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 GPLv2 or later, see file LICENSE in this source tree.
   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  * (C) 2006 Jac Goudsmit added -o option
  19  */
  20
  21 //applet:IF_GREP(APPLET(grep, BB_DIR_BIN, BB_SUID_DROP))
  22 //applet:IF_FEATURE_GREP_EGREP_ALIAS(APPLET_ODDNAME(egrep, grep, BB_DIR_BIN, BB_SUID_DROP, egrep))
  23 //applet:IF_FEATURE_GREP_FGREP_ALIAS(APPLET_ODDNAME(fgrep, grep, BB_DIR_BIN, BB_SUID_DROP, fgrep))
  24
  25 //kbuild:lib-$(CONFIG_GREP) += grep.o
  26
  27 //config:config GREP
  28 //config:       bool "grep"
  29 //config:       default y
  30 //config:       help
  31 //config:         grep is used to search files for a specified pattern.
  32 //config:
  33 //config:config FEATURE_GREP_EGREP_ALIAS
  34 //config:       bool "Enable extended regular expressions (egrep & grep -E)"
  35 //config:       default y
  36 //config:       depends on GREP
  37 //config:       help
  38 //config:         Enabled support for extended regular expressions. Extended
  39 //config:         regular expressions allow for alternation (foo|bar), grouping,
  40 //config:         and various repetition operators.
  41 //config:
  42 //config:config FEATURE_GREP_FGREP_ALIAS
  43 //config:       bool "Alias fgrep to grep -F"
  44 //config:       default y
  45 //config:       depends on GREP
  46 //config:       help
  47 //config:         fgrep sees the search pattern as a normal string rather than
  48 //config:         regular expressions.
  49 //config:         grep -F always works, this just creates the fgrep alias.
  50 //config:
  51 //config:config FEATURE_GREP_CONTEXT
  52 //config:       bool "Enable before and after context flags (-A, -B and -C)"
  53 //config:       default y
  54 //config:       depends on GREP
  55 //config:       help
  56 //config:         Print the specified number of leading (-B) and/or trailing (-A)
  57 //config:         context surrounding our matching lines.
  58 //config:         Print the specified number of context lines (-C).
  59
  60 #include "libbb.h"
  61 #include "xregex.h"
  62
  63
  64 /* options */
  65 //usage:#define grep_trivial_usage
  66 //usage:       "[-HhnlLoqvsriw"
  67 //usage:       "F"
  68 //usage:        IF_FEATURE_GREP_EGREP_ALIAS("E")
  69 //usage:        IF_EXTRA_COMPAT("z")
  70 //usage:       "] [-m N] "
  71 //usage:        IF_FEATURE_GREP_CONTEXT("[-A/B/C N] ")
  72 //usage:       "PATTERN/-e PATTERN.../-f FILE [FILE]..."
  73 //usage:#define grep_full_usage "\n\n"
  74 //usage:       "Search for PATTERN in FILEs (or stdin)\n"
  75 //usage:     "\nOptions:"
  76 //usage:     "\n        -H      Add 'filename:' prefix"
  77 //usage:     "\n        -h      Do not add 'filename:' prefix"
  78 //usage:     "\n        -n      Add 'line_no:' prefix"
  79 //usage:     "\n        -l      Show only names of files that match"
  80 //usage:     "\n        -L      Show only names of files that don't match"
  81 //usage:     "\n        -c      Show only count of matching lines"
  82 //usage:     "\n        -o      Show only the matching part of line"
  83 //usage:     "\n        -q      Quiet. Return 0 if PATTERN is found, 1 otherwise"
  84 //usage:     "\n        -v      Select non-matching lines"
  85 //usage:     "\n        -s      Suppress open and read errors"
  86 //usage:     "\n        -r      Recurse"
  87 //usage:     "\n        -i      Ignore case"
  88 //usage:     "\n        -w      Match whole words only"
  89 //usage:     "\n        -F      PATTERN is a literal (not regexp)"
  90 //usage:        IF_FEATURE_GREP_EGREP_ALIAS(
  91 //usage:     "\n        -E      PATTERN is an extended regexp"
  92 //usage:        )
  93 //usage:        IF_EXTRA_COMPAT(
  94 //usage:     "\n        -z      Input is NUL terminated"
  95 //usage:        )
  96 //usage:     "\n        -m N    Match up to N times per file"
  97 //usage:        IF_FEATURE_GREP_CONTEXT(
  98 //usage:     "\n        -A N    Print N lines of trailing context"
  99 //usage:     "\n        -B N    Print N lines of leading context"
 100 //usage:     "\n        -C N    Same as '-A N -B N'"
 101 //usage:        )
 102 //usage:     "\n        -e PTRN Pattern to match"
 103 //usage:     "\n        -f FILE Read pattern from file"
 104 //usage:
 105 //usage:#define grep_example_usage
 106 //usage:       "$ grep root /etc/passwd\n"
 107 //usage:       "root:x:0:0:root:/root:/bin/bash\n"
 108 //usage:       "$ grep ^[rR]oo. /etc/passwd\n"
