1 /* $OpenBSD: main.c,v 1.7 2007/09/12 19:32:35 otto Exp $ */ 2 /* $NetBSD: main.c,v 1.2 1995/04/20 22:39:51 cgd Exp $ */ 3 4 #include <stdio.h> 5 #include <stdlib.h> 6 #include <string.h> 7 #include <sys/types.h> 8 #include <regex.h> 9 #include <assert.h> 10 #include <unistd.h> 11 12 #include "main.ih" 13 14 char *progname; 15 int debug = 0; 16 int line = 0; 17 int status = 0; 18 19 int copts = REG_EXTENDED; 20 int eopts = 0; 21 regoff_t startoff = 0; 22 regoff_t endoff = 0; 23 24 25 extern int split(char *, char *[], int, char *); 26 extern void regprint(regex_t *, FILE *); 27 28 /* 29 - main - do the simple case, hand off to regress() for regression 30 */ 31 int 32 main(int argc, char *argv[]) 33 34 { 35 regex_t re; 36 # define NS 10 37 regmatch_t subs[NS]; 38 char erbuf[100]; 39 int err; 40 size_t len; 41 int c; 42 int errflg = 0; 43 register int i; 44 extern int optind; 45 extern char *optarg; 46 47 progname = argv[0]; 48 49 while ((c = getopt(argc, argv, "c:e:S:E:x")) != -1) 50 switch (c) { 51 case 'c': /* compile options */ 52 copts = options('c', optarg); 53 break; 54 case 'e': /* execute options */ 55 eopts = options('e', optarg); 56 break; 57 case 'S': /* start offset */ 58 startoff = (regoff_t)atoi(optarg); 59 break; 60 case 'E': /* end offset */ 61 endoff = (regoff_t)atoi(optarg); 62 break; 63 case 'x': /* Debugging. */ 64 debug++; 65 break; 66 case '?': 67 default: 68 errflg++; 69 break; 70 } 71 if (errflg) { 72 fprintf(stderr, "usage: %s ", progname); 73 fprintf(stderr, "[-c copt][-C][-d] [re]\n"); 74 exit(2); 75 } 76 77 if (optind >= argc) { 78 regress(stdin); 79 exit(status); 80 } 81 82 err = regcomp(&re, argv[optind++], copts); 83 if (err) { 84 len = regerror(err, &re, erbuf, sizeof(erbuf)); 85 fprintf(stderr, "error %s, %zu/%zu `%s'\n", 86 eprint(err), len, sizeof(erbuf), erbuf); 87 exit(status); 88 } 89 regprint(&re, stdout); 90 91 if (optind >= argc) { 92 regfree(&re); 93 exit(status); 94 } 95 96 if (eopts®_STARTEND) { 97 subs[0].rm_so = startoff; 98 subs[0].rm_eo = strlen(argv[optind]) - endoff; 99 } 100 err = regexec(&re, argv[optind], (size_t)NS, subs, eopts); 101 if (err) { 102 len = regerror(err, &re, erbuf, sizeof(erbuf)); 103 fprintf(stderr, "error %s, %zu/%zu `%s'\n", 104 eprint(err), len, sizeof(erbuf), erbuf); 105 exit(status); 106 } 107 if (!(copts®_NOSUB)) { 108 len = (size_t)(subs[0].rm_eo - subs[0].rm_so); 109 if (subs[0].rm_so != -1) { 110 if (len != 0) 111 printf("match `%.*s'\n", (int)len, 112 argv[optind] + subs[0].rm_so); 113 else 114 printf("match `'@%.1s\n", 115 argv[optind] + subs[0].rm_so); 116 } 117 for (i = 1; i < NS; i++) 118 if (subs[i].rm_so != -1) 119 printf("(%d) `%.*s'\n", i, 120 (int)(subs[i].rm_eo - subs[i].rm_so), 121 argv[optind] + subs[i].rm_so); 122 } 123 exit(status); 124 } 125 126 /* 127 - regress - main loop of regression test 128 == void regress(FILE *in); 129 */ 130 void 131 regress(in) 132 FILE *in; 133 { 134 char inbuf[1000]; 135 # define MAXF 10 136 char *f[MAXF]; 137 int nf; 138 int i; 139 char erbuf[100]; 140 size_t ne; 141 char *badpat = "invalid regular expression"; 142 # define SHORT 10 143 char *bpname = "REG_BADPAT"; 144 