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 char *compile_addr(char *, struct s_addr *); 68 static char *compile_ccl(char **, char *); 69 static char *compile_delimited(char *, char *); 70 static char *compile_flags(char *, struct s_subst *); 71 static char *compile_re(char *, regex_t **); 72 static char *compile_subst(char *, struct s_subst *); 73 static char *compile_text(void); 74 static char *compile_tr(char *, struct s_tr **); 75 static struct s_command 76 **compile_stream(struct s_command **); 77 static char *duptoeol(char *, const char *); 78 static void enterlabel(struct s_command *); 79 static struct s_command 80 *findlabel(char *); 81 static void fixuplabel(struct s_command *, 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 if (p) \ 149 while (*p && isspace((unsigned char)*p)) \ 150 p++; \ 151 } while (0) 152 153 static struct s_command ** 154 compile_stream(struct s_command **link) 155 { 156 char *p; 157 static char lbuf[_POSIX2_LINE_MAX + 1]; /* To save stack */ 158 struct s_command *cmd, *cmd2, *stack; 159 struct s_format *fp; 160 int naddr; /* Number of addresses */ 161 162 stack = 0; 163 for (;;) { 164 if ((p = cu_fgets(lbuf, sizeof(lbuf), NULL)) == NULL) { 165 if (stack != 0) 166 errx(1, "%lu: %s: unexpected EOF (pending }'s)", 167 linenum, fname); 168 return (link); 169 } 170 171 semicolon: EATSPACE(); 172 if (p) { 173 if (*p == '#' || *p == '\0') 174 continue; 175 else if (*p == ';') { 176 p++; 177 goto semicolon; 178 } 179 } 180 if ((*link = cmd = malloc(sizeof(struct s_command))) == NULL) 181 err(1, "malloc"); 182 link = &cmd->next; 183 cmd->nonsel = cmd->inrange = 0; 184 /* First parse the addresses */ 185 naddr = 0; 186 187 /* Valid characters to start an address */ 188 #define addrchar(c) (strchr("0123456789/\\$", (c))) 189 if (addrchar(*p)) { 190 naddr++; 191 if ((cmd->a1 = malloc(sizeof(struct s_addr))) == NULL) 192 err(1, "malloc"); 193 p = compile_addr(p, cmd->a1); 194 EATSPACE(); /* EXTENSION */ 195 if (*p == ',') { 196 p++; 197 EATSPACE(); /* EXTENSION */ 198 naddr++; 199 if ((cmd->a2 = malloc(sizeof(struct s_addr))) 200 == NULL) 201 err(1, "malloc"); 202 p = compile_addr(p, cmd->a2); 203 EATSPACE(); 204 } else 205 cmd->a2 = 0; 206 } else 207 cmd->a1 = cmd->a2 = 0; 208 209 nonsel: /* Now parse the command */ 210 if (!*p) 211 errx(1, "%lu: %s: command expected", linenum, fname); 212 cmd->code = *p; 213 for (fp = cmd_fmts; fp->code; fp++) 214 if (fp->code == *p) 215 break; 216 if (!fp->code) 217 errx(1, "%lu: %s: invalid command code %c", linenum, fname, *p); 218 if (naddr > fp->naddr) 219 errx(1, 220 "%lu: %s: command %c expects up to %d address(es), found %d", 221 linenum, fname, *p, fp->naddr, naddr); 222 switch (fp->args) { 223 case NONSEL: /* ! */ 224 p++; 225 EATSPACE(); 226 cmd->nonsel = ! cmd->nonsel; 227 goto nonsel; 228 case GROUP: /* { */ 229 p++; 230 EATSPACE(); 231 cmd->next = stack; 232 stack = cmd; 233 link = &cmd->u.c; 234 if (*p) 235 goto semicolon; 236 break; 237 case ENDGROUP: 238 /* 239 * Short-circuit command processing, since end of 240 * group is really just a noop. 241 */ 242 cmd->nonsel = 1; 243 if (stack == 0) 244 errx(1, "%lu: %s: unexpected }", linenum, fname); 245 cmd2 = stack; 246 stack = cmd2->next; 247 cmd2->next = cmd; 248 /*FALLTHROUGH*/ 249 case EMPTY: /* d D g G h H l n N p P q x = \0 */ 250 p++; 251 EATSPACE(); 252 