1 /* $NetBSD: compile.c,v 1.56 2026/03/12 19:13:11 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 1.56 2026/03/12 19:13:11 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 #include <util.h> 64 65 #include "defs.h" 66 #include "extern.h" 67 68 #define LHSZ 128 69 #define LHMASK (LHSZ - 1) 70 static struct labhash { 71 struct labhash *lh_next; 72 u_int lh_hash; 73 struct s_command *lh_cmd; 74 int lh_ref; 75 } *labels[LHSZ]; 76 77 static char *compile_addr(char *, struct s_addr *); 78 static char *compile_ccl(char **, char *); 79 static char *compile_delimited(char *, char *, int); 80 static char *compile_flags(char *, struct s_subst *); 81 static regex_t *compile_re(char *, int); 82 static char *compile_subst(char *, struct s_subst *); 83 static char *compile_text(void); 84 static char *compile_tr(char *, struct s_tr **); 85 static struct s_command 86 **compile_stream(struct s_command **); 87 static char *duptoeol(char *, const char *); 88 static void enterlabel(struct s_command *); 89 static struct s_command 90 *findlabel(char *); 91 static void fixuplabel(struct s_command *, struct s_command *); 92 static void uselabel(void); 93 static void parse_escapes(char *); 94 95 /* 96 * Command specification. This is used to drive the command parser. 97 */ 98 struct s_format { 99 char code; /* Command code */ 100 int naddr; /* Number of address args */ 101 enum e_args args; /* Argument type */ 102 }; 103 104 static struct s_format cmd_fmts[] = { 105 {'{', 2, GROUP}, 106 {'}', 0, ENDGROUP}, 107 {'a', 1, TEXT}, 108 {'b', 2, BRANCH}, 109 {'c', 2, TEXT}, 110 {'d', 2, EMPTY}, 111 {'D', 2, EMPTY}, 112 {'g', 2, EMPTY}, 113 {'G', 2, EMPTY}, 114 {'h', 2, EMPTY}, 115 {'H', 2, EMPTY}, 116 {'i', 1, TEXT}, 117 {'l', 2, EMPTY}, 118 {'n', 2, EMPTY}, 119 {'N', 2, EMPTY}, 120 {'p', 2, EMPTY}, 121 {'P', 2, EMPTY}, 122 {'q', 1, EMPTY}, 123 {'r', 1, RFILE}, 124 {'s', 2, SUBST}, 125 {'t', 2, BRANCH}, 126 {'w', 2, WFILE}, 127 {'x', 2, EMPTY}, 128 {'y', 2, TR}, 129 {'!', 2, NONSEL}, 130 {':', 0, LABEL}, 131 {'#', 0, COMMENT}, 132 {'=', 1, EMPTY}, 133 {'\0', 0, COMMENT}, 134 }; 135 136 /* The compiled program. */ 137 struct s_command *prog; 138 139 /* 140 * Compile the program into prog. 141 * Initialise appends. 142 */ 143 void 144 compile(void) 145 { 146 *compile_stream(&prog) = NULL; 147 fixuplabel(prog, NULL); 148 uselabel(); 149 if (appendnum > 0) 150 appends = emalloc(sizeof(struct s_appends) * appendnum); 151 match = emalloc((maxnsub + 1) * sizeof(regmatch_t)); 152 } 153 154 #define EATSPACE() do { \ 155 if (p) \ 156 while (*p && isspace((unsigned char)*p)) \ 157 p++; \ 158 } while (0) 159 160 static struct s_command ** 161 compile_stream(struct s_command **link) 162 { 163 char *p; 164 static char lbuf[_POSIX2_LINE_MAX + 1]; /* To save stack */ 165 struct s_command *cmd, *cmd2, *stack; 166 struct s_format *fp; 167 char re[_POSIX2_LINE_MAX + 1]; 168 int naddr; /* Number of addresses */ 169 170 stack = 0; 171 for (;;) { 172 if ((p = cu_fgets(lbuf, sizeof(lbuf), NULL)) == NULL) { 173 if (stack != 0) 174 sederr("unexpected EOF (pending }'s)"); 175 return (link); 176 } 177 178 semicolon: EATSPACE(); 179 if (p) { 180 if (*p == '#' || *p == '\0') 181 continue; 182 else if (*p == ';') { 183 p++; 184 goto semicolon; 185 } 186 } 187 *link = cmd = emalloc(sizeof(struct s_command)); 188 link = &cmd->next; 189 cmd->startline = cmd->nonsel = 0; 190 /* First parse the addresses */ 191 naddr = 0; 192 193 /* Valid characters to start an address */ 194 #define addrchar(c) (strchr("0123456789/\\$", (c))) 195 if (addrchar(*p)) { 196 naddr++; 197 cmd->a1 = emalloc(sizeof(struct s_addr)); 198 p = compile_addr(p, cmd->a1); 199 EATSPACE(); /* EXTENSION */ 200 if (*p == ',') { 201 p++; 202 EATSPACE(); /* EXTENSION */ 203 naddr++; 204 cmd->a2 = emalloc(sizeof(struct s_addr)); 205 p = compile_addr(p, cmd->a2); 206 EATSPACE(); 207 } else 208 cmd->a2 = 0; 209 } else 210 cmd->a1 = cmd->a2 = 0; 211 212 nonsel: /* Now parse the command */ 213 if (!*p) 214 sederr("command expected"); 215 cmd->code = *p; 216 for (fp = cmd_fmts; fp->code; fp++) 217 if (fp->code == *p) 218 break; 219 if (!fp->code) 220 sederr("invalid command code %c", *p); 221 if (naddr > fp->naddr) 222 sederr( 223 "command %c expects up to %d address(es), found %d", 224 *p, fp->naddr, naddr); 225 switch (fp->args) { 226 case NONSEL: /* ! */ 227 p++; 228 EATSPACE(); 229 cmd->nonsel = ! cmd->nonsel; 230 goto nonsel; 231 case GROUP: /* { */ 232 p++; 233 EATSPACE(); 234 cmd->next = stack; 235 stack = cmd; 236 link = &cmd->u.c; 237 if (*p) 238 goto semicolon; 239 break; 240 case ENDGROUP: 241 /* 242 * Short-circuit command processing, since end of 243 * group is really just a noop. 244 */ 245 cmd->nonsel = 1; 246 if (stack == 0) 247 sederr("unexpected }"); 248 cmd2 = stack; 249 stack = cmd2->next; 250 cmd2->next = cmd; 251 /*FALLTHROUGH*/ 252 case EMPTY: /* d D g G h H l n N p P q x = \0 */ 253 p++; 254 EATSPACE(); 255 switch (*p) { 256 case ';': 257 p++; 258 link = &cmd->next; 259 goto semicolon; 260 case '}': 261 goto semicolon; 262 case '\0': 263 break; 264 default: 265 sederr( 266 "extra characters at the end of %c command", 267 cmd->code); 268 } 269 break; 270 case TEXT: /* a c i */ 271 p++; 272 EATSPACE(); 273 if (*p != '\\') 274 sederr("command %c expects \\ followed by text", cmd->code); 275 p++; 276 EATSPACE(); 277 if (*p) 278 sederr( 279 "extra characters after \\ at the end of %c command", 280 cmd->code); 281 cmd->t = compile_text(); 282 break; 283 case COMMENT: /* \0 # */ 284 break; 285 case WFILE: /* w */ 286 p++; 287 EATSPACE(); 288 if (*p == '\0') 289 sederr("filename expected"); 290 cmd->t = duptoeol(p, "w command"); 291 if (aflag) 292 cmd->u.fd = -1; 293 else if ((cmd->u.fd = open(p, 294 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 295 DEFFILEMODE)) == -1) 296 err(1, "%s", p); 297 break; 298 case RFILE: /* r */ 299 p++; 300 EATSPACE(); 301 if (*p == '\0') 302 sederr("filename expected"); 