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 "busybox.h"
  23 #include "xregex.h"
  24
  25 /* options */
  26 #define OPTSTR_GREP \
  27         "lnqvscFiHhe:f:Lor" \
  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,
  37         OPTBIT_n,
  38         OPTBIT_q,
  39         OPTBIT_v,
  40         OPTBIT_s,
  41         OPTBIT_c,
  42         OPTBIT_F,
  43         OPTBIT_i,
  44         OPTBIT_H,
  45         OPTBIT_h,
  46         OPTBIT_e,
  47         OPTBIT_f,
  48         OPTBIT_L,
  49         OPTBIT_o,
  50         OPTBIT_r,
  51         USE_FEATURE_GREP_CONTEXT(OPTBIT_A ,)
  52         USE_FEATURE_GREP_CONTEXT(OPTBIT_B ,)
  53         USE_FEATURE_GREP_CONTEXT(OPTBIT_C ,)
  54         USE_FEATURE_GREP_CONTEXT(OPTBIT_E ,)
  55         USE_DESKTOP(             OPTBIT_w ,)
  56         OPT_l = 1 << OPTBIT_l,
  57         OPT_n = 1 << OPTBIT_n,
  58         OPT_q = 1 << OPTBIT_q,
  59         OPT_v = 1 << OPTBIT_v,
  60         OPT_s = 1 << OPTBIT_s,
  61         OPT_c = 1 << OPTBIT_c,
  62         OPT_F = 1 << OPTBIT_F,
  63         OPT_i = 1 << OPTBIT_i,
  64         OPT_H = 1 << OPTBIT_H,
  65         OPT_h = 1 << OPTBIT_h,
  66         OPT_e = 1 << OPTBIT_e,
  67         OPT_f = 1 << OPTBIT_f,
  68         OPT_L = 1 << OPTBIT_L,
  69         OPT_o = 1 << OPTBIT_o,
  70         OPT_r = 1 << OPTBIT_r,
  71         OPT_A = USE_FEATURE_GREP_CONTEXT((1 << OPTBIT_A)) + 0,
  72         OPT_B = USE_FEATURE_GREP_CONTEXT((1 << OPTBIT_B)) + 0,
  73         OPT_C = USE_FEATURE_GREP_CONTEXT((1 << OPTBIT_C)) + 0,
  74         OPT_E = USE_FEATURE_GREP_CONTEXT((1 << OPTBIT_E)) + 0,
  75         OPT_w = USE_DESKTOP(             (1 << OPTBIT_w)) + 0,
  76 };
  77
  78 #define PRINT_FILES_WITH_MATCHES (option_mask32 & OPT_l)
  79 #define PRINT_LINE_NUM (option_mask32 & OPT_n)
  80 #define BE_QUIET (option_mask32 & OPT_q)
  81 #define SUPPRESS_ERR_MSGS (option_mask32 & OPT_s)
  82 #define PRINT_MATCH_COUNTS (option_mask32 & OPT_c)
  83 #define FGREP_FLAG (option_mask32 & OPT_F)
  84 #define PRINT_FILES_WITHOUT_MATCHES (option_mask32 & OPT_L)
  85
  86 typedef unsigned char byte_t;
  87
  88 static int reflags;
  89 static byte_t invert_search;
  90 static byte_t print_filename;
  91 static byte_t open_errors;
  92
  93 #if ENABLE_FEATURE_GREP_CONTEXT
  94 static int lines_before;
  95 static int lines_after;
  96 static char **before_buf;
  97 static int last_line_printed;
  98 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
  99
 100 /* globals used internally */
 101 static llist_t *pattern_head;   /* growable list of patterns to match */
 102 static const char *cur_file;    /* the current file we are reading */
 103
 104 typedef struct grep_list_data_t {
 105         char *pattern;
 106         regex_t preg;
 107 #define PATTERN_MEM_A 1
 108 #define COMPILED 2
 109         int flg_mem_alocated_compiled;
 110 } grep_list_data_t;
 111
 112 static void print_line(const char *line, int linenum, char decoration)
 113 {
 114 #if ENABLE_FEATURE_GREP_CONTEXT
 115         /* possibly print the little '--' separator */
 116         if ((lines_before || lines_after) && last_line_printed &&
 117                         last_line_printed < linenum - 1) {
 118                 puts("--");
 119         }
 120         last_line_printed = linenum;
 121 #endif
 122         if (print_filename)
 123                 printf("%s%c", cur_file, decoration);
 124         if (PRINT_LINE_NUM)
 125                 printf("%i%c", linenum, decoration);
 126         /* Emulate weird GNU grep behavior with -ov */
 127         if ((option_mask32 & (OPT_v+OPT_o)) != (OPT_v+OPT_o))
 128                 puts(line);
 129 }
 130
 131 static int grep_file(FILE *file)
 132 {
 133         char *line;
 134         byte_t ret;
 135         int linenum = 0;
 136         int nmatches = 0;
 137         regmatch_t regmatch;
 138 #if ENABLE_FEATURE_GREP_CONTEXT
 139         int print_n_lines_after = 0;
 140         int curpos = 0; /* track where we are in the circular 'before' buffer */
 141         int idx = 0; /* used for iteration through the circular buffer */
 142 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
 143
 144         while ((line = xmalloc_getline(file)) != NULL) {
