findutils/grep.c

   1 /* vi: set sw=4 ts=4: */
   2 /*
   3  * Mini grep implementation for busybox using libc regex.
   4  *
   5  * Copyright (C) 1999,2000,2001 by Lineo, inc. and Mark Whitley
   6  * Copyright (C) 1999,2000,2001 by Mark Whitley <markw@codepoet.org>
   7  *
   8  * Licensed under the GPL v2 or later, see the file LICENSE in this tarball.
   9  */
  10 /* BB_AUDIT SUSv3 defects - unsupported option -x.  */
  11 /* BB_AUDIT GNU defects - always acts as -a.  */
  12 /* http://www.opengroup.org/onlinepubs/007904975/utilities/grep.html */
  13 /*
  14  * 2004,2006 (C) Vladimir Oleynik <dzo@simtreas.ru> -
  15  * correction "-e pattern1 -e pattern2" logic and more optimizations.
  16  * precompiled regex
  17  */
  18 /*
  19  * (C) 2006 Jac Goudsmit added -o option
  20  */
  21
  22 #include "libbb.h"
  23 #include "xregex.h"
  24
  25 /* options */
  26 #define OPTSTR_GREP \
  27         "lnqvscFiHhe:f:Lorm:" \
  28         USE_FEATURE_GREP_CONTEXT("A:B:C:") \
  29         USE_FEATURE_GREP_EGREP_ALIAS("E") \
  30         USE_DESKTOP("w") \
  31         "aI"
  32 /* ignored: -a "assume all files to be text" */
  33 /* ignored: -I "assume binary files have no matches" */
  34
  35 enum {
  36         OPTBIT_l, /* list matched file names only */
  37         OPTBIT_n, /* print line# */
  38         OPTBIT_q, /* quiet - exit(0) of first match */
  39         OPTBIT_v, /* invert the match, to select non-matching lines */
  40         OPTBIT_s, /* suppress errors about file open errors */
  41         OPTBIT_c, /* count matches per file (suppresses normal output) */
  42         OPTBIT_F, /* literal match */
  43         OPTBIT_i, /* case-insensitive */
  44         OPTBIT_H, /* force filename display */
  45         OPTBIT_h, /* inhibit filename display */
  46         OPTBIT_e, /* -e PATTERN */
  47         OPTBIT_f, /* -f FILE_WITH_PATTERNS */
  48         OPTBIT_L, /* list unmatched file names only */
  49         OPTBIT_o, /* show only matching parts of lines */
  50         OPTBIT_r, /* recurse dirs */
  51         OPTBIT_m, /* -m MAX_MATCHES */
  52         USE_FEATURE_GREP_CONTEXT(    OPTBIT_A ,) /* -A NUM: after-match context */
  53         USE_FEATURE_GREP_CONTEXT(    OPTBIT_B ,) /* -B NUM: before-match context */
  54         USE_FEATURE_GREP_CONTEXT(    OPTBIT_C ,) /* -C NUM: -A and -B combined */
  55         USE_FEATURE_GREP_EGREP_ALIAS(OPTBIT_E ,) /* extended regexp */
  56         USE_DESKTOP(                 OPTBIT_w ,) /* whole word match */
  57         OPT_l = 1 << OPTBIT_l,
  58         OPT_n = 1 << OPTBIT_n,
  59         OPT_q = 1 << OPTBIT_q,
  60         OPT_v = 1 << OPTBIT_v,
  61         OPT_s = 1 << OPTBIT_s,
  62         OPT_c = 1 << OPTBIT_c,
  63         OPT_F = 1 << OPTBIT_F,
  64         OPT_i = 1 << OPTBIT_i,
  65         OPT_H = 1 << OPTBIT_H,
  66         OPT_h = 1 << OPTBIT_h,
  67         OPT_e = 1 << OPTBIT_e,
  68         OPT_f = 1 << OPTBIT_f,
  69         OPT_L = 1 << OPTBIT_L,
  70         OPT_o = 1 << OPTBIT_o,
  71         OPT_r = 1 << OPTBIT_r,
  72         OPT_m = 1 << OPTBIT_m,
  73         OPT_A = USE_FEATURE_GREP_CONTEXT(    (1 << OPTBIT_A)) + 0,
  74         OPT_B = USE_FEATURE_GREP_CONTEXT(    (1 << OPTBIT_B)) + 0,
  75         OPT_C = USE_FEATURE_GREP_CONTEXT(    (1 << OPTBIT_C)) + 0,
  76         OPT_E = USE_FEATURE_GREP_EGREP_ALIAS((1 << OPTBIT_E)) + 0,
  77         OPT_w = USE_DESKTOP(                 (1 << OPTBIT_w)) + 0,
  78 };
  79
  80 #define PRINT_FILES_WITH_MATCHES    (option_mask32 & OPT_l)
  81 #define PRINT_LINE_NUM              (option_mask32 & OPT_n)
  82 #define BE_QUIET                    (option_mask32 & OPT_q)
  83 #define SUPPRESS_ERR_MSGS           (option_mask32 & OPT_s)
  84 #define PRINT_MATCH_COUNTS          (option_mask32 & OPT_c)
  85 #define FGREP_FLAG                  (option_mask32 & OPT_F)
  86 #define PRINT_FILES_WITHOUT_MATCHES (option_mask32 & OPT_L)
  87
  88 struct globals {
  89         int max_matches;
  90         int reflags;
  91         smalluint invert_search;
  92         smalluint print_filename;
  93         smalluint open_errors;
  94 #if ENABLE_FEATURE_GREP_CONTEXT
  95         smalluint did_print_line;
  96         int lines_before;
  97         int lines_after;