if (*p == ';') { 253 p++; 254 link = &cmd->next; 255 goto semicolon; 256 } 257 if (*p) 258 errx(1, "%lu: %s: extra characters at the end of %c command", 259 linenum, fname, cmd->code); 260 break; 261 case TEXT: /* a c i */ 262 p++; 263 EATSPACE(); 264 if (*p != '\\') 265 errx(1, 266 "%lu: %s: command %c expects \\ followed by text", linenum, fname, cmd->code); 267 p++; 268 EATSPACE(); 269 if (*p) 270 errx(1, 271 "%lu: %s: extra characters after \\ at the end of %c command", 272 linenum, fname, cmd->code); 273 cmd->t = compile_text(); 274 break; 275 case COMMENT: /* \0 # */ 276 break; 277 case WFILE: /* w */ 278 p++; 279 EATSPACE(); 280 if (*p == '\0') 281 errx(1, "%lu: %s: filename expected", linenum, fname); 282 cmd->t = duptoeol(p, "w command"); 283 if (aflag) 284 cmd->u.fd = -1; 285 else if ((cmd->u.fd = open(p, 286 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 287 DEFFILEMODE)) == -1) 288 err(1, "%s", p); 289 break; 290 case RFILE: /* r */ 291 p++; 292 EATSPACE(); 293 if (*p == '\0') 294 errx(1, "%lu: %s: filename expected", linenum, fname); 295 else 296 cmd->t = duptoeol(p, "read command"); 297 break; 298 case BRANCH: /* b t */ 299 p++; 300 EATSPACE(); 301 if (*p == '\0') 302 cmd->t = NULL; 303 else 304 cmd->t = duptoeol(p, "branch"); 305 break; 306 case LABEL: /* : */ 307 p++; 308 EATSPACE(); 309 cmd->t = duptoeol(p, "label"); 310 if (strlen(p) == 0) 311 errx(1, "%lu: %s: empty label", linenum, fname); 312 enterlabel(cmd); 313 break; 314 case SUBST: /* s */ 315 p++; 316 if (*p == '\0' || *p == '\\') 317 errx(1, 318 "%lu: %s: substitute pattern can not be delimited by newline or backslash", 319 linenum, fname); 320 if ((cmd->u.s = malloc(sizeof(struct s_subst))) == NULL) 321 err(1, "malloc"); 322 p = compile_re(p, &cmd->u.s->re); 323 if (p == NULL) 324 errx(1, 325 "%lu: %s: unterminated substitute pattern", linenum, fname); 326 --p; 327 p = compile_subst(p, cmd->u.s); 328 p = compile_flags(p, cmd->u.s); 329 EATSPACE(); 330 if (*p == ';') { 331 p++; 332 link = &cmd->next; 333 goto semicolon; 334 } 335 break; 336 case TR: /* y */ 337 p++; 338 p = compile_tr(p, &cmd->u.y); 339 EATSPACE(); 340 if (*p == ';') { 341 p++; 342 link = &cmd->next; 343 goto semicolon; 344 } 345 if (*p) 346 errx(1, 347 "%lu: %s: extra text at the end of a transform command", linenum, fname); 348 break; 349 } 350 } 351 } 352 353 /* 354 * Get a delimited string. P points to the delimeter of the string; d points 355 * to a buffer area. Newline and delimiter escapes are processed; other 356 * escapes are ignored. 357 * 358 * Returns a pointer to the first character after the final delimiter or NULL 359 * in the case of a non-terminated string. The character array d is filled 360 * with the processed string. 361 */ 362 static char * 363 compile_delimited(char *p, char *d) 364 { 365 char c; 366 367 c = *p++; 368 if (c == '\0') 369 return (NULL); 370 else if (c == '\\') 371 errx(1, "%lu: %s: \\ can not be used as a string delimiter", 372 linenum, fname); 373 else if (c == '\n') 374 errx(1, "%lu: %s: newline can not be used as a string delimiter", 375 linenum, fname); 376 while (*p) { 377 if (*p == '[') { 378 if ((d = compile_ccl(&p, d)) == NULL) 379 errx(1, "%lu: %s: unbalanced brackets ([])", linenum, fname); 380 continue; 381 } else if (*p == '\\' && p[1] == '[') { 382 *d++ = *p++; 383 } else if (*p == '\\' && p[1] == c) 