1 /* $NetBSD: compile.c,v 1.47 2016/04/05 00:13:03 christos Exp $ */ 2 3 /*- 4 * Copyright (c) 1992 Diomidis Spinellis. 5 * Copyright (c) 1992, 1993 6 * The Regents of the University of California. All rights reserved. 7 * 8 * This code is derived from software contributed to Berkeley by 9 * Diomidis Spinellis of Imperial College, University of London. 10 * 11 * Redistribution and use in source and binary forms, with or without 12 * modification, are permitted provided that the following conditions 13 * are met: 14 * 1. Redistributions of source code must retain the above copyright 15 * notice, this list of conditions and the following disclaimer. 16 * 2. Redistributions in binary form must reproduce the above copyright 17 * notice, this list of conditions and the following disclaimer in the 18 * documentation and/or other materials provided with the distribution. 19 * 3. Neither the name of the University nor the names of its contributors 20 * may be used to endorse or promote products derived from this software 21 * without specific prior written permission. 22 * 23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 33 * SUCH DAMAGE. 34 */ 35 36 #if HAVE_NBTOOL_CONFIG_H 37 #include "nbtool_config.h" 38 #endif 39 40 #include <sys/cdefs.h> 41 __RCSID("$NetBSD: compile.c,v 1.47 2016/04/05 00:13:03 christos Exp $"); 42 #ifdef __FBSDID 43 __FBSDID("$FreeBSD: head/usr.bin/sed/compile.c 259132 2013-12-09 18:57:20Z eadler $"); 44 #endif 45 46 #if 0 47 static const char sccsid[] = "@(#)compile.c 8.1 (Berkeley) 6/6/93"; 48 #endif 49 50 #include <sys/types.h> 51 #include <sys/stat.h> 52 53 #include <ctype.h> 54 #include <err.h> 55 #include <errno.h> 56 #include <fcntl.h> 57 #include <limits.h> 58 #include <regex.h> 59 #include <stdio.h> 60 #include <stdlib.h> 61 #include <string.h> 62 #include <wchar.h> 63 64 #include "defs.h" 65 #include "extern.h" 66 67 #define LHSZ 128 68 #define LHMASK (LHSZ - 1) 69 static struct labhash { 70 struct labhash *lh_next; 71 u_int lh_hash; 72 struct s_command *lh_cmd; 73 int lh_ref; 74 } *labels[LHSZ]; 75 76 static char *compile_addr(char *, struct s_addr *); 77 static char *compile_ccl(char **, char *); 78 static char *compile_delimited(char *, char *, int); 79 static char *compile_flags(char *, struct s_subst *); 80 static regex_t *compile_re(char *, int); 81 static char *compile_subst(char *, struct s_subst *); 82 static char *compile_text(void); 83 static char *compile_tr(char *, struct s_tr **); 84 static struct s_command 85 **compile_stream(struct s_command **); 86 static char *duptoeol(char *, const char *); 87 static void enterlabel(struct s_command *); 88 static struct s_command 89 *findlabel(char *); 90 static void fixuplabel(struct s_command *, struct s_command *); 91 static void uselabel(void); 92 93 /* 94 * Command specification. This is used to drive the command parser. 95 */ 96 struct s_format { 97 char code; /* Command code */ 98 int naddr; /* Number of address args */ 99 enum e_args args; /* Argument type */ 100 }; 101 102 static struct s_format cmd_fmts[] = { 103 {'{', 2, GROUP}, 104 {'}', 0, ENDGROUP}, 105 {'a', 1, TEXT}, 106 {'b', 2, BRANCH}, 107 {'c', 2, TEXT}, 108 {'d', 2, EMPTY}, 109 {'D', 2, EMPTY}, 110 {'g', 2, EMPTY}, 111 {'G', 2, EMPTY}, 112 {'h', 2, EMPTY}, 113 {'H', 2, EMPTY}, 114 {'i', 1, TEXT}, 115 {'l', 2, EMPTY}, 116 {'n', 2, EMPTY}, 117 {'N', 2, EMPTY}, 118 {'p', 2, EMPTY}, 119 {'P', 2, EMPTY}, 120 {'q', 1, EMPTY}, 121 {'r', 1, RFILE}, 122 {'s', 2, SUBST}, 123 {'t', 2, BRANCH}, 124 {'w', 2, WFILE}, 125 {'x', 2, EMPTY}, 126 {'y', 2, TR}, 127 {'!', 2, NONSEL}, 128 {':', 0, LABEL}, 129 {'#', 0, COMMENT}, 130 {'=', 1, EMPTY}, 131 {'\0', 0, COMMENT}, 132 }; 133 134 /* The compiled program. */ 135 struct s_command *prog; 136 137 /* 138 * Compile the program into prog. 139 * Initialise appends. 