1 /*- 2 * SPDX-License-Identifier: BSD-3-Clause 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 #include <sys/cdefs.h> 37 __FBSDID("$FreeBSD$"); 38 39 #ifndef lint 40 static const char sccsid[] = "@(#)compile.c 8.1 (Berkeley) 6/6/93"; 41 #endif 42 43 #include <sys/types.h> 44 #include <sys/stat.h> 45 46 #include <ctype.h> 47 #include <err.h> 48 #include <errno.h> 49 #include <fcntl.h> 50 #include <limits.h> 51 #include <regex.h> 52 #include <stdbool.h> 53 #include <stdio.h> 54 #include <stdlib.h> 55 #include <string.h> 56 #include <wchar.h> 57 58 #include "defs.h" 59 #include "extern.h" 60 61 #define LHSZ 128 62 #define LHMASK (LHSZ - 1) 63 static struct labhash { 64 struct labhash *lh_next; 65 u_int lh_hash; 66 struct s_command *lh_cmd; 67 int lh_ref; 68 } *labels[LHSZ]; 69 70 static char *compile_addr(char *, struct s_addr *); 71 static char *compile_ccl(char **, char *); 72 static char *compile_delimited(char *, char *, int); 73 static char *compile_flags(char *, struct s_subst *); 74 static regex_t *compile_re(char *, int); 75 static char *compile_subst(char *, struct s_subst *); 76 static char *compile_text(void); 77 static char *compile_tr(char *, struct s_tr **); 78 static struct s_command 79 **compile_stream(struct s_command **); 80 static char *duptoeol(char *, const char *); 81 static void enterlabel(struct s_command *); 82 static struct s_command 83 *findlabel(char *); 84 static void fixuplabel(struct s_command *, struct s_command *); 85 static void uselabel(void); 86 87 /* 88 * Command specification. This is used to drive the command parser. 89 */ 90 struct s_format { 91 char code; /* Command code */ 92 int naddr; /* Number of address args */ 93 enum e_args args; /* Argument type */ 94 }; 95 96 static struct s_format cmd_fmts[] = { 97 {'{', 2, GROUP}, 98 {'}', 0, ENDGROUP}, 99 {'a', 1, TEXT}, 100 {'b', 2, BRANCH}, 101 {'c', 2, TEXT}, 102 {'d', 2, EMPTY}, 103 {'D', 2, EMPTY}, 104 {'g', 2, EMPTY}, 105 {'G', 2, EMPTY}, 106 {'h', 2, EMPTY}, 107 {'H', 2, EMPTY}, 108 {'i', 1, TEXT}, 109 {'l', 2, EMPTY}, 110 {'n', 2, EMPTY}, 111 {'N', 2, EMPTY}, 112 {'p', 2, EMPTY}, 113 {'P', 2, EMPTY}, 114 {'q', 1, EMPTY}, 115 {'r', 1, RFILE}, 116 {'s', 2, SUBST}, 117 {'t', 2, BRANCH}, 118 {'w', 2, WFILE}, 119 {'x', 2, EMPTY}, 120 {'y', 2, TR}, 121 {'!', 2, NONSEL}, 122 {':', 0, LABEL}, 123 {'#', 0, COMMENT}, 124 {'=', 1, EMPTY}, 125 {'\0', 0, COMMENT}, 126 }; 127 128 /* The compiled program. */ 129 struct s_command *prog; 130 131 /* 132 * Compile the program into prog. 133 * Initialise appends. 134 */ 135 void 136 compile(void) 137 { 138 *compile_stream(&prog) = NULL; 139 fixuplabel(prog, NULL); 140 uselabel(); 141 if (appendnum == 0) 142 appends = NULL; 143 else if ((appends = malloc(sizeof(struct s_appends) * appendnum)) == 144 NULL) 145 err(1, "malloc"); 146 if ((match = malloc((maxnsub + 1) * sizeof(regmatch_t))) == NULL) 147 err(1, "malloc"); 148 } 149 150 #define EATSPACE() do { \ 151 if (p) \ 152 while (*p && isspace((unsigned char)*p)) \ 153 p++; \ 154 } while (0) 155 156 static struct s_command ** 157 compile_stream(struct s_command **link) 158 { 159 