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 * @(#)compile.c 8.1 (Berkeley) 6/6/93 34 * $FreeBSD: src/usr.bin/sed/compile.c,v 1.34 2009/05/25 06:45:33 brian Exp $ 35 * $DragonFly: src/usr.bin/sed/compile.c,v 1.4 2008/04/08 13:23:38 swildner Exp $ 36 */ 37 38 #include <sys/types.h> 39 #include <sys/stat.h> 40 41 #include <ctype.h> 42 #include <err.h> 43 #include <errno.h> 44 #include <fcntl.h> 45 #include <limits.h> 46 #include <regex.h> 47 #include <stdio.h> 48 #include <stdlib.h> 49 #include <string.h> 50 #include <wchar.h> 51 52 #include "defs.h" 53 #include "extern.h" 54 55 #define LHSZ 128 56 #define LHMASK (LHSZ - 1) 57 static struct labhash { 58 struct labhash *lh_next; 59 u_int lh_hash; 60 struct s_command *lh_cmd; 61 int lh_ref; 62 } *labels[LHSZ]; 63 64 static char *compile_addr(char *, struct s_addr *); 65 static char *compile_ccl(char **, char *); 66 static char *compile_delimited(char *, char *); 67 static char *compile_flags(char *, struct s_subst *); 68 static regex_t *compile_re(char *, int); 69 static char *compile_subst(char *, struct s_subst *); 70 static char *compile_text(void); 71 static char *compile_tr(char *, struct s_tr **); 72 static struct s_command 73 **compile_stream(struct s_command **); 74 static char *duptoeol(char *, const char *); 75 static void enterlabel(struct s_command *); 76 static struct s_command 77 *findlabel(char *); 78 static void fixuplabel(struct s_command *, struct s_command *); 79 static void uselabel(void); 80 81 /* 82 * Command specification. This is used to drive the command parser. 83 */ 84 struct s_format { 85 char code; /* Command code */ 86 int naddr; /* Number of address args */ 87 enum e_args args; /* Argument type */ 88 }; 89 90 static struct s_format cmd_fmts[] = { 91 {'{', 2, GROUP}, 92 {'}', 0, ENDGROUP}, 93 {'a', 1, TEXT}, 94 {'b', 2, BRANCH}, 95 {'c', 2, TEXT}, 96 {'d', 2, EMPTY}, 97 {'D', 2, EMPTY}, 98 {'g', 2, EMPTY}, 99 {'G', 2, EMPTY}, 100 {'h', 2, EMPTY}, 101 {'H', 2, EMPTY}, 102 {'i', 1, TEXT}, 103 {'l', 2, EMPTY}, 104 {'n', 2, EMPTY}, 105 {'N', 2, EMPTY}, 106 {'p', 2, EMPTY}, 107 {'P', 2, EMPTY}, 108 {'q', 1, EMPTY}, 109 {'r', 1, RFILE}, 110 {'s', 2, SUBST}, 111 {'t', 2, BRANCH}, 112 {'w', 2, WFILE}, 113 {'x', 2, EMPTY}, 114 {'y', 2, TR}, 115 {'!', 2, NONSEL}, 116 {':', 0, LABEL}, 117 {'#', 0, COMMENT}, 118 {'=', 1, EMPTY}, 119 {'\0', 0, COMMENT}, 120 }; 121 122 /* The compiled program. */ 123 struct s_command *prog; 124 125 /* 126 * Compile the program into prog. 127 * Initialise appends. 128 */ 129 void 130 compile(void) 131 { 132 *compile_stream(&prog) = NULL; 133 fixuplabel(prog, NULL); 134 uselabel(); 135 if (appendnum == 0) 136 appends = NULL; 137 else if ((appends = malloc(sizeof(struct s_appends) * appendnum)) == 138 NULL) 139 err(1, "malloc"); 140 if ((match = malloc((maxnsub + 1) * sizeof(regmatch_t))) == NULL) 141 err(1, "malloc"); 142 } 143 144 #define EATSPACE() do { \ 145 if (p) \ 146 while (*p && isspace((unsigned char)*p)) \ 147 p++; \ 148 } while (0) 149 150 static struct s_command ** 151 compile_stream(struct s_command **link) 152 { 153 char *p; 154 static char lbuf[_POSIX2_LINE_MAX + 1]; /* To save stack */ 155 struct s_command *cmd, *cmd2, *stack; 156 struct s_format *fp; 157 char re[_POSIX2_LINE_MAX + 1]; 158 int naddr; /* Number of addresses */ 159 160 stack = 0; 161 for (;;) { 162 if ((p = cu_fgets(lbuf, sizeof(lbuf), NULL)) == NULL) { 163 if (stack != 0) 164 errx(1, "%lu: %s: unexpected EOF (pending }'s)", 165 linenum, fname); 166 return (link); 167 } 168 169 semicolon: EATSPACE(); 170 if (p) { 171 if (*p == '#' || *p == '\0') 172 continue; 173 else if (*p == ';') { 174 p++; 175 goto semicolon; 176 } 177 } 178 if ((*link = cmd = malloc(sizeof(struct s_command))) == NULL) 179 err(1, "malloc"); 180 link = &cmd->next; 181 cmd->startline = cmd->nonsel = 0; 182 /* First parse the addresses */ 183 naddr = 0; 184 185 /* Valid characters to start an address */ 186 #define addrchar(c) (strchr("0123456789/\\$", (c))) 187 if (addrchar(*p)) { 188 naddr++; 189 if ((cmd->a1 = malloc(sizeof(struct s_addr))) == NULL) 190 err(1, "malloc"); 191 p = compile_addr(p, cmd->a1); 192 EATSPACE(); /* EXTENSION */ 193 if (*p == ',') { 194 p++; 195 EATSPACE(); /* EXTENSION */ 196 naddr++; 197 if ((cmd->a2 = malloc(sizeof(struct s_addr))) 198 == NULL) 199 err(1, "malloc"); 200 p = compile_addr(p, cmd->a2); 201 EATSPACE(); 202 } else 203 cmd->a2 = 0; 204 } else 205 cmd->a1 = cmd->a2 = 0; 206 207 nonsel: /* Now parse the command */ 208 if (!*p) 209 errx(1, "%lu: %s: command expected", linenum, fname); 210 cmd->code = *p; 211 for (fp = cmd_fmts; fp->code; fp++) 212 if (fp->code == *p) 213 break; 214 if (!fp->code) 215 errx(1, "%lu: %s: invalid command code %c", linenum, fname, *p); 216 if (naddr > fp->naddr) 217 errx(1, 218 "%lu: %s: command %c expects up to %d address(es), found %d", 219 linenum, fname, *p, fp->naddr, naddr); 220 switch (fp->args) { 221 case NONSEL: /* ! */ 222 p++; 223 EATSPACE(); 224 cmd->nonsel = 1; 225 goto nonsel; 226 case GROUP: /* { */ 227 p++; 228 EATSPACE(); 229 cmd->next = stack; 230 stack = cmd; 231 link = &cmd->u.c; 232 if (*p) 233 goto semicolon; 234 break; 235 case ENDGROUP: 236 /* 237 * Short-circuit command processing, since end of 238 * group is really just a noop. 239 */ 240 cmd->nonsel = 1; 241 if (stack == 0) 242 errx(1, "%lu: %s: unexpected }", linenum, fname); 243 cmd2 = stack; 244 stack = cmd2->next; 245 cmd2->next = cmd; 246 /*FALLTHROUGH*/ 247 case EMPTY: /* d D g G h H l n N p P q x = \0 */ 248 p++; 249 EATSPACE(); 250 if (*p == ';') { 251 p++; 252 link = &cmd->next; 253 goto semicolon; 254 } 255 if (*p) 256 errx(1, "%lu: %s: extra characters at the end of %c command", 257 linenum, fname, cmd->code); 258 break; 259 case TEXT: /* a c i */ 260 p++; 261 EATSPACE(); 262 if (*p != '\\') 263 errx(1, 264 "%lu: %s: command %c expects \\ followed by text", linenum, fname, cmd->code); 265 p++; 266 EATSPACE(); 267 if (*p) 268 errx(1, 269 "%lu: %s: extra characters after \\ at the end of %c command", 270 linenum, fname, cmd->code); 271 cmd->t = compile_text(); 272 break; 273 case COMMENT: /* \0 # */ 274 break; 275 case WFILE: /* w */ 276 p++; 277 EATSPACE(); 278 if (*p == '\0') 279 errx(1, "%lu: %s: filename expected", linenum, fname); 280 cmd->t = duptoeol(p, "w command"); 281 if (aflag) 282 cmd->u.fd = -1; 283 else if ((cmd->u.fd = open(p, 284 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 285 DEFFILEMODE)) == -1) 286 err(1, "%s", p); 287 break; 288 case RFILE: /* r */ 289 p++; 290 EATSPACE(); 291 if (*p == '\0') 292 errx(1, "%lu: %s: filename expected", linenum, fname); 293 else 294 cmd->t = duptoeol(p, "read command"); 295 break; 296 case BRANCH: /* b t */ 297 p++; 298 EATSPACE(); 299 if (*p == '\0') 