1 /* $NetBSD: process.c,v 1.26 1999/11/09 15:06:36 drochner Exp $ */ 2 3 /*- 4 * Copyright (c) 1992 Diomidis Spinellis. 5 * Copyright (c) 1992, 1993, 1994 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. All advertising materials mentioning features or use of this software 20 * must display the following acknowledgement: 21 * This product includes software developed by the University of 22 * California, Berkeley and its contributors. 23 * 4. Neither the name of the University nor the names of its contributors 24 * may be used to endorse or promote products derived from this software 25 * without specific prior written permission. 26 * 27 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 28 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 29 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 30 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 31 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 32 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 33 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 34 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 35 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 36 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 37 * SUCH DAMAGE. 38 */ 39 40 #include <sys/cdefs.h> 41 #ifndef lint 42 #if 0 43 static char sccsid[] = "@(#)process.c 8.6 (Berkeley) 4/20/94"; 44 #else 45 __RCSID("$NetBSD: process.c,v 1.26 1999/11/09 15:06:36 drochner Exp $"); 46 #endif 47 #endif /* not lint */ 48 49 #include <sys/types.h> 50 #include <sys/stat.h> 51 #include <sys/ioctl.h> 52 #include <sys/uio.h> 53 54 #include <ctype.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 <unistd.h> 63 64 #include "defs.h" 65 #include "extern.h" 66 67 static SPACE HS, PS, SS; 68 #define pd PS.deleted 69 #define ps PS.space 70 #define psl PS.len 71 #define hs HS.space 72 #define hsl HS.len 73 74 static inline int applies __P((struct s_command *)); 75 static void flush_appends __P((void)); 76 static void lputs __P((char *)); 77 static inline int regexec_e __P((regex_t *, const char *, int, int, size_t)); 78 static void regsub __P((SPACE *, char *, char *)); 79 static int substitute __P((struct s_command *)); 80 81 struct s_appends *appends; /* Array of pointers to strings to append. */ 82 static int appendx; /* Index into appends array. */ 83 int appendnum; /* Size of appends array. */ 84 85 static int lastaddr; /* Set by applies if last address of a range. */ 86 static int sdone; /* If any substitutes since last line input. */ 87 /* Iov structure for 'w' commands. */ 88 static regex_t *defpreg; 89 size_t maxnsub; 90 regmatch_t *match; 91 92 #define OUT(s) { fwrite(s, sizeof(u_char), psl, stdout); } 93 94 void 95 process() 96 { 97 struct s_command *cp; 98 SPACE tspace; 99 size_t len, oldpsl; 100 char *p; 101 102 oldpsl = 0; 103 for (linenum = 0; mf_fgets(&PS, REPLACE);) { 104 pd = 0; 105 top: 106 cp = prog; 107 redirect: 108 while (cp != NULL) { 109 if (!applies(cp)) { 110 cp = cp->next; 111 continue; 112 } 113 switch (cp->code) { 114 case '{': 115 cp = cp->u.c; 116 goto redirect; 117 case 'a': 118 if (appendx >= appendnum) 119 appends = xrealloc(appends, 120 sizeof(struct s_appends) * 121 (appendnum *= 2)); 122 appends[appendx].type = AP_STRING; 123 appends[appendx].s = cp->t; 124 appends[appendx].len = strlen(cp->t); 125 appendx++; 126 break; 127 case 'b': 128 cp = cp->u.c; 129 goto redirect; 130 case 'c': 131 pd = 1; 132 psl = 0; 133 if (cp->a2 == NULL || lastaddr) 134 (void)printf("%s", cp->t); 135 break; 136 case 'd': 137 pd = 1; 138 goto new; 139 case 'D': 140 if (pd) 141 goto new; 142 if ((p = memchr(ps, '\n', psl - 1)) == NULL) { 143 pd = 1; 144 goto new; 145 } else { 146 psl -= (p + 1) - ps; 147 memmove(ps, p + 1, psl); 148 goto top; 149 } 150 case 'g': 151 cspace(&PS, hs, hsl, REPLACE); 152 break; 153 case 'G': 154 if (hs == NULL) 155 cspace(&HS, "\n", 1, REPLACE); 156 cspace(&PS, hs, hsl, 0); 157 break; 158 case 'h': 159 cspace(&HS, ps, psl, REPLACE); 160 break; 161 case 'H': 162 cspace(&HS, ps, psl, 0); 163 break; 164 case 'i': 165 (void)printf("%s", cp->t); 166 break; 167 case 'l': 168 lputs(ps); 169 break; 170 case 'n': 171 if (!nflag && !pd) 172 OUT(ps) 173 flush_appends(); 174 if (!mf_fgets(&PS, REPLACE)) 175 exit(0); 176 pd = 0; 177 break; 178 case 'N': 179 flush_appends(); 180 if (!mf_fgets(&PS, 0)) { 181 if (!nflag && !pd) 182 OUT(ps) 183 exit(0); 184 } 185 break; 186 case 'p': 187 if (pd) 188 break; 189 OUT(ps) 190 break; 191 case 'P': 192 if (pd) 193 break; 194 if ((p = memchr(ps, '\n', psl - 1)) != NULL) { 195 oldpsl = psl; 196 psl = (p + 1) - ps; 197 } 198 OUT(ps) 199 if (p != NULL) 200 psl = oldpsl; 201 break; 202 case 'q': 203 if (!nflag && !pd) 204 OUT(ps) 205 flush_appends(); 206 exit(0); 207 case 'r': 208 if (appendx >= appendnum) 209 appends = xrealloc(appends, 210 sizeof(struct s_appends) * 211 (appendnum *= 2)); 212 appends[appendx].type = AP_FILE; 213 appends[appendx].s = cp->t; 214 appends[appendx].len = strlen(cp->t); 215 appendx++; 216 break; 217 case 's': 218 sdone |= substitute(cp); 219 break; 220 case 't': 221 if (sdone) { 222 sdone = 0; 223 cp = cp->u.c; 224 goto redirect; 225 } 226 break; 227 case 'w': 228 if (pd) 229 break; 230 if (cp->u.fd == -1 && (cp->u.fd = open(cp->t, 231 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, 232 DEFFILEMODE)) == -1) 233 err(FATAL, "%s: %s\n", 234 cp->t, strerror(errno)); 235 if (write(cp->u.fd, ps, psl) != psl) 236 err(FATAL, "%s: %s\n", 237 cp->t, strerror(errno)); 238 break; 239 case 'x': 240 if (hs == NULL) 241 cspace(&HS, "\n", 1, REPLACE); 242 tspace = PS; 243 PS = HS; 244 HS = tspace; 245 break; 246 case 'y': 247 if (pd) 248 break; 249 for (p = ps, len = psl; --len; ++p) 250 *p = cp->u.y[(int)*p]; 251 break; 252 case ':': 253 case '}': 254 break; 255 case '=': 256 (void)printf("%lu\n", linenum); 257 } 258 cp = cp->next; 259 } /* for all cp */ 260 261 new: if (!nflag && !pd) 262 OUT(ps) 263 flush_appends(); 264 } /* for all lines */ 265 } 266 267 /* 268 * TRUE if the address passed matches the current program state 269 * (lastline, linenumber, ps). 