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 "match whole line only". */
  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(EXIT_SUCCESS) 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() do { \
 107         struct G_sizecheck { \
 108                 char G_sizecheck[sizeof(G) > COMMON_BUFSIZE ? -1 : 1]; \
 109         }; \
 110 } while (0)
 111 #define max_matches       (G.max_matches         )
 112 #define reflags           (G.reflags             )
 113 #define invert_search     (G.invert_search       )
 114 #define print_filename    (G.print_filename      )
 115 #define open_errors       (G.open_errors         )
 116 #define did_print_line    (G.did_print_line      )
 117 #define lines_before      (G.lines_before        )
 118 #define lines_after       (G.lines_after         )
 119 #define before_buf        (G.before_buf          )
 120 #define last_line_printed (G.last_line_printed   )
 121 #define pattern_head      (G.pattern_head        )
 122 #define cur_file          (G.cur_file            )
 123
 124
 125 typedef struct grep_list_data_t {
 126         char *pattern;
 127         regex_t preg;
 128 #define ALLOCATED 1
 129 #define COMPILED 2
 130         int flg_mem_alocated_compiled;
 131 } grep_list_data_t;
 132
 133
 134 static void print_line(const char *line, int linenum, char decoration)
 135 {
 136 #if ENABLE_FEATURE_GREP_CONTEXT
 137         /* Happens when we go to next file, immediately hit match
 138          * and try to print prev context... from prev file! Don't do it */
 139         if (linenum < 1)
 140                 return;
 141         /* possibly print the little '--' separator */
 142         if ((lines_before || lines_after) && did_print_line &&
 143                         last_line_printed != linenum - 1) {
 144                 puts("--");
 145         }
 146         /* guard against printing "--" before first line of first file */
 147         did_print_line = 1;
 148         last_line_printed = linenum;
 149 #endif
 150         if (print_filename)
 151                 printf("%s%c", cur_file, decoration);
 152         if (PRINT_LINE_NUM)
 153                 printf("%i%c", linenum, decoration);
 154         /* Emulate weird GNU grep behavior with -ov */
 155         if ((option_mask32 & (OPT_v|OPT_o)) != (OPT_v|OPT_o))
 156                 puts(line);
 157 }
 158
 159 static int grep_file(FILE *file)
 160 {
 161         char *line;
 162         smalluint found;
 163         int linenum = 0;
 164         int nmatches = 0;
 165         regmatch_t regmatch;
 166 #if ENABLE_FEATURE_GREP_CONTEXT
 167         int print_n_lines_after = 0;
 168         int curpos = 0; /* track where we are in the circular 'before' buffer */
 169         int idx = 0; /* used for iteration through the circular buffer */
 170 #else
 171         enum { print_n_lines_after = 0 };
 172 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
 173
 174         while ((line = xmalloc_fgetline(file)) != NULL) {
 175                 llist_t *pattern_ptr = pattern_head;
 176                 grep_list_data_t *gl = gl; /* for gcc */
 177
 178                 linenum++;
 179                 found = 0;
 180                 while (pattern_ptr) {
 181                         gl = (grep_list_data_t *)pattern_ptr->data;
 182                         if (FGREP_FLAG) {
 183                                 found |= (strstr(line, gl->pattern) != NULL);
 184                         } else {
 185                                 if (!(gl->flg_mem_alocated_compiled & COMPILED)) {
 186                                         gl->flg_mem_alocated_compiled |= COMPILED;
 187                                         xregcomp(&(gl->preg), gl->pattern, reflags);
 188                                 }
 189                                 regmatch.rm_so = 0;
 190                                 regmatch.rm_eo = 0;
 191                                 if (regexec(&(gl->preg), line, 1, &regmatch, 0) == 0) {
 192                                         if (!(option_mask32 & OPT_w))
 193                                                 found = 1;
 194                                         else {
 195                                                 char c = ' ';
 196                                                 if (regmatch.rm_so)
 197                                                         c = line[regmatch.rm_so - 1];
 198                                                 if (!isalnum(c) && c != '_') {
 199                                                         c = line[regmatch.rm_eo];
 200                                                         if (!c || (!isalnum(c) && c != '_'))
 201                                                                 found = 1;
 202                                                 }
 203                                         }
 204                                 }
 205                         }
 206                         /* If it's non-inverted search, we can stop
 207                          * at first match */
 208                         if (found && !invert_search)
 209                                 goto do_found;
 210                         pattern_ptr = pattern_ptr->link;
 211                 } /* while (pattern_ptr) */
