1 /*- 2 * Copyright (c) 1992 Diomidis Spinellis. 3 * Copyright (c) 1992, 1993 4 * The Regents of the University of California. All rights reserved. 5 * 6 * This code is derived from software contributed to Berkeley by 7 * Diomidis Spinellis of Imperial College, University of London. 8 * 9 * Redistribution and use in source and binary forms, with or without 10 * modification, are permitted provided that the following conditions 11 * are met: 12 * 1. Redistributions of source code must retain the above copyright 13 * notice, this list of conditions and the following disclaimer. 14 * 2. Redistributions in binary form must reproduce the above copyright 15 * notice, this list of conditions and the following disclaimer in the 16 * documentation and/or other materials provided with the distribution. 17 * 4. Neither the name of the University nor the names of its contributors 18 * may be used to endorse or promote products derived from this software 19 * without specific prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 */ 33 34 #include <sys/cdefs.h> 35 __FBSDID("$FreeBSD$"); 36 37 #ifndef lint 38 static const char sccsid[] = "@(#)compile.c 8.1 (Berkeley) 6/6/93"; 39 #endif 40 41 #include <sys/types.h> 42 #include <sys/stat.h> 43 44 #include <ctype.h> 45 #include <err.h> 46 #include <errno.h> 47 #include <fcntl.h> 48 #include <limits.h> 49 #include <regex.h> 50 #include <stdio.h> 51 #include <stdlib.h> 52 #include <string.h> 53 #include <wchar.h> 54 55 #include "defs.h" 56 #include "extern.h" 57 58 #define LHSZ 128 59 #define LHMASK (LHSZ - 1) 60 static struct labhash { 61 struct labhash *lh_next; 62 u_int lh_hash; 63 struct s_command *lh_cmd; 64 int lh_ref; 65 } *labels[LHSZ]; 66 67 static const char *compile_addr(const char *, struct s_addr *); 68 static char *compile_ccl(const char **, char *); 69 static const char *compile_delimited(const char *, char *, int); 70 static const char *compile_flags(const char *, struct s_subst *); 71 static const regex_t *compile_re(const char *, int); 72 static const char *compile_subst(const char *, struct s_subst *); 73 static char *compile_text(size_t *); 74 static const char *compile_tr(const char *, struct s_tr **); 75 static struct s_command 76 **compile_stream(struct s_command **); 77 static char *duptoeol(const char *, const char *, size_t *); 78 static void enterlabel(struct s_command *); 79 static struct s_command 80 *findlabel(const char *); 81 static void fixuplabel(struct s_command *, const struct s_command *); 82 static void uselabel(void); 83 84 /* 85 * Command specification. This is used to drive the command parser. 86 */ 87 struct s_format { 88 char code; /* Command code */ 89 int naddr; /* Number of address args */ 90 enum e_args args; /* Argument type */ 91 }; 92 93 static struct s_format cmd_fmts[] = { 94 {'{', 2, GROUP}, 95 {'}', 0, ENDGROUP}, 96 {'a', 1, TEXT}, 97 {'b', 2, BRANCH}, 98 {'c', 2, TEXT}, 99 {'d', 2, EMPTY}, 100 {'D', 2, EMPTY}, 101 {'g', 2, EMPTY}, 102 {'G', 2, EMPTY}, 103 {'h', 2, EMPTY}, 104 {'H', 2, EMPTY}, 105 {'i', 1, TEXT}, 106 {'l', 2, EMPTY}, 107 {'n', 2, EMPTY}, 108 {'N', 2, EMPTY}, 109 {'p', 2, EMPTY}, 110 {'P', 2, EMPTY}, 111 {'q', 1, EMPTY}, 112 {'r', 1, RFILE}, 113 {'s', 2, SUBST}, 114 {'t', 2, BRANCH}, 115 {'w', 2, WFILE}, 116 {'x', 2, EMPTY}, 117 {'y', 2, TR}, 118 {'!', 2, NONSEL}, 119 {':', 0, LABEL}, 120 {'#', 0, COMMENT}, 121 {'=', 1, EMPTY}, 122 {'\0', 0, COMMENT}, 123 }; 124 125 /* The compiled program. */ 126 struct s_command *prog; 127 128 /* 129 * Compile the program into prog. 130 * Initialise appends. 