  98         char **before_buf;
  99         int last_line_printed;
 100 #endif
 101         /* globals used internally */
 102         llist_t *pattern_head;   /* growable list of patterns to match */
 103         const char *cur_file;    /* the current file we are reading */
 104 };
 105 #define G (*(struct globals*)&bb_common_bufsiz1)
 106 #define INIT_G() \
 107         do { \
 108                 struct G_sizecheck { \
 109                         char G_sizecheck[sizeof(G) > COMMON_BUFSIZE ? -1 : 1]; \
 110                 }; \
 111         } while (0)
 112 #define max_matches       (G.max_matches         )
 113 #define reflags           (G.reflags             )
 114 #define invert_search     (G.invert_search       )
 115 #define print_filename    (G.print_filename      )
 116 #define open_errors       (G.open_errors         )
 117 #define did_print_line    (G.did_print_line      )
 118 #define lines_before      (G.lines_before        )
 119 #define lines_after       (G.lines_after         )
 120 #define before_buf        (G.before_buf          )
 121 #define last_line_printed (G.last_line_printed   )
 122 #define pattern_head      (G.pattern_head        )
 123 #define cur_file          (G.cur_file            )
 124
 125
 126 typedef struct grep_list_data_t {
 127         char *pattern;
 128         regex_t preg;
 129 #define PATTERN_MEM_A 1
 130 #define COMPILED 2
 131         int flg_mem_alocated_compiled;
 132 } grep_list_data_t;
 133
 134
 135 static void print_line(const char *line, int linenum, char decoration)
 136 {
 137 #if ENABLE_FEATURE_GREP_CONTEXT
 138         /* Happens when we go to next file, immediately hit match
 139          * and try to print prev context... from prev file! Don't do it */
 140         if (linenum < 1)
 141                 return;
 142         /* possibly print the little '--' separator */
 143         if ((lines_before || lines_after) && did_print_line &&
 144                         last_line_printed != linenum - 1) {
 145                 puts("--");
 146         }
 147         /* guard against printing "--" before first line of first file */
 148         did_print_line = 1;
 149         last_line_printed = linenum;
 150 #endif
 151         if (print_filename)
 152                 printf("%s%c", cur_file, decoration);
 153         if (PRINT_LINE_NUM)
 154                 printf("%i%c", linenum, decoration);
 155         /* Emulate weird GNU grep behavior with -ov */
 156         if ((option_mask32 & (OPT_v|OPT_o)) != (OPT_v|OPT_o))
 157                 puts(line);
 158 }
 159
 160 static int grep_file(FILE *file)
 161 {
 162         char *line;
 163         smalluint found;
 164         int linenum = 0;
 165         int nmatches = 0;
 166         regmatch_t regmatch;
 167 #if ENABLE_FEATURE_GREP_CONTEXT
 168         int print_n_lines_after = 0;
 169         int curpos = 0; /* track where we are in the circular 'before' buffer */
 170         int idx = 0; /* used for iteration through the circular buffer */
 171 #else
 172         enum { print_n_lines_after = 0 };
 173 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
 174
 175         while ((line = xmalloc_getline(file)) != NULL) {
 176                 llist_t *pattern_ptr = pattern_head;
 177                 grep_list_data_t *gl = gl; /* for gcc */
 178
 179                 linenum++;
 180                 found = 0;
 181                 while (pattern_ptr) {
 182                         gl = (grep_list_data_t *)pattern_ptr->data;
 183                         if (FGREP_FLAG) {
 184                                 found |= (strstr(line, gl->pattern) != NULL);
 185                         } else {
 186                                 if (!(gl->flg_mem_alocated_compiled & COMPILED)) {
 187                                         gl->flg_mem_alocated_compiled |= COMPILED;
 188                                         xregcomp(&(gl->preg), gl->pattern, reflags);
 189                                 }
 190                                 regmatch.rm_so = 0;
 191                                 regmatch.rm_eo = 0;
 192                                 if (regexec(&(gl->preg), line, 1, &regmatch, 0) == 0) {
 193                                         if (!(option_mask32 & OPT_w))