 145                 llist_t *pattern_ptr = pattern_head;
 146                 grep_list_data_t * gl;
 147
 148                 linenum++;
 149                 ret = 0;
 150                 while (pattern_ptr) {
 151                         gl = (grep_list_data_t *)pattern_ptr->data;
 152                         if (FGREP_FLAG) {
 153                                 ret = strstr(line, gl->pattern) != NULL;
 154                         } else {
 155                                 /*
 156                                  * test for a postitive-assertion match (regexec returns success (0)
 157                                  * and the user did not specify invert search), or a negative-assertion
 158                                  * match (regexec returns failure (REG_NOMATCH) and the user specified
 159                                  * invert search)
 160                                  */
 161                                 if (!(gl->flg_mem_alocated_compiled & COMPILED)) {
 162                                         gl->flg_mem_alocated_compiled |= COMPILED;
 163                                         xregcomp(&(gl->preg), gl->pattern, reflags);
 164                                 }
 165                                 regmatch.rm_so = 0;
 166                                 regmatch.rm_eo = 0;
 167                                 if (regexec(&(gl->preg), line, 1, &regmatch, 0) == 0) {
 168                                         if (!(option_mask32 & OPT_w))
 169                                                 ret = 1;
 170                                         else {
 171                                                 char c = ' ';
 172                                                 if (regmatch.rm_so)
 173                                                         c = line[regmatch.rm_so - 1];
 174                                                 if (!isalnum(c) && c != '_') {
 175                                                         c = line[regmatch.rm_eo];
 176                                                         if (!c || (!isalnum(c) && c != '_'))
 177                                                                 ret = 1;
 178                                                 }
 179                                         }
 180                                 }
 181                         }
 182                         pattern_ptr = pattern_ptr->link;
 183                 } /* while (pattern_ptr) */
 184
 185                 if (ret ^ invert_search) {
 186                         if (PRINT_FILES_WITH_MATCHES || BE_QUIET)
 187                                 free(line);
 188
 189                         /* if we found a match but were told to be quiet, stop here */
 190                         if (BE_QUIET || PRINT_FILES_WITHOUT_MATCHES)
 191                                 return -1;
 192
 193                         /* keep track of matches */
 194                         nmatches++;
 195
 196                         /* if we're just printing filenames, we stop after the first match */
 197                         if (PRINT_FILES_WITH_MATCHES)
 198                                 break;
 199
 200                         /* print the matched line */
 201                         if (PRINT_MATCH_COUNTS == 0) {
 202 #if ENABLE_FEATURE_GREP_CONTEXT
 203                                 int prevpos = (curpos == 0) ? lines_before - 1 : curpos - 1;
 204
 205                                 /* if we were told to print 'before' lines and there is at least
 206                                  * one line in the circular buffer, print them */
 207                                 if (lines_before && before_buf[prevpos] != NULL) {
 208                                         int first_buf_entry_line_num = linenum - lines_before;
 209
 210                                         /* advance to the first entry in the circular buffer, and
 211                                          * figure out the line number is of the first line in the
 212                                          * buffer */
 213                                         idx = curpos;
 214                                         while (before_buf[idx] == NULL) {
