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 1.31 2008/02/09 09:12:02 dwmalone 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->nonsel = cmd->inrange = 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 = ! cmd->nonsel; 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 --p; 325 p = compile_subst(p, cmd->u.s); 326 p = compile_flags(p, cmd->u.s); 327 if (*re == '\0') 328 cmd->u.s->re = NULL; 329 else 330 cmd->u.s->re = compile_re(re, cmd->u.s->icase); 331 EATSPACE(); 332 if (*p == ';') { 333 p++; 334 link = &cmd->next; 335 goto semicolon; 336 } 337 break; 338 case TR: /* y */ 339 p++; 340 p = compile_tr(p, &cmd->u.y); 341 EATSPACE(); 342 if (*p == ';') { 343 p++; 344 link = &cmd->next; 345 goto semicolon; 346 } 347 if (*p) 348 errx(1, 349 "%lu: %s: extra text at the end of a transform command", linenum, fname); 350 break; 351 } 352 } 353 } 354 355 /* 356 * Get a delimited string. P points to the delimeter of the string; d points 357 * to a buffer area. Newline and delimiter escapes are processed; other 358 * escapes are ignored. 359 * 360 * Returns a pointer to the first character after the final delimiter or NULL 361 * in the case of a non-terminated string. The character array d is filled 362 * with the processed string. 363 */ 364 static char * 365 compile_delimited(char *p, char *d) 366 { 367 char c; 368 369 c = *p++; 370 if (c == '\0') 371 return (NULL); 372 else if (c == '\\') 373 errx(1, "%lu: %s: \\ can not be used as a string delimiter", 374 linenum, fname); 375 else if (c == '\n') 376 errx(1, "%lu: %s: newline can not be used as a string delimiter", 377 linenum, fname); 378 while (*p) { 379 if (*p == '[') { 380 if ((d = compile_ccl(&p, d)) == NULL) 381 errx(1, "%lu: %s: unbalanced brackets ([])", linenum, fname); 382 continue; 383 } else if (*p == '\\' && p[1] == '[') { 384 *d++ = *p++; 385 } else if (*p == '\\' && p[1] == c) 386 p++; 387 else if (*p == '\\' && p[1] == 'n') { 388 *d++ = '\n'; 389 p += 2; 390 continue; 391 } else if (*p == '\\' && p[1] == '\\') 392 *d++ = *p++; 393 else if (*p == c) { 394 *d = '\0'; 395 return (p + 1); 396 } 397 *d++ = *p++; 398 } 399 return (NULL); 400 } 401 402 403 /* compile_ccl: expand a POSIX character class */ 404 static char * 405 compile_ccl(char **sp, char *t) 406 { 407 int c, d; 408 char *s = *sp; 409 410 *t++ = *s++; 411 if (*s == '^') 412 *t++ = *s++; 413 if (*s == ']') 414 *t++ = *s++; 415 for (; *s && (*t = *s) != ']'; s++, t++) 416 if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) { 417 *++t = *++s, t++, s++; 418 for (c = *s; (*t = *s) != ']' || c != d; s++, t++) 419 if ((c = *s) == '\0') 420 return NULL; 421 } else if (*s == '\\' && s[1] == 'n') 422 *t = '\n', s++; 423 return (*s == ']') ? *sp = ++s, ++t : NULL; 424 } 425 426 /* 427 * Compiles the regular expression in RE and returns a pointer to the compiled 428 * regular expression. 429 * Cflags are passed to regcomp. 