300 cmd->t = NULL; 301 else 302 cmd->t = duptoeol(p, "branch"); 303 break; 304 case LABEL: /* : */ 305 p++; 306 EATSPACE(); 307 cmd->t = duptoeol(p, "label"); 308 if (strlen(p) == 0) 309 errx(1, "%lu: %s: empty label", linenum, fname); 310 enterlabel(cmd); 311 break; 312 case SUBST: /* s */ 313 p++; 314 if (*p == '\0' || *p == '\\') 315 errx(1, 316 "%lu: %s: substitute pattern can not be delimited by newline or backslash", 317 linenum, fname); 318 if ((cmd->u.s = calloc(1, sizeof(struct s_subst))) == NULL) 319 err(1, "malloc"); 320 p = compile_delimited(p, re); 321 if (p == NULL) 322 errx(1, 323 "%lu: %s: unterminated substitute pattern", linenum, fname); 324 325 /* Compile RE with no case sensitivity temporarily */ 326 if (*re == '\0') 327 cmd->u.s->re = NULL; 328 else 329 cmd->u.s->re = compile_re(re, 0); 330 --p; 331 p = compile_subst(p, cmd->u.s); 332 p = compile_flags(p, cmd->u.s); 333 334 /* Recompile RE with case sensitivity from "I" flag if any */ 335 if (*re == '\0') 336 cmd->u.s->re = NULL; 337 else 338 cmd->u.s->re = compile_re(re, cmd->u.s->icase); 339 EATSPACE(); 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 delimeter 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 char * 373 compile_delimited(char *p, char *d) 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 == '[') { 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 *d++ = *p++; 401 else if (*p == c) { 402 *d = '\0'; 403 return (p + 1); 404 } 405 *d++ = *p++; 406 } 407 return (NULL); 408 } 409 410 411 /* compile_ccl: expand a POSIX character class */ 412 static char * 413 compile_ccl(char **sp, char *t) 414 { 415 int c, d; 416 char *s = *sp; 417 418 *t++ = *s++; 419 if (*s == '^') 420 *t++ = *s++; 421 if (*s == ']') 422 *t++ = *s++; 423 for (; *s && (*t = *s) != ']'; s++, t++) 424 if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) { 425 *++t = *++s, t++, s++; 426 for (c = *s; (*t = *s) != ']' || c != d; s++, t++) 427 if ((c = *s) == '\0') 428 return NULL; 429 } else if (*s == '\\' && s[1] == 'n') 430 *t = '\n', s++; 431 return (*s == ']') ? *sp = ++s, ++t : NULL; 432 } 433 434 /* 435 * Compiles the regular expression in RE and returns a pointer to the compiled 436 * regular expression. 437 * Cflags are passed to regcomp. 438 */ 439 static regex_t * 440 compile_re(char *re, int case_insensitive) 441 { 442 regex_t *rep; 443 int eval, flags; 444 445 446 flags = rflags; 447 if (case_insensitive) 448 flags |= REG_ICASE; 449 if ((rep = malloc(sizeof(regex_t))) == NULL) 450 err(1, "malloc"); 451 if ((eval = regcomp(rep, re, flags)) != 0) 452 errx(1, "%lu: %s: RE error: %s", 453 linenum, fname, strregerror(eval, rep)); 454 if (maxnsub < rep->re_nsub) 455 maxnsub = rep->re_nsub; 456 return (rep); 457 } 458 459 /* 460 * Compile the substitution string of a regular expression and set res to 461 * point to a saved copy of it. Nsub is the number of parenthesized regular 462 * expressions. 