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 //config:config GREP
  21 //config:       bool "grep (8.6 kb)"
  22 //config:       default y
  23 //config:       help
  24 //config:       grep is used to search files for a specified pattern.
  25 //config:
  26 //config:config EGREP
  27 //config:       bool "egrep (7.8 kb)"
  28 //config:       default y
  29 //config:       help
  30 //config:       Alias to "grep -E".
  31 //config:
  32 //config:config FGREP
  33 //config:       bool "fgrep (7.8 kb)"
  34 //config:       default y
  35 //config:       help
  36 //config:       Alias to "grep -F".
  37 //config:
  38 //config:config FEATURE_GREP_CONTEXT
  39 //config:       bool "Enable before and after context flags (-A, -B and -C)"
  40 //config:       default y
  41 //config:       depends on GREP || EGREP || FGREP
  42 //config:       help
  43 //config:       Print the specified number of leading (-B) and/or trailing (-A)
  44 //config:       context surrounding our matching lines.
  45 //config:       Print the specified number of context lines (-C).
  46
  47 //applet:IF_GREP(APPLET(grep, BB_DIR_BIN, BB_SUID_DROP))
  48 //                APPLET_ODDNAME:name   main  location    suid_type     help
  49 //applet:IF_EGREP(APPLET_ODDNAME(egrep, grep, BB_DIR_BIN, BB_SUID_DROP, egrep))
  50 //applet:IF_FGREP(APPLET_ODDNAME(fgrep, grep, BB_DIR_BIN, BB_SUID_DROP, fgrep))
  51
  52 //kbuild:lib-$(CONFIG_GREP) += grep.o
  53 //kbuild:lib-$(CONFIG_EGREP) += grep.o
  54 //kbuild:lib-$(CONFIG_FGREP) += grep.o
  55
  56 #include "libbb.h"
  57 #include "common_bufsiz.h"
  58 #include "xregex.h"
  59
  60
  61 /* options */
  62 //usage:#define grep_trivial_usage
  63 //usage:       "[-HhnlLoqvsrRiwFE"
  64 //usage:        IF_EXTRA_COMPAT("z")
  65 //usage:       "] [-m N] "
  66 //usage:        IF_FEATURE_GREP_CONTEXT("[-A/B/C N] ")
  67 //usage:       "PATTERN/-e PATTERN.../-f FILE [FILE]..."
  68 //usage:#define grep_full_usage "\n\n"
  69 //usage:       "Search for PATTERN in FILEs (or stdin)\n"
  70 //usage:     "\n        -H      Add 'filename:' prefix"
  71 //usage:     "\n        -h      Do not add 'filename:' prefix"
  72 //usage:     "\n        -n      Add 'line_no:' prefix"
  73 //usage:     "\n        -l      Show only names of files that match"
  74 //usage:     "\n        -L      Show only names of files that don't match"
  75 //usage:     "\n        -c      Show only count of matching lines"
  76 //usage:     "\n        -o      Show only the matching part of line"
  77 //usage:     "\n        -q      Quiet. Return 0 if PATTERN is found, 1 otherwise"
  78 //usage:     "\n        -v      Select non-matching lines"
  79 //usage:     "\n        -s      Suppress open and read errors"
  80 //usage:     "\n        -r      Recurse"
  81 //usage:     "\n        -R      Recurse and dereference symlinks"
  82 //usage:     "\n        -i      Ignore case"
  83 //usage:     "\n        -w      Match whole words only"
  84 //usage:     "\n        -x      Match whole lines only"
  85 //usage:     "\n        -F      PATTERN is a literal (not regexp)"
  86 //usage:     "\n        -E      PATTERN is an extended regexp"
  87 //usage:        IF_EXTRA_COMPAT(
  88 //usage:     "\n        -z      Input is NUL terminated"
  89 //usage:        )
  90 //usage:     "\n        -m N    Match up to N times per file"
  91 //usage:        IF_FEATURE_GREP_CONTEXT(
  92 //usage:     "\n        -A N    Print N lines of trailing context"
  93 //usage:     "\n        -B N    Print N lines of leading context"
  94 //usage:     "\n        -C N    Same as '-A N -B N'"
  95 //usage:        )
  96 //usage:     "\n        -e PTRN Pattern to match"
  97 //usage:     "\n        -f FILE Read pattern from file"
  98 //usage:
  99 //usage:#define grep_example_usage