270 */ 271 #define MATCH(a) \ 272 (a)->type == AT_RE ? regexec_e((a)->u.r, ps, 0, 1, psl) : \ 273 (a)->type == AT_LINE ? linenum == (a)->u.l : lastline 274 275 /* 276 * Return TRUE if the command applies to the current line. Sets the inrange 277 * flag to process ranges. Interprets the non-select (``!'') flag. 278 */ 279 static inline int 280 applies(cp) 281 struct s_command *cp; 282 { 283 int r; 284 285 lastaddr = 0; 286 if (cp->a1 == NULL && cp->a2 == NULL) 287 r = 1; 288 else if (cp->a2) { 289 if (cp->inrange) { 290 if (MATCH(cp->a2)) { 291 cp->inrange = 0; 292 lastaddr = 1; 293 } 294 r = 1; 295 } else if (MATCH(cp->a1)) { 296 /* 297 * If the second address is a number less than or 298 * equal to the line number first selected, only 299 * one line shall be selected. 300 * -- POSIX 1003.2 301 */ 302 if (cp->a2->type == AT_LINE && 303 linenum >= cp->a2->u.l) 304 lastaddr = 1; 305 else 306 cp->inrange = 1; 307 r = 1; 308 } else 309 r = 0; 310 } else 311 r = MATCH(cp->a1); 312 return (cp->nonsel ? ! r : r); 313 } 314 315 /* 316 * substitute -- 317 * Do substitutions in the pattern space. Currently, we build a 318 * copy of the new pattern space in the substitute space structure 319 * and then swap them. 320 */ 321 static int 322 substitute(cp) 323 struct s_command *cp; 324 { 325 SPACE tspace; 326 regex_t *re; 327 size_t re_off, slen; 328 int lastempty, n; 329 char *s; 330 331 s = ps; 332 re = cp->u.s->re; 333 if (re == NULL) { 334 if (defpreg != NULL && cp->u.s->maxbref > defpreg->re_nsub) { 335 linenum = cp->u.s->linenum; 336 err(COMPILE, "\\%d not defined in the RE", 337 cp->u.s->maxbref); 338 } 339 } 340 if (!regexec_e(re, s, 0, 0, psl)) 341 return (0); 342 343 SS.len = 0; /* Clean substitute space. */ 344 slen = psl; 345 n = cp->u.s->n; 346 lastempty = 1; 347 348 switch (n) { 349 case 0: /* Global */ 350 do { 351 if (lastempty || match[0].rm_so != match[0].rm_eo) { 352 /* Locate start of replaced string. */ 353 re_off = match[0].rm_so; 354 /* Copy leading retained string. */ 355 cspace(&SS, s, re_off, APPEND); 356 /* Add in regular expression. */ 357 regsub(&SS, s, cp->u.s->new); 358 } 359 360 /* Move past this match. */ 361 if (match[0].rm_so != match[0].rm_eo) { 362 s += match[0].rm_eo; 363 slen -= match[0].rm_eo; 364 lastempty = 0; 365 } else { 366 if (match[0].rm_so == 0) 367 cspace(&SS, 368 s, match[0].rm_so + 1, APPEND); 369 else 370 cspace(&SS, 371 s + match[0].rm_so, 1, APPEND); 372 s += match[0].rm_so + 1; 373 slen -= match[0].rm_so + 1; 374 lastempty = 1; 375 } 376 } while (slen > 0 && regexec_e(re, s, REG_NOTBOL, 0, slen)); 377 /* Copy trailing retained string. */ 378 if (slen > 0) 379 cspace(&SS, s, slen, APPEND); 380 break; 381 default: /* Nth occurrence */ 382 while (--n) { 383 s += match[0].rm_eo; 384 slen -= match[0].rm_eo; 385 if (!regexec_e(re, s, REG_NOTBOL, 0, slen)) 386 return (0); 387 } 388 /* FALLTHROUGH */ 389 case 1: /* 1st occurrence */ 390 /* Locate start of replaced string. */ 391 re_off = match[0].rm_so + (s - ps); 392 /* Copy leading retained string. */ 393 cspace(&SS, ps, re_off, APPEND); 394 /* Add in regular expression. */ 395 regsub(&SS, s, cp->u.s->new); 396 /* Copy trailing retained string. */ 397 s += match[0].rm_eo; 398 slen -= match[0].rm_eo; 