 109 //usage:       "root:x:0:0:root:/root:/bin/bash\n"
 110 //usage:
 111 //usage:#define egrep_trivial_usage NOUSAGE_STR
 112 //usage:#define egrep_full_usage ""
 113 //usage:#define fgrep_trivial_usage NOUSAGE_STR
 114 //usage:#define fgrep_full_usage ""
 115
 116 #define OPTSTR_GREP \
 117         "lnqvscFiHhe:f:Lorm:w" \
 118         IF_FEATURE_GREP_CONTEXT("A:B:C:") \
 119         IF_FEATURE_GREP_EGREP_ALIAS("E") \
 120         IF_EXTRA_COMPAT("z") \
 121         "aI"
 122 /* ignored: -a "assume all files to be text" */
 123 /* ignored: -I "assume binary files have no matches" */
 124 enum {
 125         OPTBIT_l, /* list matched file names only */
 126         OPTBIT_n, /* print line# */
 127         OPTBIT_q, /* quiet - exit(EXIT_SUCCESS) of first match */
 128         OPTBIT_v, /* invert the match, to select non-matching lines */
 129         OPTBIT_s, /* suppress errors about file open errors */
 130         OPTBIT_c, /* count matches per file (suppresses normal output) */
 131         OPTBIT_F, /* literal match */
 132         OPTBIT_i, /* case-insensitive */
 133         OPTBIT_H, /* force filename display */
 134         OPTBIT_h, /* inhibit filename display */
 135         OPTBIT_e, /* -e PATTERN */
 136         OPTBIT_f, /* -f FILE_WITH_PATTERNS */
 137         OPTBIT_L, /* list unmatched file names only */
 138         OPTBIT_o, /* show only matching parts of lines */
 139         OPTBIT_r, /* recurse dirs */
 140         OPTBIT_m, /* -m MAX_MATCHES */
 141         OPTBIT_w, /* -w whole word match */
 142         IF_FEATURE_GREP_CONTEXT(    OPTBIT_A ,) /* -A NUM: after-match context */
 143         IF_FEATURE_GREP_CONTEXT(    OPTBIT_B ,) /* -B NUM: before-match context */
 144         IF_FEATURE_GREP_CONTEXT(    OPTBIT_C ,) /* -C NUM: -A and -B combined */
 145         IF_FEATURE_GREP_EGREP_ALIAS(OPTBIT_E ,) /* extended regexp */
 146         IF_EXTRA_COMPAT(            OPTBIT_z ,) /* input is NUL terminated */
 147         OPT_l = 1 << OPTBIT_l,
 148         OPT_n = 1 << OPTBIT_n,
 149         OPT_q = 1 << OPTBIT_q,
 150         OPT_v = 1 << OPTBIT_v,
 151         OPT_s = 1 << OPTBIT_s,
 152         OPT_c = 1 << OPTBIT_c,
 153         OPT_F = 1 << OPTBIT_F,
 154         OPT_i = 1 << OPTBIT_i,
 155         OPT_H = 1 << OPTBIT_H,
 156         OPT_h = 1 << OPTBIT_h,
 157         OPT_e = 1 << OPTBIT_e,
 158         OPT_f = 1 << OPTBIT_f,
 159         OPT_L = 1 << OPTBIT_L,
 160         OPT_o = 1 << OPTBIT_o,
 161         OPT_r = 1 << OPTBIT_r,
 162         OPT_m = 1 << OPTBIT_m,
 163         OPT_w = 1 << OPTBIT_w,
 164         OPT_A = IF_FEATURE_GREP_CONTEXT(    (1 << OPTBIT_A)) + 0,
 165         OPT_B = IF_FEATURE_GREP_CONTEXT(    (1 << OPTBIT_B)) + 0,
 166         OPT_C = IF_FEATURE_GREP_CONTEXT(    (1 << OPTBIT_C)) + 0,
 167         OPT_E = IF_FEATURE_GREP_EGREP_ALIAS((1 << OPTBIT_E)) + 0,
 168         OPT_z = IF_EXTRA_COMPAT(            (1 << OPTBIT_z)) + 0,
 169 };
 170
 171 #define PRINT_FILES_WITH_MATCHES    (option_mask32 & OPT_l)
 172 #define PRINT_LINE_NUM              (option_mask32 & OPT_n)
 173 #define BE_QUIET                    (option_mask32 & OPT_q)
 174 #define SUPPRESS_ERR_MSGS           (option_mask32 & OPT_s)
 175 #define PRINT_MATCH_COUNTS          (option_mask32 & OPT_c)
 176 #define FGREP_FLAG                  (option_mask32 & OPT_F)
 177 #define PRINT_FILES_WITHOUT_MATCHES (option_mask32 & OPT_L)
 178 #define NUL_DELIMITED               (option_mask32 & OPT_z)
 179
 180 struct globals {
 181         int max_matches;
 182 #if !ENABLE_EXTRA_COMPAT
 183         int reflags;
 184 #else
 185         RE_TRANSLATE_TYPE case_fold; /* RE_TRANSLATE_TYPE is [[un]signed] char* */
 186 #endif
 187         smalluint invert_search;
 188         smalluint print_filename;
 189         smalluint open_errors;
 190 #if ENABLE_FEATURE_GREP_CONTEXT
 191         smalluint did_print_line;
 192         int lines_before;
 193         int lines_after;
 194         char **before_buf;
 195         IF_EXTRA_COMPAT(size_t *before_buf_size;)
 196         int last_line_printed;
 197 #endif
 198         /* globals used internally */
 199         llist_t *pattern_head;   /* growable list of patterns to match */
 200         const char *cur_file;    /* the current file we are reading */
 201 } FIX_ALIASING;
 202 #define G (*(struct globals*)&bb_common_bufsiz1)
 203 #define INIT_G() do { \
 204         struct G_sizecheck { \
 205                 char G_sizecheck[sizeof(G) > COMMON_BUFSIZE ? -1 : 1]; \