 100 //usage:       "$ grep root /etc/passwd\n"
 101 //usage:       "root:x:0:0:root:/root:/bin/bash\n"
 102 //usage:       "$ grep ^[rR]oo. /etc/passwd\n"
 103 //usage:       "root:x:0:0:root:/root:/bin/bash\n"
 104 //usage:
 105 //usage:#define egrep_trivial_usage NOUSAGE_STR
 106 //usage:#define egrep_full_usage ""
 107 //usage:#define fgrep_trivial_usage NOUSAGE_STR
 108 //usage:#define fgrep_full_usage ""
 109
 110 /* -e,-f are lists; -m,-A,-B,-C have numeric param */
 111 #define OPTSTR_GREP \
 112         "lnqvscFiHhe:*f:*LorRm:+wx" \
 113         IF_FEATURE_GREP_CONTEXT("A:+B:+C:+") \
 114         "E" \
 115         IF_EXTRA_COMPAT("z") \
 116         "aI"
 117 /* ignored: -a "assume all files to be text" */
 118 /* ignored: -I "assume binary files have no matches" */
 119 enum {
 120         OPTBIT_l, /* list matched file names only */
 121         OPTBIT_n, /* print line# */
 122         OPTBIT_q, /* quiet - exit(EXIT_SUCCESS) of first match */
 123         OPTBIT_v, /* invert the match, to select non-matching lines */
 124         OPTBIT_s, /* suppress errors about file open errors */
 125         OPTBIT_c, /* count matches per file (suppresses normal output) */
 126         OPTBIT_F, /* literal match */
 127         OPTBIT_i, /* case-insensitive */
 128         OPTBIT_H, /* force filename display */
 129         OPTBIT_h, /* inhibit filename display */
 130         OPTBIT_e, /* -e PATTERN */
 131         OPTBIT_f, /* -f FILE_WITH_PATTERNS */
 132         OPTBIT_L, /* list unmatched file names only */
 133         OPTBIT_o, /* show only matching parts of lines */
 134         OPTBIT_r, /* recurse dirs */
 135         OPTBIT_R, /* recurse dirs and symlinks to dirs */
 136         OPTBIT_m, /* -m MAX_MATCHES */
 137         OPTBIT_w, /* -w whole word match */
 138         OPTBIT_x, /* -x whole line match */
 139         IF_FEATURE_GREP_CONTEXT(    OPTBIT_A ,) /* -A NUM: after-match context */
 140         IF_FEATURE_GREP_CONTEXT(    OPTBIT_B ,) /* -B NUM: before-match context */
 141         IF_FEATURE_GREP_CONTEXT(    OPTBIT_C ,) /* -C NUM: -A and -B combined */
 142         OPTBIT_E, /* extended regexp */
 143         IF_EXTRA_COMPAT(            OPTBIT_z ,) /* input is NUL terminated */
 144         OPT_l = 1 << OPTBIT_l,
 145         OPT_n = 1 << OPTBIT_n,
 146         OPT_q = 1 << OPTBIT_q,
 147         OPT_v = 1 << OPTBIT_v,
 148         OPT_s = 1 << OPTBIT_s,
 149         OPT_c = 1 << OPTBIT_c,
 150         OPT_F = 1 << OPTBIT_F,
 151         OPT_i = 1 << OPTBIT_i,
 152         OPT_H = 1 << OPTBIT_H,
 153         OPT_h = 1 << OPTBIT_h,
 154         OPT_e = 1 << OPTBIT_e,
 155         OPT_f = 1 << OPTBIT_f,
 156         OPT_L = 1 << OPTBIT_L,
 157         OPT_o = 1 << OPTBIT_o,
 158         OPT_r = 1 << OPTBIT_r,
 159         OPT_R = 1 << OPTBIT_R,
 160         OPT_m = 1 << OPTBIT_m,
 161         OPT_w = 1 << OPTBIT_w,
 162         OPT_x = 1 << OPTBIT_x,
 163         OPT_A = IF_FEATURE_GREP_CONTEXT(    (1 << OPTBIT_A)) + 0,
 164         OPT_B = IF_FEATURE_GREP_CONTEXT(    (1 << OPTBIT_B)) + 0,
 165         OPT_C = IF_FEATURE_GREP_CONTEXT(    (1 << OPTBIT_C)) + 0,
 166         OPT_E = 1 << OPTBIT_E,
 167         OPT_z = IF_EXTRA_COMPAT(            (1 << OPTBIT_z)) + 0,
 168 };
 169
 170 #define PRINT_FILES_WITH_MATCHES    (option_mask32 & OPT_l)
 171 #define PRINT_LINE_NUM              (option_mask32 & OPT_n)
 172 #define BE_QUIET                    (option_mask32 & OPT_q)
 173 #define SUPPRESS_ERR_MSGS           (option_mask32 & OPT_s)
 174 #define PRINT_MATCH_COUNTS          (option_mask32 & OPT_c)
 175 #define FGREP_FLAG                  (option_mask32 & OPT_F)
 176 #define PRINT_FILES_WITHOUT_MATCHES (option_mask32 & OPT_L)
 177 #define NUL_DELIMITED               (option_mask32 & OPT_z)
 178
 179 struct globals {
 180         int max_matches;
 181 #if !ENABLE_EXTRA_COMPAT