384 p++; 385 else if (*p == '\\' && p[1] == 'n') { 386 *d++ = '\n'; 387 p += 2; 388 continue; 389 } else if (*p == '\\' && p[1] == '\\') 390 *d++ = *p++; 391 else if (*p == c) { 392 *d = '\0'; 393 return (p + 1); 394 } 395 *d++ = *p++; 396 } 397 return (NULL); 398 } 399 400 401 /* compile_ccl: expand a POSIX character class */ 402 static char * 403 compile_ccl(char **sp, char *t) 404 { 405 int c, d; 406 char *s = *sp; 407 408 *t++ = *s++; 409 if (*s == '^') 410 *t++ = *s++; 411 if (*s == ']') 412 *t++ = *s++; 413 for (; *s && (*t = *s) != ']'; s++, t++) 414 if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) { 415 *++t = *++s, t++, s++; 416 for (c = *s; (*t = *s) != ']' || c != d; s++, t++) 417 if ((c = *s) == '\0') 418 return NULL; 419 } else if (*s == '\\' && s[1] == 'n') 420 *t = '\n', s++; 421 return (*s == ']') ? *sp = ++s, ++t : NULL; 422 } 423 424 /* 425 * Get a regular expression. P points to the delimiter of the regular 426 * expression; repp points to the address of a regexp pointer. Newline 427 * and delimiter escapes are processed; other escapes are ignored. 428 * Returns a pointer to the first character after the final delimiter 429 * or NULL in the case of a non terminated regular expression. The regexp 430 * pointer is set to the compiled regular expression. 431 * Cflags are passed to regcomp. 432 */ 433 static char * 434 compile_re(char *p, regex_t **repp) 435 { 436 int eval; 437 char re[_POSIX2_LINE_MAX + 1]; 438 439 p = compile_delimited(p, re); 440 if (p && strlen(re) == 0) { 441 *repp = NULL; 442 return (p); 443 } 444 if ((*repp = malloc(sizeof(regex_t))) == NULL) 445 err(1, "malloc"); 446 if (p && (eval = regcomp(*repp, re, rflags)) != 0) 447 errx(1, "%lu: %s: RE error: %s", 448 linenum, fname, strregerror(eval, *repp)); 449 if (maxnsub < (*repp)->re_nsub) 450 maxnsub = (*repp)->re_nsub; 451 return (p); 452 } 453 454 /* 455 * Compile the substitution string of a regular expression and set res to 456 * point to a saved copy of it. Nsub is the number of parenthesized regular 457 * expressions. 458 */ 459 static char * 460 compile_subst(char *p, struct s_subst *s) 461 { 462 static char lbuf[_POSIX2_LINE_MAX + 1]; 463 int asize, size; 464 u_char ref; 465 char c, *text, *op, *sp; 466 int more = 1, sawesc = 0; 467 468 c = *p++; /* Terminator character */ 469 if (c == '\0') 470 return (NULL); 471 472 s->maxbref = 0; 473 s->linenum = linenum; 474 asize = 2 * _POSIX2_LINE_MAX + 1; 475 if ((text = malloc(asize)) == NULL) 476 err(1, "malloc"); 477 size = 0; 478 do { 479 op = sp = text + size; 480 for (; *p; p++) { 481 if (*p == '\\' || sawesc) { 482 /* 483 * If this is a continuation from the last 484 * buffer, we won't have a character to 485 * skip over. 486 */ 487 if (sawesc) 488 sawesc = 0; 489 else 490 p++; 491 492 if (*p == '\0') { 493 /* 494 * This escaped character is continued 495 * in the next part of the line. Note 496 * this fact, then cause the loop to 497 * exit w/ normal EOL case and reenter 498 * above with the new buffer. 