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, printmatch, same_file; 205 206 if (strcmp(fn, "-") == 0) { 207 fn = label != NULL ? label : getstr(1); 208 f = grep_open(NULL); 209 } else { 210 if (!stat(fn, &sb)) { 211 /* Check if we need to process the file */ 212 s = sb.st_mode & S_IFMT; 213 if (s == S_IFDIR && dirbehave == DIR_SKIP) 214 return (0); 215 if ((s == S_IFIFO || s == S_IFCHR || s == S_IFBLK 216 || s == S_IFSOCK) && devbehave == DEV_SKIP) 217 return (0); 218 } 219 f = grep_open(fn); 220 } 221 if (f == NULL) { 222 file_err = true; 223 if (!sflag) 224 warn("%s", fn); 225 return (0); 226 } 227 228 /* Convenience */ 229 ln = &pc.ln; 230 pc.ln.file = grep_malloc(strlen(fn) + 1); 231 strcpy(pc.ln.file, fn); 232 pc.ln.line_no = 0; 233 pc.ln.len = 0; 234 pc.ln.off = -1; 235 pc.binary = f->binary; 236 tail = 0; 237 last_outed = 0; 238 same_file = false; 239 doctx = false; 240 printmatch = true; 241 if ((pc.binary && binbehave == BINFILE_BIN) || cflag || qflag || 242 lflag || Lflag) 243 printmatch = false; 244 if (printmatch && (Aflag != 0 || Bflag != 0)) 245 doctx = true; 246 mcount = mlimit; 247 248 for (c = 0; c == 0 || !(lflag || qflag); ) { 249 /* Reset match count for every line processed */ 250 pc.matchidx = 0; 251 pc.ln.off += pc.ln.len + 1; 252 if ((pc.ln.dat = grep_fgetln(f, &pc.ln.len)) == NULL || 253 pc.ln.len == 0) { 254 if (pc.ln.line_no == 0 && matchall) 255 /* 256 * An empty file with an empty pattern and the 257 * -w flag does not match 258 */ 259 exit(matchall && wflag ? 1 : 0); 260 else 261 break; 262 } 263 264 if (pc.ln.len > 0 && pc.ln.dat[pc.ln.len - 1] == fileeol) 265 --pc.ln.len; 266 pc.ln.line_no++; 267 268 /* Return if we need to skip a binary file */ 269 if (pc.binary && binbehave == BINFILE_SKIP) { 270 grep_close(f); 271 free(pc.ln.file); 272 free(f); 273 return (0); 274 } 275 276 if ((t = procline(&pc)) == 0) 277 ++c; 278 279 /* Deal with any -B context or context separators */ 280 if (t == 0 && doctx) { 281 if (!first_match && (!same_file || last_outed > 0)) 282 printf("--\n"); 283 if (Bflag > 0) 284 printqueue(); 285 tail = Aflag; 286 } 287 /* Print the matching line, but only if not quiet/binary */ 288 if (t == 0 && printmatch) { 289 printline(&pc, ':'); 290 first_match = false; 291 same_file = true; 292 last_outed = 0; 293 } 294 if (t != 0 && doctx) { 295 /* Deal with any -A context */ 296 if (tail > 0) { 297 printline(&pc, '-'); 298 tail--; 299 if (Bflag > 0) 300 clearqueue(); 301 } else { 302 /* 303 * Enqueue non-matching lines for -B context. 304 * If we're not actually doing -B context or if 305 * the enqueue resulted in a line being rotated 306 * out, then go ahead and increment last_outed 307 * to signify a gap between context/match. 308 */ 309 if (Bflag == 0 || (Bflag > 0 && enqueue(ln))) 310 ++last_outed; 311 } 312 } 313 314 /* Count the matches if we have a match limit */ 315 if (t == 0 && mflag) { 316 --mcount; 317 if (mflag && mcount <= 0) 318 break; 319 } 320 321 } 322 if (Bflag > 0) 323 clearqueue(); 324 grep_close(f); 325 326 if (cflag) { 327 if (!hflag) 328 printf("%s:", pc.ln.file); 329 printf("%u\n", c); 330 } 331 if (lflag && !qflag && c != 0) 332 printf("%s%c", fn, nullflag ? 0 : '\n'); 333 if (Lflag && !qflag && c == 0) 334 printf("%s%c", fn, nullflag ? 0 : '\n'); 335 if (c && !cflag && !lflag && !Lflag && 336 binbehave == BINFILE_BIN && f->binary && !qflag) 337 printf(getstr(8), fn); 338 339 free(pc.ln.file); 340 free(f); 341 return (c); 342 } 343 344 #define iswword(x) (iswalnum((x)) || (x) == L'_') 345 346 /* 347 * Processes a line comparing it with the specified patterns. Each pattern 348 * is looped to be compared along with the full string, saving each and every 349 * match, which is necessary to colorize the output and to count the 350 * matches. The matching lines are passed to printline() to display the 351 * appropriate output. 