463 */ 464 static char * 465 compile_subst(char *p, struct s_subst *s) 466 { 467 static char lbuf[_POSIX2_LINE_MAX + 1]; 468 int asize, size; 469 u_char ref; 470 char c, *text, *op, *sp; 471 int more = 1, sawesc = 0; 472 473 c = *p++; /* Terminator character */ 474 if (c == '\0') 475 return (NULL); 476 477 s->maxbref = 0; 478 s->linenum = linenum; 479 asize = 2 * _POSIX2_LINE_MAX + 1; 480 if ((text = malloc(asize)) == NULL) 481 err(1, "malloc"); 482 size = 0; 483 do { 484 op = sp = text + size; 485 for (; *p; p++) { 486 if (*p == '\\' || sawesc) { 487 /* 488 * If this is a continuation from the last 489 * buffer, we won't have a character to 490 * skip over. 491 */ 492 if (sawesc) 493 sawesc = 0; 494 else 495 p++; 496 497 if (*p == '\0') { 498 /* 499 * This escaped character is continued 500 * in the next part of the line. Note 501 * this fact, then cause the loop to 502 * exit w/ normal EOL case and reenter 503 * above with the new buffer. 504 */ 505 sawesc = 1; 506 p--; 507 continue; 508 } else if (strchr("123456789", *p) != NULL) { 509 *sp++ = '\\'; 510 ref = *p - '0'; 511 if (s->re != NULL && 512 ref > s->re->re_nsub) 513 errx(1, "%lu: %s: \\%c not defined in the RE", 514 linenum, fname, *p); 515 if (s->maxbref < ref) 516 s->maxbref = ref; 517 } else if (*p == '&' || *p == '\\') 518 *sp++ = '\\'; 519 } else if (*p == c) { 520 if (*++p == '\0' && more) { 521 if (cu_fgets(lbuf, sizeof(lbuf), &more)) 522 p = lbuf; 523 } 524 *sp++ = '\0'; 525 size += sp - op; 526 if ((s->new = realloc(text, size)) == NULL) 527 err(1, "realloc"); 528 return (p); 529 } else if (*p == '\n') { 530 errx(1, 531 "%lu: %s: unescaped newline inside substitute pattern", linenum, fname); 532 /* NOTREACHED */ 533 } 534 *sp++ = *p; 535 } 536 size += sp - op; 537 if (asize - size < _POSIX2_LINE_MAX + 1) { 538 asize *= 2; 539 if ((text = realloc(text, asize)) == NULL) 540 err(1, "realloc"); 541 } 542 } while (cu_fgets(p = lbuf, sizeof(lbuf), &more)); 543 errx(1, "%lu: %s: unterminated substitute in regular expression", 544 linenum, fname); 545 /* NOTREACHED */ 546 } 547 548 /* 549 * Compile the flags of the s command 550 */ 551 static char * 552 compile_flags(char *p, struct s_subst *s) 553 { 554 int gn; /* True if we have seen g or n */ 555 unsigned long nval; 556 char wfile[_POSIX2_LINE_MAX + 1], *q; 557 558 s->n = 1; /* Default */ 559 s->p = 0; 560 s->wfile = NULL; 561 s->wfd = -1; 562 s->icase = 0; 563 for (gn = 0;;) { 564 EATSPACE(); /* EXTENSION */ 565 switch (*p) { 566 case 'g': 567 if (gn) 568 errx(1, 569 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname); 570 gn = 1; 571 s->n = 0; 572 break; 573 case '\0': 574 case '\n': 575 case ';': 576 return (p); 577 case 'p': 578 s->p = 1; 579 break; 580 case 'I': 581 s->icase = 1; 582 break; 583 case '1': case '2': case '3': 584 case '4': case '5': case '6': 585 case '7': case '8': case '9': 586 if (gn) 587 errx(1, 588 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname); 589 gn = 1; 590 errno = 0; 591 nval = strtol(p, &p, 10); 592 if (errno == ERANGE || nval > INT_MAX) 593 errx(1, 594 "%lu: %s: overflow in the 'N' substitute flag", linenum, fname); 595 s->n = nval; 596 p--; 597 break; 598 case 'w': 599 p++; 600 #ifdef HISTORIC_PRACTICE 601 if (*p != ' ') { 602 warnx("%lu: %s: space missing before w wfile", linenum, fname); 603 return (p); 604 } 605 #endif 606 EATSPACE(); 607 q = wfile; 608 while (*p) { 609 if (*p == '\n') 610 break; 611 *q++ = *p++; 612 } 613 *q = '\0'; 614 if (q == wfile) 615 errx(1, "%lu: %s: no wfile specified", linenum, fname); 616 s->wfile = strdup(wfile); 617 if (!aflag && (s->wfd = open(wfile, 618 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 619 DEFFILEMODE)) == -1) 620 err(1, "%s", wfile); 621 return (p); 622 default: 623 errx(1, "%lu: %s: bad flag in substitute command: '%c'", 624 linenum, fname, *p); 625 break; 626 } 627 p++; 628 } 629 } 630 631 /* 632 * Compile a translation set of strings into a lookup table. 