303 else 304 cmd->t = duptoeol(p, "read command"); 305 break; 306 case BRANCH: /* b t */ 307 p++; 308 EATSPACE(); 309 if (*p == '\0') 310 cmd->t = NULL; 311 else 312 cmd->t = duptoeol(p, "branch"); 313 break; 314 case LABEL: /* : */ 315 p++; 316 EATSPACE(); 317 cmd->t = duptoeol(p, "label"); 318 if (strlen(p) == 0) 319 sederr("empty label"); 320 enterlabel(cmd); 321 break; 322 case SUBST: /* s */ 323 p++; 324 if (*p == '\0' || *p == '\\') 325 sederr( 326 "substitute pattern can not be delimited by newline or backslash"); 327 cmd->u.s = ecalloc(1, sizeof(struct s_subst)); 328 p = compile_delimited(p, re, 0); 329 if (p == NULL) 330 sederr("unterminated substitute pattern"); 331 332 /* Compile RE with no case sensitivity temporarily */ 333 if (*re == '\0') 334 cmd->u.s->re = NULL; 335 else 336 cmd->u.s->re = compile_re(re, 0); 337 --p; 338 p = compile_subst(p, cmd->u.s); 339 p = compile_flags(p, cmd->u.s); 340 341 /* Recompile RE with case sensitivity from "I" flag if any */ 342 if (*re == '\0') 343 cmd->u.s->re = NULL; 344 else 345 cmd->u.s->re = compile_re(re, cmd->u.s->icase); 346 EATSPACE(); 347 if (*p == ';') { 348 p++; 349 link = &cmd->next; 350 goto semicolon; 351 } 352 break; 353 case TR: /* y */ 354 p++; 355 p = compile_tr(p, &cmd->u.y); 356 EATSPACE(); 357 switch (*p) { 358 case ';': 359 p++; 360 link = &cmd->next; 361 goto semicolon; 362 case '}': 363 goto semicolon; 364 case '\0': 365 break; 366 default: 367 sederr( 368 "extra text at the end of a transform command"); 369 } 370 if (*p) 371 break; 372 } 373 } 374 } 375 376 /* 377 * Get a delimited string. P points to the delimiter of the string; d points 378 * to a buffer area. Newline and delimiter escapes are processed; other 379 * escapes are ignored. 380 * 381 * Returns a pointer to the first character after the final delimiter or NULL 382 * in the case of a non-terminated string. The character array d is filled 383 * with the processed string. 384 */ 385 static char * 386 compile_delimited(char *p, char *d, int is_tr) 387 { 388 char c; 389 390 c = *p++; 391 if (c == '\0') 392 return (NULL); 393 else if (c == '\\') 394 sederr("\\ can not be used as a string delimiter"); 395 else if (c == '\n') 396 sederr("newline can not be used as a string delimiter"); 397 while (*p) { 398 if (*p == '[' && *p != c) { 399 if ((d = compile_ccl(&p, d)) == NULL) 400 sederr("unbalanced brackets ([])"); 401 continue; 402 } else if (*p == '\\' && p[1] == '[') { 403 *d++ = *p++; 404 } else if (*p == '\\' && p[1] == c) 405 p++; 406 else if (*p == '\\' && p[1] == 'n') { 407 *d++ = '\n'; 408 p += 2; 409 continue; 410 } else if (*p == '\\' && p[1] == '\\') { 411 if (is_tr) 412 p++; 413 else 414 *d++ = *p++; 415 } else if (*p == c) { 416 *d = '\0'; 417 return (p + 1); 418 } 419 *d++ = *p++; 420 } 421 return (NULL); 422 } 423 424 425 /* compile_ccl: expand a POSIX character class */ 426 static char * 427 compile_ccl(char **sp, char *t) 428 { 429 int c, d; 430 char *s = *sp; 431 