 194                                                 found = 1;
 195                                         else {
 196                                                 char c = ' ';
 197                                                 if (regmatch.rm_so)
 198                                                         c = line[regmatch.rm_so - 1];
 199                                                 if (!isalnum(c) && c != '_') {
 200                                                         c = line[regmatch.rm_eo];
 201                                                         if (!c || (!isalnum(c) && c != '_'))
 202                                                                 found = 1;
 203                                                 }
 204                                         }
 205                                 }
 206                         }
 207                         /* If it's non-inverted search, we can stop
 208                          * at first match */
 209                         if (found && !invert_search)
 210                                 goto do_found;
 211                         pattern_ptr = pattern_ptr->link;
 212                 } /* while (pattern_ptr) */
 213
 214                 if (found ^ invert_search) {
 215  do_found:
 216                         /* keep track of matches */
 217                         nmatches++;
 218
 219                         /* quiet/print (non)matching file names only? */
 220                         if (option_mask32 & (OPT_q|OPT_l|OPT_L)) {
 221                                 free(line); /* we don't need line anymore */
 222                                 if (BE_QUIET) {
 223                                         /* manpage says about -q:
 224                                          * "exit immediately with zero status
 225                                          * if any match is found,
 226                                          * even if errors were detected" */
 227                                         exit(0);
 228                                 }
 229                                 /* if we're just printing filenames, we stop after the first match */
 230                                 if (PRINT_FILES_WITH_MATCHES) {
 231                                         puts(cur_file);
 232                                         /* fall through to "return 1" */
 233                                 }
 234                                 /* OPT_L aka PRINT_FILES_WITHOUT_MATCHES: return early */
 235                                 return 1; /* one match */
 236                         }
 237
 238 #if ENABLE_FEATURE_GREP_CONTEXT
 239                         /* Were we printing context and saw next (unwanted) match? */
 240                         if ((option_mask32 & OPT_m) && nmatches > max_matches)
 241                                 break;
 242 #endif
 243
 244                         /* print the matched line */
 245                         if (PRINT_MATCH_COUNTS == 0) {
 246 #if ENABLE_FEATURE_GREP_CONTEXT
 247                                 int prevpos = (curpos == 0) ? lines_before - 1 : curpos - 1;
 248
 249                                 /* if we were told to print 'before' lines and there is at least
 250                                  * one line in the circular buffer, print them */
 251                                 if (lines_before && before_buf[prevpos] != NULL) {
 252                                         int first_buf_entry_line_num = linenum - lines_before;
 253
 254                                         /* advance to the first entry in the circular buffer, and
 255                                          * figure out the line number is of the first line in the
 256                                          * buffer */
 257                                         idx = curpos;
 258                                         while (before_buf[idx] == NULL) {
 259                                                 idx = (idx + 1) % lines_before;
 260                                                 first_buf_entry_line_num++;
 261                                         }
 262
 263                                         /* now print each line in the buffer, clearing them as we go */