131 */ 132 void 133 compile(void) 134 { 135 *compile_stream(&prog) = NULL; 136 fixuplabel(prog, NULL); 137 uselabel(); 138 if (appendnum == 0) 139 appends = NULL; 140 else if ((appends = malloc(sizeof(struct s_appends) * appendnum)) == 141 NULL) 142 err(1, "malloc"); 143 if ((match = malloc((maxnsub + 1) * sizeof(regmatch_t))) == NULL) 144 err(1, "malloc"); 145 } 146 147 #define EATSPACE() do { \ 148 while (*p && isspace((unsigned char)*p)) \ 149 p++; \ 150 } while (0) 151 152 #define EATSPACEN() do { \ 153 while (*p && *p != '\n' && isspace((unsigned char)*p)) \ 154 p++; \ 155 } while (0) 156 157 static struct s_command ** 158 compile_stream(struct s_command **link) 159 { 160 const char *p; 161 struct s_command *cmd, *cmd2, *stack; 162 struct s_format *fp; 163 char re[_POSIX2_LINE_MAX + 1]; 164 int naddr; /* Number of addresses */ 165 166 stack = NULL; 167 for (;;) { 168 if ((p = cu_fgets(NULL)) == NULL) { 169 if (stack != NULL) 170 errx(1, "%lu: %s: unexpected EOF (pending }'s)", 171 linenum, fname); 172 return (link); 173 } 174 175 semicolon: EATSPACEN(); 176 switch (*p) { 177 case '#': case '\0': case '\n': 178 continue; /* to next command-unit */ 179 case ';': 180 p++; 181 goto semicolon; 182 } 183 184 if ((*link = cmd = malloc(sizeof(struct s_command))) == NULL) 185 err(1, "malloc"); 186 link = &cmd->next; 187 cmd->startline = cmd->nonsel = 0; 188 /* First parse the addresses */ 189 naddr = 0; 190 191 /* Valid characters to start an address */ 192 #define addrchar(c) (strchr("0123456789/\\$", (c))) 193 if (addrchar(*p)) { 194 naddr++; 195 if ((cmd->a1 = malloc(sizeof(struct s_addr))) == NULL) 196 err(1, "malloc"); 197 p = compile_addr(p, cmd->a1); 198 EATSPACE(); /* EXTENSION */ 199 if (*p == ',') { 200 p++; 201 EATSPACE(); /* EXTENSION */ 202 naddr++; 203 if ((cmd->a2 = malloc(sizeof(struct s_addr))) 204 == NULL) 205 err(1, "malloc"); 206 p = compile_addr(p, cmd->a2); 207 EATSPACE(); 208 } else 209 cmd->a2 = NULL; 210 } else 211 cmd->a1 = cmd->a2 = NULL; 212 213 nonsel: /* Now parse the command */ 214 if (*p == '\0' || *p == '\n') 215 errx(1, "%lu: %s: command expected", linenum, fname); 216 cmd->code = *p; 217 for (fp = cmd_fmts; fp->code; fp++) 218 if (fp->code == *p) 219 break; 220 if (!fp->code) 221 errx(1, "%lu: %s: invalid command code %c (%s)", linenum, fname, *p, p); 222 if (naddr > fp->naddr) 223 errx(1, 224 "%lu: %s: command %c expects up to %d address(es), found %d", 225 linenum, fname, *p, fp->naddr, naddr); 226 switch (fp->args) { 227 case NONSEL: /* ! */ 228 p++; 229 EATSPACE(); 230 cmd->nonsel = 1; 231 goto nonsel; 232 case GROUP: /* { */ 233 p++; 234 EATSPACEN(); 235 cmd->next = stack; 236 stack = cmd; 237 link = &cmd->u.c; 238 if (*p != '\0' && *p != '\n') 239 goto semicolon; 240 break; 241 case ENDGROUP: 242 /* 243 * Short-circuit command processing, since end of 244 * group is really just a noop. 245 */ 246 cmd->nonsel = 1; 247 if (stack == NULL) 248 errx(1, "%lu: %s: unexpected }", linenum, fname); 249 cmd2 = stack; 250 stack = cmd2->next; 251 cmd2->next = cmd; 252 /*FALLTHROUGH*/ 253 case EMPTY: /* d D g G h H l n N p P q x = \0 */ 254 p++; 255 EATSPACEN(); 256 if (*p == ';') { 257 p++; 258 link = &cmd->next; 259 goto semicolon; 260 } 261 if (*p != '\0' && *p != '\n') 262 errx(1, "%lu: %s: extra characters at the end of %c command", 263 linenum, fname, cmd->code); 264 break; 265 case TEXT: /* a c i */ 266 p++; 267 EATSPACE(); 268 if (*p != '\\') 269 errx(1, 270 "%lu: %s: command %c expects \\ followed by text", linenum, fname, cmd->code); 271 p++; 272 EATSPACEN(); 273 if (*p != '\n') 274 errx(1, 275 "%lu: %s: extra characters (%c) after \\ at the end of %c command", 276 linenum, fname, *p, cmd->code); 277 cmd->t = compile_text(&cmd->tlen); 278 break; 279 case COMMENT: /* \0 # */ 280 break; 281 case WFILE: /* w */ 282 p++; 283 EATSPACE(); 284 if (*p == '\0') 285 errx(1, "%lu: %s: filename expected", linenum, fname); 286 cmd->t = duptoeol(p, "w command", &cmd->tlen); 287 if (aflag) 288 cmd->u.fd = -1; 289 else if ((cmd->u.fd = open(cmd->t, 290 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 291 DEFFILEMODE)) == -1) 292 err(1, "%s", p); 293 break; 294 case RFILE: /* r */ 295 p++; 296 EATSPACE(); 297 if (*p == '\0') 298 errx(1, "%lu: %s: filename expected", linenum, fname); 299 else 300 cmd->t = duptoeol(p, "read command", &cmd->tlen); 301 break; 302 case BRANCH: /* b t */ 303 p++; 304 EATSPACEN(); 305 if (*p == '\0' || *p == '\n') 306 cmd->t = NULL; 307 else 308 cmd->t = duptoeol(p, "branch", &cmd->tlen); 309 break; 310 case LABEL: /* : */ 311 p++; 312 EATSPACE(); 313 cmd->t = duptoeol(p, "label", &cmd->tlen); 314 if (cmd->t[0] == '\0') 315 errx(1, "%lu: %s: empty label", linenum, fname); 316 enterlabel(cmd); 317 break; 318 case SUBST: /* s */ 319 p++; 320 if (*p == '\0' || *p == '\\' || *p == '\n') 321 errx(1, 322 "%lu: %s: substitute pattern can not be delimited by newline or backslash", 323 linenum, fname); 324 if ((cmd->u.s = calloc(1, sizeof(struct s_subst))) == NULL) 325 err(1, "malloc"); 326 p = compile_delimited(p, re, 0); 327 if (p == NULL) 328 errx(1, 329 "%lu: %s: unterminated substitute pattern", linenum, fname); 330 331 --p; 332 p = compile_subst(p, cmd->u.s); 333 p = compile_flags(p, cmd->u.s); 334 335 if (*re != '\0') 336 cmd->u.s->re = compile_re(re, cmd->u.s->icase); 337 338 EATSPACE(); 339 340 if (*p == ';') { 341 p++; 342 link = &cmd->next; 343 goto semicolon; 344 } 345 break; 346 case TR: /* y */ 347 p++; 348 p = compile_tr(p, &cmd->u.y); 349 EATSPACE(); 350 if (*p == ';') { 351 p++; 352 link = &cmd->next; 353 goto semicolon; 354 } 355 if (*p) 356 errx(1, 357 "%lu: %s: extra text at the end of a transform command", linenum, fname); 358 break; 359 } 360 } 361 } 362 363 /* 364 * Get a delimited string. P points to the delimiter of the string; d points 365 * to a buffer area. Newline and delimiter escapes are processed; other 366 * escapes are ignored. 367 * 368 * Returns a pointer to the first character after the final delimiter or NULL 369 * in the case of a non-terminated string. The character array d is filled 370 * with the processed string. 371 */ 372 static const char * 373 compile_delimited(const char *p, char *d, int is_tr) 374 { 375 char c; 376 377 c = *p++; 378 if (c == '\0') 379 return (NULL); 380 else if (c == '\\') 381 errx(1, "%lu: %s: \\ can not be used as a string delimiter", 382 linenum, fname); 383 else if (c == '\n') 384 errx(1, "%lu: %s: newline can not be used as a string delimiter", 385 linenum, fname); 386 while (*p) { 387 if (*p == '[' && *p != c) { 388 if ((d = compile_ccl(&p, d)) == NULL) 389 errx(1, "%lu: %s: unbalanced brackets ([])", linenum, fname); 390 continue; 391 } else if (*p == '\\' && p[1] == '[') { 392 *d++ = *p++; 393 } else if (*p == '\\' && p[1] == c) 394 p++; 395 else if (*p == '\\' && p[1] == 'n') { 396 *d++ = '\n'; 397 p += 2; 398 continue; 399 } else if (*p == '\\' && p[1] == '\\') { 400 if (is_tr) 401 p++; 402 else 403 *d++ = *p++; 404 } else if (*p == c) { 405 *d = '\0'; 406 return (p + 1); 407 } 408 *d++ = *p++; 409 } 410 return (NULL); 411 } 412 413 414 /* compile_ccl: expand a POSIX character class */ 415 static char * 416 compile_ccl(const char **sp, char *t) 417 { 418 int c, d; 419 const char *s = *sp; 420 421 *t++ = *s++; 422 if (*s == '^') 423 *t++ = *s++; 424 if (*s == ']') 425 *t++ = *s++; 426 for (; *s && (*t = *s) != ']'; s++, t++) 427 if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) { 428 *++t = *++s, t++, s++; 429 for (c = *s; (*t = *s) != ']' || c != d; s++, t++) 430 if ((c = *s) == '\0') 431 return NULL; 432 } 433 return (*s == ']') ? *sp = ++s, ++t : NULL; 434 } 435 436 /* 437 * Compiles the regular expression in RE and returns a pointer to the compiled 438 * regular expression. 439 * Cflags are passed to regcomp. 440 */ 441 static const regex_t * 442 compile_re(const char *re, int case_insensitive) 443 { 444 regex_t *rep; 445 int eval, flags; 446 447 448 flags = rflags; 449 if (case_insensitive) 450 flags |= REG_ICASE; 451 if ((rep = malloc(sizeof(regex_t))) == NULL) 452 err(1, "malloc"); 453 if ((eval = regcomp(rep, re, flags)) != 0) 454 errx(1, "%lu: %s: RE error: %s", 455 linenum, fname, strregerror(eval, rep)); 456 if (maxnsub < rep->re_nsub) 457 maxnsub = rep->re_nsub; 458 return (rep); 459 } 460 461 /* 462 * Compile the substitution string of a regular expression and set res to 463 * point to a saved copy of it. Nsub is the number of parenthesized regular 464 * expressions. 465 */ 466 static const char * 467 compile_subst(const char *p, struct s_subst *s) 468 { 469 int asize, size; 470 u_char ref; 471 char c, *text, *op, *sp; 472 int more = 0, sawesc = 0; 473 474 c = *p++; /* Terminator character */ 475 if (c == '\0') 476 return (NULL); 477 478 s->maxbref = 0; 479 s->linenum = linenum; 480 asize = 2 * _POSIX2_LINE_MAX + 1; 481 if ((text = malloc(asize)) == NULL) 482 err(1, "malloc"); 483 size = 0; 484 do { 485 op = sp = text + size; 486 for (; *p != '\0' && *p != '\n'; p++) { 487 if (*p == '\\' || sawesc) { 488 /* 489 * If this is a continuation from the last 490 * buffer, we won't have a character to 491 * skip over. 492 */ 493 if (sawesc) 494 sawesc = 0; 495 else 496 p++; 497 498 if (*p == '\0') { 499 /* 500 * This escaped character is continued 501 * in the next part of the line. Note 502 * this fact, then cause the loop to 503 * exit w/ normal EOL case and reenter 504 * above with the new buffer. 505 */ 506 sawesc = 1; 507 p--; 508 break; 509 } else if (*p == '\n') { 510 *sp++ = '\n'; 511 break; 512 } else if (strchr("123456789", *p) != NULL) { 513 *sp++ = '\\'; 514 ref = *p - '0'; 515 if (s->re != NULL && 516 ref > s->re->re_nsub) 517 errx(1, "%lu: %s: \\%c not defined in the RE", 518 linenum, fname, *p); 519 if (s->maxbref < ref) 520 s->maxbref = ref; 521 } else if (*p == '&' || *p == '\\') 522 *sp++ = '\\'; 523 } else if (*p == c) { 524 if (*++p == '\0' && more) { 525 const char *nextp; 526 527 nextp = cu_fgets(&more); 528 if (nextp != NULL) 529 p = nextp; 530 } 531 *sp++ = '\0'; 532 size += sp - op; 533 if ((s->new = realloc(text, size)) == NULL) 534 err(1, "realloc"); 535 return (p); 536 } else if (*p == '\n') { 537 errx(1, 538 "%lu: %s: unescaped newline inside substitute pattern", linenum, fname); 539 /* NOTREACHED */ 540 } 541 *sp++ = *p; 542 } 543 size += sp - op; 544 if (asize - size < _POSIX2_LINE_MAX + 1) { 545 asize *= 2; 546 if ((text = realloc(text, asize)) == NULL) 547 err(1, "realloc"); 548 } 549 } while ((p = cu_fgets(&more))); 550 errx(1, "%lu: %s: unterminated substitute in regular expression", 551 linenum, fname); 552 /* NOTREACHED */ 553 } 554 555 /* 556 * Compile the flags of the s command 557 */ 558 static const char * 559 compile_flags(const char *p, struct s_subst *s) 560 { 561 int gn; /* True if we have seen g or n */ 562 unsigned long nval; 563 char *q; 564 565 s->n = 1; /* Default */ 566 s->p = 0; 567 s->wfile = NULL; 568 s->wfd = -1; 569 s->icase = 0; 570 for (gn = 0;;) { 571 EATSPACEN(); /* EXTENSION */ 572 switch (*p) { 573 case 'g': 574 if (gn) 575 errx(1, 576 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname); 577 gn = 1; 578 s->n = 0; 579 break; 580 case '\0': 581 case '\n': 582 case ';': 583 return (p); 584 case 'p': 585 s->p = 1; 586 break; 587 case 'i': 588 case 'I': 589 s->icase = 1; 590 break; 591 case '1': case '2': case '3': 592 case '4': case '5': case '6': 593 case '7': case '8': case '9': 594 if (gn) 595 errx(1, 596 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname); 597 gn = 1; 598 errno = 0; 599 nval = strtol(p, &q, 10); 600 if (errno == ERANGE || nval > INT_MAX) 601 errx(1, 602 "%lu: %s: overflow in the 'N' substitute flag", linenum, fname); 603 s->n = nval; 604 p = q; 605 continue; 606 case 'w': 607 p++; 608 #ifdef HISTORIC_PRACTICE 609 if (*p != ' ') { 610 warnx("%lu: %s: space missing before w wfile", linenum, fname); 611 return (p); 612 } 613 #endif 614 EATSPACE(); 615 s->wfile = duptoeol(p, "w flag", NULL); 616 if (!aflag && (s->wfd = open(s->wfile, 617 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 618 DEFFILEMODE)) == -1) 619 err(1, "%s", s->wfile); 620 return (p); 621 default: 622 errx(1, "%lu: %s: bad flag in substitute command: '%c' (%.10s)", 623 linenum, fname, *p, p); 624 break; 625 } 626 p++; 627 } 628 } 629 630 /* 631 * Compile a translation set of strings into a lookup table. 632 */ 633 static const char * 634 compile_tr(const char *p, struct s_tr **py) 635 { 636 struct s_tr *y; 637 int i; 638 const char *op, *np; 639 char old[_POSIX2_LINE_MAX + 1]; 640 char new[_POSIX2_LINE_MAX + 1]; 641 size_t oclen, oldlen, nclen, newlen; 642 mbstate_t mbs1, mbs2; 643 644 if ((*py = y = malloc(sizeof(*y))) == NULL) 645 err(1, "malloc"); 646 y->multis = NULL; 647 y->nmultis = 0; 648 649 if (*p == '\0' || *p == '\\') 650 errx(1, 651 "%lu: %s: transform pattern can not be delimited by newline or backslash", 652 linenum, fname); 653 p = compile_delimited(p, old, 1); 654 if (p == NULL) 655 errx(1, "%lu: %s: unterminated transform source string", 656 linenum, fname); 657 p = compile_delimited(p - 1, new, 1); 658 if (p == NULL) 659 errx(1, "%lu: %s: unterminated transform target string", 660 linenum, fname); 661 EATSPACE(); 662 op = old; 663 oldlen = mbsrtowcs(NULL, &op, 0, NULL); 664 if (oldlen == (size_t)-1) 665 err(1, "mbsrtowcs"); 666 np = new; 667 newlen = mbsrtowcs(NULL, &np, 0, NULL); 668 if (newlen == (size_t)-1) 669 err(1, "mbsrtowcs"); 670 if (newlen != oldlen) 671 errx(1, "%lu: %s: transform strings are not the same length", 672 linenum, fname); 673 if (MB_CUR_MAX == 1) { 674 /* 675 * The single-byte encoding case is easy: generate a 676 * lookup table. 