140 */ 141 void 142 compile(void) 143 { 144 *compile_stream(&prog) = NULL; 145 fixuplabel(prog, NULL); 146 uselabel(); 147 if (appendnum > 0) 148 appends = xmalloc(sizeof(struct s_appends) * appendnum); 149 match = xmalloc((maxnsub + 1) * sizeof(regmatch_t)); 150 } 151 152 #define EATSPACE() do { \ 153 if (p) \ 154 while (*p && isspace((unsigned char)*p)) \ 155 p++; \ 156 } while (0) 157 158 static struct s_command ** 159 compile_stream(struct s_command **link) 160 { 161 char *p; 162 static char lbuf[_POSIX2_LINE_MAX + 1]; /* To save stack */ 163 struct s_command *cmd, *cmd2, *stack; 164 struct s_format *fp; 165 char re[_POSIX2_LINE_MAX + 1]; 166 int naddr; /* Number of addresses */ 167 168 stack = 0; 169 for (;;) { 170 if ((p = cu_fgets(lbuf, sizeof(lbuf), NULL)) == NULL) { 171 if (stack != 0) 172 errx(1, "%lu: %s: unexpected EOF (pending }'s)", 173 linenum, fname); 174 return (link); 175 } 176 177 semicolon: EATSPACE(); 178 if (p) { 179 if (*p == '#' || *p == '\0') 180 continue; 181 else if (*p == ';') { 182 p++; 183 goto semicolon; 184 } 185 } 186 *link = cmd = xmalloc(sizeof(struct s_command)); 187 link = &cmd->next; 188 cmd->startline = cmd->nonsel = 0; 189 /* First parse the addresses */ 190 naddr = 0; 191 192 /* Valid characters to start an address */ 193 #define addrchar(c) (strchr("0123456789/\\$", (c))) 194 if (addrchar(*p)) { 195 naddr++; 196 cmd->a1 = xmalloc(sizeof(struct s_addr)); 197 p = compile_addr(p, cmd->a1); 198 EATSPACE(); /* EXTENSION */ 199 if (*p == ',') { 200 p++; 201 EATSPACE(); /* EXTENSION */ 202 naddr++; 203 cmd->a2 = xmalloc(sizeof(struct s_addr)); 204 p = compile_addr(p, cmd->a2); 205 EATSPACE(); 206 } else 207 cmd->a2 = 0; 208 } else 209 cmd->a1 = cmd->a2 = 0; 210 211 nonsel: /* Now parse the command */ 212 if (!*p) 213 errx(1, "%lu: %s: command expected", linenum, fname); 214 cmd->code = *p; 215 for (fp = cmd_fmts; fp->code; fp++) 216 if (fp->code == *p) 217 break; 218 if (!fp->code) 219 errx(1, "%lu: %s: invalid command code %c", linenum, fname, *p); 220 if (naddr > fp->naddr) 221 errx(1, 222 "%lu: %s: command %c expects up to %d address(es), found %d", 223 linenum, fname, *p, fp->naddr, naddr); 224 switch (fp->args) { 225 case NONSEL: /* ! */ 226 p++; 227 EATSPACE(); 228 cmd->nonsel = ! cmd->nonsel; 229 goto nonsel; 230 case GROUP: /* { */ 231 p++; 232 EATSPACE(); 233 cmd->next = stack; 234 stack = cmd; 235 link = &cmd->u.c; 236 if (*p) 237 goto semicolon; 238 break; 239 case ENDGROUP: 240 /* 241 * Short-circuit command processing, since end of 242 * group is really just a noop. 243 */ 244 cmd->nonsel = 1; 245 if (stack == 0) 246 errx(1, "%lu: %s: unexpected }", linenum, fname); 247 cmd2 = stack; 248 stack = cmd2->next; 249 cmd2->next = cmd; 250 /*FALLTHROUGH*/ 251 case EMPTY: /* d D g G h H l n N p P q x = \0 */ 252 p++; 253 EATSPACE(); 254 switch (*p) { 255 case ';': 256 p++; 257 link = &cmd->next; 258 goto semicolon; 259 case '}': 260 goto semicolon; 261 