 206         }; \
 207 } while (0)
 208 #define max_matches       (G.max_matches         )
 209 #if !ENABLE_EXTRA_COMPAT
 210 # define reflags          (G.reflags             )
 211 #else
 212 # define case_fold        (G.case_fold           )
 213 /* http://www.delorie.com/gnu/docs/regex/regex_46.html */
 214 # define reflags           re_syntax_options
 215 # undef REG_NOSUB
 216 # undef REG_EXTENDED
 217 # undef REG_ICASE
 218 # define REG_NOSUB    bug:is:here /* should not be used */
 219 /* Just RE_SYNTAX_EGREP is not enough, need to enable {n[,[m]]} too */
 220 # define REG_EXTENDED (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES)
 221 # define REG_ICASE    bug:is:here /* should not be used */
 222 #endif
 223 #define invert_search     (G.invert_search       )
 224 #define print_filename    (G.print_filename      )
 225 #define open_errors       (G.open_errors         )
 226 #define did_print_line    (G.did_print_line      )
 227 #define lines_before      (G.lines_before        )
 228 #define lines_after       (G.lines_after         )
 229 #define before_buf        (G.before_buf          )
 230 #define before_buf_size   (G.before_buf_size     )
 231 #define last_line_printed (G.last_line_printed   )
 232 #define pattern_head      (G.pattern_head        )
 233 #define cur_file          (G.cur_file            )
 234
 235
 236 typedef struct grep_list_data_t {
 237         char *pattern;
 238 /* for GNU regex, matched_range must be persistent across grep_file() calls */
 239 #if !ENABLE_EXTRA_COMPAT
 240         regex_t compiled_regex;
 241         regmatch_t matched_range;
 242 #else
 243         struct re_pattern_buffer compiled_regex;
 244         struct re_registers matched_range;
 245 #endif
 246 #define ALLOCATED 1
 247 #define COMPILED 2
 248         int flg_mem_alocated_compiled;
 249 } grep_list_data_t;
 250
 251 #if !ENABLE_EXTRA_COMPAT
 252 #define print_line(line, line_len, linenum, decoration) \
 253         print_line(line, linenum, decoration)
 254 #endif
 255 static void print_line(const char *line, size_t line_len, int linenum, char decoration)
 256 {
 257 #if ENABLE_FEATURE_GREP_CONTEXT
 258         /* Happens when we go to next file, immediately hit match
 259          * and try to print prev context... from prev file! Don't do it */
 260         if (linenum < 1)
 261                 return;
 262         /* possibly print the little '--' separator */
 263         if ((lines_before || lines_after) && did_print_line
 264          && last_line_printed != linenum - 1
 265         ) {
 266                 puts("--");
 267         }
 268         /* guard against printing "--" before first line of first file */
 269         did_print_line = 1;
 270         last_line_printed = linenum;
 271 #endif
 272         if (print_filename)
 273                 printf("%s%c", cur_file, decoration);
 274         if (PRINT_LINE_NUM)
 275                 printf("%i%c", linenum, decoration);
 276         /* Emulate weird GNU grep behavior with -ov */
 277         if ((option_mask32 & (OPT_v|OPT_o)) != (OPT_v|OPT_o)) {
 278 #if !ENABLE_EXTRA_COMPAT
 279                 puts(line);
 280 #else
 281                 fwrite(line, 1, line_len, stdout);
 282                 putchar(NUL_DELIMITED ? '\0' : '\n');
 283 #endif
 284         }
 285 }
 286
 287 #if ENABLE_EXTRA_COMPAT
 288 /* Unlike getline, this one removes trailing '\n' */
 289 static ssize_t FAST_FUNC bb_getline(char **line_ptr, size_t *line_alloc_len, FILE *file)
 290 {
 291         ssize_t res_sz;
 292         char *line;
 293         int delim = (NUL_DELIMITED ? '\0' : '\n');
 294
 295         res_sz = getdelim(line_ptr, line_alloc_len, delim, file);
 296         line = *line_ptr;
 297
 298         if (res_sz > 0) {
 299                 if (line[res_sz - 1] == delim)
 300                         line[--res_sz] = '\0';
 301         } else {
 302                 free(line); /* uclibc allocates a buffer even on EOF. WTF? */
 303         }
 304         return res_sz;
 305 }
 306 #endif
 307
 308 static int grep_file(FILE *file)
 309 {
 310         smalluint found;
 311         int linenum = 0;
 312         int nmatches = 0;
 313 #if !ENABLE_EXTRA_COMPAT
 314         char *line;
 315 #else
 316         char *line = NULL;
 317         ssize_t line_len;
 318         size_t line_alloc_len;
 319 # define rm_so start[0]
 320 # define rm_eo end[0]
 321 #endif
 322 #if ENABLE_FEATURE_GREP_CONTEXT
 323         int print_n_lines_after = 0;