 264                                         while (before_buf[idx] != NULL) {
 265                                                 print_line(before_buf[idx], first_buf_entry_line_num, '-');
 266                                                 free(before_buf[idx]);
 267                                                 before_buf[idx] = NULL;
 268                                                 idx = (idx + 1) % lines_before;
 269                                                 first_buf_entry_line_num++;
 270                                         }
 271                                 }
 272
 273                                 /* make a note that we need to print 'after' lines */
 274                                 print_n_lines_after = lines_after;
 275 #endif
 276                                 if (option_mask32 & OPT_o) {
 277                                         if (FGREP_FLAG) {
 278                                                 /* -Fo just prints the pattern
 279                                                  * (unless -v: -Fov doesnt print anything at all) */
 280                                                 if (found)
 281                                                         print_line(gl->pattern, linenum, ':');
 282                                         } else {
 283                                                 line[regmatch.rm_eo] = '\0';
 284                                                 print_line(line + regmatch.rm_so, linenum, ':');
 285                                         }
 286                                 } else {
 287                                         print_line(line, linenum, ':');
 288                                 }
 289                         }
 290                 }
 291 #if ENABLE_FEATURE_GREP_CONTEXT
 292                 else { /* no match */
 293                         /* if we need to print some context lines after the last match, do so */
 294                         if (print_n_lines_after) {
 295                                 print_line(line, linenum, '-');
 296                                 print_n_lines_after--;
 297                         } else if (lines_before) {
 298                                 /* Add the line to the circular 'before' buffer */
 299                                 free(before_buf[curpos]);
 300                                 before_buf[curpos] = line;
 301                                 curpos = (curpos + 1) % lines_before;
 302                                 /* avoid free(line) - we took line */
 303                                 line = NULL;
 304                         }
 305                 }
 306
 307 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
 308                 free(line);
 309
 310                 /* Did we print all context after last requested match? */
 311                 if ((option_mask32 & OPT_m)
 312                  && !print_n_lines_after && nmatches == max_matches)
 313                         break;
 314         }
 315
 316         /* special-case file post-processing for options where we don't print line
 317          * matches, just filenames and possibly match counts */
 318
 319         /* grep -c: print [filename:]count, even if count is zero */
 320         if (PRINT_MATCH_COUNTS) {
 321                 if (print_filename)
 322                         printf("%s:", cur_file);
 323                 printf("%d\n", nmatches);
 324         }
 325
 326         /* grep -L: print just the filename */
 327         if (PRINT_FILES_WITHOUT_MATCHES) {
 328                 /* nmatches is zero, no need to check it:
 329                  * we return 1 early if we detected a match
 330                  * and PRINT_FILES_WITHOUT_MATCHES is set */
 331                 puts(cur_file);
 332         }
 333
 334         return nmatches;
 335 }
 336
 337 #if ENABLE_FEATURE_CLEAN_UP
 338 #define new_grep_list_data(p, m) add_grep_list_data(p, m)
 339 static char *add_grep_list_data(char *pattern, int flg_used_mem)
 340 #else
 341 #define new_grep_list_data(p, m) add_grep_list_data(p)
 342 static char *add_grep_list_data(char *pattern)
 343 #endif
 344 {
 345         grep_list_data_t *gl = xzalloc(sizeof(*gl));
 346         gl->pattern = pattern;