case '\0': 262 break; 263 default: 264 errx(1, "%lu: %s: extra characters at the end of %c command", 265 linenum, fname, cmd->code); 266 } 267 break; 268 case TEXT: /* a c i */ 269 p++; 270 EATSPACE(); 271 if (*p != '\\') 272 errx(1, 273 "%lu: %s: command %c expects \\ followed by text", linenum, fname, cmd->code); 274 p++; 275 EATSPACE(); 276 if (*p) 277 errx(1, 278 "%lu: %s: extra characters after \\ at the end of %c command", 279 linenum, fname, cmd->code); 280 cmd->t = compile_text(); 281 break; 282 case COMMENT: /* \0 # */ 283 break; 284 case WFILE: /* w */ 285 p++; 286 EATSPACE(); 287 if (*p == '\0') 288 errx(1, "%lu: %s: filename expected", linenum, fname); 289 cmd->t = duptoeol(p, "w command"); 290 if (aflag) 291 cmd->u.fd = -1; 292 else if ((cmd->u.fd = open(p, 293 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 294 DEFFILEMODE)) == -1) 295 err(1, "%s", p); 296 break; 297 case RFILE: /* r */ 298 p++; 299 EATSPACE(); 300 if (*p == '\0') 301 errx(1, "%lu: %s: filename expected", linenum, fname); 302 else 303 cmd->t = duptoeol(p, "read command"); 304 break; 305 case BRANCH: /* b t */ 306 p++; 307 EATSPACE(); 308 if (*p == '\0') 309 cmd->t = NULL; 310 else 311 cmd->t = duptoeol(p, "branch"); 312 break; 313 case LABEL: /* : */ 314 p++; 315 EATSPACE(); 316 cmd->t = duptoeol(p, "label"); 317 if (strlen(p) == 0) 318 errx(1, "%lu: %s: empty label", linenum, fname); 319 enterlabel(cmd); 320 break; 321 case SUBST: /* s */ 322 p++; 323 if (*p == '\0' || *p == '\\') 324 errx(1, 325 "%lu: %s: substitute pattern can not be delimited by newline or backslash", 326 linenum, fname); 327 cmd->u.s = xcalloc(1, sizeof(struct s_subst)); 328 p = compile_delimited(p, re, 0); 329 if (p == NULL) 330 errx(1, 331 "%lu: %s: unterminated substitute pattern", linenum, fname); 332 333 /* Compile RE with no case sensitivity temporarily */ 334 if (*re == '\0') 335 cmd->u.s->re = NULL; 336 else 337 cmd->u.s->re = compile_re(re, 0); 338 --p; 339 p = compile_subst(p, cmd->u.s); 340 p = compile_flags(p, cmd->u.s); 341 342 /* Recompile RE with case sensitivity from "I" flag if any */ 343 if (*re == '\0') 344 cmd->u.s->re = NULL; 345 else 346 cmd->u.s->re = compile_re(re, cmd->u.s->icase); 347 EATSPACE(); 348 if (*p == ';') { 349 p++; 350 link = &cmd->next; 351 goto semicolon; 352 } 353 break; 354 case TR: /* y */ 355 p++; 356 p = compile_tr(p, &cmd->u.y); 357 EATSPACE(); 358 switch (*p) { 359 case ';': 360 p++; 361 link = &cmd->next; 362 goto semicolon; 363 case '}': 364 goto semicolon; 365 case '\0': 366 break; 367 default: 368 errx(1, 369 "%lu: %s: extra text at the end of a transform command", linenum, fname); 370 } 371 if (*p) 372 break; 373 } 374 } 375 } 376 377 /* 378 * Get a delimited string. P points to the delimeter of the string; d points 379 * to a buffer area. Newline and delimiter escapes are processed; other 380 * escapes are ignored. 381 * 382 * Returns a pointer to the first character after the final delimiter or NULL 383 * in the case of a non-terminated string. The character array d is filled 384 * with the processed string. 