char *p; 160 static char lbuf[_POSIX2_LINE_MAX + 1]; /* To save stack */ 161 struct s_command *cmd, *cmd2, *stack; 162 struct s_format *fp; 163 char re[_POSIX2_LINE_MAX + 1]; 164 int naddr; /* Number of addresses */ 165 166 stack = NULL; 167 for (;;) { 168 if ((p = cu_fgets(lbuf, sizeof(lbuf), NULL)) == NULL) { 169 if (stack != NULL) 170 errx(1, "%lu: %s: unexpected EOF (pending }'s)", 171 linenum, fname); 172 return (link); 173 } 174 175 semicolon: EATSPACE(); 176 if (p) { 177 if (*p == '#' || *p == '\0') 178 continue; 179 else if (*p == ';') { 180 p++; 181 goto semicolon; 182 } 183 } 184 if ((*link = cmd = malloc(sizeof(struct s_command))) == NULL) 185 err(1, "malloc"); 186 link = &cmd->next; 187 cmd->startline = cmd->nonsel = 0; 188 /* First parse the addresses */ 189 naddr = 0; 190 191 /* Valid characters to start an address */ 192 #define addrchar(c) (strchr("0123456789/\\$", (c))) 193 if (addrchar(*p)) { 194 naddr++; 195 if ((cmd->a1 = malloc(sizeof(struct s_addr))) == NULL) 196 err(1, "malloc"); 197 p = compile_addr(p, cmd->a1); 198 EATSPACE(); /* EXTENSION */ 199 if (*p == ',') { 200 p++; 201 EATSPACE(); /* EXTENSION */ 202 naddr++; 203 if ((cmd->a2 = malloc(sizeof(struct s_addr))) 204 == NULL) 205 err(1, "malloc"); 206 p = compile_addr(p, cmd->a2); 207 EATSPACE(); 208 } else 209 cmd->a2 = NULL; 210 } else 211 cmd->a1 = cmd->a2 = NULL; 212 213 nonsel: /* Now parse the command */ 214 if (!*p) 215 errx(1, "%lu: %s: command expected", linenum, fname); 216 cmd->code = *p; 217 for (fp = cmd_fmts; fp->code; fp++) 218 if (fp->code == *p) 219 break; 220 if (!fp->code) 221 errx(1, "%lu: %s: invalid command code %c", linenum, fname, *p); 222 if (naddr > fp->naddr) 223 errx(1, 224 "%lu: %s: command %c expects up to %d address(es), found %d", 225 linenum, fname, *p, fp->naddr, naddr); 226 switch (fp->args) { 227 case NONSEL: /* ! */ 228 p++; 229 EATSPACE(); 230 cmd->nonsel = 1; 231 goto nonsel; 232 case GROUP: /* { */ 233 p++; 234 EATSPACE(); 235 cmd->next = stack; 236 stack = cmd; 237 link = &cmd->u.c; 238 if (*p) 239 goto semicolon; 240 break; 241 case ENDGROUP: 242 /* 243 * Short-circuit command processing, since end of 244 * group is really just a noop. 245 */ 246 cmd->nonsel = 1; 247 if (stack == NULL) 248 errx(1, "%lu: %s: unexpected }", linenum, fname); 249 cmd2 = stack; 250 stack = cmd2->next; 251 cmd2->next = cmd; 252 /*FALLTHROUGH*/ 253 case EMPTY: /* d D g G h H l n N p P q x = \0 */ 254 p++; 255 EATSPACE(); 256 if (*p == ';') { 257 p++; 258 link = &cmd->next; 259 goto semicolon; 260 } 261 if (*p) 262 errx(1, "%lu: %s: extra characters at the end of %c command", 263 linenum, fname, cmd->code); 264 break; 265 case TEXT: /* a c i */ 266 p++; 267 EATSPACE(); 268 if (*p != '\\') 269 errx(1, 270 "%lu: %s: command %c expects \\ followed by text", linenum, fname, cmd->code); 271 p++; 272 EATSPACE(); 273 if (*p) 274 errx(1, 275 "%lu: %s: extra characters after \\ at the end of %c command", 276 linenum, fname, cmd->code); 277 cmd->t = compile_text(); 278 break; 279 case COMMENT: /* \0 # */ 280 break; 281 case WFILE: /* w */ 282 p++; 283 EATSPACE(); 284 if (*p == '\0') 285 errx(1, "%lu: %s: filename expected", linenum, fname); 286 cmd->t = duptoeol(p, "w command"); 287 if (aflag) 288 cmd->u.fd = -1; 289 else if ((cmd->u.fd = open(p, 290 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 291 DEFFILEMODE)) == -1) 292 err(1, "%s", p); 293 break; 294 case RFILE: /* r */ 295 p++; 296 EATSPACE(); 297 if (*p == '\0') 298 errx(1, "%lu: %s: filename expected", linenum, fname); 299 else 300 cmd->t = duptoeol(p, "read command"); 301 break; 302 case BRANCH: /* b t */ 303 p++; 304 EATSPACE(); 305 if (*p == '\0') 306 cmd->t = NULL; 307 else 308 cmd->t = duptoeol(p, "branch"); 309 break; 310 case LABEL: /* : */ 311 p++; 312 EATSPACE(); 313 cmd->t = duptoeol(p, "label"); 314 if (strlen(p) == 0) 315 errx(1, "%lu: %s: empty label", linenum, fname); 316 enterlabel(cmd); 317 break; 318 case SUBST: /* s */ 319 p++; 320 if (*p == '\0' || *p == '\\') 321 errx(1, 322 "%lu: %s: substitute pattern can not be delimited by newline or backslash", 323 linenum, fname); 324 if ((cmd->u.s = calloc(1, sizeof(struct s_subst))) == NULL) 325 err(1, "malloc"); 326 p = compile_delimited(p, re, 0); 327 if (p == NULL) 328 errx(1, 329 "%lu: %s: unterminated substitute pattern", linenum, fname); 330 331 /* Compile RE with no case sensitivity temporarily */ 332 if (*re == '\0') 333 cmd->u.s->re = NULL; 334 else 335 cmd->u.s->re = compile_re(re, 0); 336 --p; 337 p = compile_subst(p, cmd->u.s); 338 p = compile_flags(p, cmd->u.s); 339 340 /* Recompile RE with case sensitivity from "I" flag if any */ 341 if (*re == '\0') 342 cmd->u.s->re = NULL; 343 else 344 cmd->u.s->re = compile_re(re, cmd->u.s->icase); 345 EATSPACE(); 346 if (*p == ';') { 347 p++; 348 link = &cmd->next; 349 goto semicolon; 350 } 351 break; 352 case TR: /* y */ 353 p++; 354 p = compile_tr(p, &cmd->u.y); 355 EATSPACE(); 356 if (*p == ';') { 357 p++; 358 link = &cmd->next; 359 goto semicolon; 360 } 361 if (*p) 362 errx(1, 363 "%lu: %s: extra text at the end of a transform command", linenum, fname); 364 break; 365 } 366 } 367 } 368 369 static int 370 hex2char(const char *in, char *out, int len) 371 { 372 long ord; 373 char *endptr, hexbuf[3]; 374 375 hexbuf[0] = in[0]; 376 hexbuf[1] = len > 1 ? in[1] : '\0'; 377 hexbuf[2] = '\0'; 378 379 errno = 0; 380 ord = strtol(hexbuf, &endptr, 16); 381 if (*endptr != '\0' || errno != 0) 382 return (ERANGE); 383 *out = (char)ord; 384 return (0); 385 } 386 387 static bool 388 hexdigit(char c) 389 { 390 int lc; 391 392 lc = tolower(c); 393 return isdigit(lc) || (lc >= 'a' && lc <= 'f'); 394 } 395 396 static bool 397 dohex(const char *in, char *out, int *len) 398 { 399 int tmplen; 400 401 if (!hexdigit(in[0])) 402 return (false); 403 tmplen = 1; 404 if (hexdigit(in[1])) 405 ++tmplen; 406 if (hex2char(in, out, tmplen) == 0) { 407 *len = tmplen; 408 return (true); 409 } 410 411 return (false); 412 } 413 414 /* 415 * Get a delimited string. P points to the delimiter of the string; d points 416 * to a buffer area. Newline and delimiter escapes are processed; other 417 * escapes are ignored. 