 212
 213                 if (found ^ invert_search) {
 214  do_found:
 215                         /* keep track of matches */
 216                         nmatches++;
 217
 218                         /* quiet/print (non)matching file names only? */
 219                         if (option_mask32 & (OPT_q|OPT_l|OPT_L)) {
 220                                 free(line); /* we don't need line anymore */
 221                                 if (BE_QUIET) {
 222                                         /* manpage says about -q:
 223                                          * "exit immediately with zero status
 224                                          * if any match is found,
 225                                          * even if errors were detected" */
 226                                         exit(EXIT_SUCCESS);
 227                                 }
 228                                 /* if we're just printing filenames, we stop after the first match */
 229                                 if (PRINT_FILES_WITH_MATCHES) {
 230                                         puts(cur_file);
 231                                         /* fall through to "return 1" */
 232                                 }
 233                                 /* OPT_L aka PRINT_FILES_WITHOUT_MATCHES: return early */
 234                                 return 1; /* one match */
 235                         }
 236
 237 #if ENABLE_FEATURE_GREP_CONTEXT
 238                         /* Were we printing context and saw next (unwanted) match? */
 239                         if ((option_mask32 & OPT_m) && nmatches > max_matches)
 240                                 break;
 241 #endif
 242
 243                         /* print the matched line */
 244                         if (PRINT_MATCH_COUNTS == 0) {
 245 #if ENABLE_FEATURE_GREP_CONTEXT
 246                                 int prevpos = (curpos == 0) ? lines_before - 1 : curpos - 1;
 247
 248                                 /* if we were told to print 'before' lines and there is at least
 249                                  * one line in the circular buffer, print them */
 250                                 if (lines_before && before_buf[prevpos] != NULL) {
 251                                         int first_buf_entry_line_num = linenum - lines_before;
 252
 253                                         /* advance to the first entry in the circular buffer, and
 254                                          * figure out the line number is of the first line in the
 255                                          * buffer */
 256                                         idx = curpos;
 257                                         while (before_buf[idx] == NULL) {
 258                                                 idx = (idx + 1) % lines_before;
 259                                                 first_buf_entry_line_num++;
 260                                         }
 261
 262                                         /* now print each line in the buffer, clearing them as we go */
 263                                         while (before_buf[idx] != NULL) {
 264                                                 print_line(before_buf[idx], first_buf_entry_line_num, '-');
 265                                                 free(before_buf[idx]);
 266                                                 before_buf[idx] = NULL;
 267                                                 idx = (idx + 1) % lines_before;
 268                                                 first_buf_entry_line_num++;
 269                                         }
 270                                 }
 271
 272                                 /* make a note that we need to print 'after' lines */
 273                                 print_n_lines_after = lines_after;
 274 #endif
 275                                 if (option_mask32 & OPT_o) {
 276                                         if (FGREP_FLAG) {
 277                                                 /* -Fo just prints the pattern
 278                                                  * (unless -v: -Fov doesnt print anything at all) */
 279                                                 if (found)
 280                                                         print_line(gl->pattern, linenum, ':');
 281                                         } else {
 282                                                 line[regmatch.rm_eo] = '\0';
 283                                                 print_line(line + regmatch.rm_so, linenum, ':');
 284                                         }
 285                                 } else {
 286                                         print_line(line, linenum, ':');
 287                                 }
 288                         }
 289                 }
 290 #if ENABLE_FEATURE_GREP_CONTEXT
 291                 else { /* no match */
 292                         /* if we need to print some context lines after the last match, do so */
 293                         if (print_n_lines_after) {
 294                                 print_line(line, linenum, '-');
 295                                 print_n_lines_after--;
 296                         } else if (lines_before) {
