "ERE" : "BRE"; 218 char f0copy[1000]; 219 char f2copy[1000]; 220 221 strcpy(f0copy, f0); 222 re.re_endp = (opts®_PEND) ? f0copy + strlen(f0copy) : NULL; 223 fixstr(f0copy); 224 err = regcomp(&re, f0copy, opts); 225 if (err != 0 && (!opt('C', f1) || err != efind(f2))) { 226 /* unexpected error or wrong error */ 227 len = regerror(err, &re, erbuf, sizeof(erbuf)); 228 fprintf(stderr, "%d: %s error %s, %zu/%zu `%s'\n", 229 line, type, eprint(err), len, sizeof(erbuf), erbuf); 230 status = 1; 231 } else if (err == 0 && opt('C', f1)) { 232 /* unexpected success */ 233 fprintf(stderr, "%d: %s should have given REG_%s\n", 234 line, type, f2); 235 status = 1; 236 err = 1; /* so we won't try regexec */ 237 } 238 239 if (err != 0) { 240 regfree(&re); 241 return; 242 } 243 244 strcpy(f2copy, f2); 245 fixstr(f2copy); 246 247 if (options('e', f1)®_STARTEND) { 248 if (strchr(f2, '(') == NULL || strchr(f2, ')') == NULL) 249 fprintf(stderr, "%d: bad STARTEND syntax\n", line); 250 subs[0].rm_so = strchr(f2, '(') - f2 + 1; 251 subs[0].rm_eo = strchr(f2, ')') - f2; 252 } 253 err = regexec(&re, f2copy, NSUBS, subs, options('e', f1)); 254 255 if (err != 0 && (f3 != NULL || err != REG_NOMATCH)) { 256 /* unexpected error or wrong error */ 257 len = regerror(err, &re, erbuf, sizeof(erbuf)); 258 fprintf(stderr, "%d: %s exec error %s, %zu/%zu `%s'\n", 259 line, type, eprint(err), len, sizeof(erbuf), erbuf); 260 status = 1; 261 } else if (err != 0) { 262 /* nothing more to check */ 263 } else if (f3 == NULL) { 264 /* unexpected success */ 265 fprintf(stderr, "%d: %s exec should have failed\n", 266 line, type); 267 status = 1; 268 err = 1; /* just on principle */ 269 } else if (opts®_NOSUB) { 270 /* nothing more to check */ 271 } else if ((grump = check(f2, subs[0], f3)) != NULL) { 272 fprintf(stderr, "%d: %s %s\n", line, type, grump); 273 status = 1; 274 err = 1; 275 } 276 277 if (err != 0 || f4 == NULL) { 278 regfree(&re); 279 return; 280 } 281 282 for (i = 1; i < NSHOULD; i++) 283 should[i] = NULL; 284 nshould = split(f4, should+1, NSHOULD-1, ","); 285 if (nshould == 0) { 286 nshould = 1; 287 should[1] = ""; 288 } 289 for (i = 1; i < NSUBS; i++) { 290 grump = check(f2, subs[i], should[i]); 291 if (grump != NULL) { 292 fprintf(stderr, "%d: %s $%d %s\n", line, 293 type, i, grump); 294 status = 1; 295 err = 1; 296 } 297 } 298 299 regfree(&re); 300 } 301 302 /* 303 - options - pick options out of a regression-test string 304 - type: 'c' - compile, 'e' - exec 305 == int options(int type, char *s); 306 */ 307 int 308 options(int type, char *s) 309 { 310 char *p; 311 int o = (type == 'c') ? copts : eopts; 312 char *legal = (type == 'c') ? "bisnmp" : "^$#tl"; 313 314 for (p = s; *p != '\0'; p++) 315 if (strchr(legal, *p) != NULL) 316 switch (*p) { 317 case 'b': 318 o &= ~REG_EXTENDED; 319 break; 320 case 'i': 321 o |= REG_ICASE; 322 break; 323 case 's': 324 o |= REG_NOSUB; 325 break; 326 case 'n': 327 o |= REG_NEWLINE; 328 break; 329 case 'm': 330 o &= ~REG_EXTENDED; 331 o |= REG_NOSPEC; 332 break; 333 case 'p': 334 o |= REG_PEND; 335 break; 336 case '^': 337 o |= REG_NOTBOL; 338 break; 339 case '$': 340 o |= REG_NOTEOL; 341 break; 342 case '#': 343 o |= REG_STARTEND; 344 break; 345 case 't': /* trace */ 346 o |= REG_TRACE; 347 break; 348 case 'l': /* force long representation */ 349 o |= REG_LARGE; 350 break; 351 case 'r': /* force backref use */ 352 o |= REG_BACKR; 353 break; 354 } 355 return(o); 356 } 357 358 /* 359 - opt - is a particular option in a regression string? 