1 /* $NetBSD: util.c,v 1.9 2011/02/27 17:33:37 joerg Exp $ */ 2 /* $FreeBSD$ */ 3 /* $OpenBSD: util.c,v 1.39 2010/07/02 22:18:03 tedu Exp $ */ 4 5 /*- 6 * Copyright (c) 1999 James Howard and Dag-Erling Coïdan Smørgrav 7 * Copyright (C) 2008-2010 Gabor Kovesdan <gabor@FreeBSD.org> 8 * All rights reserved. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 */ 31 32 #include <sys/cdefs.h> 33 __FBSDID("$FreeBSD$"); 34 35 #include <sys/stat.h> 36 #include <sys/types.h> 37 38 #include <ctype.h> 39 #include <err.h> 40 #include <errno.h> 41 #include <fnmatch.h> 42 #include <fts.h> 43 #include <libgen.h> 44 #include <stdbool.h> 45 #include <stdio.h> 46 #include <stdlib.h> 47 #include <string.h> 48 #include <unistd.h> 49 #include <wchar.h> 50 #include <wctype.h> 51 52 #ifndef WITHOUT_FASTMATCH 53 #include "fastmatch.h" 54 #endif 55 #include "grep.h" 56 57 static bool first_match = true; 58 59 /* 60 * Parsing context; used to hold things like matches made and 61 * other useful bits 62 */ 63 struct parsec { 64 regmatch_t matches[MAX_LINE_MATCHES]; /* Matches made */ 65 struct str ln; /* Current line */ 66 size_t matchidx; /* Latest used match index */ 67 bool binary; /* Binary file? */ 68 }; 69 70 71 static int procline(struct parsec *pc); 72 static void printline(struct parsec *pc, int sep); 73 static void printline_metadata(struct str *line, int sep); 74 75 bool 76 file_matching(const char *fname) 77 { 78 char *fname_base, *fname_buf; 79 bool ret; 80 81 ret = finclude ? false : true; 82 fname_buf = strdup(fname); 83 if (fname_buf == NULL) 84 err(2, "strdup"); 85 fname_base = basename(fname_buf); 86 87 for (unsigned int i = 0; i < fpatterns; ++i) { 88 if (fnmatch(fpattern[i].pat, fname, 0) == 0 || 89 fnmatch(fpattern[i].pat, fname_base, 0) == 0) { 90 if (fpattern[i].mode == EXCL_PAT) { 91 ret = false; 92 break; 93 } else 94 ret = true; 95 } 96 } 97 free(fname_buf); 98 return (ret); 99 } 100 101 static inline bool 102 dir_matching(const char *dname) 103 { 104 bool ret; 105 106 ret = dinclude ? false : true; 107 108 for (unsigned int i = 0; i < dpatterns; ++i) { 109 if (dname != NULL && 110 fnmatch(dpattern[i].pat, dname, 0) == 0) { 111 if (dpattern[i].mode == EXCL_PAT) 112 return (false); 113 else 114 ret = true; 115 } 116 } 117 return (ret); 118 } 119 120 /* 121 * Processes a directory when a recursive search is performed with 122 * the -R option. Each appropriate file is passed to procfile(). 123 */ 124 int 125 grep_tree(char **argv) 126 { 127 FTS *fts; 128 FTSENT *p; 129 int c, fts_flags; 130 bool ok; 131 const char *wd[] = { ".", NULL }; 132 133 c = fts_flags = 0; 134 135 switch(linkbehave) { 136 case LINK_EXPLICIT: 137 fts_flags = FTS_COMFOLLOW; 138 break; 139 case LINK_SKIP: 140 fts_flags = FTS_PHYSICAL; 141 break; 142 default: 143 fts_flags = FTS_LOGICAL; 144 145 } 146 147 fts_flags |= FTS_NOSTAT | FTS_NOCHDIR; 148 149 fts = fts_open((argv[0] == NULL) ? 