1 /* $NetBSD: compile.c,v 1.55 2025/06/03 19:02:29 martin 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 1.55 2025/06/03 19:02:29 martin 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 static void parse_escapes(char *); 93 94 /* 95 * Command specification. This is used to drive the command parser. 96 */ 97 struct s_format { 98 char code; /* Command code */ 99 int naddr; /* Number of address args */ 100 enum e_args args; /* Argument type */ 101 }; 102 103 static struct s_format cmd_fmts[] = { 104 {'{', 2, GROUP}, 105 {'}', 0, ENDGROUP}, 106 {'a', 1, TEXT}, 107 {'b', 2, BRANCH}, 108 {'c', 2, TEXT}, 109 {'d', 2, EMPTY}, 110 {'D', 2, EMPTY}, 111 {'g', 2, EMPTY}, 112 {'G', 2, EMPTY}, 113 {'h', 2, EMPTY}, 114 {'H', 2, EMPTY}, 115 {'i', 1, TEXT}, 116 {'l', 2, EMPTY}, 117 {'n', 2, EMPTY}, 118 {'N', 2, EMPTY}, 119 {'p', 2, EMPTY}, 120 {'P', 2, EMPTY}, 121 {'q', 1, EMPTY}, 122 {'r', 1, RFILE}, 123 {'s', 2, SUBST}, 124 {'t', 2, BRANCH}, 125 {'w', 2, WFILE}, 126 {'x', 2, EMPTY}, 127 {'y', 2, TR}, 128 {'!', 2, NONSEL}, 129 {':', 0, LABEL}, 130 {'#', 0, COMMENT}, 131 {'=', 1, EMPTY}, 132 {'\0', 0, COMMENT}, 133 }; 134 135 /* The compiled program. */ 136 struct s_command *prog; 137 138 /* 139 * Compile the program into prog. 140 * Initialise appends. 141 */ 142 void 143 compile(void) 144 { 145 *compile_stream(&prog) = NULL; 146 fixuplabel(prog, NULL); 147 uselabel(); 148 if (appendnum > 0) 149 appends = xmalloc(sizeof(struct s_appends) * appendnum); 150 match = xmalloc((maxnsub + 1) * sizeof(regmatch_t)); 151 } 152 153 #define EATSPACE() do { \ 154 if (p) \ 155 while (*p && isspace((unsigned char)*p)) \ 156 p++; \ 157 } while (0) 158 159 static struct s_command ** 160 compile_stream(struct s_command **link) 161 { 162 char *p; 163 static char lbuf[_POSIX2_LINE_MAX + 1]; /* To save stack */ 164 struct s_command *cmd, *cmd2, *stack; 165 struct s_format *fp; 166 char re[_POSIX2_LINE_MAX + 1]; 167 int naddr; /* Number of addresses */ 168 169 stack = 0; 170 for (;;) { 171 if ((p = cu_fgets(lbuf, sizeof(lbuf), NULL)) == NULL) { 172 if (stack != 0) 173 errx(1, "%lu: %s: unexpected EOF (pending }'s)", 174 linenum, fname); 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 = xmalloc(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 = xmalloc(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 = xmalloc(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 errx(1, "%lu: %s: command expected", linenum, fname); 215 cmd->code = *p; 216 for (fp = cmd_fmts; fp->code; fp++) 217 if (fp->code == *p) 218 break; 219 if (!fp->code) 220 errx(1, "%lu: %s: invalid command code %c", linenum, fname, *p); 221 if (naddr > fp->naddr) 222 errx(1, 223 "%lu: %s: command %c expects up to %d address(es), found %d", 224 linenum, fname, *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 errx(1, "%lu: %s: unexpected }", linenum, fname); 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 errx(1, "%lu: %s: extra characters at the end of %c command", 266 linenum, fname, cmd->code); 267 } 268 break; 269 case TEXT: /* a c i */ 270 p++; 271 EATSPACE(); 272 if (*p != '\\') 273 errx(1, 274 "%lu: %s: command %c expects \\ followed by text", linenum, fname, cmd->code); 275 p++; 276 EATSPACE(); 277 if (*p) 278 errx(1, 279 "%lu: %s: extra characters after \\ at the end of %c command", 280 linenum, fname, 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 errx(1, "%lu: %s: filename expected", linenum, fname); 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 errx(1, "%lu: %s: filename expected", linenum, fname); 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 errx(1, "%lu: %s: empty label", linenum, fname); 320 enterlabel(cmd); 321 break; 322 case SUBST: /* s */ 323 p++; 324 if (*p == '\0' || *p == '\\') 325 errx(1, 326 "%lu: %s: substitute pattern can not be delimited by newline or backslash", 327 linenum, fname); 328 cmd->u.s = xcalloc(1, sizeof(struct s_subst)); 329 p = compile_delimited(p, re, 0); 330 if (p == NULL) 331 errx(1, 332 "%lu: %s: unterminated substitute pattern", linenum, fname); 333 334 /* Compile RE with no case sensitivity temporarily */ 335 if (*re == '\0') 336 cmd->u.s->re = NULL; 337 else 338 cmd->u.s->re = compile_re(re, 0); 339 --p; 340 p = compile_subst(p, cmd->u.s); 341 p = compile_flags(p, cmd->u.s); 342 343 /* Recompile RE with case sensitivity from "I" flag if any */ 344 if (*re == '\0') 345 cmd->u.s->re = NULL; 346 else 347 cmd->u.s->re = compile_re(re, cmd->u.s->icase); 348 EATSPACE(); 349 if (*p == ';') { 350 p++; 351 link = &cmd->next; 352 goto semicolon; 353 } 354 break; 355 case TR: /* y */ 356 p++; 357 p = compile_tr(p, &cmd->u.y); 358 EATSPACE(); 359 switch (*p) { 360 case ';': 361 p++; 362 link = &cmd->next; 363 goto semicolon; 364 case '}': 365 goto semicolon; 366 case '\0': 367 break; 368 default: 369 errx(1, 370 "%lu: %s: extra text at the end of a transform command", linenum, fname); 371 } 372 if (*p) 373 break; 374 } 375 } 376 } 377 378 /* 379 * Get a delimited string. P points to the delimiter of the string; d points 380 * to a buffer area. Newline and delimiter escapes are processed; other 381 * escapes are ignored. 382 * 383 * Returns a pointer to the first character after the final delimiter or NULL 384 * in the case of a non-terminated string. The character array d is filled 385 * with the processed string. 386 */ 387 static char * 388 compile_delimited(char *p, char *d, int is_tr) 389 { 390 char c; 391 392 c = *p++; 393 if (c == '\0') 394 return (NULL); 395 else if (c == '\\') 396 errx(1, "%lu: %s: \\ can not be used as a string delimiter", 397 linenum, fname); 398 else if (c == '\n') 399 errx(1, "%lu: %s: newline can not be used as a string delimiter", 400 linenum, fname); 401 while (*p) { 402 if (*p == '[' && *p != c) { 403 if ((d = compile_ccl(&p, d)) == NULL) 404 errx(1, "%lu: %s: unbalanced brackets ([])", linenum, fname); 405 continue; 406 } else if (*p == '\\' && p[1] == '[') { 407 *d++ = *p++; 408 } else if (*p == '\\' && p[1] == c) 409 p++; 410 else if (*p == '\\' && p[1] == 'n') { 411 *d++ = '\n'; 412 p += 2; 413 continue; 414 } else if (*p == '\\' && p[1] == '\\') { 415 if (is_tr) 416 p++; 417 else 418 *d++ = *p++; 419 } else if (*p == c) { 420 *d = '\0'; 421 return (p + 1); 422 } 423 *d++ = *p++; 424 } 425 return (NULL); 426 } 427 428 429 /* compile_ccl: expand a POSIX character class */ 430 static char * 431 compile_ccl(char **sp, char *t) 432 { 433 int c, d; 434 char *s = *sp; 435 436 *t++ = *s++; 437 if (*s == '^') 438 *t++ = *s++; 439 if (*s == ']') 440 *t++ = *s++; 441 for (; *s && (*t = *s) != ']'; s++, t++) 442 if (*s == '[' && ((d = *(s+1)) == '.' || d == ':' || d == '=')) { 443 *++t = *++s, t++, s++; 444 for (c = *s; (*t = *s) != ']' || c != d; s++, t++) 445 if ((c = *s) == '\0') 446 return NULL; 447 } 448 return (*s == ']') ? *sp = ++s, ++t : NULL; 449 } 450 451 /* 452 * Compiles the regular expression in RE and returns a pointer to the compiled 453 * regular expression. 454 * Cflags are passed to regcomp. 455 */ 456 static regex_t * 457 compile_re(char *re, int case_insensitive) 458 { 459 regex_t *rep; 460 int eval, flags; 461 462 463 flags = rflags; 464 if (case_insensitive) 465 flags |= REG_ICASE; 466 rep = xmalloc(sizeof(regex_t)); 467 parse_escapes(re); 468 if ((eval = regcomp(rep, re, flags)) != 0) 469 errx(1, "%lu: %s: RE error: %s", 470 linenum, fname, strregerror(eval, rep)); 471 if (maxnsub < rep->re_nsub) 472 maxnsub = rep->re_nsub; 473 return (rep); 474 } 475 476 static char 477 cton(char c, int base) 478 { 479 switch (c) { 480 case '0': case '1': case '2': case '3': case '4': 481 case '5': case '6': case '7': 482 return (char)(c - '0'); 483 case '8': case '9': 484 return base == 8 ? '?' : (char)(c - '0'); 485 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': 486 return base == 16 ? (char)(c - 'a' + 10) : '?'; 487 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': 488 return base == 16 ? (char)(c - 'A' + 10) : '?'; 489 default: 490 return '?'; 491 } 492 } 493 494 static int 495 ston(char **pp, char *sp, int base) 496 { 497 char *p = *pp, n; 498 int r = cton(p[1], base); 499 char *ep = p + (base == 16 ? 2 : 3); 500 501 if (r == '?') 502 return 0; 503 for (p++; p < ep && (n = cton(p[1], base)) != '?'; p++) { 504 int nr = r * base + n; 505 if (nr > 255) 506 break; 507 r = nr; 508 } 509 *sp = (char)r; 510 *pp = p; 511 return 1; 512 } 513 514 static int 515 unescape(char **pp, char **spp) 516 { 517 char *p = *pp; 518 char *sp = *spp; 519 520 switch (*p) { 521 case 'o': 522 if (!ston(&p, sp, 8)) 523 return 0; 524 break; 525 case 'd': 526 if (!ston(&p, sp, 10)) 527 return 0; 528 break; 529 case 'x': 530 if (!ston(&p, sp, 16)) 531 return 0; 532 break; 533 case 'a': 534 *sp = '\a'; 535 break; 536 #if 0 537 // No, \b RE 538 case 'b': 539 *sp = '\b'; 540 break; 541 #endif 542 case 'f': 543 *sp = '\f'; 544 break; 545 case 'n': 546 *sp = '\n'; 547 break; 548 case 'r': 549 *sp = '\r'; 550 break; 551 case 't': 552 *sp = '\t'; 553 break; 554 case 'v': 555 *sp = '\v'; 556 break; 557 default: 558 return 0; 559 } 560 *spp = sp + 1; 561 *pp = p; 562 return 1; 563 } 564 565 static void 566 parse_escapes(char *buf) 567 { 568 char bracket = '\0'; 569 char *p, *q; 570 571 p = q = buf; 572 573 for (p = q = buf; *p; p++) { 574 if (*p == '\\' && p[1] && !bracket) { 575 p++; 576 if (unescape(&p, &q)) 577 continue; 578 *q++ = '\\'; 579 } 