385 */ 386 static char * 387 compile_delimited(char *p, char *d, int is_tr) 388 { 389 char c; 390 391 c = *p++; 392 if (c == '\0') 393 return (NULL); 394 else if (c == '\\') 395 errx(1, "%lu: %s: \\ can not be used as a string delimiter", 396 linenum, fname); 397 else if (c == '\n') 398 errx(1, "%lu: %s: newline can not be used as a string delimiter", 399 linenum, fname); 400 while (*p) { 401 if (*p == '[' && *p != c) { 402 if ((d = compile_ccl(&p, d)) == NULL) 403 errx(1, "%lu: %s: unbalanced brackets ([])", linenum, fname); 404 continue; 405 } else if (*p == '\\' && p[1] == '[') { 406 *d++ = *p++; 407 } else if (*p == '\\' && p[1] == c) 408 p++; 409 else if (*p == '\\' && p[1] == 'n') { 410 *d++ = '\n'; 411 p += 2; 412 continue; 413 } else if (*p == '\\' && p[1] == '\\') { 414 if (is_tr) 415 p++; 416 else 417 *d++ = *p++; 418 } else if (*p == c) { 419 *d = '\0'; 420 return (p + 1); 421 } 422 *d++ = *p++; 423 } 424 return (NULL); 425 } 426 427 428 /* compile_ccl: expand a POSIX character class */ 429 static char * 430 compile_ccl(char **sp, char *t) 431 { 432 int c, d; 433 char *s = *sp; 434 435 *t++ = *s++; 436 if (*s == '^') 437 *t++ = *s++; 438 if (*s == ']') 439 *t++ = *s++; 440 for (; *s && (*t = *s) != ']'; s++, t++) 441 if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) { 442 *++t = *++s, t++, s++; 443 for (c = *s; (*t = *s) != ']' || c != d; s++, t++) 444 if ((c = *s) == '\0') 445 return NULL; 446 } 447 return (*s == ']') ? *sp = ++s, ++t : NULL; 448 } 449 450 /* 451 * Compiles the regular expression in RE and returns a pointer to the compiled 452 * regular expression. 453 * Cflags are passed to regcomp. 454 */ 455 static regex_t * 456 compile_re(char *re, int case_insensitive) 457 { 458 regex_t *rep; 459 int eval, flags; 460 461 462 flags = rflags; 463 if (case_insensitive) 464 flags |= REG_ICASE; 465 rep = xmalloc(sizeof(regex_t)); 466 if ((eval = regcomp(rep, re, flags)) != 0) 467 errx(1, "%lu: %s: RE error: %s", 468 linenum, fname, strregerror(eval, rep)); 469 if (maxnsub < rep->re_nsub) 470 maxnsub = rep->re_nsub; 471 return (rep); 472 } 473 474 /* 475 * Compile the substitution string of a regular expression and set res to 476 * point to a saved copy of it. Nsub is the number of parenthesized regular 477 * expressions. 478 */ 479 static char * 480 compile_subst(char *p, struct s_subst *s) 481 { 482 static char lbuf[_POSIX2_LINE_MAX + 1]; 483 size_t asize, size; 484 u_char ref; 485 char c, *text, *op, *sp; 486 int more = 1, sawesc = 0; 487 488 c = *p++; /* Terminator character */ 489 if (c == '\0') 490 return (NULL); 491 492 s->maxbref = 0; 493 s->linenum = linenum; 494 asize = 2 * _POSIX2_LINE_MAX + 1; 495 text = xmalloc(asize); 496 size = 0; 497 do { 498 op = sp = text + size; 499 for (; *p; p++) { 500 if (*p == '\\' || sawesc) { 501 /* 502 * If this is a continuation from the last 503 * buffer, we won't have a character to 504 * skip over. 505 */ 506 if (sawesc) 507 sawesc = 0; 508 else 509 p++; 510 511 if (*p == '\0') { 512 /* 513 * This escaped character is continued 514 * in the next part of the line. Note 515 * this fact, then cause the loop to 516 * exit w/ normal EOL case and reenter 517 * above with the new buffer. 