150 __DECONST(char * const *, wd) : argv, fts_flags, NULL); 151 if (fts == NULL) 152 err(2, "fts_open"); 153 while ((p = fts_read(fts)) != NULL) { 154 switch (p->fts_info) { 155 case FTS_DNR: 156 /* FALLTHROUGH */ 157 case FTS_ERR: 158 file_err = true; 159 if(!sflag) 160 warnx("%s: %s", p->fts_path, strerror(p->fts_errno)); 161 break; 162 case FTS_D: 163 /* FALLTHROUGH */ 164 case FTS_DP: 165 if (dexclude || dinclude) 166 if (!dir_matching(p->fts_name) || 167 !dir_matching(p->fts_path)) 168 fts_set(fts, p, FTS_SKIP); 169 break; 170 case FTS_DC: 171 /* Print a warning for recursive directory loop */ 172 warnx("warning: %s: recursive directory loop", 173 p->fts_path); 174 break; 175 default: 176 /* Check for file exclusion/inclusion */ 177 ok = true; 178 if (fexclude || finclude) 179 ok &= file_matching(p->fts_path); 180 181 if (ok) 182 c += procfile(p->fts_path); 183 break; 184 } 185 } 186 187 fts_close(fts); 188 return (c); 189 } 190 191 /* 192 * Opens a file and processes it. Each file is processed line-by-line 193 * passing the lines to procline(). 194 */ 195 int 196 procfile(const char *fn) 197 { 198 struct parsec pc; 199 struct file *f; 200 struct stat sb; 201 struct str *ln; 202 mode_t s; 203 int c, last_outed, t, tail; 204 bool doctx, same_file; 205 206 doctx = false; 207 if ((!pc.binary || binbehave != BINFILE_BIN) && !cflag && !qflag && 208 !lflag && !Lflag && (Aflag != 0 || Bflag != 0)) 209 doctx = true; 210 mcount = mlimit; 211 212 if (strcmp(fn, "-") == 0) { 213 fn = label != NULL ? label : getstr(1); 214 f = grep_open(NULL); 215 } else { 216 if (!stat(fn, &sb)) { 217 /* Check if we need to process the file */ 218 s = sb.st_mode & S_IFMT; 219 if (s == S_IFDIR && dirbehave == DIR_SKIP) 220 return (0); 221 if ((s == S_IFIFO || s == S_IFCHR || s == S_IFBLK 222 || s == S_IFSOCK) && devbehave == DEV_SKIP) 223 return (0); 224 } 225 f = grep_open(fn); 226 } 227 if (f == NULL) { 228 file_err = true; 229 if (!sflag) 230 warn("%s", fn); 231 return (0); 232 } 233 234 /* Convenience */ 235 ln = &pc.ln; 236 pc.ln.file = grep_malloc(strlen(fn) + 1); 237 strcpy(pc.ln.file, fn); 238 pc.ln.line_no = 0; 239 pc.ln.len = 0; 240 pc.ln.off = -1; 241 pc.binary = f->binary; 242 tail = 0; 243 last_outed = 0; 244 same_file = false; 245 246 for (c = 0; c == 0 || !(lflag || qflag); ) { 247 /* Reset match count for every line processed */ 248 pc.matchidx = 0; 249 pc.ln.off += pc.ln.len + 1; 250 if ((pc.ln.dat = grep_fgetln(f, &pc.ln.len)) == NULL || 251 pc.ln.len == 0) { 252 if (pc.ln.line_no == 0 && matchall) 253 /* 254 * An empty file with an empty pattern and the 255 * -w flag does not match 256 */ 257 exit(matchall && wflag ? 