499 */ 500 sawesc = 1; 501 p--; 502 continue; 503 } else if (strchr("123456789", *p) != NULL) { 504 *sp++ = '\\'; 505 ref = *p - '0'; 506 if (s->re != NULL && 507 ref > s->re->re_nsub) 508 errx(1, "%lu: %s: \\%c not defined in the RE", 509 linenum, fname, *p); 510 if (s->maxbref < ref) 511 s->maxbref = ref; 512 } else if (*p == '&' || *p == '\\') 513 *sp++ = '\\'; 514 } else if (*p == c) { 515 if (*++p == '\0' && more) { 516 if (cu_fgets(lbuf, sizeof(lbuf), &more)) 517 p = lbuf; 518 } 519 *sp++ = '\0'; 520 size += sp - op; 521 if ((s->new = realloc(text, size)) == NULL) 522 err(1, "realloc"); 523 return (p); 524 } else if (*p == '\n') { 525 errx(1, 526 "%lu: %s: unescaped newline inside substitute pattern", linenum, fname); 527 /* NOTREACHED */ 528 } 529 *sp++ = *p; 530 } 531 size += sp - op; 532 if (asize - size < _POSIX2_LINE_MAX + 1) { 533 asize *= 2; 534 if ((text = realloc(text, asize)) == NULL) 535 err(1, "realloc"); 536 } 537 } while (cu_fgets(p = lbuf, sizeof(lbuf), &more)); 538 errx(1, "%lu: %s: unterminated substitute in regular expression", 539 linenum, fname); 540 /* NOTREACHED */ 541 } 542 543 /* 544 * Compile the flags of the s command 545 */ 546 static char * 547 compile_flags(char *p, struct s_subst *s) 548 { 549 int gn; /* True if we have seen g or n */ 550 unsigned long nval; 551 char wfile[_POSIX2_LINE_MAX + 1], *q; 552 553 s->n = 1; /* Default */ 554 s->p = 0; 555 s->wfile = NULL; 556 s->wfd = -1; 557 for (gn = 0;;) { 558 EATSPACE(); /* EXTENSION */ 559 switch (*p) { 560 case 'g': 561 if (gn) 562 errx(1, 563 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname); 564 gn = 1; 565 s->n = 0; 566 break; 567 case '\0': 568 case '\n': 569 case ';': 570 return (p); 571 case 'p': 572 s->p = 1; 573 break; 574 case '1': case '2': case '3': 575 case '4': case '5': case '6': 576 case '7': case '8': case '9': 577 if (gn) 578 errx(1, 579 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname); 580 gn = 1; 581 errno = 0; 582 nval = strtol(p, &p, 10); 583 if (errno == ERANGE || nval > INT_MAX) 584 errx(1, 585 "%lu: %s: overflow in the 'N' substitute flag", linenum, fname); 586 s->n = nval; 587 p--; 588 break; 589 case 'w': 590 p++; 591 #ifdef HISTORIC_PRACTICE 592 if (*p != ' ') { 593 warnx("%lu: %s: space missing before w wfile", linenum, fname); 594 return (p); 595 } 596 #endif 597 EATSPACE(); 598 q = wfile; 599 while (*p) { 600 if (*p == '\n') 601 break; 602 *q++ = *p++; 603 } 604 *q = '\0'; 605 if (q == wfile) 606 errx(1, "%lu: %s: no wfile specified", linenum, fname); 607 s->wfile = strdup(wfile); 608 if (!aflag && (s->wfd = open(wfile, 609 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 610 DEFFILEMODE)) == -1) 611 err(1, "%s", wfile); 612 return (p); 613 default: 614 errx(1, "%lu: %s: bad flag in substitute command: '%c'", 615 linenum, fname, *p); 616 break; 617 } 618 p++; 619 } 620 } 621 622 /* 623 * Compile a translation set of strings into a lookup table. 624 */ 625 static char * 626 compile_tr(char *p, struct s_tr **py) 627 { 628 struct s_tr *y; 629 int i; 630 const char *op, *np; 631 char old[_POSIX2_LINE_MAX + 1]; 632 char new[_POSIX2_LINE_MAX + 1]; 633 size_t oclen, oldlen, nclen, newlen; 634 mbstate_t mbs1, mbs2; 635 636 if ((*py = y = malloc(sizeof(*y))) == NULL) 637 err(1, NULL); 638 y->multis = NULL; 639 y->nmultis = 0; 640 641 if (*p == '\0' || *p == '\\') 642 errx(1, 643 "%lu: %s: transform pattern can not be delimited by newline or backslash", 644 linenum, fname); 645 p = compile_delimited(p, old); 646 if (p == NULL) 647 errx(1, "%lu: %s: unterminated transform source string", 648 linenum, fname); 649 p = compile_delimited(p - 1, new); 650 if (p == NULL) 651 errx(1, "%lu: %s: unterminated transform target string", 652 linenum, fname); 653 EATSPACE(); 654 op = old; 655 oldlen = mbsrtowcs(NULL, &op, 0, NULL); 656 if (oldlen == (size_t)-1) 657 err(1, NULL); 658 np = new; 659 newlen = mbsrtowcs(NULL, &np, 0, NULL); 660 if (newlen == (size_t)-1) 661 err(1, NULL); 662 if (newlen != oldlen) 663 errx(1, "%lu: %s: transform strings are not the same length", 664 linenum, fname); 665 if (MB_CUR_MAX == 1) { 666 /* 667 * The single-byte encoding case is easy: generate a 668 * lookup table. 