518 */ 519 sawesc = 1; 520 p--; 521 continue; 522 } else if (strchr("123456789", *p) != NULL) { 523 *sp++ = '\\'; 524 ref = (u_char)(*p - '0'); 525 if (s->re != NULL && 526 ref > s->re->re_nsub) 527 errx(1, "%lu: %s: \\%c not defined in the RE", 528 linenum, fname, *p); 529 if (s->maxbref < ref) 530 s->maxbref = ref; 531 } else if (*p == '&' || *p == '\\') 532 *sp++ = '\\'; 533 } else if (*p == c) { 534 if (*++p == '\0' && more) { 535 if (cu_fgets(lbuf, sizeof(lbuf), &more)) 536 p = lbuf; 537 } 538 *sp++ = '\0'; 539 size += (size_t)(sp - op); 540 s->new = xrealloc(text, size); 541 return (p); 542 } else if (*p == '\n') { 543 errx(1, 544 "%lu: %s: unescaped newline inside substitute pattern", linenum, fname); 545 /* NOTREACHED */ 546 } 547 *sp++ = *p; 548 } 549 size += (size_t)(sp - op); 550 if (asize - size < _POSIX2_LINE_MAX + 1) { 551 asize *= 2; 552 text = xrealloc(text, asize); 553 } 554 } while (cu_fgets(p = lbuf, sizeof(lbuf), &more)); 555 errx(1, "%lu: %s: unterminated substitute in regular expression", 556 linenum, fname); 557 /* NOTREACHED */ 558 } 559 560 /* 561 * Compile the flags of the s command 562 */ 563 static char * 564 compile_flags(char *p, struct s_subst *s) 565 { 566 int gn; /* True if we have seen g or n */ 567 unsigned long nval; 568 char wfile[_POSIX2_LINE_MAX + 1], *q; 569 570 s->n = 1; /* Default */ 571 s->p = 0; 572 s->wfile = NULL; 573 s->wfd = -1; 574 s->icase = 0; 575 for (gn = 0;;) { 576 EATSPACE(); /* EXTENSION */ 577 switch (*p) { 578 case 'g': 579 if (gn) 580 errx(1, 581 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname); 582 gn = 1; 583 s->n = 0; 584 break; 585 case '\0': 586 case '\n': 587 case ';': 588 return (p); 589 case 'p': 590 s->p = 1; 591 break; 592 case 'i': 593 case 'I': 594 s->icase = 1; 595 break; 596 case '1': case '2': case '3': 597 case '4': case '5': case '6': 598 case '7': case '8': case '9': 599 if (gn) 600 errx(1, 601 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname); 602 gn = 1; 603 errno = 0; 604 nval = strtoul(p, &p, 10); 605 if (errno == ERANGE || nval > INT_MAX) 606 errx(1, 607 "%lu: %s: overflow in the 'N' substitute flag", linenum, fname); 608 s->n = (int)nval; 609 p--; 610 break; 611 case 'w': 612 p++; 613 #ifdef HISTORIC_PRACTICE 614 if (*p != ' ') { 615 warnx("%lu: %s: space missing before w wfile", linenum, fname); 616 return (p); 617 } 618 #endif 619 EATSPACE(); 620 q = wfile; 621 while (*p) { 622 if (*p == '\n') 623 break; 624 *q++ = *p++; 625 } 626 *q = '\0'; 627 if (q == wfile) 628 errx(1, "%lu: %s: no wfile specified", linenum, fname); 629 s->wfile = strdup(wfile); 630 if (!aflag && (s->wfd = open(wfile, 631 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 632 DEFFILEMODE)) == -1) 633 err(1, "%s", wfile); 634 return (p); 635 default: 636 errx(1, "%lu: %s: bad flag in substitute command: '%c'", 637 linenum, fname, *p); 638 break; 639 } 640 p++; 641 } 642 } 643 644 /* 645 * Compile a translation set of strings into a lookup table. 646 */ 647 static char * 648 compile_tr(char *p, struct s_tr **py) 649 { 650 struct s_tr *y; 651 size_t i; 652 const char *op, *np; 653 char old[_POSIX2_LINE_MAX + 1]; 654 char new[_POSIX2_LINE_MAX + 1]; 655 size_t oclen, oldlen, nclen, newlen; 656 mbstate_t mbs1, mbs2; 657 658 *py = y = xmalloc(sizeof(*y)); 659 y->multis = NULL; 660 y->nmultis = 0; 661 662 if (*p == '\0' || *p == '\\') 663 errx(1, 664 "%lu: %s: transform pattern can not be delimited by newline or backslash", 665 linenum, fname); 666 p = compile_delimited(p, old, 1); 667 if (p == NULL) 668 errx(1, "%lu: %s: unterminated transform source string", 669 linenum, fname); 670 p = compile_delimited(p - 1, new, 1); 671 if (p == NULL) 672 errx(1, "%lu: %s: unterminated transform target string", 673 linenum, fname); 674 EATSPACE(); 675 op = old; 676 oldlen = mbsrtowcs(NULL, &op, 0, NULL); 677 if (oldlen == (size_t)-1) 678 err(1, NULL); 679 np = new; 680 newlen = mbsrtowcs(NULL, &np, 0, NULL); 681 if (newlen == (size_t)-1) 682 err(1, NULL); 683 if (newlen != oldlen) 684 errx(1, "%lu: %s: transform strings are not the same length", 685 linenum, fname); 686 if (MB_CUR_MAX == 1) { 687 /* 688 * The single-byte encoding case is easy: generate a 689 * lookup table. 