 182         int reflags;
 183 #else
 184         RE_TRANSLATE_TYPE case_fold; /* RE_TRANSLATE_TYPE is [[un]signed] char* */
 185 #endif
 186         smalluint invert_search;
 187         smalluint print_filename;
 188         smalluint open_errors;
 189 #if ENABLE_FEATURE_GREP_CONTEXT
 190         smalluint did_print_line;
 191         int lines_before;
 192         int lines_after;
 193         char **before_buf;
 194         IF_EXTRA_COMPAT(size_t *before_buf_size;)
 195         int last_line_printed;
 196 #endif
 197         /* globals used internally */
 198         llist_t *pattern_head;   /* growable list of patterns to match */
 199         const char *cur_file;    /* the current file we are reading */
 200 } FIX_ALIASING;
 201 #define G (*(struct globals*)bb_common_bufsiz1)
 202 #define INIT_G() do { \
 203         setup_common_bufsiz(); \
 204         BUILD_BUG_ON(sizeof(G) > COMMON_BUFSIZE); \
 205 } while (0)
 206 #define max_matches       (G.max_matches         )
 207 #if !ENABLE_EXTRA_COMPAT
 208 # define reflags          (G.reflags             )
 209 #else
 210 # define case_fold        (G.case_fold           )
 211 /* http://www.delorie.com/gnu/docs/regex/regex_46.html */
 212 # define reflags           re_syntax_options
 213 # undef REG_NOSUB
 214 # undef REG_EXTENDED
 215 # undef REG_ICASE
 216 # define REG_NOSUB    bug:is:here /* should not be used */
 217 /* Just RE_SYNTAX_EGREP is not enough, need to enable {n[,[m]]} too */
 218 # define REG_EXTENDED (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES)
 219 # define REG_ICASE    bug:is:here /* should not be used */
 220 #endif
 221 #define invert_search     (G.invert_search       )
 222 #define print_filename    (G.print_filename      )
 223 #define open_errors       (G.open_errors         )
 224 #define did_print_line    (G.did_print_line      )
 225 #define lines_before      (G.lines_before        )
 226 #define lines_after       (G.lines_after         )
 227 #define before_buf        (G.before_buf          )
 228 #define before_buf_size   (G.before_buf_size     )
 229 #define last_line_printed (G.last_line_printed   )
 230 #define pattern_head      (G.pattern_head        )
 231 #define cur_file          (G.cur_file            )
 232
 233
 234 typedef struct grep_list_data_t {
 235         char *pattern;
 236 /* for GNU regex, matched_range must be persistent across grep_file() calls */
 237 #if !ENABLE_EXTRA_COMPAT
 238         regex_t compiled_regex;
 239         regmatch_t matched_range;
 240 #else
 241         struct re_pattern_buffer compiled_regex;
 242         struct re_registers matched_range;
 243 #endif
 244 #define ALLOCATED 1
 245 #define COMPILED 2
 246         int flg_mem_allocated_compiled;
 247 } grep_list_data_t;
 248
 249 #if !ENABLE_EXTRA_COMPAT
 250 #define print_line(line, line_len, linenum, decoration) \
 251         print_line(line, linenum, decoration)
 252 #endif
 253 static void print_line(const char *line, size_t line_len, int linenum, char decoration)
 254 {
 255 #if ENABLE_FEATURE_GREP_CONTEXT
 256         /* Happens when we go to next file, immediately hit match
 257          * and try to print prev context... from prev file! Don't do it */
 258         if (linenum < 1)
 259                 return;
 260         /* possibly print the little '--' separator */
 261         if ((lines_before || lines_after) && did_print_line
 262          && last_line_printed != linenum - 1
 263         ) {
 264                 puts("--");
 265         }
 266         /* guard against printing "--" before first line of first file */
 267         did_print_line = 1;
 268         last_line_printed = linenum;
 269 #endif
 270         if (print_filename)
 271                 printf("%s%c", cur_file, decoration);
 272         if (PRINT_LINE_NUM)