669 */ 670 for (i = 0; i <= UCHAR_MAX; i++) 671 y->bytetab[i] = (char)i; 672 for (; *op; op++, np++) 673 y->bytetab[(u_char)*op] = *np; 674 } else { 675 /* 676 * Multi-byte encoding case: generate a lookup table as 677 * above, but only for single-byte characters. The first 678 * bytes of multi-byte characters have their lookup table 679 * entries set to 0, which causes do_tr() to search through 680 * an auxiliary vector of multi-byte mappings. 681 */ 682 memset(&mbs1, 0, sizeof(mbs1)); 683 memset(&mbs2, 0, sizeof(mbs2)); 684 for (i = 0; i <= UCHAR_MAX; i++) 685 y->bytetab[i] = (btowc(i) != WEOF) ? i : 0; 686 while (*op != '\0') { 687 oclen = mbrlen(op, MB_LEN_MAX, &mbs1); 688 if (oclen == (size_t)-1 || oclen == (size_t)-2) 689 errc(1, EILSEQ, NULL); 690 nclen = mbrlen(np, MB_LEN_MAX, &mbs2); 691 if (nclen == (size_t)-1 || nclen == (size_t)-2) 692 errc(1, EILSEQ, NULL); 693 if (oclen == 1 && nclen == 1) 694 y->bytetab[(u_char)*op] = *np; 695 else { 696 y->bytetab[(u_char)*op] = 0; 697 y->multis = realloc(y->multis, 698 (y->nmultis + 1) * sizeof(*y->multis)); 699 if (y->multis == NULL) 700 err(1, NULL); 701 i = y->nmultis++; 702 y->multis[i].fromlen = oclen; 703 memcpy(y->multis[i].from, op, oclen); 704 y->multis[i].tolen = nclen; 705 memcpy(y->multis[i].to, np, nclen); 706 } 707 op += oclen; 708 np += nclen; 709 } 710 } 711 return (p); 712 } 713 714 /* 715 * Compile the text following an a or i command. 716 */ 717 static char * 718 compile_text(void) 719 { 720 int asize, esc_nl, size; 721 char *text, *p, *op, *s; 722 char lbuf[_POSIX2_LINE_MAX + 1]; 723 724 asize = 2 * _POSIX2_LINE_MAX + 1; 725 if ((text = malloc(asize)) == NULL) 726 err(1, "malloc"); 727 size = 0; 728 while (cu_fgets(lbuf, sizeof(lbuf), NULL)) { 729 op = s = text + size; 730 p = lbuf; 731 EATSPACE(); 732 for (esc_nl = 0; *p != '\0'; p++) { 733 if (*p == '\\' && p[1] != '\0' && *++p == '\n') 734 esc_nl = 1; 735 *s++ = *p; 736 } 737 size += s - op; 738 if (!esc_nl) { 739 *s = '\0'; 740 break; 741 } 742 if (asize - size < _POSIX2_LINE_MAX + 1) { 743 asize *= 2; 744 if ((text = realloc(text, asize)) == NULL) 745 err(1, "realloc"); 746 } 747 } 748 text[size] = '\0'; 749 if ((p = realloc(text, size + 1)) == NULL) 750 err(1, "realloc"); 751 return (p); 752 } 753 754 /* 755 * Get an address and return a pointer to the first character after 756 * it. Fill the structure pointed to according to the address. 757 */ 758 static char * 759 compile_addr(char *p, struct s_addr *a) 760 { 761 char *end; 762 763 switch (*p) { 764 case '\\': /* Context address */ 765 ++p; 766 /* FALLTHROUGH */ 767 case '/': /* Context address */ 768 p = compile_re(p, &a->u.r); 769 if (p == NULL) 770 errx(1, "%lu: %s: unterminated regular expression", linenum, fname); 771 a->type = AT_RE; 772 return (p); 773 774 case '$': /* Last line */ 775 a->type = AT_LAST; 776 return (p + 1); 777 /* Line number */ 778 case '0': case '1': case '2': case '3': case '4': 779 case '5': case '6': case '7': case '8': case '9': 780 a->type = AT_LINE; 781 a->u.l = strtol(p, &end, 10); 782 return (end); 783 default: 784 errx(1, "%lu: %s: expected context address", linenum, fname); 785 return (NULL); 786 } 787 } 788 789 /* 790 * duptoeol -- 791 * Return a copy of all the characters up to \n or \0. 