399 cspace(&SS, s, slen, APPEND); 400 break; 401 } 402 403 /* 404 * Swap the substitute space and the pattern space, and make sure 405 * that any leftover pointers into stdio memory get lost. 406 */ 407 tspace = PS; 408 PS = SS; 409 SS = tspace; 410 SS.space = SS.back; 411 412 /* Handle the 'p' flag. */ 413 if (cp->u.s->p) 414 OUT(ps) 415 416 /* Handle the 'w' flag. */ 417 if (cp->u.s->wfile && !pd) { 418 if (cp->u.s->wfd == -1 && (cp->u.s->wfd = open(cp->u.s->wfile, 419 O_WRONLY|O_APPEND|O_CREAT|O_TRUNC, DEFFILEMODE)) == -1) 420 err(FATAL, "%s: %s\n", cp->u.s->wfile, strerror(errno)); 421 if (write(cp->u.s->wfd, ps, psl) != psl) 422 err(FATAL, "%s: %s\n", cp->u.s->wfile, strerror(errno)); 423 } 424 return (1); 425 } 426 427 /* 428 * Flush append requests. Always called before reading a line, 429 * therefore it also resets the substitution done (sdone) flag. 430 */ 431 static void 432 flush_appends() 433 { 434 FILE *f; 435 int count, i; 436 char buf[8 * 1024]; 437 438 for (i = 0; i < appendx; i++) 439 switch (appends[i].type) { 440 case AP_STRING: 441 fwrite(appends[i].s, sizeof(char), appends[i].len, 442 stdout); 443 break; 444 case AP_FILE: 445 /* 446 * Read files probably shouldn't be cached. Since 447 * it's not an error to read a non-existent file, 448 * it's possible that another program is interacting 449 * with the sed script through the file system. It 450 * would be truly bizarre, but possible. It's probably 451 * not that big a performance win, anyhow. 452 */ 453 if ((f = fopen(appends[i].s, "r")) == NULL) 454 break; 455 while ((count = 456 fread(buf, sizeof(char), sizeof(buf), f)) > 0) 457 (void)fwrite(buf, sizeof(char), count, stdout); 458 (void)fclose(f); 459 break; 460 } 461 if (ferror(stdout)) 462 err(FATAL, "stdout: %s", strerror(errno ? errno : EIO)); 463 appendx = sdone = 0; 464 } 465 466 static void 467 lputs(s) 468 char *s; 469 { 470 int count; 471 char *escapes, *p; 472 struct winsize win; 473 static int termwidth = -1; 474 475 if (termwidth == -1) { 476 if ((p = getenv("COLUMNS")) != NULL) 477 termwidth = atoi(p); 478 else if (ioctl(STDOUT_FILENO, TIOCGWINSZ, &win) == 0 && 479 win.ws_col > 0) 480 termwidth = win.ws_col; 481 else 482 termwidth = 60; 483 } 484 for (count = 0; *s; ++s) { 485 if (count >= termwidth) { 486 (void)printf("\\\n"); 487 count = 0; 488 } 489 if (isascii((unsigned char)*s) && isprint((unsigned char)*s) && 490 *s != '\\') { 491 (void)putchar(*s); 492 count++; 493 } else { 494 escapes = "\\\a\b\f\n\r\t\v"; 495 (void)putchar('\\'); 496 if ((p = strchr(escapes, *s)) != NULL) { 497 (void)putchar("\\abfnrtv"[p - escapes]); 498 count += 2; 499 } else { 500 (void)printf("%03o", *(u_char *)s); 501 count += 4; 502 } 503 } 504 } 505 (void)putchar('$'); 506 (void)putchar('\n'); 507 if (ferror(stdout)) 508 err(FATAL, "stdout: %s", strerror(errno ? errno : EIO)); 509 } 510 511 static inline int 512 regexec_e(preg, string, eflags, nomatch, slen) 513 regex_t *preg; 514 const char *string; 515 int eflags, nomatch; 516 size_t slen; 517 { 518 int eval; 519 #ifndef REG_STARTEND 520 char *buf; 521 #endif 522 523 if (preg == NULL) { 524 if (defpreg == NULL) 525 err(FATAL, "first RE may not be empty"); 526 } else 527 defpreg = preg; 528 529 /* Set anchors, discounting trailing newline (if any). */ 530 if (slen > 0 && string[slen - 1] == '\n') 531 slen--; 532 533 #ifndef REG_STARTEND 534 if ((buf = malloc(slen + 1)) == NULL) 535 err(1, NULL); 536 (void)memcpy(buf, string, slen); 537 buf[slen] = '\0'; 538 eval = regexec(defpreg, buf, 539 nomatch ? 