432 *t++ = *s++; 433 if (*s == '^') 434 *t++ = *s++; 435 if (*s == ']') 436 *t++ = *s++; 437 for (; *s && (*t = *s) != ']'; s++, t++) 438 if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) { 439 *++t = *++s, t++, s++; 440 for (c = *s; (*t = *s) != ']' || c != d; s++, t++) 441 if ((c = *s) == '\0') 442 return NULL; 443 } 444 return (*s == ']') ? *sp = ++s, ++t : NULL; 445 } 446 447 /* 448 * Compiles the regular expression in RE and returns a pointer to the compiled 449 * regular expression. 450 * Cflags are passed to regcomp. 451 */ 452 static regex_t * 453 compile_re(char *re, int case_insensitive) 454 { 455 regex_t *rep; 456 int eval, flags; 457 458 459 flags = rflags; 460 if (case_insensitive) 461 flags |= REG_ICASE; 462 rep = emalloc(sizeof(regex_t)); 463 parse_escapes(re); 464 if ((eval = regcomp(rep, re, flags)) != 0) 465 sederr("RE error: %s", 466 strregerror(eval, rep)); 467 if (maxnsub < rep->re_nsub) 468 maxnsub = rep->re_nsub; 469 return (rep); 470 } 471 472 static char 473 cton(char c, int base) 474 { 475 switch (c) { 476 case '0': case '1': case '2': case '3': case '4': 477 case '5': case '6': case '7': 478 return (char)(c - '0'); 479 case '8': case '9': 480 return base == 8 ? '?' : (char)(c - '0'); 481 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': 482 return base == 16 ? (char)(c - 'a' + 10) : '?'; 483 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': 484 return base == 16 ? (char)(c - 'A' + 10) : '?'; 485 default: 486 return '?'; 487 } 488 } 489 490 static int 491 ston(char **pp, char *sp, int base) 492 { 493 char *p = *pp, n; 494 int r = cton(p[1], base); 495 char *ep = p + (base == 16 ? 2 : 3); 496 497 if (r == '?') 498 return 0; 499 for (p++; p < ep && (n = cton(p[1], base)) != '?'; p++) { 500 int nr = r * base + n; 501 if (nr > 255) 502 break; 503 r = nr; 504 } 505 *sp = (char)r; 506 *pp = p; 507 return 1; 508 } 509 510 static int 511 unescape(char **pp, char **spp) 512 { 513 char *p = *pp; 514 char *sp = *spp; 515 516 switch (*p) { 517 case 'o': 518 if (!ston(&p, sp, 8)) 519 return 0; 520 break; 521 case 'd': 522 if (!ston(&p, sp, 10)) 523 return 0; 524 break; 525 case 'x': 526 if (!ston(&p, sp, 16)) 527 return 0; 528 break; 529 case 'a': 530 *sp = '\a'; 531 break; 532 #if 0 533 // No, \b RE 534 case 'b': 535 *sp = '\b'; 536 break; 537 #endif 538 case 'f': 539 *sp = '\f'; 540 break; 541 case 'n': 542 *sp = '\n'; 543 break; 544 case 'r': 545 *sp = '\r'; 546 break; 547 case 't': 548 *sp = '\t'; 549 break; 550 case 'v': 551 *sp = '\v'; 552 break; 553 default: 554 return 0; 555 } 556 *spp = sp + 1; 557 *pp = p; 558 return 1; 559 } 560 561 static void 562 parse_escapes(char *buf) 563 { 564 char bracket = '\0'; 565 char *p, *q; 566 567 p = q = buf; 568 569 for (p = q = buf; *p; p++) { 570 if (*p == '\\' && p[1] && !bracket) { 571 p++; 572 if (unescape(&p, &q)) 573 continue; 574 *q++ = '\\'; 575 } 576 switch (*p) { 577 case '[': 578 if (!bracket) 579 bracket = *p; 580 break; 581 