690 */ 691 for (i = 0; i <= UCHAR_MAX; i++) 692 y->bytetab[i] = (u_char)i; 693 for (; *op; op++, np++) 694 y->bytetab[(u_char)*op] = (u_char)*np; 695 } else { 696 /* 697 * Multi-byte encoding case: generate a lookup table as 698 * above, but only for single-byte characters. The first 699 * bytes of multi-byte characters have their lookup table 700 * entries set to 0, which causes do_tr() to search through 701 * an auxiliary vector of multi-byte mappings. 702 */ 703 memset(&mbs1, 0, sizeof(mbs1)); 704 memset(&mbs2, 0, sizeof(mbs2)); 705 for (i = 0; i <= UCHAR_MAX; i++) 706 y->bytetab[i] = (u_char)((btowc((int)i) != WEOF) ? i : 0); 707 while (*op != '\0') { 708 oclen = mbrlen(op, MB_LEN_MAX, &mbs1); 709 if (oclen == (size_t)-1 || oclen == (size_t)-2) 710 errc(1, EILSEQ, NULL); 711 nclen = mbrlen(np, MB_LEN_MAX, &mbs2); 712 if (nclen == (size_t)-1 || nclen == (size_t)-2) 713 errc(1, EILSEQ, NULL); 714 if (oclen == 1 && nclen == 1) 715 y->bytetab[(u_char)*op] = (u_char)*np; 716 else { 717 y->bytetab[(u_char)*op] = 0; 718 y->multis = xrealloc(y->multis, 719 (y->nmultis + 1) * sizeof(*y->multis)); 720 i = y->nmultis++; 721 y->multis[i].fromlen = oclen; 722 memcpy(y->multis[i].from, op, oclen); 723 y->multis[i].tolen = nclen; 724 memcpy(y->multis[i].to, np, nclen); 725 } 726 op += oclen; 727 np += nclen; 728 } 729 } 730 return (p); 731 } 732 733 /* 734 * Compile the text following an a or i command. 735 */ 736 static char * 737 compile_text(void) 738 { 739 size_t asize, size; 740 int esc_nl; 741 char *text, *p, *op, *s; 742 char lbuf[_POSIX2_LINE_MAX + 1]; 743 744 asize = 2 * _POSIX2_LINE_MAX + 1; 745 text = xmalloc(asize); 746 size = 0; 747 while (cu_fgets(lbuf, sizeof(lbuf), NULL)) { 748 op = s = text + size; 749 p = lbuf; 750 for (esc_nl = 0; *p != '\0'; p++) { 751 if (*p == '\\' && p[1] != '\0' && *++p == '\n') 752 esc_nl = 1; 753 *s++ = *p; 754 } 755 size += (size_t)(s - op); 756 if (!esc_nl) { 757 *s = '\0'; 758 break; 759 } 760 if (asize - size < _POSIX2_LINE_MAX + 1) { 761 asize *= 2; 762 text = xrealloc(text, asize); 763 } 764 } 765 text[size] = '\0'; 766 p = xrealloc(text, size + 1); 767 return (p); 768 } 769 770 /* 771 * Get an address and return a pointer to the first character after 772 * it. Fill the structure pointed to according to the address. 773 */ 774 static char * 775 compile_addr(char *p, struct s_addr *a) 776 { 777 char *end, re[_POSIX2_LINE_MAX + 1]; 778 int icase; 779 780 icase = 0; 781 782 a->type = 0; 783 switch (*p) { 784 case '\\': /* Context address */ 785 ++p; 786 /* FALLTHROUGH */ 787 case '/': /* Context address */ 788 p = compile_delimited(p, re, 0); 789 if (p == NULL) 790 errx(1, "%lu: %s: unterminated regular expression", linenum, fname); 791 /* Check for case insensitive regexp flag */ 792 if (*p == 'I') { 793 icase = 1; 794 p++; 795 } 796 if (*re == '\0') 797 a->u.r = NULL; 798 else 799 a->u.r = compile_re(re, icase); 800 a->type = AT_RE; 801 return (p); 802 803 case '$': /* Last line */ 804 a->type = AT_LAST; 805 return (p + 1); 806 807 case '+': /* Relative line number */ 808 a->type = AT_RELLINE; 809 p++; 810 /* FALLTHROUGH */ 811 /* Line number */ 812 case '0': case '1': case '2': case '3': case '4': 813 case '5': case '6': case '7': case '8': case '9': 814 if (a->type == 0) 815 a->type = AT_LINE; 816 a->u.l = strtoul(p, &end, 10); 817 return (end); 818 default: 819 errx(1, "%lu: %s: expected context address", linenum, fname); 820 return (NULL); 821 } 822 } 823 824 /* 825 * duptoeol -- 826 * Return a copy of all the characters up to \n or \0. 