1 /*- 2 * Copyright (c) 1992 Diomidis Spinellis. 3 * Copyright (c) 1992, 1993, 1994 4 * The Regents of the University of California. All rights reserved. 5 * 6 * This code is derived from software contributed to Berkeley by 7 * Diomidis Spinellis of Imperial College, University of London. 8 * 9 * Redistribution and use in source and binary forms, with or without 10 * modification, are permitted provided that the following conditions 11 * are met: 12 * 1. Redistributions of source code must retain the above copyright 13 * notice, this list of conditions and the following disclaimer. 14 * 2. Redistributions in binary form must reproduce the above copyright 15 * notice, this list of conditions and the following disclaimer in the 16 * documentation and/or other materials provided with the distribution. 17 * 4. Neither the name of the University nor the names of its contributors 18 * may be used to endorse or promote products derived from this software 19 * without specific prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 */ 33 34 #include <sys/cdefs.h> 35 __FBSDID("$FreeBSD$"); 36 37 #ifndef lint 38 static const char sccsid[] = "@(#)process.c 8.6 (Berkeley) 4/20/94"; 39 #endif 40 41 #include <sys/types.h> 42 #include <sys/stat.h> 43 #include <sys/ioctl.h> 44 #include <sys/uio.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 <stdio.h> 53 #include <stdlib.h> 54 #include <string.h> 55 #include <unistd.h> 56 #include <wchar.h> 57 #include <wctype.h> 58 59 #include "defs.h" 60 #include "extern.h" 61 62 static SPACE HS, PS, SS, YS; 63 #define pd PS.deleted 64 #define ps PS.space 65 #define psl PS.len 66 #define hs HS.space 67 #define hsl HS.len 68 69 static __inline int applies(struct s_command *); 70 static void do_tr(struct s_tr *); 71 static void flush_appends(void); 72 static void lputs(char *, size_t); 73 static __inline int regexec_e(regex_t *, const char *, int, int, size_t); 74 static void regsub(SPACE *, char *, char *); 75 static int substitute(struct s_command *); 76 77 struct s_appends *appends; /* Array of pointers to strings to append. */ 78 static int appendx; /* Index into appends array. */ 79 int appendnum; /* Size of appends array. */ 80 81 static int lastaddr; /* Set by applies if last address of a range. */ 82 static int sdone; /* If any substitutes since last line input. */ 83 /* Iov structure for 'w' commands. */ 84 static regex_t *defpreg; 85 size_t maxnsub; 86 regmatch_t *match; 87 88 #define OUT(s) { fwrite(s, sizeof(u_char), psl, outfile); fputc('\n', outfile); } 89 90 void 91 process(void) 92 { 93 struct s_command *cp; 94 SPACE tspace; 95 size_t oldpsl = 0; 96 char *p; 97 98 p = NULL; 99 100 for (linenum = 0; mf_fgets(&PS, REPLACE);) { 101 pd = 0; 102 top: 103 cp = prog; 104 redirect: 105 while (cp != NULL) { 106 if (!applies(cp)) { 107 cp = cp->next; 108 continue; 109 } 110 switch (cp->code) { 111 case '{': 112 cp = cp->u.c; 113 goto redirect; 114 case 'a': 115 if (appendx >= appendnum) 116 if ((appends = realloc(appends, 117 sizeof(struct s_appends) * 118 (appendnum *= 2))) == NULL) 119 err(1, "realloc"); 120 appends[appendx].type = AP_STRING; 121 appends[appendx].s = cp->t; 122 appends[appendx].len = strlen(cp->t); 123 appendx++; 124 break; 125 case 'b': 126 cp = cp->u.c; 127 goto redirect; 128 case 'c': 129 pd = 1; 130 psl = 0; 131 if (cp->a2 == NULL || lastaddr) 132 (void)fprintf(outfile, "%s", cp->t); 133 break; 134 case 'd': 135 pd = 1; 136 goto