677 */ 678 for (i = 0; i <= UCHAR_MAX; i++) 679 y->bytetab[i] = (char)i; 680 for (; *op; op++, np++) 681 y->bytetab[(u_char)*op] = *np; 682 } else { 683 /* 684 * Multi-byte encoding case: generate a lookup table as 685 * above, but only for single-byte characters. The first 686 * bytes of multi-byte characters have their lookup table 687 * entries set to 0, which causes do_tr() to search through 688 * an auxiliary vector of multi-byte mappings. 689 */ 690 memset(&mbs1, 0, sizeof(mbs1)); 691 memset(&mbs2, 0, sizeof(mbs2)); 692 for (i = 0; i <= UCHAR_MAX; i++) 693 y->bytetab[i] = (btowc(i) != WEOF) ? i : 0; 694 while (*op != '\0') { 695 oclen = mbrlen(op, MB_LEN_MAX, &mbs1); 696 if (oclen == (size_t)-1 || oclen == (size_t)-2) 697 errc(1, EILSEQ, NULL); 698 nclen = mbrlen(np, MB_LEN_MAX, &mbs2); 699 if (nclen == (size_t)-1 || nclen == (size_t)-2) 700 errc(1, EILSEQ, NULL); 701 if (oclen == 1 && nclen == 1) 702 y->bytetab[(u_char)*op] = *np; 703 else { 704 y->bytetab[(u_char)*op] = 0; 705 y->multis = realloc(y->multis, 706 (y->nmultis + 1) * sizeof(*y->multis)); 707 if (y->multis == NULL) 708 err(1, "realloc"); 709 i = y->nmultis++; 710 y->multis[i].fromlen = oclen; 711 memcpy(y->multis[i].from, op, oclen); 712 y->multis[i].tolen = nclen; 713 memcpy(y->multis[i].to, np, nclen); 714 } 715 op += oclen; 716 np += nclen; 717 } 718 } 719 return (p); 720 } 721 722 /* 723 * Compile the text following an a, c, or i command. 724 */ 725 static char * 726 compile_text(size_t *ptlen) 727 { 728 int asize, esc_nl, size; 729 char *text, *s; 730 const char *p, *op; 731 732 asize = 2 * _POSIX2_LINE_MAX + 1; 733 if ((text = malloc(asize)) == NULL) 734 err(1, "malloc"); 735 size = 0; 736 while ((p = cu_fgets(NULL))) { 737 op = s = text + size; 738 for (esc_nl = 0; *p != '\0'; p++) { 739 if (*p == '\\' && p[1] != '\0' && *++p == '\n') 740 esc_nl = 1; 741 *s++ = *p; 742 if (*p == '\n') 743 break; 744 } 745 size += s - op; 746 if (!esc_nl) { 747 *s = '\0'; 748 break; 749 } 750 if (asize - size < _POSIX2_LINE_MAX + 1) { 751 asize *= 2; 752 if ((text = realloc(text, asize)) == NULL) 753 err(1, "realloc"); 754 } 755 } 756 text[size] = '\0'; 757 if ((text = realloc(text, size + 1)) == NULL) 758 err(1, "realloc"); 759 *ptlen = size; 760 return (text); 761 } 762 763 /* 764 * Get an address and return a pointer to the first character after 765 * it. Fill the structure pointed to according to the address. 766 */ 767 static const char * 768 compile_addr(const char *p, struct s_addr *a) 769 { 770 char *end, re[_POSIX2_LINE_MAX + 1]; 771 int icase; 772 773 icase = 0; 774 775 a->type = 0; 776 switch (*p) { 777 case '\\': /* Context address */ 778 ++p; 779 /* FALLTHROUGH */ 780 case '/': /* Context address */ 781 p = compile_delimited(p, re, 0); 782 if (p == NULL) 783 errx(1, "%lu: %s: unterminated regular expression", linenum, fname); 784 /* Check for case insensitive regexp flag */ 785 if (*p == 'I') { 786 icase = 1; 787 p++; 788 } 789 if (*re == '\0') 790 a->u.r = NULL; 791 else 792 a->u.r = compile_re(re, icase); 793 a->type = AT_RE; 794 return (p); 795 796 case '$': /* Last line */ 797 a->type = AT_LAST; 798 return (p + 1); 799 800 case '+': /* Relative line number */ 801 a->type = AT_RELLINE; 802 p++; 803 /* FALLTHROUGH */ 804 /* Line number */ 805 case '0': case '1': case '2': case '3': case '4': 806 case '5': case '6': case '7': case '8': case '9': 807 if (a->type == 0) 808 a->type = AT_LINE; 809 a->u.l = strtol(p, &end, 10); 810 return (end); 811 default: 812 errx(1, "%lu: %s: expected context address", linenum, fname); 813 return (NULL); 814 } 815 } 816 817 /* 818 * duptoeol -- 819 * Return a copy of all the characters up to \n or \0. 