 324         int curpos = 0; /* track where we are in the circular 'before' buffer */
 325         int idx = 0; /* used for iteration through the circular buffer */
 326 #else
 327         enum { print_n_lines_after = 0 };
 328 #endif
 329
 330         while (
 331 #if !ENABLE_EXTRA_COMPAT
 332                 (line = xmalloc_fgetline(file)) != NULL
 333 #else
 334                 (line_len = bb_getline(&line, &line_alloc_len, file)) >= 0
 335 #endif
 336         ) {
 337                 llist_t *pattern_ptr = pattern_head;
 338                 grep_list_data_t *gl = gl; /* for gcc */
 339
 340                 linenum++;
 341                 found = 0;
 342                 while (pattern_ptr) {
 343                         gl = (grep_list_data_t *)pattern_ptr->data;
 344                         if (FGREP_FLAG) {
 345                                 found |= (((option_mask32 & OPT_i)
 346                                         ? strcasestr(line, gl->pattern)
 347                                         : strstr(line, gl->pattern)
 348                                         ) != NULL);
 349                         } else {
 350                                 if (!(gl->flg_mem_alocated_compiled & COMPILED)) {
 351                                         gl->flg_mem_alocated_compiled |= COMPILED;
 352 #if !ENABLE_EXTRA_COMPAT
 353                                         xregcomp(&gl->compiled_regex, gl->pattern, reflags);
 354 #else
 355                                         memset(&gl->compiled_regex, 0, sizeof(gl->compiled_regex));
 356                                         gl->compiled_regex.translate = case_fold; /* for -i */
 357                                         if (re_compile_pattern(gl->pattern, strlen(gl->pattern), &gl->compiled_regex))
 358                                                 bb_error_msg_and_die("bad regex '%s'", gl->pattern);
 359 #endif
 360                                 }
 361 #if !ENABLE_EXTRA_COMPAT
 362                                 gl->matched_range.rm_so = 0;
 363                                 gl->matched_range.rm_eo = 0;
 364 #endif
 365                                 if (
 366 #if !ENABLE_EXTRA_COMPAT
 367                                         regexec(&gl->compiled_regex, line, 1, &gl->matched_range, 0) == 0
 368 #else
 369                                         re_search(&gl->compiled_regex, line, line_len,
 370                                                         /*start:*/ 0, /*range:*/ line_len,
 371                                                         &gl->matched_range) >= 0
 372 #endif
 373                                 ) {
 374                                         if (!(option_mask32 & OPT_w))
 375                                                 found = 1;
 376                                         else {
 377                                                 char c = ' ';
 378                                                 if (gl->matched_range.rm_so)
 379                                                         c = line[gl->matched_range.rm_so - 1];
 380                                                 if (!isalnum(c) && c != '_') {
 381                                                         c = line[gl->matched_range.rm_eo];
 382                                                         if (!c || (!isalnum(c) && c != '_'))
 383                                                                 found = 1;
 384                                                 }
 385                                         }
 386                                 }
 387                         }
 388                         /* If it's non-inverted search, we can stop
 389                          * at first match */
 390                         if (found && !invert_search)
 391                                 goto do_found;
 392                         pattern_ptr = pattern_ptr->link;
 393                 } /* while (pattern_ptr) */
 394
 395                 if (found ^ invert_search) {
 396  do_found:
 397                         /* keep track of matches */
 398                         nmatches++;
 399
 400                         /* quiet/print (non)matching file names only? */
 401                         if (option_mask32 & (OPT_q|OPT_l|OPT_L)) {
 402                                 free(line); /* we don't need line anymore */
 403                                 if (BE_QUIET) {
 404                                         /* manpage says about -q:
 405                                          * "exit immediately with zero status
 406                                          * if any match is found,
 407                                          * even if errors were detected" */
 408                                         exit(EXIT_SUCCESS);
 409                                 }
 410                                 /* if we're just printing filenames, we stop after the first match */
 411                                 if (PRINT_FILES_WITH_MATCHES) {
 412                                         puts(cur_file);
 413                                         /* fall through to "return 1" */
 414                                 }
 415                                 /* OPT_L aka PRINT_FILES_WITHOUT_MATCHES: return early */
 416                                 return 1; /* one match */
 417                         }
 418
 419 #if ENABLE_FEATURE_GREP_CONTEXT
 420                         /* Were we printing context and saw next (unwanted) match? */
 421                         if ((option_mask32 & OPT_m) && nmatches > max_matches)
 422                                 break;
 423 #endif
 424
 425                         /* print the matched line */
 426                         if (PRINT_MATCH_COUNTS == 0) {
 427 #if ENABLE_FEATURE_GREP_CONTEXT
 428                                 int prevpos = (curpos == 0) ? lines_before - 1 : curpos - 1;
 429
 430                                 /* if we were told to print 'before' lines and there is at least
 431                                  * one line in the circular buffer, print them */
 432                                 if (lines_before && before_buf[prevpos] != NULL) {
 433                                         int first_buf_entry_line_num = linenum - lines_before;
 434
 435                                         /* advance to the first entry in the circular buffer, and
 436                                          * figure out the line number is of the first line in the
 437                                          * buffer */
 438                                         idx = curpos;
 439                                         while (before_buf[idx] == NULL) {
 440                                                 idx = (idx + 1) % lines_before;
 441                                                 first_buf_entry_line_num++;
 442                                         }
 443
 444                                         /* now print each line in the buffer, clearing them as we go */
 445                                         while (before_buf[idx] != NULL) {
 446                                                 print_line(before_buf[idx], before_buf_size[idx], first_buf_entry_line_num, '-');
 447                                                 free(before_buf[idx]);
 448                                                 before_buf[idx] = NULL;
 449                                                 idx = (idx + 1) % lines_before;
 450                                                 first_buf_entry_line_num++;
 451                                         }
 452                                 }
 453
 454                                 /* make a note that we need to print 'after' lines */
 455                                 print_n_lines_after = lines_after;
 456 #endif
 457                                 if (option_mask32 & OPT_o) {
 458                                         if (FGREP_FLAG) {
 459                                                 /* -Fo just prints the pattern
 460                                                  * (unless -v: -Fov doesnt print anything at all) */
 461                                                 if (found)
 462                                                         print_line(gl->pattern, strlen(gl->pattern), linenum, ':');
 463                                         } else while (1) {
 464                                                 unsigned start = gl->matched_range.rm_so;
 465                                                 unsigned end = gl->matched_range.rm_eo;
 466                                                 unsigned len = end - start;
 467                                                 char old = line[end];
 468                                                 line[end] = '\0';
 469                                                 /* Empty match is not printed: try "echo test | grep -o ''" */
 470                                                 if (len != 0)