new; 137 case 'D': 138 if (pd) 139 goto new; 140 if (psl == 0 || 141 (p = memchr(ps, '\n', psl)) == NULL) { 142 pd = 1; 143 goto new; 144 } else { 145 psl -= (p + 1) - ps; 146 memmove(ps, p + 1, psl); 147 goto top; 148 } 149 case 'g': 150 cspace(&PS, hs, hsl, REPLACE); 151 break; 152 case 'G': 153 cspace(&PS, "\n", 1, 0); 154 cspace(&PS, hs, hsl, 0); 155 break; 156 case 'h': 157 cspace(&HS, ps, psl, REPLACE); 158 break; 159 case 'H': 160 cspace(&HS, "\n", 1, 0); 161 cspace(&HS, ps, psl, 0); 162 break; 163 case 'i': 164 (void)fprintf(outfile, "%s", cp->t); 165 break; 166 case 'l': 167 lputs(ps, psl); 168 break; 169 case 'n': 170 if (!nflag && !pd) 171 OUT(ps) 172 flush_appends(); 173 if (!mf_fgets(&PS, REPLACE)) 174 exit(0); 175 pd = 0; 176 break; 177 case 'N': 178 flush_appends(); 179 cspace(&PS, "\n", 1, 0); 180 if (!mf_fgets(&PS, 0)) 181 exit(0); 182 break; 183 case 'p': 184 if (pd) 185 break; 186 OUT(ps) 187 break; 188 case 'P': 189 if (pd) 190 break; 191 if (psl != 0 && 192 (p = memchr(ps, '\n', psl)) != NULL) { 193 oldpsl = psl; 194 psl = p - ps; 195 } 196 OUT(ps) 197 if (p != NULL) 198 psl = oldpsl; 199 break; 200 case 'q': 201 if (!nflag && !pd) 202 OUT(ps) 203 flush_appends(); 204 exit(0); 205 case 'r': 206 if (appendx >= appendnum) 207 if ((appends = realloc(appends, 208 sizeof(struct s_appends) * 209 (appendnum *= 2))) == NULL) 210 err(1, "realloc"); 211 appends[appendx].type = AP_FILE; 212 appends[appendx].s = cp->t; 213 appends[appendx].len = strlen(cp->t); 214 appendx++; 215 break; 216 case 's': 217 sdone |= substitute(cp); 218 break; 219 case 't': 220 if (sdone) { 221 sdone = 0; 222 cp = cp->u.c; 223 goto redirect; 224 } 225 break; 226 case 'w': 227 if (pd) 228 break; 229 if (cp->u.fd == -1 && (cp->u.fd = open(cp->t, 230 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 231 DEFFILEMODE)) == -1) 232 err(1, "%s", cp->t); 233 if (write(cp->u.fd, ps, psl) != psl || 234 write(cp->u.fd, "\n", 1) != 1) 235 err(1, "%s", cp->t); 236 break; 237 case 'x': 238 if (hs == NULL) 239 cspace(&HS, "", 0, REPLACE); 240 tspace = PS; 241 PS = HS; 242 HS = tspace; 243 break; 244 case 'y': 245 if (pd || psl == 0) 246 break; 247 do_tr(cp->u.y); 248 break; 249 case ':': 250 case '}': 251 break; 252 case '=': 253 (void)fprintf(outfile, "%lu\n", linenum); 254 } 255 cp = cp->next; 256 } /* for all cp */ 257 258 new: if (!nflag && !pd) 259 OUT(ps) 260 flush_appends(); 261 } /* for all lines */ 262 } 263 264 /* 265 * TRUE if the address passed matches the current program state 266 * (lastline, linenumber, ps). 267 */ 268 #define MATCH(a) \ 269 (a)->type == AT_RE ? regexec_e((a)->u.r, ps, 0, 1, psl) : \ 270 (a)->type == AT_LINE ? linenum == (a)->u.l : lastline() 271 272 /* 273 * Return TRUE if the command applies to the current line. Sets the inrange 274 * flag to process ranges. Interprets the non-select (``!'') flag. 275 */ 276 static __inline int 277 applies(struct s_command *cp) 278 { 279 int r; 280 281 lastaddr = 0; 282 if (cp->a1 == NULL && cp->a2 == NULL) 283 r = 1; 284 else if (cp->a2) 285 if (cp->inrange) { 286 if (MATCH(cp->a2)) { 287 cp->inrange = 0; 288 lastaddr = 1; 289 } 290 r = 1; 291 } else if (MATCH(cp->a1)) { 292 /* 293 * If the second address is a number less than or 294 * equal to the line number first selected, only 295 * one line shall be selected. 