580 switch (*p) { 581 case '[': 582 if (!bracket) 583 bracket = *p; 584 break; 585 case '.': 586 case ':': 587 case '=': 588 if (bracket == '[' && p[-1] == '[') 589 bracket = *p; 590 break; 591 case ']': 592 if (!bracket) 593 break; 594 if (bracket == '[') 595 bracket = '\0'; 596 else if (p[-2] != bracket && p[-1] == bracket) 597 bracket = '['; 598 break; 599 default: 600 break; 601 } 602 *q++ = *p; 603 } 604 *q = '\0'; 605 } 606 607 /* 608 * Compile the substitution string of a regular expression and set res to 609 * point to a saved copy of it. Nsub is the number of parenthesized regular 610 * expressions. 611 */ 612 static char * 613 compile_subst(char *p, struct s_subst *s) 614 { 615 static char lbuf[_POSIX2_LINE_MAX + 1]; 616 size_t asize, size; 617 u_char ref; 618 char c, *text, *op, *sp; 619 int more = 1, sawesc = 0; 620 621 c = *p++; /* Terminator character */ 622 if (c == '\0') 623 return (NULL); 624 625 s->maxbref = 0; 626 s->linenum = linenum; 627 asize = 2 * _POSIX2_LINE_MAX + 1; 628 text = xmalloc(asize); 629 size = 0; 630 do { 631 op = sp = text + size; 632 for (; *p; p++) { 633 if (*p == '\\' || sawesc) { 634 /* 635 * If this is a continuation from the last 636 * buffer, we won't have a character to 637 * skip over. 638 */ 639 if (sawesc) 640 sawesc = 0; 641 else 642 p++; 643 644 switch (*p) { 645 case '\0': 646 /* 647 * This escaped character is continued 648 * in the next part of the line. Note 649 * this fact, then cause the loop to 650 * exit w/ normal EOL case and reenter 651 * above with the new buffer. 652 */ 653 sawesc = 1; 654 p--; 655 continue; 656 case '0': case '1': case '2': case '3': 657 case '4': case '5': case '6': case '7': 658 case '8': case '9': 659 *sp++ = '\\'; 660 ref = (u_char)(*p - '0'); 661 if (s->re != NULL && 662 ref > s->re->re_nsub) 663 errx(1, "%lu: %s: \\%c not defined in the RE", 664 linenum, fname, *p); 665 if (s->maxbref < ref) 666 s->maxbref = ref; 667 break; 668 case '&': 669 case '\\': 670 *sp++ = '\\'; 671 break; 672 default: 673 if (unescape(&p, &sp)) 674 continue; 675 break; 676 } 677 } else if (*p == c) { 678 if (*++p == '\0' && more) { 679 if (cu_fgets(lbuf, sizeof(lbuf), &more)) 680 p = lbuf; 681 } 682 *sp++ = '\0'; 683 size += (size_t)(sp - op); 684 s->new = xrealloc(text, size); 685 return (p); 686 } else if (*p == '\n') { 687 errx(1, 688 "%lu: %s: unescaped newline inside substitute pattern", linenum, fname); 689 /* NOTREACHED */ 690 } 691 *sp++ = *p; 692 } 693 size += (size_t)(sp - op); 694 if (asize - size < _POSIX2_LINE_MAX + 1) { 695 asize *= 2; 696 text = xrealloc(text, asize); 697 } 698 } while (cu_fgets(p = lbuf, sizeof(lbuf), &more)); 699 errx(1, "%lu: %s: unterminated substitute in regular expression", 700 linenum, fname); 701 /* NOTREACHED */ 702 } 703 704 /* 705 * Compile the flags of the s command 706 */ 707 static char * 708 compile_flags(char *p, struct s_subst *s) 709 { 710 int gn; /* True if we have seen g or n */ 711 unsigned long nval; 712 char wfile[_POSIX2_LINE_MAX + 1], *q; 713 714 s->n = 1; /* Default */ 715 s->p = 0; 716 s->wfile = NULL; 717 s->wfd = -1; 718 s->icase = 0; 719 for (gn = 0;;) { 720 EATSPACE(); /* EXTENSION */ 721 switch (*p) { 722 case 'g': 723 if (gn) 724 errx(1, 725 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname); 726 gn = 1; 727 s->n = 0; 728 break; 729 case '\0': 730 case '\n': 731 case ';': 732 return (p); 733 case 'p': 734 s->p = 1; 735 break; 736 case 'i': 737 case 'I': 738 s->icase = 1; 739 break; 740 case '1': case '2': case '3': 741 case '4': case '5': case '6': 742 case '7': case '8': case '9': 743 if (gn) 744 errx(1, 745 "%lu: %s: more than one number or 'g' in substitute flags", linenum, fname); 746 gn = 1; 747 errno = 0; 748 nval = strtoul(p, &p, 10); 749 if (errno == ERANGE || nval > INT_MAX) 750 errx(1, 751 "%lu: %s: overflow in the 'N' substitute flag", linenum, fname); 752 s->n = (int)nval; 753 p--; 754 break; 755 case 'w': 756 p++; 757 #ifdef HISTORIC_PRACTICE 758 if (*p != ' ') { 759 warnx("%lu: %s: space missing before w wfile", linenum, fname); 760 return (p); 761 } 762 #endif 763 EATSPACE(); 764 q = wfile; 765 while (*p) { 766 if (*p == '\n') 767 break; 768 *q++ = *p++; 769 } 770 *q = '\0'; 771 if (q == wfile) 772 errx(1, "%lu: %s: no wfile specified", linenum, fname); 773 s->wfile = strdup(wfile); 774 if (!aflag && (s->wfd = open(wfile, 775 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 776 DEFFILEMODE)) == -1) 777 err(1, "%s", wfile); 778 return (p); 779 default: 780 errx(1, "%lu: %s: bad flag in substitute command: '%c'", 781 linenum, fname, *p); 782 break; 783 } 784 p++; 785 } 786 } 787 788 /* 789 * Compile a translation set of strings into a lookup table. 790 */ 791 static char * 792 compile_tr(char *p, struct s_tr **py) 793 { 794 struct s_tr *y; 795 size_t i; 796 const char *op, *np; 797 char old[_POSIX2_LINE_MAX + 1]; 798 char new[_POSIX2_LINE_MAX + 1]; 799 size_t oclen, oldlen, nclen, newlen; 800 mbstate_t mbs1, mbs2; 801 802 *py = y = xmalloc(sizeof(*y)); 803 y->multis = NULL; 804 y->nmultis = 0; 805 806 if (*p == '\0' || *p == '\\') 807 errx(1, 808 "%lu: %s: transform pattern can not be delimited by newline or backslash", 809 linenum, fname); 810 p = compile_delimited(p, old, 1); 811 if (p == NULL) 812 errx(1, "%lu: %s: unterminated transform source string", 813 linenum, fname); 814 p = compile_delimited(p - 1, new, 1); 815 if (p == NULL) 816 errx(1, "%lu: %s: unterminated transform target string", 817 linenum, fname); 818 EATSPACE(); 819 op = old; 820 oldlen = mbsrtowcs(NULL, &op, 0, NULL); 821 if (oldlen == (size_t)-1) 822 err(1, NULL); 823 np = new; 824 newlen = mbsrtowcs(NULL, &np, 0, NULL); 825 if (newlen == (size_t)-1) 826 err(1, NULL); 827 if (newlen != oldlen) 828 errx(1, "%lu: %s: transform strings are not the same length", 829 linenum, fname); 830 if (MB_CUR_MAX == 1) { 831 /* 832 * The single-byte encoding case is easy: generate a 833 * lookup table. 834 */ 835 for (i = 0; i <= UCHAR_MAX; i++) 836 y->bytetab[i] = (u_char)i; 837 for (; *op; op++, np++) 838 y->bytetab[(u_char)*op] = (u_char)*np; 839 } else { 840 /* 841 * Multi-byte encoding case: generate a lookup table as 842 * above, but only for single-byte characters. The first 843 * bytes of multi-byte characters have their lookup table 844 * entries set to 0, which causes do_tr() to search through 845 * an auxiliary vector of multi-byte mappings. 