 215                                                 idx = (idx + 1) % lines_before;
 216                                                 first_buf_entry_line_num++;
 217                                         }
 218
 219                                         /* now print each line in the buffer, clearing them as we go */
 220                                         while (before_buf[idx] != NULL) {
 221                                                 print_line(before_buf[idx], first_buf_entry_line_num, '-');
 222                                                 free(before_buf[idx]);
 223                                                 before_buf[idx] = NULL;
 224                                                 idx = (idx + 1) % lines_before;
 225                                                 first_buf_entry_line_num++;
 226                                         }
 227                                 }
 228
 229                                 /* make a note that we need to print 'after' lines */
 230                                 print_n_lines_after = lines_after;
 231 #endif
 232                                 if (option_mask32 & OPT_o) {
 233                                         line[regmatch.rm_eo] = '\0';
 234                                         print_line(line + regmatch.rm_so, linenum, ':');
 235                                 } else {
 236                                         print_line(line, linenum, ':');
 237                                 }
 238                         }
 239                 }
 240 #if ENABLE_FEATURE_GREP_CONTEXT
 241                 else { /* no match */
 242                         /* Add the line to the circular 'before' buffer */
 243                         if (lines_before) {
 244                                 free(before_buf[curpos]);
 245                                 before_buf[curpos] = xstrdup(line);
 246                                 curpos = (curpos + 1) % lines_before;
 247                         }
 248                 }
 249
 250                 /* if we need to print some context lines after the last match, do so */
 251                 if (print_n_lines_after && (last_line_printed != linenum)) {
 252                         print_line(line, linenum, '-');
 253                         print_n_lines_after--;
 254                 }
 255 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
 256                 free(line);
 257         }
 258
 259         /* special-case file post-processing for options where we don't print line
 260          * matches, just filenames and possibly match counts */
 261
 262         /* grep -c: print [filename:]count, even if count is zero */
 263         if (PRINT_MATCH_COUNTS) {
 264                 if (print_filename)
 265                         printf("%s:", cur_file);
 266                 printf("%d\n", nmatches);
 267         }
 268
 269         /* grep -l: print just the filename, but only if we grepped the line in the file  */
 270         if (PRINT_FILES_WITH_MATCHES && nmatches > 0) {
 271                 puts(cur_file);
 272         }
 273
 274         /* grep -L: print just the filename, but only if we didn't grep the line in the file  */
 275         if (PRINT_FILES_WITHOUT_MATCHES && nmatches == 0) {
 276                 puts(cur_file);
 277         }
 278
 279         return nmatches;
 280 }
 281
 282 #if ENABLE_FEATURE_CLEAN_UP
 283 #define new_grep_list_data(p, m) add_grep_list_data(p, m)
 284 static char * add_grep_list_data(char *pattern, int flg_used_mem)
 285 #else
 286 #define new_grep_list_data(p, m) add_grep_list_data(p)
 287 static char * add_grep_list_data(char *pattern)
 288 #endif
 289 {
 290         grep_list_data_t *gl = xmalloc(sizeof(grep_list_data_t));
 291         gl->pattern = pattern;
 292 #if ENABLE_FEATURE_CLEAN_UP
 293         gl->flg_mem_alocated_compiled = flg_used_mem;
 294 #else
 295         gl->flg_mem_alocated_compiled = 0;
 296 #endif
 297         return (char *)gl;
 298 }
 299
 300
 301 static void load_regexes_from_file(llist_t *fopt)
 302 {
 303         char *line;
 304         FILE *f;
 305
 306         while (fopt) {
 307                 llist_t *cur = fopt;