352 */ 353 static int 354 procline(struct parsec *pc) 355 { 356 regmatch_t pmatch, lastmatch, chkmatch; 357 wchar_t wbegin, wend; 358 size_t st = 0, nst = 0; 359 unsigned int i; 360 int c = 0, r = 0, lastmatches = 0, leflags = eflags; 361 size_t startm = 0, matchidx; 362 int retry; 363 364 matchidx = pc->matchidx; 365 366 /* Special case: empty pattern with -w flag, check first character */ 367 if (matchall && wflag) { 368 if (pc->ln.len == 0) 369 return (0); 370 wend = L' '; 371 if (sscanf(&pc->ln.dat[0], "%lc", &wend) != 1 || iswword(wend)) 372 return (1); 373 else 374 return (0); 375 } else if (matchall) 376 return (0); 377 378 /* Initialize to avoid a false positive warning from GCC. */ 379 lastmatch.rm_so = lastmatch.rm_eo = 0; 380 381 /* Loop to process the whole line */ 382 while (st <= pc->ln.len) { 383 lastmatches = 0; 384 startm = matchidx; 385 retry = 0; 386 if (st > 0) 387 leflags |= REG_NOTBOL; 388 /* Loop to compare with all the patterns */ 389 for (i = 0; i < patterns; i++) { 390 pmatch.rm_so = st; 391 pmatch.rm_eo = pc->ln.len; 392 #ifndef WITHOUT_FASTMATCH 393 if (fg_pattern[i].pattern) 394 r = fastexec(&fg_pattern[i], 395 pc->ln.dat, 1, &pmatch, leflags); 396 else 397 #endif 398 r = regexec(&r_pattern[i], pc->ln.dat, 1, 399 &pmatch, leflags); 400 if (r != 0) 401 continue; 402 /* Check for full match */ 403 if (xflag && (pmatch.rm_so != 0 || 404 (size_t)pmatch.rm_eo != pc->ln.len)) 405 continue; 406 /* Check for whole word match */ 407 #ifndef WITHOUT_FASTMATCH 408 if (wflag || fg_pattern[i].word) { 409 #else 410 if (wflag) { 411 #endif 412 wbegin = wend = L' '; 413 if (pmatch.rm_so != 0 && 414 sscanf(&pc->ln.dat[pmatch.rm_so - 1], 415 "%lc", &wbegin) != 1) 416 r = REG_NOMATCH; 417 else if ((size_t)pmatch.rm_eo != 418 pc->ln.len && 419 sscanf(&pc->ln.dat[pmatch.rm_eo], 420 "%lc", &wend) != 1) 421 r = REG_NOMATCH; 422 else if (iswword(wbegin) || 423 iswword(wend)) 424 r = REG_NOMATCH; 425 /* 426 * If we're doing whole word matching and we 427 * matched once, then we should try the pattern 428 * again after advancing just past the start of 429 * the earliest match. This allows the pattern 430 * to match later on in the line and possibly 431 * still match a whole word. 432 */ 433 if (r == REG_NOMATCH && 434 (retry == 0 || pmatch.rm_so + 1 < retry)) 435 retry = pmatch.rm_so + 1; 436 if (r == REG_NOMATCH) 437 continue; 438 } 439 440 lastmatches++; 441 lastmatch = pmatch; 442 443 if (matchidx == 0) 444 c++; 445 446 /* 447 * Replace previous match if the new one is earlier 448 * and/or longer. This will lead to some amount of 449 * extra work if -o/--color are specified, but it's 450 * worth it from a correctness point of view. 451 */ 452 if (matchidx > startm) { 453 chkmatch = pc->matches[matchidx - 1]; 454 if (pmatch.rm_so < chkmatch.rm_so || 455 (pmatch.rm_so == chkmatch.rm_so && 456 (pmatch.rm_eo - pmatch.rm_so) > 457 (chkmatch.rm_eo - chkmatch.rm_so))) { 458 pc->matches[matchidx - 1] = pmatch; 459 nst = pmatch.rm_eo; 460 } 461 } else { 462 /* Advance as normal if not */ 463 pc->matches[matchidx++] = pmatch; 464 nst = pmatch.rm_eo; 465 } 466 /* avoid excessive matching - skip further patterns */ 467 if ((color == NULL && !oflag) || qflag || lflag || 468 matchidx >= MAX_LINE_MATCHES) 469 break; 470 } 471 472 /* 473 * Advance to just past the start of the earliest match, try 474 * again just in case we still have a chance to match later in 475 * the string. 