296 * -- POSIX 1003.2 297 */ 298 if (cp->a2->type == AT_LINE && 299 linenum >= cp->a2->u.l) 300 lastaddr = 1; 301 else 302 cp->inrange = 1; 303 r = 1; 304 } else 305 r = 0; 306 else 307 r = MATCH(cp->a1); 308 return (cp->nonsel ? ! r : r); 309 } 310 311 /* 312 * substitute -- 313 * Do substitutions in the pattern space. Currently, we build a 314 * copy of the new pattern space in the substitute space structure 315 * and then swap them. 316 */ 317 static int 318 substitute(struct s_command *cp) 319 { 320 SPACE tspace; 321 regex_t *re; 322 regoff_t re_off, slen; 323 int lastempty, n; 324 char *s; 325 326 s = ps; 327 re = cp->u.s->re; 328 if (re == NULL) { 329 if (defpreg != NULL && cp->u.s->maxbref > defpreg->re_nsub) { 330 linenum = cp->u.s->linenum; 331 errx(1, "%lu: %s: \\%d not defined in the RE", 332 linenum, fname, cp->u.s->maxbref); 333 } 334 } 335 if (!regexec_e(re, s, 0, 0, psl)) 336 return (0); 337 338 SS.len = 0; /* Clean substitute space. */ 339 slen = psl; 340 n = cp->u.s->n; 341 lastempty = 1; 342 343 switch (n) { 344 case 0: /* Global */ 345 do { 346 if (lastempty || match[0].rm_so != match[0].rm_eo) { 347 /* Locate start of replaced string. */ 348 re_off = match[0].rm_so; 349 /* Copy leading retained string. */ 350 cspace(&SS, s, re_off, APPEND); 351 /* Add in regular expression. */ 352 regsub(&SS, s, cp->u.s->new); 353 } 354 355 /* Move past this match. */ 356 if (match[0].rm_so != match[0].rm_eo) { 357 s += match[0].rm_eo; 358 slen -= match[0].rm_eo; 359 lastempty = 0; 360 } else { 361 if (match[0].rm_so < slen) 362 cspace(&SS, s + match[0].rm_so, 1, 363 APPEND); 364 s += match[0].rm_so + 1; 365 slen -= match[0].rm_so + 1; 366 lastempty = 1; 367 } 368 } while (slen >= 0 && regexec_e(re, s, REG_NOTBOL, 0, slen)); 369 /* Copy trailing retained string. */ 370 if (slen > 0) 371 cspace(&SS, s, slen, APPEND); 372 break; 373 default: /* Nth occurrence */ 374 while (--n) { 375 if (match[0].rm_eo == match[0].rm_so) 376 match[0].rm_eo = match[0].rm_so + 1; 377 s += match[0].rm_eo; 378 slen -= match[0].rm_eo; 379 if (slen < 0) 380 return (0); 381 if (!regexec_e(re, s, REG_NOTBOL, 0, slen)) 382 return (0); 383 } 384 /* FALLTHROUGH */ 385 case 1: /* 1st occurrence */ 386 /* Locate start of replaced string. */ 387 re_off = match[0].rm_so + (s - ps); 388 /* Copy leading retained string. */ 389 cspace(&SS, ps, re_off, APPEND); 390 /* Add in regular expression. */ 391 regsub(&SS, s, cp->u.s->new); 392 /* Copy trailing retained string. */ 393 s += match[0].rm_eo; 394 slen -= match[0].rm_eo; 395 cspace(&SS, s, slen, APPEND); 396 break; 397 } 398 399 /* 400 * Swap the substitute space and the pattern space, and make sure 401 * that any leftover pointers into stdio memory get lost. 402 */ 403 tspace = PS; 404 PS = SS; 405 SS = tspace; 406 SS.space = SS.back; 407 408 /* Handle the 'p' flag. */ 409 if (cp->u.s->p) 410 OUT(ps) 411 412 /* Handle the 'w' flag. */ 413 if (cp->u.s->wfile && !pd) { 414 if (cp->u.s->wfd == -1 && (cp->u.s->wfd = open(cp->u.s->wfile, 415 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, DEFFILEMODE)) == -1) 416 err(1, "%s", cp->u.s->wfile); 417 if (write(cp->u.s->wfd, ps, psl) != psl || 418 write(cp->u.s->wfd, "\n", 1) != 1) 419 err(1, "%s", cp->u.s->wfile); 420 } 421 return (1); 422 } 423 424 /* 425 * do_tr -- 426 * Perform translation ('y' command) in the pattern space. 427 */ 428 static void 429 do_tr(struct s_tr *y) 430 { 431 SPACE tmp; 432 char c, *p; 433 size_t clen, left; 434 int i; 435 436 if (MB_CUR_MAX == 1) { 437 /* 438 * Single-byte encoding: perform in-place translation 439 * of the pattern space. 