846 */ 847 memset(&mbs1, 0, sizeof(mbs1)); 848 memset(&mbs2, 0, sizeof(mbs2)); 849 for (i = 0; i <= UCHAR_MAX; i++) 850 y->bytetab[i] = (u_char)((btowc((int)i) != WEOF) ? i : 0); 851 while (*op != '\0') { 852 oclen = mbrlen(op, MB_LEN_MAX, &mbs1); 853 if (oclen == (size_t)-1 || oclen == (size_t)-2) 854 errc(1, EILSEQ, NULL); 855 nclen = mbrlen(np, MB_LEN_MAX, &mbs2); 856 if (nclen == (size_t)-1 || nclen == (size_t)-2) 857 errc(1, EILSEQ, NULL); 858 if (oclen == 1 && nclen == 1) 859 y->bytetab[(u_char)*op] = (u_char)*np; 860 else { 861 y->bytetab[(u_char)*op] = 0; 862 y->multis = xrealloc(y->multis, 863 (y->nmultis + 1) * sizeof(*y->multis)); 864 i = y->nmultis++; 865 y->multis[i].fromlen = oclen; 866 memcpy(y->multis[i].from, op, oclen); 867 y->multis[i].tolen = nclen; 868 memcpy(y->multis[i].to, np, nclen); 869 } 870 op += oclen; 871 np += nclen; 872 } 873 } 874 return (p); 875 } 876 877 /* 878 * Compile the text following an a or i command. 879 */ 880 static char * 881 compile_text(void) 882 { 883 size_t asize, size; 884 int esc_nl; 885 char *text, *p, *op, *s; 886 char lbuf[_POSIX2_LINE_MAX + 1]; 887 888 asize = 2 * _POSIX2_LINE_MAX + 1; 889 text = xmalloc(asize); 890 size = 0; 891 while (cu_fgets(lbuf, sizeof(lbuf), NULL)) { 892 op = s = text + size; 893 p = lbuf; 894 for (esc_nl = 0; *p != '\0'; p++) { 895 if (*p == '\\' && p[1] != '\0' && *++p == '\n') 896 esc_nl = 1; 897 *s++ = *p; 898 } 899 size += (size_t)(s - op); 900 if (!esc_nl) { 901 *s = '\0'; 902 break; 903 } 904 if (asize - size < _POSIX2_LINE_MAX + 1) { 905 asize *= 2; 906 text = xrealloc(text, asize); 907 } 908 } 909 text[size] = '\0'; 910 p = xrealloc(text, size + 1); 911 return (p); 912 } 913 914 /* 915 * Get an address and return a pointer to the first character after 916 * it. Fill the structure pointed to according to the address. 917 */ 918 static char * 919 compile_addr(char *p, struct s_addr *a) 920 { 921 char *end, re[_POSIX2_LINE_MAX + 1]; 922 int icase; 923 924 icase = 0; 925 926 a->type = 0; 927 switch (*p) { 928 case '\\': /* Context address */ 929 ++p; 930 /* FALLTHROUGH */ 931 case '/': /* Context address */ 932 p = compile_delimited(p, re, 0); 933 if (p == NULL) 934 errx(1, "%lu: %s: unterminated regular expression", linenum, fname); 935 /* Check for case insensitive regexp flag */ 936 if (*p == 'I') { 937 icase = 1; 938 p++; 939 } 940 if (*re == '\0') 941 a->u.r = NULL; 942 else 943 a->u.r = compile_re(re, icase); 944 a->type = AT_RE; 945 return (p); 946 947 case '$': /* Last line */ 948 a->type = AT_LAST; 949 return (p + 1); 950 951 case '+': /* Relative line number */ 952 a->type = AT_RELLINE; 953 p++; 954 /* FALLTHROUGH */ 955 /* Line number */ 956 case '0': case '1': case '2': case '3': case '4': 957 case '5': case '6': case '7': case '8': case '9': 958 if (a->type == 0) 959 a->type = AT_LINE; 960 a->u.l = strtoul(p, &end, 10); 961 return (end); 962 default: 963 errx(1, "%lu: %s: expected context address", linenum, fname); 964 return (NULL); 965 } 966 } 967 968 /* 969 * duptoeol -- 970 * Return a copy of all the characters up to \n or \0. 