case '.': 582 case ':': 583 case '=': 584 if (bracket == '[' && p[-1] == '[') 585 bracket = *p; 586 break; 587 case ']': 588 if (!bracket) 589 break; 590 if (bracket == '[') 591 bracket = '\0'; 592 else if (p[-2] != bracket && p[-1] == bracket) 593 bracket = '['; 594 break; 595 default: 596 break; 597 } 598 *q++ = *p; 599 } 600 *q = '\0'; 601 } 602 603 /* 604 * Compile the substitution string of a regular expression and set res to 605 * point to a saved copy of it. Nsub is the number of parenthesized regular 606 * expressions. 607 */ 608 static char * 609 compile_subst(char *p, struct s_subst *s) 610 { 611 static char lbuf[_POSIX2_LINE_MAX + 1]; 612 size_t asize, size; 613 u_char ref; 614 char c, *text, *op, *sp; 615 int more = 1, sawesc = 0; 616 617 c = *p++; /* Terminator character */ 618 if (c == '\0') 619 return (NULL); 620 621 s->maxbref = 0; 622 s->linenum = linenum; 623 asize = 2 * _POSIX2_LINE_MAX + 1; 624 text = emalloc(asize); 625 size = 0; 626 do { 627 op = sp = text + size; 628 for (; *p; p++) { 629 if (*p == '\\' || sawesc) { 630 /* 631 * If this is a continuation from the last 632 * buffer, we won't have a character to 633 * skip over. 634 */ 635 if (sawesc) 636 sawesc = 0; 637 else 638 p++; 639 640 switch (*p) { 641 case '\0': 642 /* 643 * This escaped character is continued 644 * in the next part of the line. Note 645 * this fact, then cause the loop to 646 * exit w/ normal EOL case and reenter 647 * above with the new buffer. 648 */ 649 sawesc = 1; 650 p--; 651 continue; 652 case '0': case '1': case '2': case '3': 653 case '4': case '5': case '6': case '7': 654 case '8': case '9': 655 *sp++ = '\\'; 656 ref = (u_char)(*p - '0'); 657 if (s->re != NULL && 658 ref > s->re->re_nsub) 659 sederr("\\%c not defined in the RE", 660 *p); 661 if (s->maxbref < ref) 662 s->maxbref = ref; 663 break; 664 case '&': 665 case '\\': 666 *sp++ = '\\'; 667 break; 668 default: 669 if (unescape(&p, &sp)) 670 continue; 671 break; 672 } 673 } else if (*p == c) { 674 if (*++p == '\0' && more) { 675 if (cu_fgets(lbuf, sizeof(lbuf), &more)) 676 p = lbuf; 677 } 678 *sp++ = '\0'; 679 size += (size_t)(sp - op); 680 s->new = erealloc(text, size); 681 return (p); 682 } else if (*p == '\n') { 683 sederr( 684 "unescaped newline inside substitute pattern"); 685 /* NOTREACHED */ 686 } 687 *sp++ = *p; 688 } 689 size += (size_t)(sp - op); 690 if (asize - size < _POSIX2_LINE_MAX + 1) { 691 asize *= 2; 692 text = erealloc(text, asize); 693 } 694 } while (cu_fgets(p = lbuf, sizeof(lbuf), &more)); 695 sederr("unterminated substitute in regular expression"); 696 /* NOTREACHED */ 697 } 698 699 /* 700 * Compile the flags of the s command 701 */ 702 static char * 703 compile_flags(char *p, struct s_subst *s) 704 { 705 int gn; /* True if we have seen g or n */ 706 unsigned long nval; 707 char wfile[_POSIX2_LINE_MAX + 1], *q; 708 709 s->n = 1; /* Default */ 710 s->p = 0; 711 s->wfile = NULL; 712 s->wfd = -1; 713 s->icase = 0; 714 for (gn = 0;;) { 