 347 #if ENABLE_FEATURE_CLEAN_UP
 348         gl->flg_mem_alocated_compiled = flg_used_mem;
 349 #else
 350         /*gl->flg_mem_alocated_compiled = 0;*/
 351 #endif
 352         return (char *)gl;
 353 }
 354
 355 static void load_regexes_from_file(llist_t *fopt)
 356 {
 357         char *line;
 358         FILE *f;
 359
 360         while (fopt) {
 361                 llist_t *cur = fopt;
 362                 char *ffile = cur->data;
 363
 364                 fopt = cur->link;
 365                 free(cur);
 366                 f = xfopen(ffile, "r");
 367                 while ((line = xmalloc_getline(f)) != NULL) {
 368                         llist_add_to(&pattern_head,
 369                                 new_grep_list_data(line, PATTERN_MEM_A));
 370                 }
 371         }
 372 }
 373
 374 static int file_action_grep(const char *filename, struct stat *statbuf, void* matched, int depth)
 375 {
 376         FILE *file = fopen(filename, "r");
 377         if (file == NULL) {
 378                 if (!SUPPRESS_ERR_MSGS)
 379                         bb_simple_perror_msg(filename);
 380                 open_errors = 1;
 381                 return 0;
 382         }
 383         cur_file = filename;
 384         *(int*)matched += grep_file(file);
 385         fclose(file);
 386         return 1;
 387 }
 388
 389 static int grep_dir(const char *dir)
 390 {
 391         int matched = 0;
 392         recursive_action(dir,
 393                 /* recurse=yes */ ACTION_RECURSE |
 394                 /* followLinks=no */
 395                 /* depthFirst=yes */ ACTION_DEPTHFIRST,
 396                 /* fileAction= */ file_action_grep,
 397                 /* dirAction= */ NULL,
 398                 /* userData= */ &matched,
 399                 /* depth= */ 0);
 400         return matched;
 401 }
 402
 403 int grep_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 404 int grep_main(int argc, char **argv)
 405 {
 406         FILE *file;
 407         int matched;
 408         char *mopt;
 409         llist_t *fopt = NULL;
 410
 411         /* do normal option parsing */
 412 #if ENABLE_FEATURE_GREP_CONTEXT
 413         char *slines_after;
 414         char *slines_before;
 415         char *Copt;
 416
 417         opt_complementary = "H-h:e::f::C-AB";
 418         getopt32(argv,
 419                 OPTSTR_GREP,
 420                 &pattern_head, &fopt, &mopt,
 421                 &slines_after, &slines_before, &Copt);
 422
 423         if (option_mask32 & OPT_C) {
 424                 /* -C unsets prev -A and -B, but following -A or -B
 425                    may override it */
 426                 if (!(option_mask32 & OPT_A)) /* not overridden */
 427                         slines_after = Copt;
 428                 if (!(option_mask32 & OPT_B)) /* not overridden */
 429                         slines_before = Copt;
 430                 option_mask32 |= OPT_A|OPT_B; /* for parser */
 431         }
 432         if (option_mask32 & OPT_A) {
 433                 lines_after = xatoi_u(slines_after);
 434         }
 435         if (option_mask32 & OPT_B) {
 436                 lines_before = xatoi_u(slines_before);
 437         }
 438         /* sanity checks */
 439         if (option_mask32 & (OPT_c|OPT_q|OPT_l|OPT_L)) {
 440                 option_mask32 &= ~OPT_n;
 441                 lines_before = 0;
 442                 lines_after = 0;
 443         } else if (lines_before > 0)
 444                 before_buf = xzalloc(lines_before * sizeof(char *));
 445 #else
 446         /* with auto sanity checks */
 447         opt_complementary = "H-h:e::f::c-n:q-n:l-n";
 448         getopt32(argv, OPTSTR_GREP,
 449                 &pattern_head, &fopt, &mopt);
 450 #endif
 451         if (option_mask32 & OPT_m) {
 452                 max_matches = xatoi_u(mopt);
 453         }
 454         invert_search = ((option_mask32 & OPT_v) != 0); /* 0 | 1 */
 455
 456         if (pattern_head != NULL) {
 457                 /* convert char **argv to grep_list_data_t */
 458                 llist_t *cur;
 459
 460                 for (cur = pattern_head; cur; cur = cur->link)
 461                         cur->data = new_grep_list_data(cur->data, 0);
 462         }