regex_t re; 145 146 while (fgets(inbuf, sizeof(inbuf), in) != NULL) { 147 line++; 148 if (inbuf[0] == '#' || inbuf[0] == '\n') 149 continue; /* NOTE CONTINUE */ 150 inbuf[strcspn(inbuf, "\n")] = '\0'; /* get rid of stupid \n */ 151 if (debug) 152 fprintf(stdout, "%d:\n", line); 153 nf = split(inbuf, f, MAXF, "\t\t"); 154 if (nf < 3) { 155 fprintf(stderr, "bad input, line %d\n", line); 156 exit(1); 157 } 158 for (i = 0; i < nf; i++) 159 if (strcmp(f[i], "\"\"") == 0) 160 f[i] = ""; 161 if (nf <= 3) 162 f[3] = NULL; 163 if (nf <= 4) 164 f[4] = NULL; 165 try(f[0], f[1], f[2], f[3], f[4], options('c', f[1])); 166 if (opt('&', f[1])) /* try with either type of RE */ 167 try(f[0], f[1], f[2], f[3], f[4], 168 options('c', f[1]) &~ REG_EXTENDED); 169 } 170 171 ne = regerror(REG_BADPAT, (regex_t *)NULL, erbuf, sizeof(erbuf)); 172 if (strcmp(erbuf, badpat) != 0 || ne != strlen(badpat)+1) { 173 fprintf(stderr, "end: regerror() test gave `%s' not `%s'\n", 174 erbuf, badpat); 175 status = 1; 176 } 177 ne = regerror(REG_BADPAT, (regex_t *)NULL, erbuf, (size_t)SHORT); 178 if (strncmp(erbuf, badpat, SHORT-1) != 0 || erbuf[SHORT-1] != '\0' || 179 ne != strlen(badpat)+1) { 180 fprintf(stderr, "end: regerror() short test gave `%s' not `%.*s'\n", 181 erbuf, SHORT-1, badpat); 182 status = 1; 183 } 184 ne = regerror(REG_ITOA|REG_BADPAT, (regex_t *)NULL, erbuf, sizeof(erbuf)); 185 if (strcmp(erbuf, bpname) != 0 || ne != strlen(bpname)+1) { 186 fprintf(stderr, "end: regerror() ITOA test gave `%s' not `%s'\n", 187 erbuf, bpname); 188 status = 1; 189 } 190 re.re_endp = bpname; 191 ne = regerror(REG_ATOI, &re, erbuf, sizeof(erbuf)); 192 if (atoi(erbuf) != (int)REG_BADPAT) { 193 fprintf(stderr, "end: regerror() ATOI test gave `%s' not `%ld'\n", 194 erbuf, (long)REG_BADPAT); 195 status = 1; 196 } else if (ne != strlen(erbuf)+1) { 197 fprintf(stderr, "end: regerror() ATOI test len(`%s') = %ld\n", 198 erbuf, (long)REG_BADPAT); 199 status = 1; 200 } 201 } 202 203 /* 204 - try - try it, and report on problems 205 == void try(char *f0, char *f1, char *f2, char *f3, char *f4, int opts); 206 */ 207 void 208 try(f0, f1, f2, f3, f4, opts) 209 char *f0; 210 char *f1; 211 char *f2; 212 char *f3; 213 char *f4; 214 int opts; /* may not match f1 */ 215 { 216 regex_t re; 217 # define NSUBS 10 218 regmatch_t subs[NSUBS]; 219 # define NSHOULD 15 220 char *should[NSHOULD]; 221 int nshould; 222 char erbuf[100]; 223 int err; 224 int len; 225 char *type = (opts & REG_EXTENDED) ? "ERE" : "BRE"; 226 register int i; 227 char *grump; 228 char f0copy[1000]; 229 char f2copy[1000]; 230 231 strlcpy(f0copy, f0, sizeof f0copy); 232 re.re_endp = (opts®_PEND) ? f0copy + strlen(f0copy) : NULL; 233 fixstr(f0copy); 234 err = regcomp(&re, f0copy, opts); 235 if (err != 0 && (!opt('C', f1) || err != efind(f2))) { 236 /* unexpected error or wrong error */ 237 len = regerror(err, &re, erbuf, sizeof(erbuf)); 238 fprintf(stderr, "%d: %s error %s, %d/%zu `%s'\n", 239 line, type, eprint(err), len, 240 sizeof(erbuf), erbuf); 241 status = 1; 242 } else if (err == 0 && opt('C', f1)) { 243 /* unexpected success */ 244 fprintf(stderr, "%d: %s should have given REG_%s\n", 