715 EATSPACE(); /* EXTENSION */ 716 switch (*p) { 717 case 'g': 718 if (gn) 719 sederr( 720 "more than one number or 'g' in substitute flags"); 721 gn = 1; 722 s->n = 0; 723 break; 724 case '\0': 725 case '\n': 726 case ';': 727 return (p); 728 case 'p': 729 s->p = 1; 730 break; 731 case 'i': 732 case 'I': 733 s->icase = 1; 734 break; 735 case '1': case '2': case '3': 736 case '4': case '5': case '6': 737 case '7': case '8': case '9': 738 if (gn) 739 sederr( 740 "more than one number or 'g' in substitute flags"); 741 gn = 1; 742 errno = 0; 743 nval = strtoul(p, &p, 10); 744 if (errno == ERANGE || nval > INT_MAX) 745 sederr( 746 "overflow in the 'N' substitute flag"); 747 s->n = (int)nval; 748 p--; 749 break; 750 case 'w': 751 p++; 752 #ifdef HISTORIC_PRACTICE 753 if (*p != ' ') { 754 sedwarn("space missing before w wfile"); 755 return (p); 756 } 757 #endif 758 EATSPACE(); 759 q = wfile; 760 while (*p) { 761 if (*p == '\n') 762 break; 763 *q++ = *p++; 764 } 765 *q = '\0'; 766 if (q == wfile) 767 sederr("no wfile specified"); 768 s->wfile = strdup(wfile); 769 if (!aflag && (s->wfd = open(wfile, 770 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 771 DEFFILEMODE)) == -1) 772 err(1, "%s", wfile); 773 return (p); 774 default: 775 sederr("bad flag in substitute command: '%c'", 776 *p); 777 } 778 p++; 779 } 780 } 781 782 /* 783 * Compile a translation set of strings into a lookup table. 784 */ 785 static char * 786 compile_tr(char *p, struct s_tr **py) 787 { 788 struct s_tr *y; 789 size_t i; 790 const char *op, *np; 791 char old[_POSIX2_LINE_MAX + 1]; 792 char new[_POSIX2_LINE_MAX + 1]; 793 size_t oclen, oldlen, nclen, newlen; 794 mbstate_t mbs1, mbs2; 795 796 *py = y = emalloc(sizeof(*y)); 797 y->multis = NULL; 798 y->nmultis = 0; 799 800 if (*p == '\0' || *p == '\\') 801 sederr( 802 "transform pattern can not be delimited by newline or backslash"); 803 p = compile_delimited(p, old, 1); 804 if (p == NULL) 805 sederr("unterminated transform source string"); 806 p = compile_delimited(p - 1, new, 1); 807 if (p == NULL) 808 sederr("unterminated transform target string"); 809 EATSPACE(); 810 op = old; 811 oldlen = mbsrtowcs(NULL, &op, 0, NULL); 812 if (oldlen == (size_t)-1) 813 err(1, NULL); 814 np = new; 815 newlen = mbsrtowcs(NULL, &np, 0, NULL); 816 if (newlen == (size_t)-1) 817 err(1, NULL); 818 if (newlen != oldlen) 819 sederr("transform strings are not the same length"); 820 if (MB_CUR_MAX == 1) { 821 /* 822 * The single-byte encoding case is easy: generate a 823 * lookup table. 824 */ 825 for (i = 0; i <= UCHAR_MAX; i++) 826 y->bytetab[i] = (u_char)i; 827 for (; *op; op++, np++) 828 y->bytetab[(u_char)*op] = (u_char)*np; 829 } else { 830 /* 831 * Multi-byte encoding case: generate a lookup table as 832 * above, but only for single-byte characters. The first 833 * bytes of multi-byte characters have their lookup table 834 * entries set to 0, which causes do_tr() to search through 835 * an auxiliary vector of multi-byte mappings. 