 471                                                         print_line(line + start, len, linenum, ':');
 472                                                 if (old == '\0')
 473                                                         break;
 474                                                 line[end] = old;
 475                                                 if (len == 0)
 476                                                         end++;
 477 #if !ENABLE_EXTRA_COMPAT
 478                                                 if (regexec(&gl->compiled_regex, line + end,
 479                                                                 1, &gl->matched_range, REG_NOTBOL) != 0)
 480                                                         break;
 481                                                 gl->matched_range.rm_so += end;
 482                                                 gl->matched_range.rm_eo += end;
 483 #else
 484                                                 if (re_search(&gl->compiled_regex, line, line_len,
 485                                                                 end, line_len - end,
 486                                                                 &gl->matched_range) < 0)
 487                                                         break;
 488 #endif
 489                                         }
 490                                 } else {
 491                                         print_line(line, line_len, linenum, ':');
 492                                 }
 493                         }
 494                 }
 495 #if ENABLE_FEATURE_GREP_CONTEXT
 496                 else { /* no match */
 497                         /* if we need to print some context lines after the last match, do so */
 498                         if (print_n_lines_after) {
 499                                 print_line(line, strlen(line), linenum, '-');
 500                                 print_n_lines_after--;
 501                         } else if (lines_before) {
 502                                 /* Add the line to the circular 'before' buffer */
 503                                 free(before_buf[curpos]);
 504                                 before_buf[curpos] = line;
 505                                 IF_EXTRA_COMPAT(before_buf_size[curpos] = line_len;)
 506                                 curpos = (curpos + 1) % lines_before;
 507                                 /* avoid free(line) - we took the line */
 508                                 line = NULL;
 509                         }
 510                 }
 511
 512 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
 513 #if !ENABLE_EXTRA_COMPAT
 514                 free(line);
 515 #endif
 516                 /* Did we print all context after last requested match? */
 517                 if ((option_mask32 & OPT_m)
 518                  && !print_n_lines_after
 519                  && nmatches == max_matches
 520                 ) {
 521                         break;
 522                 }
 523         } /* while (read line) */
 524
 525         /* special-case file post-processing for options where we don't print line
 526          * matches, just filenames and possibly match counts */
 527
 528         /* grep -c: print [filename:]count, even if count is zero */
 529         if (PRINT_MATCH_COUNTS) {
 530                 if (print_filename)
 531                         printf("%s:", cur_file);
 532                 printf("%d\n", nmatches);
 533         }
 534
 535         /* grep -L: print just the filename */
 536         if (PRINT_FILES_WITHOUT_MATCHES) {
 537                 /* nmatches is zero, no need to check it:
 538                  * we return 1 early if we detected a match
 539                  * and PRINT_FILES_WITHOUT_MATCHES is set */
 540                 puts(cur_file);
 541         }
 542
 543         return nmatches;
 544 }
 545
 546 #if ENABLE_FEATURE_CLEAN_UP
 547 #define new_grep_list_data(p, m) add_grep_list_data(p, m)
 548 static char *add_grep_list_data(char *pattern, int flg_used_mem)
 549 #else
 550 #define new_grep_list_data(p, m) add_grep_list_data(p)
 551 static char *add_grep_list_data(char *pattern)
 552 #endif
 553 {
 554         grep_list_data_t *gl = xzalloc(sizeof(*gl));
 555         gl->pattern = pattern;
 556 #if ENABLE_FEATURE_CLEAN_UP
 557         gl->flg_mem_alocated_compiled = flg_used_mem;
 558 #else
 559         /*gl->flg_mem_alocated_compiled = 0;*/
 560 #endif
 561         return (char *)gl;
