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