245 line, type, f2); 246 status = 1; 247 err = 1; /* so we won't try regexec */ 248 } 249 250 if (err != 0) { 251 regfree(&re); 252 return; 253 } 254 255 strlcpy(f2copy, f2, sizeof f2copy); 256 fixstr(f2copy); 257 258 if (options('e', f1)®_STARTEND) { 259 if (strchr(f2, '(') == NULL || strchr(f2, ')') == NULL) 260 fprintf(stderr, "%d: bad STARTEND syntax\n", line); 261 subs[0].rm_so = strchr(f2, '(') - f2 + 1; 262 subs[0].rm_eo = strchr(f2, ')') - f2; 263 } 264 err = regexec(&re, f2copy, NSUBS, subs, options('e', f1)); 265 266 if (err != 0 && (f3 != NULL || err != REG_NOMATCH)) { 267 /* unexpected error or wrong error */ 268 len = regerror(err, &re, erbuf, sizeof(erbuf)); 269 fprintf(stderr, "%d: %s exec error %s, %d/%zu `%s'\n", 270 line, type, eprint(err), len, 271 sizeof(erbuf), erbuf); 272 status = 1; 273 } else if (err != 0) { 274 /* nothing more to check */ 275 } else if (f3 == NULL) { 276 /* unexpected success */ 277 fprintf(stderr, "%d: %s exec should have failed\n", 278 line, type); 279 status = 1; 280 err = 1; /* just on principle */ 281 } else if (opts®_NOSUB) { 282 /* nothing more to check */ 283 } else if ((grump = check(f2, subs[0], f3)) != NULL) { 284 fprintf(stderr, "%d: %s %s\n", line, type, grump); 285 status = 1; 286 err = 1; 287 } 288 289 if (err != 0 || f4 == NULL) { 290 regfree(&re); 291 return; 292 } 293 294 for (i = 1; i < NSHOULD; i++) 295 should[i] = NULL; 296 nshould = split(f4, should+1, NSHOULD-1, ","); 297 if (nshould == 0) { 298 nshould = 1; 299 should[1] = ""; 300 } 301 for (i = 1; i < NSUBS; i++) { 302 grump = check(f2, subs[i], should[i]); 303 if (grump != NULL) { 304 fprintf(stderr, "%d: %s $%d %s\n", line, 305 type, i, grump); 306 status = 1; 307 err = 1; 308 } 309 } 310 311 regfree(&re); 312 } 313 314 /* 315 - options - pick options out of a regression-test string 316 == int options(int type, char *s); 317 */ 318 int 319 options(type, s) 320 int type; /* 'c' compile, 'e' exec */ 321 char *s; 322 { 323 register char *p; 324 register int o = (type == 'c') ? copts : eopts; 325 register char *legal = (type == 'c') ? "bisnmp" : "^$#tl"; 326 327 for (p = s; *p != '\0'; p++) 328 if (strchr(legal, *p) != NULL) 329 switch (*p) { 330 case 'b': 331 o &= ~REG_EXTENDED; 332 break; 333 case 'i': 334 o |= REG_ICASE; 335 break; 336 case 's': 337 o |= REG_NOSUB; 338 break; 339 case 'n': 340 o |= REG_NEWLINE; 341 break; 342 case 'm': 343 o &= ~REG_EXTENDED; 344 o |= REG_NOSPEC; 345 break; 346 case 'p': 347 o |= REG_PEND; 348 break; 349 case '^': 350 o |= REG_NOTBOL; 351 break; 352 case '$': 353 o |= REG_NOTEOL; 354 break; 355 case '#': 356 o |= REG_STARTEND; 357 break; 358 case 't': /* trace */ 359 o |= REG_TRACE; 360 break; 361 case 'l': /* force long representation */ 362 o |= REG_LARGE; 363 break; 364 case 'r': /* force backref use */ 365 o |= REG_BACKR; 366 break; 367 } 368 return(o); 369 } 370 371 /* 372 - opt - is a particular option in a regression string? 