836 */ 837 memset(&mbs1, 0, sizeof(mbs1)); 838 memset(&mbs2, 0, sizeof(mbs2)); 839 for (i = 0; i <= UCHAR_MAX; i++) 840 y->bytetab[i] = (u_char)((btowc((int)i) != WEOF) ? i : 0); 841 while (*op != '\0') { 842 oclen = mbrlen(op, MB_LEN_MAX, &mbs1); 843 if (oclen == (size_t)-1 || oclen == (size_t)-2) 844 errc(1, EILSEQ, NULL); 845 nclen = mbrlen(np, MB_LEN_MAX, &mbs2); 846 if (nclen == (size_t)-1 || nclen == (size_t)-2) 847 errc(1, EILSEQ, NULL); 848 if (oclen == 1 && nclen == 1) 849 y->bytetab[(u_char)*op] = (u_char)*np; 850 else { 851 y->bytetab[(u_char)*op] = 0; 852 y->multis = erealloc(y->multis, 853 (y->nmultis + 1) * sizeof(*y->multis)); 854 i = y->nmultis++; 855 y->multis[i].fromlen = oclen; 856 memcpy(y->multis[i].from, op, oclen); 857 y->multis[i].tolen = nclen; 858 memcpy(y->multis[i].to, np, nclen); 859 } 860 op += oclen; 861 np += nclen; 862 } 863 } 864 return (p); 865 } 866 867 /* 868 * Compile the text following an a or i command. 869 */ 870 static char * 871 compile_text(void) 872 { 873 size_t asize, size; 874 int esc_nl; 875 char *text, *p, *op, *s; 876 char lbuf[_POSIX2_LINE_MAX + 1]; 877 878 asize = 2 * _POSIX2_LINE_MAX + 1; 879 text = emalloc(asize); 880 size = 0; 881 while (cu_fgets(lbuf, sizeof(lbuf), NULL)) { 882 op = s = text + size; 883 p = lbuf; 884 for (esc_nl = 0; *p != '\0'; p++) { 885 if (*p == '\\' && p[1] != '\0' && *++p == '\n') 886 esc_nl = 1; 887 *s++ = *p; 888 } 889 size += (size_t)(s - op); 890 if (!esc_nl) { 891 *s = '\0'; 892 break; 893 } 894 if (asize - size < _POSIX2_LINE_MAX + 1) { 895 asize *= 2; 896 text = erealloc(text, asize); 897 } 898 } 899 text[size] = '\0'; 900 p = erealloc(text, size + 1); 901 return (p); 902 } 903 904 /* 905 * Get an address and return a pointer to the first character after 906 * it. Fill the structure pointed to according to the address. 907 */ 908 static char * 909 compile_addr(char *p, struct s_addr *a) 910 { 911 char *end, re[_POSIX2_LINE_MAX + 1]; 912 int icase; 913 914 icase = 0; 915 916 a->type = 0; 917 switch (*p) { 918 case '\\': /* Context address */ 919 ++p; 920 /* FALLTHROUGH */ 921 case '/': /* Context address */ 922 p = compile_delimited(p, re, 0); 923 if (p == NULL) 924 sederr("unterminated regular expression"); 925 /* Check for case insensitive regexp flag */ 926 if (*p == 'I') { 927 icase = 1; 928 p++; 929 } 930 if (*re == '\0') 931 a->u.r = NULL; 932 else 933 a->u.r = compile_re(re, icase); 934 a->type = AT_RE; 935 return (p); 936 937 case '$': /* Last line */ 938 a->type = AT_LAST; 939 return (p + 1); 940 941 case '+': /* Relative line number */ 942 a->type = AT_RELLINE; 943 p++; 944 /* FALLTHROUGH */ 945 /* Line number */ 946 case '0': case '1': case '2': case '3': case '4': 947 case '5': case '6': case '7': case '8': case '9': 948 if (a->type == 0) 949 a->type = AT_LINE; 950 a->u.l = strtoul(p, &end, 10); 951 return (end); 952 default: 953 sederr("expected context address"); 954 } 955 } 956 957 /* 958 * duptoeol -- 959 * Return a copy of all the characters up to \n or \0. 