440 */ 441 for (p = ps; p < &ps[psl]; p++) 442 *p = y->bytetab[(u_char)*p]; 443 } else { 444 /* 445 * Multi-byte encoding: perform translation into the 446 * translation space, then swap the translation and 447 * pattern spaces. 448 */ 449 /* Clean translation space. */ 450 YS.len = 0; 451 for (p = ps, left = psl; left > 0; p += clen, left -= clen) { 452 if ((c = y->bytetab[(u_char)*p]) != '\0') { 453 cspace(&YS, &c, 1, APPEND); 454 clen = 1; 455 continue; 456 } 457 for (i = 0; i < y->nmultis; i++) 458 if (left >= y->multis[i].fromlen && 459 memcmp(p, y->multis[i].from, 460 y->multis[i].fromlen) == 0) 461 break; 462 if (i < y->nmultis) { 463 cspace(&YS, y->multis[i].to, 464 y->multis[i].tolen, APPEND); 465 clen = y->multis[i].fromlen; 466 } else { 467 cspace(&YS, p, 1, APPEND); 468 clen = 1; 469 } 470 } 471 /* Swap the translation space and the pattern space. */ 472 tmp = PS; 473 PS = YS; 474 YS = tmp; 475 YS.space = YS.back; 476 } 477 } 478 479 /* 480 * Flush append requests. Always called before reading a line, 481 * therefore it also resets the substitution done (sdone) flag. 482 */ 483 static void 484 flush_appends(void) 485 { 486 FILE *f; 487 int count, i; 488 char buf[8 * 1024]; 489 490 for (i = 0; i < appendx; i++) 491 switch (appends[i].type) { 492 case AP_STRING: 493 fwrite(appends[i].s, sizeof(char), appends[i].len, 494 outfile); 495 break; 496 case AP_FILE: 497 /* 498 * Read files probably shouldn't be cached. Since 499 * it's not an error to read a non-existent file, 500 * it's possible that another program is interacting 501 * with the sed script through the filesystem. It 502 * would be truly bizarre, but possible. It's probably 503 * not that big a performance win, anyhow. 504 */ 505 if ((f = fopen(appends[i].s, "r")) == NULL) 506 break; 507 while ((count = fread(buf, sizeof(char), sizeof(buf), f))) 508 (void)fwrite(buf, sizeof(char), count, outfile); 509 (void)fclose(f); 510 break; 511 } 512 if (ferror(outfile)) 513 errx(1, "%s: %s", outfname, strerror(errno ? errno : EIO)); 514 appendx = sdone = 0; 515 } 516 517 static void 518 lputs(char *s, size_t len) 519 { 520 static const char escapes[] = "\\\a\b\f\r\t\v"; 521 int c, col, width; 522 char *p; 523 struct winsize win; 524 static int termwidth = -1; 525 size_t clen, i; 526 wchar_t wc; 527 mbstate_t mbs; 528 529 if (outfile != stdout) 530 termwidth = 60; 531 if (termwidth == -1) { 532 if ((p = getenv("COLUMNS")) && *p != '\0') 533 termwidth = atoi(p); 534 else if (ioctl(STDOUT_FILENO, TIOCGWINSZ, &win) == 0 && 535 win.ws_col > 0) 536 termwidth = win.ws_col; 537 else 538 termwidth = 60; 539 } 540 541 memset(&mbs, 0, sizeof(mbs)); 542 col = 0; 543 while (len != 0) { 544 clen = mbrtowc(&wc, s, len, &mbs); 545 if (clen == 0) 546 clen = 1; 547 if (clen == (size_t)-1 || clen == (size_t)-2) { 548 wc = (unsigned char)*s; 549 clen = 1; 550 memset(&mbs, 0, sizeof(mbs)); 551 } 552 if (wc == '\n') { 553 if (col + 1 >= termwidth) 554 fprintf(outfile, "\\\n"); 555 fputc('$', outfile); 556 fputc('\n', outfile); 557 col = 0; 558 } else if (iswprint(wc)) { 559 width = wcwidth(wc); 560 if (col + width >= termwidth) { 561 fprintf(outfile, "\\\n"); 562 col = 0; 563 } 564 fwrite(s, 1, clen, outfile); 565 col += width; 566 } else if (wc != L'\0' && (c = wctob(wc)) != EOF && 567 (p = strchr(escapes, c)) != NULL) { 568 if (col + 2 >= termwidth) { 569 fprintf(outfile, "\\\n"); 570 col = 0; 571 } 572 