0 : maxnsub + 1, match, eflags); 540 free(buf); 541 #else 542 match[0].rm_so = 0; 543 match[0].rm_eo = slen; 544 eval = regexec(defpreg, string, 545 nomatch ? 0 : maxnsub + 1, match, eflags | REG_STARTEND); 546 #endif 547 switch(eval) { 548 case 0: 549 return (1); 550 case REG_NOMATCH: 551 return (0); 552 } 553 err(FATAL, "RE error: %s", strregerror(eval, defpreg)); 554 /* NOTREACHED */ 555 return (0); 556 } 557 558 /* 559 * regsub - perform substitutions after a regexp match 560 * Based on a routine by Henry Spencer 561 */ 562 static void 563 regsub(sp, string, src) 564 SPACE *sp; 565 char *string, *src; 566 { 567 int len, no; 568 char c, *dst; 569 570 #define NEEDSP(reqlen) \ 571 if (sp->len >= sp->blen - (reqlen) - 1) { \ 572 sp->blen += (reqlen) + 1024; \ 573 sp->space = sp->back = xrealloc(sp->back, sp->blen); \ 574 dst = sp->space + sp->len; \ 575 } 576 577 dst = sp->space + sp->len; 578 while ((c = *src++) != '\0') { 579 if (c == '&') 580 no = 0; 581 else if (c == '\\' && isdigit((unsigned char)*src)) 582 no = *src++ - '0'; 583 else 584 no = -1; 585 if (no < 0) { /* Ordinary character. */ 586 if (c == '\\' && (*src == '\\' || *src == '&')) 587 c = *src++; 588 NEEDSP(1); 589 *dst++ = c; 590 ++sp->len; 591 } else if (match[no].rm_so != -1 && match[no].rm_eo != -1) { 592 len = match[no].rm_eo - match[no].rm_so; 593 NEEDSP(len); 594 memmove(dst, string + match[no].rm_so, len); 595 dst += len; 596 sp->len += len; 597 } 598 } 599 NEEDSP(1); 600 *dst = '\0'; 601 } 602 603 /* 604 * aspace -- 605 * Append the source space to the destination space, allocating new 606 * space as necessary. 607 */ 608 void 609 cspace(sp, p, len, spflag) 610 SPACE *sp; 611 char *p; 612 size_t len; 613 enum e_spflag spflag; 614 { 615 size_t tlen; 616 617 /* Make sure SPACE has enough memory and ramp up quickly. */ 618 tlen = sp->len + len + 1; 619 if (tlen > sp->blen) { 620 sp->blen = tlen + 1024; 621 sp->space = sp->back = xrealloc(sp->back, sp->blen); 622 } 623 624 if (spflag == REPLACE) 625 sp->len = 0; 626 627 memmove(sp->space + sp->len, p, len); 628 629 sp->space[sp->len += len] = '\0'; 630 } 631 632 /* 633 * Close all cached opened files and report any errors 634 */ 635 void 636 cfclose(cp, end) 637 struct s_command *cp, *end; 638 { 639 640 for (; cp != end; cp = cp->next) 641 switch(cp->code) { 642 case 's': 643 if (cp->u.s->wfd != -1 && close(cp->u.s->wfd)) 644 err(FATAL, 645 "%s: %s", cp->u.s->wfile, strerror(errno)); 646 cp->u.s->wfd = -1; 647 break; 648 case 'w': 649 if (cp->u.fd != -1 && close(cp->u.fd)) 650 err(FATAL, "%s: %s", cp->t, strerror(errno)); 651 cp->u.fd = -1; 652 break; 653 case '{': 654 cfclose(cp->u.c, cp->next); 655 break; 656 } 657 } 658