360 == int opt(int c, char *s); 361 */ 362 int /* predicate */ 363 opt(int c, char *s) 364 { 365 return(strchr(s, c) != NULL); 366 } 367 368 /* 369 - fixstr - transform magic characters in strings 370 == void fixstr(char *p); 371 */ 372 void 373 fixstr(char *p) 374 { 375 if (p == NULL) 376 return; 377 378 for (; *p != '\0'; p++) 379 if (*p == 'N') 380 *p = '\n'; 381 else if (*p == 'T') 382 *p = '\t'; 383 else if (*p == 'S') 384 *p = ' '; 385 else if (*p == 'Z') 386 *p = '\0'; 387 } 388 389 /* 390 - check - check a substring match 391 == char *check(char *str, regmatch_t sub, char *should); 392 */ 393 char * /* NULL or complaint */ 394 check(char *str, regmatch_t sub, char *should) 395 { 396 int len; 397 int shlen; 398 char *p; 399 static char grump[500]; 400 char *at = NULL; 401 402 if (should != NULL && strcmp(should, "-") == 0) 403 should = NULL; 404 if (should != NULL && should[0] == '@') { 405 at = should + 1; 406 should = ""; 407 } 408 409 /* check rm_so and rm_eo for consistency */ 410 if (sub.rm_so > sub.rm_eo || (sub.rm_so == -1 && sub.rm_eo != -1) || 411 (sub.rm_so != -1 && sub.rm_eo == -1) || 412 (sub.rm_so != -1 && sub.rm_so < 0) || 413 (sub.rm_eo != -1 && sub.rm_eo < 0) ) { 414 sprintf(grump, "start %ld end %ld", (long)sub.rm_so, 415 (long)sub.rm_eo); 416 return(grump); 417 } 418 419 /* check for no match */ 420 if (sub.rm_so == -1 && should == NULL) 421 return(NULL); 422 if (sub.rm_so == -1) 423 return("did not match"); 424 425 /* check for in range */ 426 if (sub.rm_eo > strlen(str)) { 427 sprintf(grump, "start %ld end %ld, past end of string", 428 (long)sub.rm_so, (long)sub.rm_eo); 429 return(grump); 430 } 431 432 len = (int)(sub.rm_eo - sub.rm_so); 433 shlen = (int)strlen(should); 434 p = str + sub.rm_so; 435 436 /* check for not supposed to match */ 437 if (should == NULL) { 438 sprintf(grump, "matched `%.*s'", len, p); 439 return(grump); 440 } 441 442 /* check for wrong match */ 443 if (len != shlen || strncmp(p, should, (size_t)shlen) != 0) { 444 sprintf(grump, "matched `%.*s' instead", len, p); 445 return(grump); 446 } 447 if (shlen > 0) 448 return(NULL); 449 450 /* check null match in right place */ 451 if (at == NULL) 452 return(NULL); 453 shlen = strlen(at); 454 if (shlen == 0) 455 shlen = 1; /* force check for end-of-string */ 456 if (strncmp(p, at, shlen) != 0) { 457 sprintf(grump, "matched null at `%.20s'", p); 458 return(grump); 459 } 460 return(NULL); 461 } 462 463 /* 464 - eprint - convert error number to name 465 == static char *eprint(int err); 466 */ 467 static char * 468 eprint(int err) 469 { 470 static char epbuf[100]; 471 size_t len; 472 473 len = regerror(REG_ITOA|err, (regex_t *)NULL, epbuf, sizeof(epbuf)); 474 assert(len <= sizeof(epbuf)); 475 return(epbuf); 476 } 477 478 /* 479 - efind - convert error name to number 480 == static int efind(char *name); 481 */ 482 static int 483 efind(char *name) 484 { 485 static char efbuf[100]; 486 size_t n; 487 regex_t re; 488 489 sprintf(efbuf, "REG_%s", name); 490 assert(strlen(efbuf) < sizeof(efbuf)); 491 re.re_endp = efbuf; 492 (void) regerror(REG_ATOI, &re, efbuf, sizeof(efbuf)); 493 return(atoi(efbuf)); 494 } 495