430 */ 431 static regex_t * 432 compile_re(char *re, int case_insensitive) 433 { 434 regex_t *rep; 435 int eval, flags; 436 437 438 flags = rflags; 439 if (case_insensitive) 440 flags |= REG_ICASE; 441 if ((rep = malloc(sizeof(regex_t))) == NULL) 442 err(1, "malloc"); 443 if ((eval = regcomp(rep, re, flags)) != 0) 444 errx(1, "%lu: %s: RE error: %s", 445 linenum, fname, strregerror(eval, rep)); 446 if (maxnsub < rep->re_nsub) 447 maxnsub = rep->re_nsub; 448 return (rep); 449 } 450 451 /* 452 * Compile the substitution string of a regular expression and set res to 453 * point to a saved copy of it. Nsub is the number of parenthesized regular 454 * expressions. 455 */ 456 static char * 457 compile_subst(char *p, struct s_subst *s) 458 { 459 static char lbuf[_POSIX2_LINE_MAX + 1]; 460 int asize, size; 461 u_char ref; 462 char c, *text, *op, *sp; 463 int more = 1, sawesc = 0; 464 465 c = *p++; /* Terminator character */ 466 if (c == '\0') 467 return (NULL); 468 469 s->maxbref = 0; 470 s->linenum = linenum; 471 asize = 2 * _POSIX2_LINE_MAX + 1; 472 if ((text = malloc(asize)) == NULL) 473 err(1, "malloc"); 474 size = 0; 475 do { 476 op = sp = text + size; 477 for (; *p; p++) { 478 if (*p == '\\' || sawesc) { 479 /* 480 * If this is a continuation from the last 481 * buffer, we won't have a character to 482 * skip over. 483 */ 484 if (sawesc) 485 sawesc = 0; 486 else 487 p++; 488 489 if (*p == '\0') { 490 /* 491 * This escaped character is continued 492 * in the next part of the line. Note 493 * this fact, then cause the loop to 494 * exit w/ normal EOL case and reenter 495 * above with the new buffer. 496 */ 497 sawesc = 1; 498 p--; 499 continue; 500 } else if (strchr("123456789", *p) != NULL) { 501 *sp++ = '\\'; 502 ref = *p - '0'; 503 if (s->re != NULL && 504 ref > s->re->re_nsub) 505 errx(1, "%lu: %s: \\%c not defined in the RE", 506 linenum, fname, *p); 507 if (s->maxbref < ref) 508 s->maxbref = ref; 509 } else if (*p == '&' || *p == '\\') 510 *sp++ = '\\'; 511 } else if (*p == c) { 512 if (*++p == '\0' && more) { 513 if (cu_fgets(lbuf, sizeof(lbuf), &more)) 514 p = lbuf; 515 } 516 *sp++ = '\0'; 517 size += sp - op; 518 if ((s->new = realloc(text, size)) == NULL) 519 err(1, "realloc"); 520 return (p); 521 } else if (*p == '\n') { 522 errx(1, 523 "%lu: %s: unescaped newline inside substitute pattern", linenum, fname); 524 /* NOTREACHED */ 525 } 526 *sp++ = *p; 527 } 528 size += sp - op; 529 if (asize - size < _POSIX2_LINE_MAX + 1) { 530 asize *= 2; 531 if ((text = realloc(text, asize)) == NULL) 532 err(1, "realloc"); 533 } 534 } while (cu_fgets(p = lbuf, sizeof(lbuf), &more)); 535 errx(1, "%lu: %s: unterminated substitute in regular expression", 536 linenum, fname); 537 /* NOTREACHED */ 538 } 539 540 /* 541 * Compile the flags of the s command 542 */ 543 static char * 544 compile_flags(char *p, struct s_subst *s) 545 { 546 int gn; /* True if we have seen g or n */ 547 unsigned long nval; 548 char wfile[_POSIX2_LINE_MAX + 1], *q; 549 550 s->n = 1; /* Default */ 551 s->p = 0; 552 s->wfile = NULL; 553 s->wfd = -1; 554 s->icase = 0; 555 for (gn = 0;;) { 556 EATSPACE(); /* EXTENSION */ 557 switch (*p) { 558 case 'g': 559 if (gn) 560 