792 */ 793 static char * 794 duptoeol(char *s, const char *ctype) 795 { 796 size_t len; 797 int ws; 798 char *p, *start; 799 800 ws = 0; 801 for (start = s; *s != '\0' && *s != '\n'; ++s) 802 ws = isspace((unsigned char)*s); 803 *s = '\0'; 804 if (ws) 805 warnx("%lu: %s: whitespace after %s", linenum, fname, ctype); 806 len = s - start + 1; 807 if ((p = malloc(len)) == NULL) 808 err(1, "malloc"); 809 return (memmove(p, start, len)); 810 } 811 812 /* 813 * Convert goto label names to addresses, and count a and r commands, in 814 * the given subset of the script. Free the memory used by labels in b 815 * and t commands (but not by :). 816 * 817 * TODO: Remove } nodes 818 */ 819 static void 820 fixuplabel(struct s_command *cp, struct s_command *end) 821 { 822 823 for (; cp != end; cp = cp->next) 824 switch (cp->code) { 825 case 'a': 826 case 'r': 827 appendnum++; 828 break; 829 case 'b': 830 case 't': 831 /* Resolve branch target. */ 832 if (cp->t == NULL) { 833 cp->u.c = NULL; 834 break; 835 } 836 if ((cp->u.c = findlabel(cp->t)) == NULL) 837 errx(1, "%lu: %s: undefined label '%s'", linenum, fname, cp->t); 838 free(cp->t); 839 break; 840 case '{': 841 /* Do interior commands. */ 842 fixuplabel(cp->u.c, cp->next); 843 break; 844 } 845 } 846 847 /* 848 * Associate the given command label for later lookup. 849 */ 850 static void 851 enterlabel(struct s_command *cp) 852 { 853 struct labhash **lhp, *lh; 854 u_char *p; 855 u_int h, c; 856 857 for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++) 858 h = (h << 5) + h + c; 859 lhp = &labels[h & LHMASK]; 860 for (lh = *lhp; lh != NULL; lh = lh->lh_next) 861 if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0) 862 errx(1, "%lu: %s: duplicate label '%s'", linenum, fname, cp->t); 863 if ((lh = malloc(sizeof *lh)) == NULL) 864 err(1, "malloc"); 865 lh->lh_next = *lhp; 866 lh->lh_hash = h; 867 lh->lh_cmd = cp; 868 lh->lh_ref = 0; 869 *lhp = lh; 870 } 871 872 /* 873 * Find the label contained in the command l in the command linked 874 * list cp. L is excluded from the search. Return NULL if not found. 875 */ 876 static struct s_command * 877 findlabel(char *name) 878 { 879 struct labhash *lh; 880 u_char *p; 881 u_int h, c; 882 883 for (h = 0, p = (u_char *)name; (c = *p) != 0; p++) 884 h = (h << 5) + h + c; 885 for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) { 886 if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) { 887 lh->lh_ref = 1; 888 return (lh->lh_cmd); 889 } 890 } 891 return (NULL); 892 } 893 894 /* 895 * Warn about any unused labels. As a side effect, release the label hash 896 * table space. 897 */ 898 static void 899 uselabel(void) 900 { 901 struct labhash *lh, *next; 902 int i; 903 904 for (i = 0; i < LHSZ; i++) { 905 for (lh = labels[i]; lh != NULL; lh = next) { 906 next = lh->lh_next; 907 if (!lh->lh_ref) 908 warnx("%lu: %s: unused label '%s'", 909 linenum, fname, lh->lh_cmd->t); 910 free(lh); 911 } 912 } 913 } 914