418 * 419 * Returns a pointer to the first character after the final delimiter or NULL 420 * in the case of a non-terminated string. The character array d is filled 421 * with the processed string. 422 */ 423 static char * 424 compile_delimited(char *p, char *d, int is_tr) 425 { 426 int hexlen; 427 char c; 428 429 c = *p++; 430 if (c == '\0') 431 return (NULL); 432 else if (c == '\\') 433 errx(1, "%lu: %s: \\ can not be used as a string delimiter", 434 linenum, fname); 435 else if (c == '\n') 436 errx(1, "%lu: %s: newline can not be used as a string delimiter", 437 linenum, fname); 438 while (*p) { 439 if (*p == '[' && *p != c) { 440 if ((d = compile_ccl(&p, d)) == NULL) 441 errx(1, "%lu: %s: unbalanced brackets ([])", linenum, fname); 442 continue; 443 } else if (*p == '\\' && p[1] == '[') { 444 *d++ = *p++; 445 } else if (*p == '\\' && p[1] == c) { 446 p++; 447 } else if (*p == '\\' && 448 (p[1] == 'n' || p[1] == 'r' || p[1] == 't')) { 449 switch (p[1]) { 450 case 'n': 451 *d++ = '\n'; 452 break; 453 case 'r': 454 *d++ = '\r'; 455 break; 456 case 't': 457 *d++ = '\t'; 458 break; 459 } 460 p += 2; 461 continue; 462 } else if (*p == '\\' && p[1] == 'x') { 463 if (dohex(&p[2], d, &hexlen)) { 464 ++d; 465 p += hexlen + 2; 466 continue; 467 } 468 } else if (*p == '\\' && p[1] == '\\') { 469 if (is_tr) 470 p++; 471 else 472 *d++ = *p++; 473 } else if (*p == c) { 474 *d = '\0'; 475 return (p + 1); 476 } 477 *d++ = *p++; 478 } 479 return (NULL); 480 } 481 482 483 /* compile_ccl: expand a POSIX character class */ 484 static char * 485 compile_ccl(char **sp, char *t) 486 { 487 int c, d, hexlen; 488 char *s = *sp; 489 490 *t++ = *s++; 491 if (*s == '^') 492 *t++ = *s++; 493 if (*s == ']') 494 *t++ = *s++; 495 for (; *s && (*t = *s) != ']'; s++, t++) { 496 if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) { 497 *++t = *++s, t++, s++; 498 for (c = *s; (*t = *s) != ']' || c != d; s++, t++) 499 if ((c = *s) == '\0') 500 return NULL; 501 } else if (*s == '\\') { 502 switch (s[1]) { 503 case 'n': 504 *t = '\n'; 505 s++; 506 break; 507 case 'r': 508 *t = '\r'; 509 s++; 510 break; 511 case 't': 512 *t = '\t'; 513 s++; 514 break; 515 case 'x': 516 if (dohex(&s[2], t, &hexlen)) 517 s += hexlen + 1; 518 break; 519 } 520 } 521 } 522 return (*s == ']') ? *sp = ++s, ++t : NULL; 523 } 524 525 /* 526 * Compiles the regular expression in RE and returns a pointer to the compiled 527 * regular expression. 528 * Cflags are passed to regcomp. 529 */ 530 static regex_t * 531 compile_re(char *re, int case_insensitive) 532 { 533 regex_t *rep; 534 int eval, flags; 535 536 537 flags = rflags; 538 if (case_insensitive) 539 flags |= REG_ICASE; 540 if ((rep = malloc(sizeof(regex_t))) == NULL) 541 err(1, "malloc"); 542 if ((eval = regcomp(rep, re, flags)) != 0) 543 errx(1, "%lu: %s: RE error: %s", 544 linenum, fname, strregerror(eval, rep)); 545 if (maxnsub < rep->re_nsub) 546 maxnsub = rep->re_nsub; 547 return (rep); 548 } 549 550 /* 551 * Compile the substitution string of a regular expression and set res to 552 * point to a saved copy of it. Nsub is the number of parenthesized regular 553 * expressions. 