 308                 char *ffile = cur->data;
 309
 310                 fopt = cur->link;
 311                 free(cur);
 312                 f = xfopen(ffile, "r");
 313                 while ((line = xmalloc_getline(f)) != NULL) {
 314                         llist_add_to(&pattern_head,
 315                                 new_grep_list_data(line, PATTERN_MEM_A));
 316                 }
 317         }
 318 }
 319
 320 static int file_action_grep(const char *filename, struct stat *statbuf, void* matched, int depth)
 321 {
 322         FILE *file = fopen(filename, "r");
 323         if (file == NULL) {
 324                 if (!SUPPRESS_ERR_MSGS)
 325                         bb_perror_msg("%s", cur_file);
 326                 open_errors = 1;
 327                 return 0;
 328         }
 329         cur_file = filename;
 330         *(int*)matched += grep_file(file);
 331         fclose(file);
 332         return 1;
 333 }
 334
 335 static int grep_dir(const char *dir)
 336 {
 337         int matched = 0;
 338         recursive_action(dir,
 339                 /* recurse= */ action_recurse |
 340                 /* followLinks= */ /* no. 0 | */
 341                 /* depthFirst= */ action_depthFirst,
 342                 /* fileAction= */ file_action_grep,
 343                 /* dirAction= */ NULL,
 344                 /* userData= */ &matched,
 345                 /* depth= */ 0);
 346         return matched;
 347 }
 348
 349 int grep_main(int argc, char **argv);
 350 int grep_main(int argc, char **argv)
 351 {
 352         FILE *file;
 353         int matched;
 354         llist_t *fopt = NULL;
 355
 356         /* do normal option parsing */
 357 #if ENABLE_FEATURE_GREP_CONTEXT
 358         char *slines_after;
 359         char *slines_before;
 360         char *Copt;
 361
 362         opt_complementary = "H-h:e::f::C-AB";
 363         getopt32(argc, argv,
 364                 OPTSTR_GREP,
 365                 &pattern_head, &fopt,
 366                 &slines_after, &slines_before, &Copt);
 367
 368         if (option_mask32 & OPT_C) {
 369                 /* -C unsets prev -A and -B, but following -A or -B
 370                    may override it */
 371                 if (!(option_mask32 & OPT_A)) /* not overridden */
 372                         slines_after = Copt;
 373                 if (!(option_mask32 & OPT_B)) /* not overridden */
 374                         slines_before = Copt;
 375                 option_mask32 |= OPT_A|OPT_B; /* for parser */
 376         }
 377         if (option_mask32 & OPT_A) {
 378                 lines_after = xatoi_u(slines_after);
 379         }
 380         if (option_mask32 & OPT_B) {
 381                 lines_before = xatoi_u(slines_before);
 382         }
 383         /* sanity checks */
 384         if (option_mask32 & (OPT_c|OPT_q|OPT_l|OPT_L)) {
 385                 option_mask32 &= ~OPT_n;
 386                 lines_before = 0;
 387                 lines_after = 0;
 388         } else if (lines_before > 0)
 389                 before_buf = xzalloc(lines_before * sizeof(char *));
 390 #else
 391         /* with auto sanity checks */
 392         opt_complementary = "H-h:e::f::c-n:q-n:l-n";
 393         getopt32(argc, argv, OPTSTR_GREP,
 394                 &pattern_head, &fopt);
 395 #endif
 396         invert_search = ((option_mask32 & OPT_v) != 0); /* 0 | 1 */
 397
 398         if (pattern_head != NULL) {
 399                 /* convert char **argv to grep_list_data_t */
 400                 llist_t *cur;
 401
 402                 for (cur = pattern_head; cur; cur = cur->link)
 403                         cur->data = new_grep_list_data(cur->data, 0);
 404         }
 405         if (option_mask32 & OPT_f)
 406                 load_regexes_from_file(fopt);
 407
 408         if (ENABLE_FEATURE_GREP_FGREP_ALIAS && applet_name[0] == 'f')
 409                 option_mask32 |= OPT_F;
 410
 411         if (!(option_mask32 & (OPT_o | OPT_w)))
 412                 reflags = REG_NOSUB;
 413
 414         if (ENABLE_FEATURE_GREP_EGREP_ALIAS &&
 415                         (applet_name[0] == 'e' || (option_mask32 & OPT_E)))
 416                 reflags |= REG_EXTENDED;
 417
 418         if (option_mask32 & OPT_i)