 463         if (option_mask32 & OPT_f)
 464                 load_regexes_from_file(fopt);
 465
 466         if (ENABLE_FEATURE_GREP_FGREP_ALIAS && applet_name[0] == 'f')
 467                 option_mask32 |= OPT_F;
 468
 469         if (!(option_mask32 & (OPT_o | OPT_w)))
 470                 reflags = REG_NOSUB;
 471
 472         if (ENABLE_FEATURE_GREP_EGREP_ALIAS
 473          && (applet_name[0] == 'e' || (option_mask32 & OPT_E))
 474         ) {
 475                 reflags |= REG_EXTENDED;
 476         }
 477
 478         if (option_mask32 & OPT_i)
 479                 reflags |= REG_ICASE;
 480
 481         argv += optind;
 482         argc -= optind;
 483
 484         /* if we didn't get a pattern from a -e and no command file was specified,
 485          * argv[optind] should be the pattern. no pattern, no worky */
 486         if (pattern_head == NULL) {
 487                 char *pattern;
 488                 if (*argv == NULL)
 489                         bb_show_usage();
 490                 pattern = new_grep_list_data(*argv++, 0);
 491                 llist_add_to(&pattern_head, pattern);
 492                 argc--;
 493         }
 494
 495         /* argv[(optind)..(argc-1)] should be names of file to grep through. If
 496          * there is more than one file to grep, we will print the filenames. */
 497         if (argc > 1)
 498                 print_filename = 1;
 499         /* -H / -h of course override */
 500         if (option_mask32 & OPT_H)
 501                 print_filename = 1;
 502         if (option_mask32 & OPT_h)
 503                 print_filename = 0;
 504
 505         /* If no files were specified, or '-' was specified, take input from
 506          * stdin. Otherwise, we grep through all the files specified. */
 507         matched = 0;
 508         do {
 509                 cur_file = *argv++;
 510                 file = stdin;
 511                 if (!cur_file || (*cur_file == '-' && !cur_file[1])) {
 512                         cur_file = "(standard input)";
 513                 } else {
 514                         if (option_mask32 & OPT_r) {
 515                                 struct stat st;
 516                                 if (stat(cur_file, &st) == 0 && S_ISDIR(st.st_mode)) {
 517                                         if (!(option_mask32 & OPT_h))
 518                                                 print_filename = 1;
 519                                         matched += grep_dir(cur_file);
 520                                         goto grep_done;
 521                                 }
 522                         }
 523                         /* else: fopen(dir) will succeed, but reading won't */
 524                         file = fopen(cur_file, "r");
 525                         if (file == NULL) {
 526                                 if (!SUPPRESS_ERR_MSGS)
 527                                         bb_simple_perror_msg(cur_file);
 528                                 open_errors = 1;
 529                                 continue;
 530                         }
 531                 }
 532                 matched += grep_file(file);
 533                 fclose_if_not_stdin(file);
 534  grep_done: ;
 535         } while (--argc > 0);
 536
 537         /* destroy all the elments in the pattern list */
 538         if (ENABLE_FEATURE_CLEAN_UP) {
 539                 while (pattern_head) {
 540                         llist_t *pattern_head_ptr = pattern_head;
 541                         grep_list_data_t *gl = (grep_list_data_t *)pattern_head_ptr->data;
 542
 543                         pattern_head = pattern_head->link;
 544                         if ((gl->flg_mem_alocated_compiled & PATTERN_MEM_A))
 545                                 free(gl->pattern);
 546                         if ((gl->flg_mem_alocated_compiled & COMPILED))
 547                                 regfree(&(gl->preg));
 548                         free(gl);
 549                         free(pattern_head_ptr);
 550                 }
 551         }
 552         /* 0 = success, 1 = failed, 2 = error */
 553         if (open_errors)
 554                 return 2;
 555         return !matched; /* invert return value: 0 = success, 1 = failed */
 556 }