633 */ 634 static char * 635 compile_tr(char *p, struct s_tr **py) 636 { 637 struct s_tr *y; 638 int i; 639 const char *op, *np; 640 char old[_POSIX2_LINE_MAX + 1]; 641 char new[_POSIX2_LINE_MAX + 1]; 642 size_t oclen, oldlen, nclen, newlen; 643 mbstate_t mbs1, mbs2; 644 645 if ((*py = y = malloc(sizeof(*y))) == NULL) 646 err(1, NULL); 647 y->multis = NULL; 648 y->nmultis = 0; 649 650 if (*p == '\0' || *p == '\\') 651 errx(1, 652 "%lu: %s: transform pattern can not be delimited by newline or backslash", 653 linenum, fname); 654 p = compile_delimited(p, old); 655 if (p == NULL) 656 errx(1, "%lu: %s: unterminated transform source string", 657 linenum, fname); 658 p = compile_delimited(p - 1, new); 659 if (p == NULL) 660 errx(1, "%lu: %s: unterminated transform target string", 661 linenum, fname); 662 EATSPACE(); 663 op = old; 664 oldlen = mbsrtowcs(NULL, &op, 0, NULL); 665 if (oldlen == (size_t)-1) 666 err(1, NULL); 667 np = new; 668 newlen = mbsrtowcs(NULL, &np, 0, NULL); 669 if (newlen == (size_t)-1) 670 err(1, NULL); 671 if (newlen != oldlen) 672 errx(1, "%lu: %s: transform strings are not the same length", 673 linenum, fname); 674 if (MB_CUR_MAX == 1) { 675 /* 676 * The single-byte encoding case is easy: generate a 677 * lookup table. 678 */ 679 for (i = 0; i <= UCHAR_MAX; i++) 680 y->bytetab[i] = (char)i; 681 for (; *op; op++, np++) 682 y->bytetab[(u_char)*op] = *np; 683 } else { 684 /* 685 * Multi-byte encoding case: generate a lookup table as 686 * above, but only for single-byte characters. The first 687 * bytes of multi-byte characters have their lookup table 688 * entries set to 0, which causes do_tr() to search through 689 * an auxiliary vector of multi-byte mappings. 690 */ 691 memset(&mbs1, 0, sizeof(mbs1)); 692 memset(&mbs2, 0, sizeof(mbs2)); 693 for (i = 0; i <= UCHAR_MAX; i++) 694 y->bytetab[i] = (btowc(i) != WEOF) ? i : 0; 695 while (*op != '\0') { 696 oclen = mbrlen(op, MB_LEN_MAX, &mbs1); 697 if (oclen == (size_t)-1 || oclen == (size_t)-2) 698 errc(1, EILSEQ, NULL); 699 nclen = mbrlen(np, MB_LEN_MAX, &mbs2); 700 if (nclen == (size_t)-1 || nclen == (size_t)-2) 701 errc(1, EILSEQ, NULL); 702 if (oclen == 1 && nclen == 1) 703 y->bytetab[(u_char)*op] = *np; 704 else { 705 y->bytetab[(u_char)*op] = 0; 706 y->multis = realloc(y->multis, 707 (y->nmultis + 1) * sizeof(*y->multis)); 708 if (y->multis == NULL) 709 err(1, NULL); 710 i = y->nmultis++; 711 y->multis[i].fromlen = oclen; 712 memcpy(y->multis[i].from, op, oclen); 713 y->multis[i].tolen = nclen; 714 memcpy(y->multis[i].to, np, nclen); 715 } 716 op += oclen; 717 np += nclen; 718 } 719 } 720 return (p); 721 } 722 723 /* 724 * Compile the text following an a or i command. 725 */ 726 static char * 727 compile_text(void) 728 { 729 int asize, esc_nl, size; 730 char *text, *p, *op, *s; 731 char lbuf[_POSIX2_LINE_MAX + 1]; 732 733 asize = 2 * _POSIX2_LINE_MAX + 1; 734 if ((text = malloc(asize)) == NULL) 735 err(1, "malloc"); 736 size = 0; 737 while (cu_fgets(lbuf, sizeof(lbuf), NULL)) { 738 op = s = text + size; 739 p = lbuf; 740 EATSPACE(); 741 for (esc_nl = 0; *p != '\0'; p++) { 742 if (*p == '\\' && p[1] != '\0' && *++p == '\n') 743 esc_nl = 1; 744 *s++ = *p; 745 } 746 size += s - op; 747 if (!esc_nl) { 748 *s = '\0'; 749 break; 750 } 751 if (asize - size < _POSIX2_LINE_MAX + 1) { 752 asize *= 2; 753 if ((text = realloc(text, asize)) == NULL) 754 err(1, "realloc"); 755 } 756 } 757 text[size] = '\0'; 758 if ((p = realloc(text, size + 1)) == NULL) 759 err(1, "realloc"); 760 return (p); 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 char * 768 compile_addr(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); 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(char *s, const char *ctype) 823 { 824 size_t len; 825 int ws; 826 char *p, *start; 827 828 ws = 0; 829 for (start = s; *s != '\0' && *s != '\n'; ++s) 830 ws = isspace((unsigned char)*s); 831 *s = '\0'; 832 if (ws) 833 warnx("%lu: %s: whitespace after %s", linenum, fname, ctype); 834 len = s - start + 1; 835 if ((p = malloc(len)) == NULL) 836 err(1, "malloc"); 837 return (memmove(p, start, len)); 838 } 839 840 /* 841 * Convert goto label names to addresses, and count a and r commands, in 842 * the given subset of the script. Free the memory used by labels in b 843 * and t commands (but not by :). 844 * 845 * TODO: Remove } nodes 846 */ 847 static void 848 fixuplabel(struct s_command *cp, struct s_command *end) 849 { 850 851 for (; cp != end; cp = cp->next) 852 switch (cp->code) { 853 case 'a': 854 case 'r': 855 appendnum++; 856 break; 857 case 'b': 858 case 't': 859 /* Resolve branch target. */ 860 if (cp->t == NULL) { 861 cp->u.c = NULL; 862 break; 863 } 864 if ((cp->u.c = findlabel(cp->t)) == NULL) 865 errx(1, "%lu: %s: undefined label '%s'", linenum, fname, cp->t); 866 free(cp->t); 867 break; 868 case '{': 869 /* Do interior commands. */ 870 fixuplabel(cp->u.c, cp->next); 871 break; 872 } 873 } 874 875 /* 876 * Associate the given command label for later lookup. 877 */ 878 static void 879 enterlabel(struct s_command *cp) 880 { 881 struct labhash **lhp, *lh; 882 u_char *p; 883 u_int h, c; 884 885 for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++) 886 h = (h << 5) + h + c; 887 lhp = &labels[h & LHMASK]; 888 for (lh = *lhp; lh != NULL; lh = lh->lh_next) 889 if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0) 890 errx(1, "%lu: %s: duplicate label '%s'", linenum, fname, cp->t); 891 if ((lh = malloc(sizeof *lh)) == NULL) 892 err(1, "malloc"); 893 lh->lh_next = *lhp; 894 lh->lh_hash = h; 895 lh->lh_cmd = cp; 896 lh->lh_ref = 0; 897 *lhp = lh; 898 } 899 900 /* 901 * Find the label contained in the command l in the command linked 902 * list cp. L is excluded from the search. Return NULL if not found. 903 */ 904 static struct s_command * 905 findlabel(char *name) 906 { 907 struct labhash *lh; 908 u_char *p; 909 u_int h, c; 910 911 for (h = 0, p = (u_char *)name; (c = *p) != 0; p++) 912 h = (h << 5) + h + c; 913 for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) { 914 if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) { 915 lh->lh_ref = 1; 916 return (lh->lh_cmd); 917 } 918 } 919 return (NULL); 920 } 921 922 /* 923 * Warn about any unused labels. As a side effect, release the label hash 924 * table space. 925 */ 926 static void 927 uselabel(void) 928 { 929 struct labhash *lh, *next; 930 int i; 931 932 for (i = 0; i < LHSZ; i++) { 933 for (lh = labels[i]; lh != NULL; lh = next) { 934 next = lh->lh_next; 935 if (!lh->lh_ref) 936 warnx("%lu: %s: unused label '%s'", 937 linenum, fname, lh->lh_cmd->t); 938 free(lh); 939 } 940 } 941 } 942