 562 }
 563
 564 static void load_regexes_from_file(llist_t *fopt)
 565 {
 566         char *line;
 567         FILE *f;
 568
 569         while (fopt) {
 570                 llist_t *cur = fopt;
 571                 char *ffile = cur->data;
 572
 573                 fopt = cur->link;
 574                 free(cur);
 575                 f = xfopen_stdin(ffile);
 576                 while ((line = xmalloc_fgetline(f)) != NULL) {
 577                         llist_add_to(&pattern_head,
 578                                 new_grep_list_data(line, ALLOCATED));
 579                 }
 580         }
 581 }
 582
 583 static int FAST_FUNC file_action_grep(const char *filename,
 584                         struct stat *statbuf UNUSED_PARAM,
 585                         void* matched,
 586                         int depth UNUSED_PARAM)
 587 {
 588         FILE *file = fopen_for_read(filename);
 589         if (file == NULL) {
 590                 if (!SUPPRESS_ERR_MSGS)
 591                         bb_simple_perror_msg(filename);
 592                 open_errors = 1;
 593                 return 0;
 594         }
 595         cur_file = filename;
 596         *(int*)matched += grep_file(file);
 597         fclose(file);
 598         return 1;
 599 }
 600
 601 static int grep_dir(const char *dir)
 602 {
 603         int matched = 0;
 604         recursive_action(dir,
 605                 /* recurse=yes */ ACTION_RECURSE |
 606                 /* followLinks=no */
 607                 /* depthFirst=yes */ ACTION_DEPTHFIRST,
 608                 /* fileAction= */ file_action_grep,
 609                 /* dirAction= */ NULL,
 610                 /* userData= */ &matched,
 611                 /* depth= */ 0);
 612         return matched;
 613 }
 614
 615 int grep_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 616 int grep_main(int argc UNUSED_PARAM, char **argv)
 617 {
 618         FILE *file;
 619         int matched;
 620         llist_t *fopt = NULL;
 621
 622         /* do normal option parsing */
 623 #if ENABLE_FEATURE_GREP_CONTEXT
 624         int Copt, opts;
 625
 626         /* -H unsets -h; -C unsets -A,-B; -e,-f are lists;
 627          * -m,-A,-B,-C have numeric param */
 628         opt_complementary = "H-h:C-AB:e::f::m+:A+:B+:C+";
 629         opts = getopt32(argv,
 630                 OPTSTR_GREP,
 631                 &pattern_head, &fopt, &max_matches,
 632                 &lines_after, &lines_before, &Copt);
 633
 634         if (opts & OPT_C) {
 635                 /* -C unsets prev -A and -B, but following -A or -B
 636                    may override it */
 637                 if (!(opts & OPT_A)) /* not overridden */
 638                         lines_after = Copt;
 639                 if (!(opts & OPT_B)) /* not overridden */
 640                         lines_before = Copt;
 641         }
 642         /* sanity checks */
 643         if (opts & (OPT_c|OPT_q|OPT_l|OPT_L)) {
 644                 option_mask32 &= ~OPT_n;
 645                 lines_before = 0;
 646                 lines_after = 0;
 647         } else if (lines_before > 0) {
 648                 if (lines_before > INT_MAX / sizeof(long long))
 649                         lines_before = INT_MAX / sizeof(long long);
 650                 /* overflow in (lines_before * sizeof(x)) is prevented (above) */
 651                 before_buf = xzalloc(lines_before * sizeof(before_buf[0]));
 652                 IF_EXTRA_COMPAT(before_buf_size = xzalloc(lines_before * sizeof(before_buf_size[0]));)
 653         }
 654 #else
 655         /* with auto sanity checks */
 656         /* -H unsets -h; -c,-q or -l unset -n; -e,-f are lists; -m N */
 657         opt_complementary = "H-h:c-n:q-n:l-n:e::f::m+";
 658         getopt32(argv, OPTSTR_GREP,
 659                 &pattern_head, &fopt, &max_matches);
 660 #endif
 661         invert_search = ((option_mask32 & OPT_v) != 0); /* 0 | 1 */
 662
 663         if (pattern_head != NULL) {
 664                 /* convert char **argv to grep_list_data_t */
 665                 llist_t *cur;
 666
 667                 for (cur = pattern_head; cur; cur = cur->link)
 668                         cur->data = new_grep_list_data(cur->data, 0);
 669         }
 670         if (option_mask32 & OPT_f)
 671                 load_regexes_from_file(fopt);
 672
 673         if (ENABLE_FEATURE_GREP_FGREP_ALIAS && applet_name[0] == 'f')
 674                 option_mask32 |= OPT_F;
 675
 676 #if !ENABLE_EXTRA_COMPAT