476 */ 477 if (lastmatches == 0 && retry > 0) { 478 st = retry; 479 continue; 480 } 481 482 /* One pass if we are not recording matches */ 483 if (!wflag && ((color == NULL && !oflag) || qflag || lflag || Lflag)) 484 break; 485 486 /* If we didn't have any matches or REG_NOSUB set */ 487 if (lastmatches == 0 || (cflags & REG_NOSUB)) 488 nst = pc->ln.len; 489 490 if (lastmatches == 0) 491 /* No matches */ 492 break; 493 else if (st == nst && lastmatch.rm_so == lastmatch.rm_eo) 494 /* Zero-length match -- advance one more so we don't get stuck */ 495 nst++; 496 497 /* Advance st based on previous matches */ 498 st = nst; 499 } 500 501 /* Reflect the new matchidx in the context */ 502 pc->matchidx = matchidx; 503 if (vflag) 504 c = !c; 505 return (c ? 0 : 1); 506 } 507 508 /* 509 * Safe malloc() for internal use. 510 */ 511 void * 512 grep_malloc(size_t size) 513 { 514 void *ptr; 515 516 if ((ptr = malloc(size)) == NULL) 517 err(2, "malloc"); 518 return (ptr); 519 } 520 521 /* 522 * Safe calloc() for internal use. 523 */ 524 void * 525 grep_calloc(size_t nmemb, size_t size) 526 { 527 void *ptr; 528 529 if ((ptr = calloc(nmemb, size)) == NULL) 530 err(2, "calloc"); 531 return (ptr); 532 } 533 534 /* 535 * Safe realloc() for internal use. 536 */ 537 void * 538 grep_realloc(void *ptr, size_t size) 539 { 540 541 if ((ptr = realloc(ptr, size)) == NULL) 542 err(2, "realloc"); 543 return (ptr); 544 } 545 546 /* 547 * Safe strdup() for internal use. 548 */ 549 char * 550 grep_strdup(const char *str) 551 { 552 char *ret; 553 554 if ((ret = strdup(str)) == NULL) 555 err(2, "strdup"); 556 return (ret); 557 } 558 559 /* 560 * Print an entire line as-is, there are no inline matches to consider. This is 561 * used for printing context. 562 */ 563 void grep_printline(struct str *line, int sep) { 564 printline_metadata(line, sep); 565 fwrite(line->dat, line->len, 1, stdout); 566 putchar(fileeol); 567 } 568 569 static void 570 printline_metadata(struct str *line, int sep) 571 { 572 bool printsep; 573 574 printsep = false; 575 if (!hflag) { 576 if (!nullflag) { 577 fputs(line->file, stdout); 578 printsep = true; 579 } else { 580 printf("%s", line->file); 581 putchar(0); 582 } 583 } 584 if (nflag) { 585 if (printsep) 586 putchar(sep); 587 printf("%d", line->line_no); 588 printsep = true; 589 } 590 if (bflag) { 591 if (printsep) 592 putchar(sep); 593 printf("%lld", (long long)line->off); 594 printsep = true; 595 } 596 if (printsep) 597 putchar(sep); 598 } 599 600 /* 601 * Prints a matching line according to the command line options. 602 */ 603 static void 604 printline(struct parsec *pc, int sep) 605 { 606 size_t a = 0; 607 size_t i, matchidx; 608 regmatch_t match; 609 610 /* If matchall, everything matches but don't actually print for -o */ 611 if (oflag && matchall) 612 return; 613 614 matchidx = pc->matchidx; 615 616 /* --color and -o */ 617 if ((oflag || color) && matchidx > 0) { 618 printline_metadata(&pc->ln, sep); 619 for (i = 0; i < matchidx; i++) { 620 match = pc->matches[i]; 621 /* Don't output zero length matches */ 622 if (match.rm_so == match.rm_eo) 623 continue; 624 if (!oflag) 625 fwrite(pc->ln.dat + a, match.rm_so - a, 1, 626 stdout); 627 if (color) 628 fprintf(stdout, "\33[%sm\33[K", color); 629 fwrite(pc->ln.dat + match.rm_so, 630 match.rm_eo - match.rm_so, 1, stdout); 631 if (color) 632 fprintf(stdout, "\33[m\33[K"); 633 a = match.rm_eo; 634 if (oflag) 635 putchar('\n'); 636 } 637 if (!oflag) { 638 if (pc->ln.len - a > 0) 639 fwrite(pc->ln.dat + a, pc->ln.len - a, 1, 640 stdout); 641 putchar('\n'); 642 } 643 } else 644 grep_printline(&pc->ln, sep); 645 } 646