errx(1, 561 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname); 562 gn = 1; 563 s->n = 0; 564 break; 565 case '\0': 566 case '\n': 567 case ';': 568 return (p); 569 case 'p': 570 s->p = 1; 571 break; 572 case 'I': 573 s->icase = 1; 574 break; 575 case '1': case '2': case '3': 576 case '4': case '5': case '6': 577 case '7': case '8': case '9': 578 if (gn) 579 errx(1, 580 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname); 581 gn = 1; 582 errno = 0; 583 nval = strtol(p, &p, 10); 584 if (errno == ERANGE || nval > INT_MAX) 585 errx(1, 586 "%lu: %s: overflow in the 'N' substitute flag", linenum, fname); 587 s->n = nval; 588 p--; 589 break; 590 case 'w': 591 p++; 592 #ifdef HISTORIC_PRACTICE 593 if (*p != ' ') { 594 warnx("%lu: %s: space missing before w wfile", linenum, fname); 595 return (p); 596 } 597 #endif 598 EATSPACE(); 599 q = wfile; 600 while (*p) { 601 if (*p == '\n') 602 break; 603 *q++ = *p++; 604 } 605 *q = '\0'; 606 if (q == wfile) 607 errx(1, "%lu: %s: no wfile specified", linenum, fname); 608 s->wfile = strdup(wfile); 609 if (!aflag && (s->wfd = open(wfile, 610 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 611 DEFFILEMODE)) == -1) 612 err(1, "%s", wfile); 613 return (p); 614 default: 615 errx(1, "%lu: %s: bad flag in substitute command: '%c'", 616 linenum, fname, *p); 617 break; 618 } 619 p++; 620 } 621 } 622 623 /* 624 * Compile a translation set of strings into a lookup table. 625 */ 626 static char * 627 compile_tr(char *p, struct s_tr **py) 628 { 629 struct s_tr *y; 630 int i; 631 const char *op, *np; 632 char old[_POSIX2_LINE_MAX + 1]; 633 char new[_POSIX2_LINE_MAX + 1]; 634 size_t oclen, oldlen, nclen, newlen; 635 mbstate_t mbs1, mbs2; 636 637 if ((*py = y = malloc(sizeof(*y))) == NULL) 638 err(1, NULL); 639 y->multis = NULL; 640 y->nmultis = 0; 641 642 if (*p == '\0' || *p == '\\') 643 errx(1, 644 "%lu: %s: transform pattern can not be delimited by newline or backslash", 645 linenum, fname); 646 p = compile_delimited(p, old); 647 if (p == NULL) 648 errx(1, "%lu: %s: unterminated transform source string", 649 linenum, fname); 650 p = compile_delimited(p - 1, new); 651 if (p == NULL) 652 errx(1, "%lu: %s: unterminated transform target string", 653 linenum, fname); 654 EATSPACE(); 655 op = old; 656 oldlen = mbsrtowcs(NULL, &op, 0, NULL); 657 if (oldlen == (size_t)-1) 658 err(1, NULL); 659 np = new; 660 newlen = mbsrtowcs(NULL, &np, 0, NULL); 661 if (newlen == (size_t)-1) 662 err(1, NULL); 663 if (newlen != oldlen) 664 errx(1, "%lu: %s: transform strings are not the same length", 665 linenum, fname); 666 if (MB_CUR_MAX == 1) { 667 /* 668 * The single-byte encoding case is easy: generate a 669 * lookup table. 670 */ 671 for (i = 0; i <= UCHAR_MAX; i++) 672 y->bytetab[i] = (char)i; 673 for (; *op; op++, np++) 674 y->bytetab[(u_char)*op] = *np; 675 } else { 676 /* 677 * Multi-byte encoding case: generate a lookup table as 678 * above, but only for single-byte characters. The first 679 * bytes of multi-byte characters have their lookup table 680 * entries set to 0, which causes do_tr() to search through 681 * an auxiliary vector of multi-byte mappings. 