960 */ 961 static char * 962 duptoeol(char *s, const char *ctype) 963 { 964 size_t len; 965 int ws; 966 char *p, *start; 967 968 ws = 0; 969 for (start = s; *s != '\0' && *s != '\n'; ++s) 970 ws = isspace((unsigned char)*s); 971 *s = '\0'; 972 if (ws) 973 sedwarn("whitespace after %s", ctype); 974 len = (size_t)(s - start + 1); 975 p = emalloc(len); 976 return (memmove(p, start, len)); 977 } 978 979 /* 980 * Convert goto label names to addresses, and count a and r commands, in 981 * the given subset of the script. Free the memory used by labels in b 982 * and t commands (but not by :). 983 * 984 * TODO: Remove } nodes 985 */ 986 static void 987 fixuplabel(struct s_command *cp, struct s_command *end) 988 { 989 990 for (; cp != end; cp = cp->next) 991 switch (cp->code) { 992 case 'a': 993 case 'r': 994 appendnum++; 995 break; 996 case 'b': 997 case 't': 998 /* Resolve branch target. */ 999 if (cp->t == NULL) { 1000 cp->u.c = NULL; 1001 break; 1002 } 1003 if ((cp->u.c = findlabel(cp->t)) == NULL) 1004 sederr("undefined label '%s'", cp->t); 1005 free(cp->t); 1006 break; 1007 case '{': 1008 /* Do interior commands. */ 1009 fixuplabel(cp->u.c, cp->next); 1010 break; 1011 } 1012 } 1013 1014 /* 1015 * Associate the given command label for later lookup. 1016 */ 1017 static void 1018 enterlabel(struct s_command *cp) 1019 { 1020 struct labhash **lhp, *lh; 1021 u_char *p; 1022 u_int h, c; 1023 1024 for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++) 1025 h = (h << 5) + h + c; 1026 lhp = &labels[h & LHMASK]; 1027 for (lh = *lhp; lh != NULL; lh = lh->lh_next) 1028 if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0) 1029 sederr("duplicate label '%s'", cp->t); 1030 lh = emalloc(sizeof *lh); 1031 lh->lh_next = *lhp; 1032 lh->lh_hash = h; 1033 lh->lh_cmd = cp; 1034 lh->lh_ref = 0; 1035 *lhp = lh; 1036 } 1037 1038 /* 1039 * Find the label contained in the command l in the command linked 1040 * list cp. L is excluded from the search. Return NULL if not found. 1041 */ 1042 static struct s_command * 1043 findlabel(char *name) 1044 { 1045 struct labhash *lh; 1046 u_char *p; 1047 u_int h, c; 1048 1049 for (h = 0, p = (u_char *)name; (c = *p) != 0; p++) 1050 h = (h << 5) + h + c; 1051 for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) { 1052 if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) { 1053 lh->lh_ref = 1; 1054 return (lh->lh_cmd); 1055 } 1056 } 1057 return (NULL); 1058 } 1059 1060 /* 1061 * Warn about any unused labels. As a side effect, release the label hash 1062 * table space. 1063 */ 1064 static void 1065 uselabel(void) 1066 { 1067 struct labhash *lh, *next; 1068 int i; 1069 1070 for (i = 0; i < LHSZ; i++) { 1071 for (lh = labels[i]; lh != NULL; lh = next) { 1072 next = lh->lh_next; 1073 if (!lh->lh_ref) 1074 sedwarn("unused label '%s'", lh->lh_cmd->t); 1075 free(lh); 1076 } 1077 } 1078 } 1079
Indexes created Sun Oct 25 00:26:45 UTC 2026