820 */ 821 static char * 822 duptoeol(const char *s, const char *ctype, size_t *ptlen) 823 { 824 size_t len; 825 int ws; 826 char *p; 827 const char *start; 828 829 ws = 0; 830 for (start = s; *s != '\0' && *s != '\n'; ++s) 831 ws = isspace((unsigned char)*s); 832 if (ws) 833 warnx("%lu: %s: whitespace after %s", linenum, fname, ctype); 834 len = s - start; 835 if ((p = malloc(len + 1)) == NULL) 836 err(1, "malloc"); 837 memmove(p, start, len); 838 p[len] = '\0'; 839 if (ptlen != NULL) 840 *ptlen = len; 841 return p; 842 } 843 844 /* 845 * Convert goto label names to addresses, and count a and r commands, in 846 * the given subset of the script. Free the memory used by labels in b 847 * and t commands (but not by :). 848 * 849 * TODO: Remove } nodes 850 */ 851 static void 852 fixuplabel(struct s_command *cp, const struct s_command *end) 853 { 854 855 for (; cp != end; cp = cp->next) 856 switch (cp->code) { 857 case 'a': 858 case 'r': 859 appendnum++; 860 break; 861 case 'b': 862 case 't': 863 /* Resolve branch target. */ 864 if (cp->t == NULL) { 865 cp->u.c = NULL; 866 break; 867 } 868 if ((cp->u.c = findlabel(cp->t)) == NULL) 869 errx(1, "%lu: %s: %c: undefined label '%s'", linenum, fname, cp->code, cp->t); 870 free(cp->t); 871 break; 872 case '{': 873 /* Do interior commands. */ 874 fixuplabel(cp->u.c, cp->next); 875 break; 876 } 877 } 878 879 /* 880 * Associate the given command label for later lookup. 881 */ 882 static void 883 enterlabel(struct s_command *cp) 884 { 885 struct labhash **lhp, *lh; 886 u_char *p; 887 u_int h, c; 888 889 for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++) 890 h = (h << 5) + h + c; 891 lhp = &labels[h & LHMASK]; 892 for (lh = *lhp; lh != NULL; lh = lh->lh_next) 893 if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0) 894 errx(1, "%lu: %s: duplicate label '%s'", linenum, fname, cp->t); 895 if ((lh = malloc(sizeof *lh)) == NULL) 896 err(1, "malloc"); 897 lh->lh_next = *lhp; 898 lh->lh_hash = h; 899 lh->lh_cmd = cp; 900 lh->lh_ref = 0; 901 *lhp = lh; 902 } 903 904 /* 905 * Find the label contained in the command l in the command linked 906 * list cp. L is excluded from the search. Return NULL if not found. 907 */ 908 static struct s_command * 909 findlabel(const char *name) 910 { 911 struct labhash *lh; 912 const u_char *p; 913 u_int h, c; 914 915 for (h = 0, p = (const u_char *)name; (c = *p) != 0; p++) 916 h = (h << 5) + h + c; 917 for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) { 918 if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) { 919 lh->lh_ref = 1; 920 return (lh->lh_cmd); 921 } 922 } 923 return (NULL); 924 } 925 926 /* 927 * Warn about any unused labels. As a side effect, release the label hash 928 * table space. 929 */ 930 static void 931 uselabel(void) 932 { 933 struct labhash *lh, *next; 934 int i; 935 936 for (i = 0; i < LHSZ; i++) { 937 for (lh = labels[i]; lh != NULL; lh = next) { 938 next = lh->lh_next; 939 if (!lh->lh_ref) 940 warnx("%lu: %s: unused label '%s'", 941 linenum, fname, lh->lh_cmd->t); 942 free(lh); 943 } 944 } 945 } 946