682 */ 683 memset(&mbs1, 0, sizeof(mbs1)); 684 memset(&mbs2, 0, sizeof(mbs2)); 685 for (i = 0; i <= UCHAR_MAX; i++) 686 y->bytetab[i] = (btowc(i) != WEOF) ? i : 0; 687 while (*op != '\0') { 688 oclen = mbrlen(op, MB_LEN_MAX, &mbs1); 689 if (oclen == (size_t)-1 || oclen == (size_t)-2) 690 errc(1, EILSEQ, NULL); 691 nclen = mbrlen(np, MB_LEN_MAX, &mbs2); 692 if (nclen == (size_t)-1 || nclen == (size_t)-2) 693 errc(1, EILSEQ, NULL); 694 if (oclen == 1 && nclen == 1) 695 y->bytetab[(u_char)*op] = *np; 696 else { 697 y->bytetab[(u_char)*op] = 0; 698 y->multis = realloc(y->multis, 699 (y->nmultis + 1) * sizeof(*y->multis)); 700 if (y->multis == NULL) 701 err(1, NULL); 702 i = y->nmultis++; 703 y->multis[i].fromlen = oclen; 704 memcpy(y->multis[i].from, op, oclen); 705 y->multis[i].tolen = nclen; 706 memcpy(y->multis[i].to, np, nclen); 707 } 708 op += oclen; 709 np += nclen; 710 } 711 } 712 return (p); 713 } 714 715 /* 716 * Compile the text following an a or i command. 717 */ 718 static char * 719 compile_text(void) 720 { 721 int asize, esc_nl, size; 722 char *text, *p, *op, *s; 723 char lbuf[_POSIX2_LINE_MAX + 1]; 724 725 asize = 2 * _POSIX2_LINE_MAX + 1; 726 if ((text = malloc(asize)) == NULL) 727 err(1, "malloc"); 728 size = 0; 729 while (cu_fgets(lbuf, sizeof(lbuf), NULL)) { 730 op = s = text + size; 731 p = lbuf; 732 EATSPACE(); 733 for (esc_nl = 0; *p != '\0'; p++) { 734 if (*p == '\\' && p[1] != '\0' && *++p == '\n') 735 esc_nl = 1; 736 *s++ = *p; 737 } 738 size += s - op; 739 if (!esc_nl) { 740 *s = '\0'; 741 break; 742 } 743 if (asize - size < _POSIX2_LINE_MAX + 1) { 744 asize *= 2; 745 if ((text = realloc(text, asize)) == NULL) 746 err(1, "realloc"); 747 } 748 } 749 text[size] = '\0'; 750 if ((p = realloc(text, size + 1)) == NULL) 751 err(1, "realloc"); 752 return (p); 753 } 754 755 /* 756 * Get an address and return a pointer to the first character after 757 * it. Fill the structure pointed to according to the address. 758 */ 759 static char * 760 compile_addr(char *p, struct s_addr *a) 761 { 762 char *end, re[_POSIX2_LINE_MAX + 1]; 763 int icase; 764 765 icase = 0; 766 767 switch (*p) { 768 case '\\': /* Context address */ 769 ++p; 770 /* FALLTHROUGH */ 771 case '/': /* Context address */ 772 p = compile_delimited(p, re); 773 if (p == NULL) 774 errx(1, "%lu: %s: unterminated regular expression", linenum, fname); 775 /* Check for case insensitive regexp flag */ 776 if (*p == 'I') { 777 icase = 1; 778 p++; 779 } 780 if (*re == '\0') 781 a->u.r = NULL; 782 else 783 a->u.r = compile_re(re, icase); 784 a->type = AT_RE; 785 return (p); 786 787 case '$': /* Last line */ 788 a->type = AT_LAST; 789 return (p + 1); 790 /* Line number */ 791 case '0': case '1': case '2': case '3': case '4': 792 case '5': case '6': case '7': case '8': case '9': 793 a->type = AT_LINE; 794 a->u.l = strtol(p, &end, 10); 795 return (end); 796 default: 797 errx(1, "%lu: %s: expected context address", linenum, fname); 798 return (NULL); 799 } 800 } 801 802 /* 803 * duptoeol -- 804 * Return a copy of all the characters up to \n or \0. 