 297                                 /* Add the line to the circular 'before' buffer */
 298                                 free(before_buf[curpos]);
 299                                 before_buf[curpos] = line;
 300                                 curpos = (curpos + 1) % lines_before;
 301                                 /* avoid free(line) - we took the line */
 302                                 line = NULL;
 303                         }
 304                 }
 305
 306 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
 307                 free(line);
 308
 309                 /* Did we print all context after last requested match? */
 310                 if ((option_mask32 & OPT_m)
 311                  && !print_n_lines_after && nmatches == max_matches)
 312                         break;
 313         }
 314
 315         /* special-case file post-processing for options where we don't print line
 316          * matches, just filenames and possibly match counts */
 317
 318         /* grep -c: print [filename:]count, even if count is zero */
 319         if (PRINT_MATCH_COUNTS) {
 320                 if (print_filename)
 321                         printf("%s:", cur_file);
 322                 printf("%d\n", nmatches);
 323         }
 324
 325         /* grep -L: print just the filename */
 326         if (PRINT_FILES_WITHOUT_MATCHES) {
 327                 /* nmatches is zero, no need to check it:
 328                  * we return 1 early if we detected a match
 329                  * and PRINT_FILES_WITHOUT_MATCHES is set */
 330                 puts(cur_file);
 331         }
 332
 333         return nmatches;
 334 }
 335
 336 #if ENABLE_FEATURE_CLEAN_UP
 337 #define new_grep_list_data(p, m) add_grep_list_data(p, m)
 338 static char *add_grep_list_data(char *pattern, int flg_used_mem)
 339 #else
 340 #define new_grep_list_data(p, m) add_grep_list_data(p)
 341 static char *add_grep_list_data(char *pattern)
 342 #endif
 343 {
 344         grep_list_data_t *gl = xzalloc(sizeof(*gl));
 345         gl->pattern = pattern;
 346 #if ENABLE_FEATURE_CLEAN_UP
 347         gl->flg_mem_alocated_compiled = flg_used_mem;
 348 #else
 349         /*gl->flg_mem_alocated_compiled = 0;*/
 350 #endif
 351         return (char *)gl;
 352 }
 353
 354 static void load_regexes_from_file(llist_t *fopt)
 355 {
 356         char *line;
 357         FILE *f;
 358
 359         while (fopt) {
 360                 llist_t *cur = fopt;
 361                 char *ffile = cur->data;
 362
 363                 fopt = cur->link;
 364                 free(cur);
 365                 f = xfopen_stdin(ffile);
 366                 while ((line = xmalloc_fgetline(f)) != NULL) {
 367                         llist_add_to(&pattern_head,
 368                                 new_grep_list_data(line, ALLOCATED));
 369                 }
 370         }
 371 }
 372
 373 static int FAST_FUNC file_action_grep(const char *filename,
 374                         struct stat *statbuf UNUSED_PARAM,
 375                         void* matched,
 376                         int depth UNUSED_PARAM)
 377 {
 378         FILE *file = fopen_for_read(filename);
 379         if (file == NULL) {
 380                 if (!SUPPRESS_ERR_MSGS)
 381                         bb_simple_perror_msg(filename);
 382                 open_errors = 1;
 383                 return 0;
 384         }
 385         cur_file = filename;
 386         *(int*)matched += grep_file(file);
 387         fclose(file);
 388         return 1;
 389 }
 390
 391 static int grep_dir(const char *dir)
 392 {
 393         int matched = 0;
 394         recursive_action(dir,
 395                 /* recurse=yes */ ACTION_RECURSE |
 396                 /* followLinks=no */
 397                 /* depthFirst=yes */ ACTION_DEPTHFIRST,
 398                 /* fileAction= */ file_action_grep,
 399                 /* dirAction= */ NULL,
 400                 /* userData= */ &matched,
 401                 /* depth= */ 0);
 402         return matched;
 403 }
 404
 405 int grep_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 406 int grep_main(int argc, char **argv)
 407 {
 408         FILE *file;
 409         int matched;
 410         llist_t *fopt = NULL;
 411
 412         /* do normal option parsing */
 413 #if ENABLE_FEATURE_GREP_CONTEXT
 414         int Copt;
 415
 416         /* -H unsets -h; -C unsets -A,-B; -e,-f are lists;
 417          * -m,-A,-B,-C have numeric param */
 418         opt_complementary = "H-h:C-AB:e::f::m+:A+:B+:C+";
 419         getopt32(argv,
 420                 OPTSTR_GREP,
 421                 &pattern_head, &fopt, &max_matches,
 422                 &lines_after, &lines_before, &Copt);
 423
 424         if (option_mask32 & OPT_C) {
 425                 /* -C unsets prev -A and -B, but following -A or -B
 426                    may override it */
 427                 if (!(option_mask32 & OPT_A)) /* not overridden */
 428                         lines_after = Copt;
 429                 if (!(option_mask32 & OPT_B)) /* not overridden */
 430                         lines_before = Copt;
 431                 //option_mask32 |= OPT_A|OPT_B; /* for parser */
 432         }
 433         /* sanity checks */
