vcnet/regex/main.c

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