554 */ 555 static char * 556 compile_subst(char *p, struct s_subst *s) 557 { 558 static char lbuf[_POSIX2_LINE_MAX + 1]; 559 int asize, hexlen, size; 560 u_char ref; 561 char c, *text, *op, *sp; 562 int more = 1, sawesc = 0; 563 564 c = *p++; /* Terminator character */ 565 if (c == '\0') 566 return (NULL); 567 568 s->maxbref = 0; 569 s->linenum = linenum; 570 asize = 2 * _POSIX2_LINE_MAX + 1; 571 if ((text = malloc(asize)) == NULL) 572 err(1, "malloc"); 573 size = 0; 574 do { 575 op = sp = text + size; 576 for (; *p; p++) { 577 if (*p == '\\' || sawesc) { 578 /* 579 * If this is a continuation from the last 580 * buffer, we won't have a character to 581 * skip over. 582 */ 583 if (sawesc) 584 sawesc = 0; 585 else 586 p++; 587 588 if (*p == '\0') { 589 /* 590 * This escaped character is continued 591 * in the next part of the line. Note 592 * this fact, then cause the loop to 593 * exit w/ normal EOL case and reenter 594 * above with the new buffer. 595 */ 596 sawesc = 1; 597 p--; 598 continue; 599 } else if (strchr("123456789", *p) != NULL) { 600 *sp++ = '\\'; 601 ref = *p - '0'; 602 if (s->re != NULL && 603 ref > s->re->re_nsub) 604 errx(1, "%lu: %s: \\%c not defined in the RE", 605 linenum, fname, *p); 606 if (s->maxbref < ref) 607 s->maxbref = ref; 608 } else { 609 switch (*p) { 610 case '&': 611 case '\\': 612 *sp++ = '\\'; 613 break; 614 case 'n': 615 *p = '\n'; 616 break; 617 case 'r': 618 *p = '\r'; 619 break; 620 case 't': 621 *p = '\t'; 622 break; 623 case 'x': 624 #define ADVANCE_N(s, n) \ 625 do { \ 626 char *adv = (s); \ 627 while (*(adv + (n) - 1) != '\0') { \ 628 *adv = *(adv + (n)); \ 629 ++adv; \ 630 } \ 631 *adv = '\0'; \ 632 } while (0); 633 if (dohex(&p[1], p, &hexlen)) { 634 ADVANCE_N(p + 1, 635 hexlen); 636 } 637 break; 638 } 639 } 640 } else if (*p == c) { 641 if (*++p == '\0' && more) { 642 if (cu_fgets(lbuf, sizeof(lbuf), &more)) 643 p = lbuf; 644 } 645 *sp++ = '\0'; 646 size += sp - op; 647 if ((s->new = realloc(text, size)) == NULL) 648 err(1, "realloc"); 649 return (p); 650 } else if (*p == '\n') { 651 errx(1, 652 "%lu: %s: unescaped newline inside substitute pattern", linenum, fname); 653 /* NOTREACHED */ 654 } 655 *sp++ = *p; 656 } 657 size += sp - op; 658 if (asize - size < _POSIX2_LINE_MAX + 1) { 659 asize *= 2; 660 if ((text = realloc(text, asize)) == NULL) 661 err(1, "realloc"); 662 } 663 } while (cu_fgets(p = lbuf, sizeof(lbuf), &more) != NULL); 664 errx(1, "%lu: %s: unterminated substitute in regular expression", 665 linenum, fname); 666 /* NOTREACHED */ 667 } 668 669 /* 670 * Compile the flags of the s command 671 */ 672 static char * 673 compile_flags(char *p, struct s_subst *s) 674 { 675 int gn; /* True if we have seen g or n */ 676 unsigned long nval; 677 char wfile[_POSIX2_LINE_MAX + 1], *q, *eq; 678 679 s->n = 1; /* Default */ 680 s->p = 0; 681 s->wfile = NULL; 682 s->wfd = -1; 683 s->icase = 0; 684 for (gn = 0;;) { 685 EATSPACE(); /* EXTENSION */ 686 switch (*p) { 687 case 'g': 688 if (gn) 689 errx(1, 690 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname); 691 gn = 1; 692 s->n = 0; 693 break; 694 case '\0': 695 case '\n': 696 case ';': 697 return (p); 698 case 