 434         if (option_mask32 & (OPT_c|OPT_q|OPT_l|OPT_L)) {
 435                 option_mask32 &= ~OPT_n;
 436                 lines_before = 0;
 437                 lines_after = 0;
 438         } else if (lines_before > 0)
 439                 before_buf = xzalloc(lines_before * sizeof(char *));
 440 #else
 441         /* with auto sanity checks */
 442         /* -H unsets -h; -c,-q or -l unset -n; -e,-f are lists; -m N */
 443         opt_complementary = "H-h:c-n:q-n:l-n:e::f::m+";
 444         getopt32(argv, OPTSTR_GREP,
 445                 &pattern_head, &fopt, &max_matches);
 446 #endif
 447         invert_search = ((option_mask32 & OPT_v) != 0); /* 0 | 1 */
 448
 449         if (pattern_head != NULL) {
 450                 /* convert char **argv to grep_list_data_t */
 451                 llist_t *cur;
 452
 453                 for (cur = pattern_head; cur; cur = cur->link)
 454                         cur->data = new_grep_list_data(cur->data, 0);
 455         }
 456         if (option_mask32 & OPT_f)
 457                 load_regexes_from_file(fopt);
 458
 459         if (ENABLE_FEATURE_GREP_FGREP_ALIAS && applet_name[0] == 'f')
 460                 option_mask32 |= OPT_F;
 461
 462         if (!(option_mask32 & (OPT_o | OPT_w)))
 463                 reflags = REG_NOSUB;
 464
 465         if (ENABLE_FEATURE_GREP_EGREP_ALIAS
 466          && (applet_name[0] == 'e' || (option_mask32 & OPT_E))
 467         ) {
 468                 reflags |= REG_EXTENDED;
 469         }
 470
 471         if (option_mask32 & OPT_i)
 472                 reflags |= REG_ICASE;
 473
 474         argv += optind;
 475         argc -= optind;
 476
 477         /* if we didn't get a pattern from -e and no command file was specified,
 478          * first parameter should be the pattern. no pattern, no worky */
 479         if (pattern_head == NULL) {
 480                 char *pattern;
 481                 if (*argv == NULL)
 482                         bb_show_usage();
 483                 pattern = new_grep_list_data(*argv++, 0);
 484                 llist_add_to(&pattern_head, pattern);
 485                 argc--;
 486         }
 487
 488         /* argv[0..(argc-1)] should be names of file to grep through. If
 489          * there is more than one file to grep, we will print the filenames. */
 490         if (argc > 1)
 491                 print_filename = 1;
 492         /* -H / -h of course override */
 493         if (option_mask32 & OPT_H)
 494                 print_filename = 1;
 495         if (option_mask32 & OPT_h)
 496                 print_filename = 0;
 497
 498         /* If no files were specified, or '-' was specified, take input from
 499          * stdin. Otherwise, we grep through all the files specified. */
 500         matched = 0;
 501         do {
 502                 cur_file = *argv++;
 503                 file = stdin;
 504                 if (!cur_file || LONE_DASH(cur_file)) {
 505                         cur_file = "(standard input)";
 506                 } else {
 507                         if (option_mask32 & OPT_r) {
 508                                 struct stat st;
 509                                 if (stat(cur_file, &st) == 0 && S_ISDIR(st.st_mode)) {
 510                                         if (!(option_mask32 & OPT_h))
 511                                                 print_filename = 1;
 512                                         matched += grep_dir(cur_file);
 513                                         goto grep_done;
 514                                 }
 515                         }
 516                         /* else: fopen(dir) will succeed, but reading won't */
 517                         file = fopen_for_read(cur_file);
 518                         if (file == NULL) {
 519                                 if (!SUPPRESS_ERR_MSGS)
 520                                         bb_simple_perror_msg(cur_file);
 521                                 open_errors = 1;
 522                                 continue;
 523                         }
 524                 }
 525                 matched += grep_file(file);
 526                 fclose_if_not_stdin(file);
 527  grep_done: ;
 528         } while (--argc > 0);
 529
 530         /* destroy all the elments in the pattern list */
 531         if (ENABLE_FEATURE_CLEAN_UP) {
 532                 while (pattern_head) {
 533                         llist_t *pattern_head_ptr = pattern_head;
 534                         grep_list_data_t *gl = (grep_list_data_t *)pattern_head_ptr->data;
 535
 536                         pattern_head = pattern_head->link;
 537                         if (gl->flg_mem_alocated_compiled & ALLOCATED)
 538                                 free(gl->pattern);
 539                         if (gl->flg_mem_alocated_compiled & COMPILED)
 540                                 regfree(&(gl->preg));
 541                         free(gl);
 542                         free(pattern_head_ptr);
 543                 }
 544         }
 545         /* 0 = success, 1 = failed, 2 = error */
 546         if (open_errors)
 547                 return 2;
 548         return !matched; /* invert return value: 0 = success, 1 = failed */
 549 }