373 == int opt(int c, char *s); 374 */ 375 int /* predicate */ 376 opt(c, s) 377 int c; 378 char *s; 379 { 380 return(strchr(s, c) != NULL); 381 } 382 383 /* 384 - fixstr - transform magic characters in strings 385 == void fixstr(register char *p); 386 */ 387 void 388 fixstr(p) 389 register char *p; 390 { 391 if (p == NULL) 392 return; 393 394 for (; *p != '\0'; p++) 395 if (*p == 'N') 396 *p = '\n'; 397 else if (*p == 'T') 398 *p = '\t'; 399 else if (*p == 'S') 400 *p = ' '; 401 else if (*p == 'Z') 402 *p = '\0'; 403 } 404 405 /* 406 - check - check a substring match 407 == char *check(char *str, regmatch_t sub, char *should); 408 */ 409 char * /* NULL or complaint */ 410 check(str, sub, should) 411 char *str; 412 regmatch_t sub; 413 char *should; 414 { 415 register int len; 416 register int shlen; 417 register char *p; 418 static char grump[500]; 419 register char *at = NULL; 420 421 if (should != NULL && strcmp(should, "-") == 0) 422 should = NULL; 423 if (should != NULL && should[0] == '@') { 424 at = should + 1; 425 should = ""; 426 } 427 428 /* check rm_so and rm_eo for consistency */ 429 if (sub.rm_so > sub.rm_eo || (sub.rm_so == -1 && sub.rm_eo != -1) || 430 (sub.rm_so != -1 && sub.rm_eo == -1) || 431 (sub.rm_so != -1 && sub.rm_so < 0) || 432 (sub.rm_eo != -1 && sub.rm_eo < 0) ) { 433 snprintf(grump, sizeof grump, 434 "start %ld end %ld", (long)sub.rm_so, 435 (long)sub.rm_eo); 436 return(grump); 437 } 438 439 /* check for no match */ 440 if (sub.rm_so == -1 && should == NULL) 441 return(NULL); 442 if (sub.rm_so == -1) 443 return("did not match"); 444 445 /* check for in range */ 446 if (sub.rm_eo > strlen(str)) { 447 snprintf(grump, sizeof grump, 448 "start %ld end %ld, past end of string", 449 (long)sub.rm_so, (long)sub.rm_eo); 450 return(grump); 451 } 452 453 len = (int)(sub.rm_eo - sub.rm_so); 454 shlen = (int)strlen(should); 455 p = str + sub.rm_so; 456 457 /* check for not supposed to match */ 458 if (should == NULL) { 459 snprintf(grump, sizeof grump, "matched `%.*s'", len, p); 460 return(grump); 461 } 462 463 /* check for wrong match */ 464 if (len != shlen || strncmp(p, should, (size_t)shlen) != 0) { 465 snprintf(grump, sizeof grump, "matched `%.*s' instead", len, p); 466 return(grump); 467 } 468 if (shlen > 0) 469 return(NULL); 470 471 /* check null match in right place */ 472 if (at == NULL) 473 return(NULL); 474 shlen = strlen(at); 475 if (shlen == 0) 476 shlen = 1; /* force check for end-of-string */ 477 if (strncmp(p, at, shlen) != 0) { 478 snprintf(grump, sizeof grump, "matched null at `%.20s'", p); 479 return(grump); 480 } 481 return(NULL); 482 } 483 484 /* 485 - eprint - convert error number to name 486 == static char *eprint(int err); 487 */ 488 static char * 489 eprint(err) 490 int err; 491 { 492 static char epbuf[100]; 493 size_t len; 494 495 len = regerror(REG_ITOA|err, (regex_t *)NULL, epbuf, sizeof(epbuf)); 496 assert(len <= sizeof(epbuf)); 497 return(epbuf); 498 } 499 500 /* 501 - efind - convert error name to number 502 == static int efind(char *name); 503 */ 504 static int 505 efind(name) 506 char *name; 507 { 508 static char efbuf[100]; 509 regex_t re; 510 511 snprintf(efbuf, sizeof efbuf, "REG_%s", name); 512 assert(strlen(efbuf) < sizeof(efbuf)); 513 re.re_endp = efbuf; 514 (void) regerror(REG_ATOI, &re, efbuf, sizeof(efbuf)); 515 return(atoi(efbuf)); 516 } 517