 273                 printf("%i%c", linenum, decoration);
 274         /* Emulate weird GNU grep behavior with -ov */
 275         if ((option_mask32 & (OPT_v|OPT_o)) != (OPT_v|OPT_o)) {
 276 #if !ENABLE_EXTRA_COMPAT
 277                 puts(line);
 278 #else
 279                 fwrite(line, 1, line_len, stdout);
 280                 putchar(NUL_DELIMITED ? '\0' : '\n');
 281 #endif
 282         }
 283 }
 284
 285 #if ENABLE_EXTRA_COMPAT
 286 /* Unlike getline, this one removes trailing '\n' */
 287 static ssize_t FAST_FUNC bb_getline(char **line_ptr, size_t *line_alloc_len, FILE *file)
 288 {
 289         ssize_t res_sz;
 290         char *line;
 291         int delim = (NUL_DELIMITED ? '\0' : '\n');
 292
 293         res_sz = getdelim(line_ptr, line_alloc_len, delim, file);
 294         line = *line_ptr;
 295
 296         if (res_sz > 0) {
 297                 if (line[res_sz - 1] == delim)
 298                         line[--res_sz] = '\0';
 299         } else {
 300                 free(line); /* uclibc allocates a buffer even on EOF. WTF? */
 301         }
 302         return res_sz;
 303 }
 304 #endif
 305
 306 static int grep_file(FILE *file)
 307 {
 308         smalluint found;
 309         int linenum = 0;
 310         int nmatches = 0;
 311 #if !ENABLE_EXTRA_COMPAT
 312         char *line;
 313 #else
 314         char *line = NULL;
 315         ssize_t line_len;
 316         size_t line_alloc_len;
 317 # define rm_so start[0]
 318 # define rm_eo end[0]
 319 #endif
 320 #if ENABLE_FEATURE_GREP_CONTEXT
 321         int print_n_lines_after = 0;
 322         int curpos = 0; /* track where we are in the circular 'before' buffer */
 323         int idx = 0; /* used for iteration through the circular buffer */
 324 #else
 325         enum { print_n_lines_after = 0 };
 326 #endif
 327
 328         while (
 329 #if !ENABLE_EXTRA_COMPAT
 330                 (line = xmalloc_fgetline(file)) != NULL
 331 #else
 332                 (line_len = bb_getline(&line, &line_alloc_len, file)) >= 0
 333 #endif
 334         ) {
 335                 llist_t *pattern_ptr = pattern_head;
 336                 grep_list_data_t *gl = gl; /* for gcc */
 337
 338                 linenum++;
 339                 found = 0;
 340                 while (pattern_ptr) {
 341                         gl = (grep_list_data_t *)pattern_ptr->data;
 342                         if (FGREP_FLAG) {
 343                                 char *match;
 344                                 char *str = line;
 345  opt_f_again:
 346                                 match = ((option_mask32 & OPT_i)
 347                                         ? strcasestr(str, gl->pattern)
 348                                         : strstr(str, gl->pattern)
 349                                         );
 350                                 if (match) {
 351                                         if (option_mask32 & OPT_x) {
 352                                                 if (match != str)
 353                                                         goto opt_f_not_found;
 354                                                 if (str[strlen(gl->pattern)] != '\0')
 355                                                         goto opt_f_not_found;
 356                                         } else
 357                                         if (option_mask32 & OPT_w) {
 358                                                 char c = (match != line) ? match[-1] : ' ';
 359                                                 if (!isalnum(c) && c != '_') {
 360                                                         c = match[strlen(gl->pattern)];
 361                                                         if (!c || (!isalnum(c) && c != '_'))
 362                                                                 goto opt_f_found;