 419                 reflags |= REG_ICASE;
 420
 421         argv += optind;
 422         argc -= optind;
 423
 424         /* if we didn't get a pattern from a -e and no command file was specified,
 425          * argv[optind] should be the pattern. no pattern, no worky */
 426         if (pattern_head == NULL) {
 427                 char *pattern;
 428                 if (*argv == NULL)
 429                         bb_show_usage();
 430                 pattern = new_grep_list_data(*argv++, 0);
 431                 llist_add_to(&pattern_head, pattern);
 432                 argc--;
 433         }
 434
 435         /* argv[(optind)..(argc-1)] should be names of file to grep through. If
 436          * there is more than one file to grep, we will print the filenames. */
 437         if (argc > 1)
 438                 print_filename = 1;
 439         /* -H / -h of course override */
 440         if (option_mask32 & OPT_H)
 441                 print_filename = 1;
 442         if (option_mask32 & OPT_h)
 443                 print_filename = 0;
 444
 445         /* If no files were specified, or '-' was specified, take input from
 446          * stdin. Otherwise, we grep through all the files specified. */
 447         if (argc == 0)
 448                 argc++;
 449         matched = 0;
 450         while (argc--) {
 451                 cur_file = *argv++;
 452                 file = stdin;
 453                 if (!cur_file || (*cur_file == '-' && !cur_file[1])) {
 454                         cur_file = "(standard input)";
 455                 } else {
 456                         if (option_mask32 & OPT_r) {
 457                                 struct stat st;
 458                                 if (stat(cur_file, &st) == 0 && S_ISDIR(st.st_mode)) {
 459                                         if (!(option_mask32 & OPT_h))
 460                                                 print_filename = 1;
 461                                         matched += grep_dir(cur_file);
 462                                         goto grep_done;
 463                                 }
 464                         }
 465                         /* else: fopen(dir) will succeed, but reading won't */
 466                         file = fopen(cur_file, "r");
 467                         if (file == NULL) {
 468                                 if (!SUPPRESS_ERR_MSGS)
 469                                         bb_perror_msg("%s", cur_file);
 470                                 open_errors = 1;
 471                                 continue;
 472                         }
 473                 }
 474                 matched += grep_file(file);
 475                 fclose_if_not_stdin(file);
 476  grep_done:
 477                 if (matched < 0) {
 478                         /* we found a match but were told to be quiet, stop here and
 479                         * return success */
 480                         break;
 481                 }
 482         }
 483
 484         /* destroy all the elments in the pattern list */
 485         if (ENABLE_FEATURE_CLEAN_UP) {
 486                 while (pattern_head) {
 487                         llist_t *pattern_head_ptr = pattern_head;
 488                         grep_list_data_t *gl =
 489                                 (grep_list_data_t *)pattern_head_ptr->data;
 490
 491                         pattern_head = pattern_head->link;
 492                         if ((gl->flg_mem_alocated_compiled & PATTERN_MEM_A))
 493                                 free(gl->pattern);
 494                         if ((gl->flg_mem_alocated_compiled & COMPILED))
 495                                 regfree(&(gl->preg));
 496                         free(pattern_head_ptr);
 497                 }
 498         }
 499         /* 0 = success, 1 = failed, 2 = error */
 500         /* If the -q option is specified, the exit status shall be zero
 501          * if an input line is selected, even if an error was detected.  */
 502         if (BE_QUIET && matched)
 503                 return 0;
 504         if (open_errors)
 505                 return 2;
 506         return !matched; /* invert return value 0 = success, 1 = failed */
 507 }