827 */ 828 static char * 829 duptoeol(char *s, const char *ctype) 830 { 831 size_t len; 832 int ws; 833 char *p, *start; 834 835 ws = 0; 836 for (start = s; *s != '\0' && *s != '\n'; ++s) 837 ws = isspace((unsigned char)*s); 838 *s = '\0'; 839 if (ws) 840 warnx("%lu: %s: whitespace after %s", linenum, fname, ctype); 841 len = (size_t)(s - start + 1); 842 p = xmalloc(len); 843 return (memmove(p, start, len)); 844 } 845 846 /* 847 * Convert goto label names to addresses, and count a and r commands, in 848 * the given subset of the script. Free the memory used by labels in b 849 * and t commands (but not by :). 850 * 851 * TODO: Remove } nodes 852 */ 853 static void 854 fixuplabel(struct s_command *cp, struct s_command *end) 855 { 856 857 for (; cp != end; cp = cp->next) 858 switch (cp->code) { 859 case 'a': 860 case 'r': 861 appendnum++; 862 break; 863 case 'b': 864 case 't': 865 /* Resolve branch target. */ 866 if (cp->t == NULL) { 867 cp->u.c = NULL; 868 break; 869 } 870 if ((cp->u.c = findlabel(cp->t)) == NULL) 871 errx(1, "%lu: %s: undefined label '%s'", linenum, fname, cp->t); 872 free(cp->t); 873 break; 874 case '{': 875 /* Do interior commands. */ 876 fixuplabel(cp->u.c, cp->next); 877 break; 878 } 879 } 880 881 /* 882 * Associate the given command label for later lookup. 883 */ 884 static void 885 enterlabel(struct s_command *cp) 886 { 887 struct labhash **lhp, *lh; 888 u_char *p; 889 u_int h, c; 890 891 for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++) 892 h = (h << 5) + h + c; 893 lhp = &labels[h & LHMASK]; 894 for (lh = *lhp; lh != NULL; lh = lh->lh_next) 895 if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0) 896 errx(1, "%lu: %s: duplicate label '%s'", linenum, fname, cp->t); 897 lh = xmalloc(sizeof *lh); 898 lh->lh_next = *lhp; 899 lh->lh_hash = h; 900 lh->lh_cmd = cp; 901 lh->lh_ref = 0; 902 *lhp = lh; 903 } 904 905 /* 906 * Find the label contained in the command l in the command linked 907 * list cp. L is excluded from the search. Return NULL if not found. 908 */ 909 static struct s_command * 910 findlabel(char *name) 911 { 912 struct labhash *lh; 913 u_char *p; 914 u_int h, c; 915 916 for (h = 0, p = (u_char *)name; (c = *p) != 0; p++) 917 h = (h << 5) + h + c; 918 for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) { 919 if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) { 920 lh->lh_ref = 1; 921 return (lh->lh_cmd); 922 } 923 } 924 return (NULL); 925 } 926 927 /* 928 * Warn about any unused labels. As a side effect, release the label hash 929 * table space. 930 */ 931 static void 932 uselabel(void) 933 { 934 struct labhash *lh, *next; 935 int i; 936 937 for (i = 0; i < LHSZ; i++) { 938 for (lh = labels[i]; lh != NULL; lh = next) { 939 next = lh->lh_next; 940 if (!lh->lh_ref) 941 warnx("%lu: %s: unused label '%s'", 942 linenum, fname, lh->lh_cmd->t); 943 free(lh); 944 } 945 } 946 } 947