'p': 699 s->p = 1; 700 break; 701 case 'i': 702 case 'I': 703 s->icase = 1; 704 break; 705 case '1': case '2': case '3': 706 case '4': case '5': case '6': 707 case '7': case '8': case '9': 708 if (gn) 709 errx(1, 710 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname); 711 gn = 1; 712 errno = 0; 713 nval = strtol(p, &p, 10); 714 if (errno == ERANGE || nval > INT_MAX) 715 errx(1, 716 "%lu: %s: overflow in the 'N' substitute flag", linenum, fname); 717 s->n = nval; 718 p--; 719 break; 720 case 'w': 721 p++; 722 #ifdef HISTORIC_PRACTICE 723 if (*p != ' ') { 724 warnx("%lu: %s: space missing before w wfile", linenum, fname); 725 return (p); 726 } 727 #endif 728 EATSPACE(); 729 q = wfile; 730 eq = wfile + sizeof(wfile) - 1; 731 while (*p) { 732 if (*p == '\n') 733 break; 734 if (q >= eq) 735 err(1, "wfile too long"); 736 *q++ = *p++; 737 } 738 *q = '\0'; 739 if (q == wfile) 740 errx(1, "%lu: %s: no wfile specified", linenum, fname); 741 s->wfile = strdup(wfile); 742 if (!aflag && (s->wfd = open(wfile, 743 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 744 DEFFILEMODE)) == -1) 745 err(1, "%s", wfile); 746 return (p); 747 default: 748 errx(1, "%lu: %s: bad flag in substitute command: '%c'", 749 linenum, fname, *p); 750 break; 751 } 752 p++; 753 } 754 } 755 756 /* 757 * Compile a translation set of strings into a lookup table. 758 */ 759 static char * 760 compile_tr(char *p, struct s_tr **py) 761 { 762 struct s_tr *y; 763 int i; 764 const char *op, *np; 765 char old[_POSIX2_LINE_MAX + 1]; 766 char new[_POSIX2_LINE_MAX + 1]; 767 size_t oclen, oldlen, nclen, newlen; 768 mbstate_t mbs1, mbs2; 769 770 if ((*py = y = malloc(sizeof(*y))) == NULL) 771 err(1, NULL); 772 y->multis = NULL; 773 y->nmultis = 0; 774 775 if (*p == '\0' || *p == '\\') 776 errx(1, 777 "%lu: %s: transform pattern can not be delimited by newline or backslash", 778 linenum, fname); 779 p = compile_delimited(p, old, 1); 780 if (p == NULL) 781 errx(1, "%lu: %s: unterminated transform source string", 782 linenum, fname); 783 p = compile_delimited(p - 1, new, 1); 784 if (p == NULL) 785 errx(1, "%lu: %s: unterminated transform target string", 786 linenum, fname); 787 EATSPACE(); 788 op = old; 789 oldlen = mbsrtowcs(NULL, &op, 0, NULL); 790 if (oldlen == (size_t)-1) 791 err(1, NULL); 792 np = new; 793 newlen = mbsrtowcs(NULL, &np, 0, NULL); 794 if (newlen == (size_t)-1) 795 err(1, NULL); 796 if (newlen != oldlen) 797 errx(1, "%lu: %s: transform strings are not the same length", 798 linenum, fname); 799 if (MB_CUR_MAX == 1) { 800 /* 801 * The single-byte encoding case is easy: generate a 802 * lookup table. 803 */ 804 for (i = 0; i <= UCHAR_MAX; i++) 805 y->bytetab[i] = (char)i; 806 for (; *op; op++, np++) 807 y->bytetab[(u_char)*op] = *np; 808 } else { 809 /* 810 * Multi-byte encoding case: generate a lookup table as 811 * above, but only for single-byte characters. The first 812 * bytes of multi-byte characters have their lookup table 813 * entries set to 0, which causes do_tr() to search through 814 * an auxiliary vector of multi-byte mappings. 