fprintf(outfile, "\\%c", "\\abfrtv"[p - escapes]); 573 col += 2; 574 } else { 575 if (col + 4 * clen >= termwidth) { 576 fprintf(outfile, "\\\n"); 577 col = 0; 578 } 579 for (i = 0; i < clen; i++) 580 fprintf(outfile, "\\%03o", 581 (int)(unsigned char)s[i]); 582 col += 4 * clen; 583 } 584 s += clen; 585 len -= clen; 586 } 587 if (col + 1 >= termwidth) 588 fprintf(outfile, "\\\n"); 589 (void)fputc('$', outfile); 590 (void)fputc('\n', outfile); 591 if (ferror(outfile)) 592 errx(1, "%s: %s", outfname, strerror(errno ? errno : EIO)); 593 } 594 595 static __inline int 596 regexec_e(regex_t *preg, const char *string, int eflags, int nomatch, 597 size_t slen) 598 { 599 int eval; 600 601 if (preg == NULL) { 602 if (defpreg == NULL) 603 errx(1, "first RE may not be empty"); 604 } else 605 defpreg = preg; 606 607 /* Set anchors */ 608 match[0].rm_so = 0; 609 match[0].rm_eo = slen; 610 611 eval = regexec(defpreg, string, 612 nomatch ? 0 : maxnsub + 1, match, eflags | REG_STARTEND); 613 switch(eval) { 614 case 0: 615 return (1); 616 case REG_NOMATCH: 617 return (0); 618 } 619 errx(1, "RE error: %s", strregerror(eval, defpreg)); 620 /* NOTREACHED */ 621 } 622 623 /* 624 * regsub - perform substitutions after a regexp match 625 * Based on a routine by Henry Spencer 626 */ 627 static void 628 regsub(SPACE *sp, char *string, char *src) 629 { 630 int len, no; 631 char c, *dst; 632 633 #define NEEDSP(reqlen) \ 634 /* XXX What is the +1 for? */ \ 635 if (sp->len + (reqlen) + 1 >= sp->blen) { \ 636 sp->blen += (reqlen) + 1024; \ 637 if ((sp->space = sp->back = realloc(sp->back, sp->blen)) \ 638 == NULL) \ 639 err(1, "realloc"); \ 640 dst = sp->space + sp->len; \ 641 } 642 643 dst = sp->space + sp->len; 644 while ((c = *src++) != '\0') { 645 if (c == '&') 646 no = 0; 647 else if (c == '\\' && isdigit((unsigned char)*src)) 648 no = *src++ - '0'; 649 else 650 no = -1; 651 if (no < 0) { /* Ordinary character. */ 652 if (c == '\\' && (*src == '\\' || *src == '&')) 653 c = *src++; 654 NEEDSP(1); 655 *dst++ = c; 656 ++sp->len; 657 } else if (match[no].rm_so != -1 && match[no].rm_eo != -1) { 658 len = match[no].rm_eo - match[no].rm_so; 659 NEEDSP(len); 660 memmove(dst, string + match[no].rm_so, len); 661 dst += len; 662 sp->len += len; 663 } 664 } 665 NEEDSP(1); 666 *dst = '\0'; 667 } 668 669 /* 670 * aspace -- 671 * Append the source space to the destination space, allocating new 672 * space as necessary. 673 */ 674 void 675 cspace(SPACE *sp, const char *p, size_t len, enum e_spflag spflag) 676 { 677 size_t tlen; 678 679 /* Make sure SPACE has enough memory and ramp up quickly. */ 680 tlen = sp->len + len + 1; 681 if (tlen > sp->blen) { 682 sp->blen = tlen + 1024; 683 if ((sp->space = sp->back = realloc(sp->back, sp->blen)) == 684 NULL) 685 err(1, "realloc"); 686 } 687 688 if (spflag == REPLACE) 689 sp->len = 0; 690 691 memmove(sp->space + sp->len, p, len); 692 693 sp->space[sp->len += len] = '\0'; 694 } 695 696 /* 697 * Close all cached opened files and report any errors 698 */ 699 void 700 cfclose(struct s_command *cp, struct s_command *end) 701 { 702 703 for (; cp != end; cp = cp->next) 704 switch(cp->code) { 705 case 's': 706 if (cp->u.s->wfd != -1 && close(cp->u.s->wfd)) 707 err(1, "%s", cp->u.s->wfile); 708 cp->u.s->wfd = -1; 709 break; 710 case 'w': 711 if (cp->u.fd != -1 && close(cp->u.fd)) 712 err(1, "%s", cp->t); 713 cp->u.fd = -1; 714 break; 715 case '{': 716 cfclose(cp->u.c, cp->next); 717 break; 718 } 719 } 720