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