 677         if (!(option_mask32 & (OPT_o | OPT_w)))
 678                 reflags = REG_NOSUB;
 679 #endif
 680
 681         if (ENABLE_FEATURE_GREP_EGREP_ALIAS
 682          && (applet_name[0] == 'e' || (option_mask32 & OPT_E))
 683         ) {
 684                 reflags |= REG_EXTENDED;
 685         }
 686 #if ENABLE_EXTRA_COMPAT
 687         else {
 688                 reflags = RE_SYNTAX_GREP;
 689         }
 690 #endif
 691
 692         if (option_mask32 & OPT_i) {
 693 #if !ENABLE_EXTRA_COMPAT
 694                 reflags |= REG_ICASE;
 695 #else
 696                 int i;
 697                 case_fold = xmalloc(256);
 698                 for (i = 0; i < 256; i++)
 699                         case_fold[i] = (unsigned char)i;
 700                 for (i = 'a'; i <= 'z'; i++)
 701                         case_fold[i] = (unsigned char)(i - ('a' - 'A'));
 702 #endif
 703         }
 704
 705         argv += optind;
 706
 707         /* if we didn't get a pattern from -e and no command file was specified,
 708          * first parameter should be the pattern. no pattern, no worky */
 709         if (pattern_head == NULL) {
 710                 char *pattern;
 711                 if (*argv == NULL)
 712                         bb_show_usage();
 713                 pattern = new_grep_list_data(*argv++, 0);
 714                 llist_add_to(&pattern_head, pattern);
 715         }
 716
 717         /* argv[0..(argc-1)] should be names of file to grep through. If
 718          * there is more than one file to grep, we will print the filenames. */
 719         if (argv[0] && argv[1])
 720                 print_filename = 1;
 721         /* -H / -h of course override */
 722         if (option_mask32 & OPT_H)
 723                 print_filename = 1;
 724         if (option_mask32 & OPT_h)
 725                 print_filename = 0;
 726
 727         /* If no files were specified, or '-' was specified, take input from
 728          * stdin. Otherwise, we grep through all the files specified. */
 729         matched = 0;
 730         do {
 731                 cur_file = *argv;
 732                 file = stdin;
 733                 if (!cur_file || LONE_DASH(cur_file)) {
 734                         cur_file = "(standard input)";
 735                 } else {
 736                         if (option_mask32 & OPT_r) {
 737                                 struct stat st;
 738                                 if (stat(cur_file, &st) == 0 && S_ISDIR(st.st_mode)) {
 739                                         if (!(option_mask32 & OPT_h))
 740                                                 print_filename = 1;
 741                                         matched += grep_dir(cur_file);
 742                                         goto grep_done;
 743                                 }
 744                         }
 745                         /* else: fopen(dir) will succeed, but reading won't */
 746                         file = fopen_for_read(cur_file);
 747                         if (file == NULL) {
 748                                 if (!SUPPRESS_ERR_MSGS)
 749                                         bb_simple_perror_msg(cur_file);
 750                                 open_errors = 1;
 751                                 continue;
 752                         }
 753                 }
 754                 matched += grep_file(file);
 755                 fclose_if_not_stdin(file);
 756  grep_done: ;
 757         } while (*argv && *++argv);
 758
 759         /* destroy all the elments in the pattern list */
 760         if (ENABLE_FEATURE_CLEAN_UP) {
 761                 while (pattern_head) {
 762                         llist_t *pattern_head_ptr = pattern_head;
 763                         grep_list_data_t *gl = (grep_list_data_t *)pattern_head_ptr->data;
 764
 765                         pattern_head = pattern_head->link;
 766                         if (gl->flg_mem_alocated_compiled & ALLOCATED)
 767                                 free(gl->pattern);
 768                         if (gl->flg_mem_alocated_compiled & COMPILED)
 769                                 regfree(&gl->compiled_regex);
 770                         free(gl);
 771                         free(pattern_head_ptr);
 772                 }
 773         }
 774         /* 0 = success, 1 = failed, 2 = error */
 775         if (open_errors)
 776                 return 2;
 777         return !matched; /* invert return value: 0 = success, 1 = failed */
 778 }