 363                                                 }
 364                                                 str = match + 1;
 365                                                 goto opt_f_again;
 366                                         }
 367  opt_f_found:
 368                                         found = 1;
 369  opt_f_not_found: ;
 370                                 }
 371                         } else {
 372 #if ENABLE_EXTRA_COMPAT
 373                                 unsigned start_pos;
 374 #else
 375                                 int match_flg;
 376 #endif
 377                                 char *match_at;
 378
 379                                 if (!(gl->flg_mem_allocated_compiled & COMPILED)) {
 380                                         gl->flg_mem_allocated_compiled |= COMPILED;
 381 #if !ENABLE_EXTRA_COMPAT
 382                                         xregcomp(&gl->compiled_regex, gl->pattern, reflags);
 383 #else
 384                                         memset(&gl->compiled_regex, 0, sizeof(gl->compiled_regex));
 385                                         gl->compiled_regex.translate = case_fold; /* for -i */
 386                                         if (re_compile_pattern(gl->pattern, strlen(gl->pattern), &gl->compiled_regex))
 387                                                 bb_error_msg_and_die("bad regex '%s'", gl->pattern);
 388 #endif
 389                                 }
 390 #if !ENABLE_EXTRA_COMPAT
 391                                 gl->matched_range.rm_so = 0;
 392                                 gl->matched_range.rm_eo = 0;
 393                                 match_flg = 0;
 394 #else
 395                                 start_pos = 0;
 396 #endif
 397                                 match_at = line;
 398  opt_w_again:
 399 //bb_error_msg("'%s' start_pos:%d line_len:%d", match_at, start_pos, line_len);
 400                                 if (
 401 #if !ENABLE_EXTRA_COMPAT
 402                                         regexec(&gl->compiled_regex, match_at, 1, &gl->matched_range, match_flg) == 0
 403 #else
 404                                         re_search(&gl->compiled_regex, match_at, line_len,
 405                                                         start_pos, /*range:*/ line_len,
 406                                                         &gl->matched_range) >= 0
 407 #endif
 408                                 ) {
 409                                         if (option_mask32 & OPT_x) {
 410                                                 found |= (gl->matched_range.rm_so == 0
 411                                                          && match_at[gl->matched_range.rm_eo] == '\0');
 412                                         } else
 413                                         if (!(option_mask32 & OPT_w)) {
 414                                                 found = 1;
 415                                         } else {
 416                                                 char c = ' ';
 417                                                 if (match_at > line || gl->matched_range.rm_so != 0) {
 418                                                         c = match_at[gl->matched_range.rm_so - 1];
 419                                                 }
 420                                                 if (!isalnum(c) && c != '_') {
 421                                                         c = match_at[gl->matched_range.rm_eo];
 422                                                 }
 423                                                 if (!isalnum(c) && c != '_') {
 424                                                         found = 1;
 425                                                 } else {
 426                         /*
 427                          * Why check gl->matched_range.rm_eo?