1 : 0); 258 else 259 break; 260 } 261 262 if (pc.ln.len > 0 && pc.ln.dat[pc.ln.len - 1] == fileeol) 263 --pc.ln.len; 264 pc.ln.line_no++; 265 266 /* Return if we need to skip a binary file */ 267 if (pc.binary && binbehave == BINFILE_SKIP) { 268 grep_close(f); 269 free(pc.ln.file); 270 free(f); 271 return (0); 272 } 273 274 if ((t = procline(&pc)) == 0) 275 ++c; 276 277 /* Deal with any -B context or context separators */ 278 if (t == 0 && doctx) { 279 if (!first_match && (!same_file || last_outed > 0)) 280 printf("--\n"); 281 if (Bflag > 0) 282 printqueue(); 283 tail = Aflag; 284 } 285 /* Print the matching line, but only if not quiet/binary */ 286 if (t == 0 && !qflag && !pc.binary) { 287 printline(&pc, ':'); 288 first_match = false; 289 same_file = true; 290 last_outed = 0; 291 } 292 if (t != 0 && doctx) { 293 /* Deal with any -A context */ 294 if (tail > 0) { 295 printline(&pc, '-'); 296 tail--; 297 if (Bflag > 0) 298 clearqueue(); 299 } else { 300 /* 301 * Enqueue non-matching lines for -B context. 302 * If we're not actually doing -B context or if 303 * the enqueue resulted in a line being rotated 304 * out, then go ahead and increment last_outed 305 * to signify a gap between context/match. 306 */ 307 if (Bflag == 0 || (Bflag > 0 && enqueue(ln))) 308 ++last_outed; 309 } 310 } 311 312 /* Count the matches if we have a match limit */ 313 if (t == 0 && mflag) { 314 --mcount; 315 if (mflag && mcount <= 0) 316 break; 317 } 318 319 } 320 if (Bflag > 0) 321 clearqueue(); 322 grep_close(f); 323 324 if (cflag) { 325 if (!hflag) 326 printf("%s:", pc.ln.file); 327 printf("%u\n", c); 328 } 329 if (lflag && !qflag && c != 0) 330 printf("%s%c", fn, nullflag ? 0 : '\n'); 331 if (Lflag && !qflag && c == 0) 332 printf("%s%c", fn, nullflag ? 0 : '\n'); 333 if (c && !cflag && !lflag && !Lflag && 334 binbehave == BINFILE_BIN && f->binary && !qflag) 335 printf(getstr(8), fn); 336 337 free(pc.ln.file); 338 free(f); 339 return (c); 340 } 341 342 #define iswword(x) (iswalnum((x)) || (x) == L'_') 343 344 /* 345 * Processes a line comparing it with the specified patterns. Each pattern 346 * is looped to be compared along with the full string, saving each and every 347 * match, which is necessary to colorize the output and to count the 348 * matches. The matching lines are passed to printline() to display the 349 * appropriate output. 350 */ 351 static int 352 procline(struct parsec *pc) 353 { 354 regmatch_t pmatch, lastmatch, chkmatch; 355 wchar_t wbegin, wend; 356 size_t st = 0, nst = 0; 357 unsigned int i; 358 int c = 0, r = 0, lastmatches = 0, leflags = eflags; 359 size_t startm = 0, matchidx; 360 int retry; 361 362 matchidx = pc->matchidx; 363 364 /* Special case: empty pattern with -w flag, check first character */ 365 if (matchall && wflag) { 366 if (pc->ln.len == 0) 367 return (0); 368 wend = L' '; 369 if (sscanf(&pc->ln.dat[0], "%lc", &wend) != 1 || iswword(wend)) 370 return (1); 371 else 372 return (0); 373 } else if (matchall) 374 return (0); 375 376 /* Initialize to avoid a false positive warning from GCC. */ 377 lastmatch.rm_so = lastmatch.rm_eo = 0; 378 379 /* Loop to process the whole line */ 380 while (st <= pc->ln.len) { 381 lastmatches = 0; 382 startm = matchidx; 383 retry = 0; 384 if (st > 0) 385 leflags |= REG_NOTBOL; 386 /* Loop to compare with all the patterns */ 387 for (i = 0; i < patterns; i++) { 388 pmatch.rm_so = st; 389 pmatch.rm_eo = pc->ln.len; 390 #ifndef WITHOUT_FASTMATCH 391 if (fg_pattern[i].pattern) 392 r = fastexec(&fg_pattern[i], 393 pc->ln.dat, 1, &pmatch, leflags); 394 else 395 #endif 396 r = regexec(&r_pattern[i], pc->ln.dat, 1, 397 &pmatch, leflags); 398 if (r != 0) 399 continue; 400 /* Check for full match */ 401 if (xflag && (pmatch.rm_so != 0 || 402 (size_t)pmatch.rm_eo != pc->ln.len)) 403 continue; 404 /* Check for whole word match */ 405 #ifndef WITHOUT_FASTMATCH 406 if (wflag || fg_pattern[i].word) { 407 #else 408 if (wflag) { 409 #endif 410 wbegin = wend = L' '; 411 if (pmatch.rm_so != 0 && 412 sscanf(&pc->ln.dat[pmatch.rm_so - 1], 413 "%lc", &wbegin) != 1) 414 r = REG_NOMATCH; 415 else if ((size_t)pmatch.rm_eo != 416 pc->ln.len && 417 sscanf(&pc->ln.dat[pmatch.rm_eo], 418 "%lc", &wend) != 1) 419 r = REG_NOMATCH; 420 else if (iswword(wbegin) || 421 iswword(wend)) 422 r = REG_NOMATCH; 423 /* 424 * If we're doing whole word matching and we 425 * matched once, then we should try the pattern 426 * again after advancing just past the start of 427 * the earliest match. This allows the pattern 428 * to match later on in the line and possibly 429 * still match a whole word. 430 */ 431 if (r == REG_NOMATCH && 432 (retry == 0 || pmatch.rm_so + 1 < retry)) 433 retry = pmatch.rm_so + 1; 434 if (r == REG_NOMATCH) 435 continue; 436 } 437 438 lastmatches++; 439 lastmatch = pmatch; 440 441 if (matchidx == 0) 442 c++; 443 444 /* 445 * Replace previous match if the new one is earlier 446 * and/or longer. This will lead to some amount of 447 * extra work if -o/--color are specified, but it's 448 * worth it from a correctness point of view. 449 */ 450 if (matchidx > startm) { 451 chkmatch = pc->matches[matchidx - 1]; 452 if (pmatch.rm_so < chkmatch.rm_so || 453 (pmatch.rm_so == chkmatch.rm_so && 454 (pmatch.rm_eo - pmatch.rm_so) > 455 (chkmatch.rm_eo - chkmatch.rm_so))) { 456 pc->matches[matchidx - 1] = pmatch; 457 nst = pmatch.rm_eo; 458 } 459 } else { 460 /* Advance as normal if not */ 461 pc->matches[matchidx++] = pmatch; 462 nst = pmatch.rm_eo; 463 } 464 /* avoid excessive matching - skip further patterns */ 465 if ((color == NULL && !oflag) || qflag || lflag || 466 matchidx >= MAX_LINE_MATCHES) 467 break; 468 } 469 470 /* 471 * Advance to just past the start of the earliest match, try 472 * again just in case we still have a chance to match later in 473 * the string. 474 */ 475 if (lastmatches == 0 && retry > 0) { 476 st = retry; 477 continue; 478 } 479 480 /* One pass if we are not recording matches */ 481 if (!wflag && ((color == NULL && !oflag) || qflag || lflag || Lflag)) 482 break; 483 484 /* If we didn't have any matches or REG_NOSUB set */ 485 if (lastmatches == 0 || (cflags & REG_NOSUB)) 486 nst = pc->ln.len; 487 488 if (lastmatches == 0) 489 /* No matches */ 490 break; 491 else if (st == nst && lastmatch.rm_so == lastmatch.rm_eo) 492 /* Zero-length match -- advance one more so we don't get stuck */ 493 nst++; 494 495 /* Advance st based on previous matches */ 496 st = nst; 497 } 498 499 /* Reflect the new matchidx in the context */ 500 pc->matchidx = matchidx; 501 if (vflag) 502 c = !c; 503 return (c ? 