805 */ 806 static char * 807 duptoeol(char *s, const char *ctype) 808 { 809 size_t len; 810 int ws; 811 char *p, *start; 812 813 ws = 0; 814 for (start = s; *s != '\0' && *s != '\n'; ++s) 815 ws = isspace((unsigned char)*s); 816 *s = '\0'; 817 if (ws) 818 warnx("%lu: %s: whitespace after %s", linenum, fname, ctype); 819 len = s - start + 1; 820 if ((p = malloc(len)) == NULL) 821 err(1, "malloc"); 822 return (memmove(p, start, len)); 823 } 824 825 /* 826 * Convert goto label names to addresses, and count a and r commands, in 827 * the given subset of the script. Free the memory used by labels in b 828 * and t commands (but not by :). 829 * 830 * TODO: Remove } nodes 831 */ 832 static void 833 fixuplabel(struct s_command *cp, struct s_command *end) 834 { 835 836 for (; cp != end; cp = cp->next) 837 switch (cp->code) { 838 case 'a': 839 case 'r': 840 appendnum++; 841 break; 842 case 'b': 843 case 't': 844 /* Resolve branch target. */ 845 if (cp->t == NULL) { 846 cp->u.c = NULL; 847 break; 848 } 849 if ((cp->u.c = findlabel(cp->t)) == NULL) 850 errx(1, "%lu: %s: undefined label '%s'", linenum, fname, cp->t); 851 free(cp->t); 852 break; 853 case '{': 854 /* Do interior commands. */ 855 fixuplabel(cp->u.c, cp->next); 856 break; 857 } 858 } 859 860 /* 861 * Associate the given command label for later lookup. 862 */ 863 static void 864 enterlabel(struct s_command *cp) 865 { 866 struct labhash **lhp, *lh; 867 u_char *p; 868 u_int h, c; 869 870 for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++) 871 h = (h << 5) + h + c; 872 lhp = &labels[h & LHMASK]; 873 for (lh = *lhp; lh != NULL; lh = lh->lh_next) 874 if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0) 875 errx(1, "%lu: %s: duplicate label '%s'", linenum, fname, cp->t); 876 if ((lh = malloc(sizeof *lh)) == NULL) 877 err(1, "malloc"); 878 lh->lh_next = *lhp; 879 lh->lh_hash = h; 880 lh->lh_cmd = cp; 881 lh->lh_ref = 0; 882 *lhp = lh; 883 } 884 885 /* 886 * Find the label contained in the command l in the command linked 887 * list cp. L is excluded from the search. Return NULL if not found. 888 */ 889 static struct s_command * 890 findlabel(char *name) 891 { 892 struct labhash *lh; 893 u_char *p; 894 u_int h, c; 895 896 for (h = 0, p = (u_char *)name; (c = *p) != 0; p++) 897 h = (h << 5) + h + c; 898 for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) { 899 if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) { 900 lh->lh_ref = 1; 901 return (lh->lh_cmd); 902 } 903 } 904 return (NULL); 905 } 906 907 /* 908 * Warn about any unused labels. As a side effect, release the label hash 909 * table space. 910 */ 911 static void 912 uselabel(void) 913 { 914 struct labhash *lh, *next; 915 int i; 916 917 for (i = 0; i < LHSZ; i++) { 918 for (lh = labels[i]; lh != NULL; lh = next) { 919 next = lh->lh_next; 920 if (!lh->lh_ref) 921 warnx("%lu: %s: unused label '%s'", 922 linenum, fname, lh->lh_cmd->t); 923 free(lh); 924 } 925 } 926 } 927