815 */ 816 memset(&mbs1, 0, sizeof(mbs1)); 817 memset(&mbs2, 0, sizeof(mbs2)); 818 for (i = 0; i <= UCHAR_MAX; i++) 819 y->bytetab[i] = (btowc(i) != WEOF) ? i : 0; 820 while (*op != '\0') { 821 oclen = mbrlen(op, MB_LEN_MAX, &mbs1); 822 if (oclen == (size_t)-1 || oclen == (size_t)-2) 823 errc(1, EILSEQ, NULL); 824 nclen = mbrlen(np, MB_LEN_MAX, &mbs2); 825 if (nclen == (size_t)-1 || nclen == (size_t)-2) 826 errc(1, EILSEQ, NULL); 827 if (oclen == 1 && nclen == 1) 828 y->bytetab[(u_char)*op] = *np; 829 else { 830 y->bytetab[(u_char)*op] = 0; 831 y->multis = realloc(y->multis, 832 (y->nmultis + 1) * sizeof(*y->multis)); 833 if (y->multis == NULL) 834 err(1, NULL); 835 i = y->nmultis++; 836 y->multis[i].fromlen = oclen; 837 memcpy(y->multis[i].from, op, oclen); 838 y->multis[i].tolen = nclen; 839 memcpy(y->multis[i].to, np, nclen); 840 } 841 op += oclen; 842 np += nclen; 843 } 844 } 845 return (p); 846 } 847 848 /* 849 * Compile the text following an a, c, or i command. 850 */ 851 static char * 852 compile_text(void) 853 { 854 int asize, esc_nl, size; 855 char *text, *p, *op, *s; 856 char lbuf[_POSIX2_LINE_MAX + 1]; 857 858 asize = 2 * _POSIX2_LINE_MAX + 1; 859 if ((text = malloc(asize)) == NULL) 860 err(1, "malloc"); 861 size = 0; 862 while (cu_fgets(lbuf, sizeof(lbuf), NULL) != NULL) { 863 op = s = text + size; 864 p = lbuf; 865 #ifdef LEGACY_BSDSED_COMPAT 866 EATSPACE(); 867 #endif 868 for (esc_nl = 0; *p != '\0'; p++) { 869 if (*p == '\\' && p[1] != '\0' && *++p == '\n') 870 esc_nl = 1; 871 *s++ = *p; 872 } 873 size += s - op; 874 if (!esc_nl) { 875 *s = '\0'; 876 break; 877 } 878 if (asize - size < _POSIX2_LINE_MAX + 1) { 879 asize *= 2; 880 if ((text = realloc(text, asize)) == NULL) 881 err(1, "realloc"); 882 } 883 } 884 text[size] = '\0'; 885 if ((p = realloc(text, size + 1)) == NULL) 886 err(1, "realloc"); 887 return (p); 888 } 889 890 /* 891 * Get an address and return a pointer to the first character after 892 * it. Fill the structure pointed to according to the address. 893 */ 894 static char * 895 compile_addr(char *p, struct s_addr *a) 896 { 897 char *end, re[_POSIX2_LINE_MAX + 1]; 898 int icase; 899 900 icase = 0; 901 902 a->type = 0; 903 switch (*p) { 904 case '\\': /* Context address */ 905 ++p; 906 /* FALLTHROUGH */ 907 case '/': /* Context address */ 908 p = compile_delimited(p, re, 0); 909 if (p == NULL) 910 errx(1, "%lu: %s: unterminated regular expression", linenum, fname); 911 /* Check for case insensitive regexp flag */ 912 if (*p == 'I') { 913 icase = 1; 914 p++; 915 } 916 if (*re == '\0') 917 a->u.r = NULL; 918 else 919 a->u.r = compile_re(re, icase); 920 a->type = AT_RE; 921 return (p); 922 923 case '$': /* Last line */ 924 a->type = AT_LAST; 925 return (p + 1); 926 927 case '+': /* Relative line number */ 928 a->type = AT_RELLINE; 929 p++; 930 /* FALLTHROUGH */ 931 /* Line number */ 932 case '0': case '1': case '2': case '3': case '4': 933 case '5': case '6': case '7': case '8': case '9': 934 if (a->type == 0) 935 a->type = AT_LINE; 936 a->u.l = strtol(p, &end, 10); 937 return (end); 938 default: 939 errx(1, "%lu: %s: expected context address", linenum, fname); 940 return (NULL); 941 } 942 } 943 944 /* 945 * duptoeol -- 946 * Return a copy of all the characters up to \n or \0. 