0 : 1); 504 } 505 506 /* 507 * Safe malloc() for internal use. 508 */ 509 void * 510 grep_malloc(size_t size) 511 { 512 void *ptr; 513 514 if ((ptr = malloc(size)) == NULL) 515 err(2, "malloc"); 516 return (ptr); 517 } 518 519 /* 520 * Safe calloc() for internal use. 521 */ 522 void * 523 grep_calloc(size_t nmemb, size_t size) 524 { 525 void *ptr; 526 527 if ((ptr = calloc(nmemb, size)) == NULL) 528 err(2, "calloc"); 529 return (ptr); 530 } 531 532 /* 533 * Safe realloc() for internal use. 534 */ 535 void * 536 grep_realloc(void *ptr, size_t size) 537 { 538 539 if ((ptr = realloc(ptr, size)) == NULL) 540 err(2, "realloc"); 541 return (ptr); 542 } 543 544 /* 545 * Safe strdup() for internal use. 546 */ 547 char * 548 grep_strdup(const char *str) 549 { 550 char *ret; 551 552 if ((ret = strdup(str)) == NULL) 553 err(2, "strdup"); 554 return (ret); 555 } 556 557 /* 558 * Print an entire line as-is, there are no inline matches to consider. This is 559 * used for printing context. 560 */ 561 void grep_printline(struct str *line, int sep) { 562 printline_metadata(line, sep); 563 fwrite(line->dat, line->len, 1, stdout); 564 putchar(fileeol); 565 } 566 567 static void 568 printline_metadata(struct str *line, int sep) 569 { 570 bool printsep; 571 572 printsep = false; 573 if (!hflag) { 574 if (!nullflag) { 575 fputs(line->file, stdout); 576 printsep = true; 577 } else { 578 printf("%s", line->file); 579 putchar(0); 580 } 581 } 582 if (nflag) { 583 if (printsep) 584 putchar(sep); 585 printf("%d", line->line_no); 586 printsep = true; 587 } 588 if (bflag) { 589 if (printsep) 590 putchar(sep); 591 printf("%lld", (long long)line->off); 592 printsep = true; 593 } 594 if (printsep) 595 putchar(sep); 596 } 597 598 /* 599 * Prints a matching line according to the command line options. 600 */ 601 static void 602 printline(struct parsec *pc, int sep) 603 { 604 size_t a = 0; 605 size_t i, matchidx; 606 regmatch_t match; 607 608 /* If matchall, everything matches but don't actually print for -o */ 609 if (oflag && matchall) 610 return; 611 612 matchidx = pc->matchidx; 613 614 /* --color and -o */ 615 if ((oflag || color) && matchidx > 0) { 616 printline_metadata(&pc->ln, sep); 617 for (i = 0; i < matchidx; i++) { 618 match = pc->matches[i]; 619 /* Don't output zero length matches */ 620 if (match.rm_so == match.rm_eo) 621 continue; 622 if (!oflag) 623 fwrite(pc->ln.dat + a, match.rm_so - a, 1, 624 stdout); 625 if (color) 626 fprintf(stdout, "\33[%sm\33[K", color); 627 fwrite(pc->ln.dat + match.rm_so, 628 match.rm_eo - match.rm_so, 1, stdout); 629 if (color) 630 fprintf(stdout, "\33[m\33[K"); 631 a = match.rm_eo; 632 if (oflag) 633 putchar('\n'); 634 } 635 if (!oflag) { 636 if (pc->ln.len - a > 0) 637 fwrite(pc->ln.dat + a, pc->ln.len - a, 1, 638 stdout); 639 putchar('\n'); 640 } 641 } else 642 grep_printline(&pc->ln, sep); 643 } 644