971 */ 972 static char * 973 duptoeol(char *s, const char *ctype) 974 { 975 size_t len; 976 int ws; 977 char *p, *start; 978 979 ws = 0; 980 for (start = s; *s != '\0' && *s != '\n'; ++s) 981 ws = isspace((unsigned char)*s); 982 *s = '\0'; 983 if (ws) 984 warnx("%lu: %s: whitespace after %s", linenum, fname, ctype); 985 len = (size_t)(s - start + 1); 986 p = xmalloc(len); 987 return (memmove(p, start, len)); 988 } 989 990 /* 991 * Convert goto label names to addresses, and count a and r commands, in 992 * the given subset of the script. Free the memory used by labels in b 993 * and t commands (but not by :). 994 * 995 * TODO: Remove } nodes 996 */ 997 static void 998 fixuplabel(struct s_command *cp, struct s_command *end) 999 { 1000 1001 for (; cp != end; cp = cp->next) 1002 switch (cp->code) { 1003 case 'a': 1004 case 'r': 1005 appendnum++; 1006 break; 1007 case 'b': 1008 case 't': 1009 /* Resolve branch target. */ 1010 if (cp->t == NULL) { 1011 cp->u.c = NULL; 1012 break; 1013 } 1014 if ((cp->u.c = findlabel(cp->t)) == NULL) 1015 errx(1, "%lu: %s: undefined label '%s'", linenum, fname, cp->t); 1016 free(cp->t); 1017 break; 1018 case '{': 1019 /* Do interior commands. */ 1020 fixuplabel(cp->u.c, cp->next); 1021 break; 1022 } 1023 } 1024 1025 /* 1026 * Associate the given command label for later lookup. 1027 */ 1028 static void 1029 enterlabel(struct s_command *cp) 1030 { 1031 struct labhash **lhp, *lh; 1032 u_char *p; 1033 u_int h, c; 1034 1035 for (h = 0, p = (u_char *)cp->t; (c = *p) != 0; p++) 1036 h = (h << 5) + h + c; 1037 lhp = &labels[h & LHMASK]; 1038 for (lh = *lhp; lh != NULL; lh = lh->lh_next) 1039 if (lh->lh_hash == h && strcmp(cp->t, lh->lh_cmd->t) == 0) 1040 errx(1, "%lu: %s: duplicate label '%s'", linenum, fname, cp->t); 1041 lh = xmalloc(sizeof *lh); 1042 lh->lh_next = *lhp; 1043 lh->lh_hash = h; 1044 lh->lh_cmd = cp; 1045 lh->lh_ref = 0; 1046 *lhp = lh; 1047 } 1048 1049 /* 1050 * Find the label contained in the command l in the command linked 1051 * list cp. L is excluded from the search. Return NULL if not found. 1052 */ 1053 static struct s_command * 1054 findlabel(char *name) 1055 { 1056 struct labhash *lh; 1057 u_char *p; 1058 u_int h, c; 1059 1060 for (h = 0, p = (u_char *)name; (c = *p) != 0; p++) 1061 h = (h << 5) + h + c; 1062 for (lh = labels[h & LHMASK]; lh != NULL; lh = lh->lh_next) { 1063 if (lh->lh_hash == h && strcmp(name, lh->lh_cmd->t) == 0) { 1064 lh->lh_ref = 1; 1065 return (lh->lh_cmd); 1066 } 1067 } 1068 return (NULL); 1069 } 1070 1071 /* 1072 * Warn about any unused labels. As a side effect, release the label hash 1073 * table space. 1074 */ 1075 static void 1076 uselabel(void) 1077 { 1078 struct labhash *lh, *next; 1079 int i; 1080 1081 for (i = 0; i < LHSZ; i++) { 1082 for (lh = labels[i]; lh != NULL; lh = next) { 1083 next = lh->lh_next; 1084 if (!lh->lh_ref) 1085 warnx("%lu: %s: unused label '%s'", 1086 linenum, fname, lh->lh_cmd->t); 1087 free(lh); 1088 } 1089 } 1090 } 1091
Indexes created Sun Sep 21 20:09:37 GMT 2025