947 */ 948 static char * 949 duptoeol(char *s, const char *ctype) 950 { 951 size_t len; 952 int ws; 953 char *p, *start; 954 955 ws = 0; 956 for (start = s; *s != '\0' && *s != '\n'; ++s) 957 ws = isspace((unsigned char)*s); 958 *s = '\0'; 959 if (ws) 960 warnx("%lu: %s: whitespace after %s", linenum, fname, ctype); 961 len = s - start + 1; 962 if ((p = malloc(len)) == NULL) 963 err(1, "malloc"); 964 return (memmove(p, start, len)); 965 } 966 967 /* 968 * Convert goto label names to addresses, and count a and r commands, in 969 * the given subset of the script. Free the memory used by labels in b 970 * and t commands (but not by :). 971 * 972 * TODO: Remove } nodes 973 */ 974 static void 975 fixuplabel(struct s_command *cp, struct s_command *end) 976 { 977 978 for (; cp != end; cp = cp->next) 979 switch (cp->code) { 980 case 'a': 981 case 'r': 982 appendnum++; 983 break; 984 case 'b': 985 case 't': 986 /* Resolve branch target. */ 987 if (cp->t == NULL) { 988 cp->u.c = NULL; 989 break; 990 } 991 if ((cp->u.c = findlabel(cp->t)) == NULL) 992 errx(1, "%lu: %s: undefined label '%s'", linenum, fname, cp->t); 993 free(cp->t); 994 break; 995 case '{': 996 /* Do interior commands. */ 997 fixuplabel(cp->u.c, cp->next); 998 break; 999 } 1000 } 1001 1002 /* 1003 * Associate the given command label for later lookup. 1004 */ 1005 static void 1006 enterlabel(struct s_command *cp) 1007 { 1008 struct labhash **lhp, *lh; 1009 u_char *p; 1010 u_int h, c; 1011 1012 for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++) 1013 h = (h << 5) + h + c; 1014 lhp = &labels[h & LHMASK]; 1015 for (lh = *lhp; lh != NULL; lh = lh->lh_next) 1016 if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0) 1017 errx(1, "%lu: %s: duplicate label '%s'", linenum, fname, cp->t); 1018 if ((lh = malloc(sizeof *lh)) == NULL) 1019 err(1, "malloc"); 1020 lh->lh_next = *lhp; 1021 lh->lh_hash = h; 1022 lh->lh_cmd = cp; 1023 lh->lh_ref = 0; 1024 *lhp = lh; 1025 } 1026 1027 /* 1028 * Find the label contained in the command l in the command linked 1029 * list cp. L is excluded from the search. Return NULL if not found. 1030 */ 1031 static struct s_command * 1032 findlabel(char *name) 1033 { 1034 struct labhash *lh; 1035 u_char *p; 1036 u_int h, c; 1037 1038 for (h = 0, p = (u_char *)name; (c = *p) != 0; p++) 1039 h = (h << 5) + h + c; 1040 for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) { 1041 if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) { 1042 lh->lh_ref = 1; 1043 return (lh->lh_cmd); 1044 } 1045 } 1046 return (NULL); 1047 } 1048 1049 /* 1050 * Warn about any unused labels. As a side effect, release the label hash 1051 * table space. 1052 */ 1053 static void 1054 uselabel(void) 1055 { 1056 struct labhash *lh, *next; 1057 int i; 1058 1059 for (i = 0; i < LHSZ; i++) { 1060 for (lh = labels[i]; lh != NULL; lh = next) { 1061 next = lh->lh_next; 1062 if (!lh->lh_ref) 1063 warnx("%lu: %s: unused label '%s'", 1064 linenum, fname, lh->lh_cmd->t); 1065 free(lh); 1066 } 1067 } 1068 } 1069