 428                          * Zero-length match makes -w skip the line:
 429                          * "echo foo | grep ^" prints "foo",
 430                          * "echo foo | grep -w ^" prints nothing.
 431                          * Without such check, we can loop forever.
 432                          */
 433 #if !ENABLE_EXTRA_COMPAT
 434                                                         if (gl->matched_range.rm_eo != 0) {
 435                                                                 match_at += gl->matched_range.rm_eo;
 436                                                                 match_flg |= REG_NOTBOL;
 437                                                                 goto opt_w_again;
 438                                                         }
 439 #else
 440                                                         if (gl->matched_range.rm_eo > start_pos) {
 441                                                                 start_pos = gl->matched_range.rm_eo;
 442                                                                 goto opt_w_again;
 443                                                         }
 444 #endif
 445                                                 }
 446                                         }
 447                                 }
 448                         }
 449                         /* If it's a non-inverted search, we can stop
 450                          * at first match and report it.
 451                          * If it's an inverted search, we can move on
 452                          * to the next line of input, ignoring the
 453                          * rest of the patterns.
 454                          */
 455                         if (found) {
 456                                 //if (invert_search)
 457                                 //      goto do_not_found;
 458                                 //goto do_found;
 459                                 break; // this accomplishes both
 460                         }
 461                         pattern_ptr = pattern_ptr->link;
 462                 } /* while (pattern_ptr) */
 463
 464                 if (found ^ invert_search) {
 465  //do_found:
 466                         /* keep track of matches */
 467                         nmatches++;
 468
 469                         /* quiet/print (non)matching file names only? */
 470                         if (option_mask32 & (OPT_q|OPT_l|OPT_L)) {
 471                                 free(line); /* we don't need line anymore */
 472                                 if (BE_QUIET) {
 473                                         /* manpage says about -q:
 474                                          * "exit immediately with zero status
 475                                          * if any match is found,
 476                                          * even if errors were detected" */
 477                                         exit(EXIT_SUCCESS);
 478                                 }
 479                                 /* if we're just printing filenames, we stop after the first match */
 480                                 if (PRINT_FILES_WITH_MATCHES) {
 481                                         puts(cur_file);
 482                                         /* fall through to "return 1" */
 483                                 }
 484                                 /* OPT_L aka PRINT_FILES_WITHOUT_MATCHES: return early */
 485                                 return 1; /* one match */
 486                         }
 487
 488 #if ENABLE_FEATURE_GREP_CONTEXT
 489                         /* Were we printing context and saw next (unwanted) match? */
 490                         if ((option_mask32 & OPT_m) && nmatches > max_matches)
 491                                 break;
 492 #endif
 493
 494                         /* print the matched line */
 495                         if (PRINT_MATCH_COUNTS == 0) {
 496 #if ENABLE_FEATURE_GREP_CONTEXT
 497                                 int prevpos = (curpos == 0) ? lines_before - 1 : curpos - 1;
 498
 499                                 /* if we were told to print 'before' lines and there is at least
 500                                  * one line in the circular buffer, print them */
 501                                 if (lines_before && before_buf[prevpos] != NULL) {
 502                                         int first_buf_entry_line_num = linenum - lines_before;
 503
 504                                         /* advance to the first entry in the circular buffer, and
 505                                          * figure out the line number is of the first line in the
 506                                          * buffer */
 507                                         idx = curpos;
 508                                         while (before_buf[idx] == NULL) {
 509                                                 idx = (idx + 1) % lines_before;
 510                                                 first_buf_entry_line_num++;
 511                                         }
 512
 513                                         /* now print each line in the buffer, clearing them as we go */
 514                                         while (before_buf[idx] != NULL) {
 515                                                 print_line(before_buf[idx], before_buf_size[idx], first_buf_entry_line_num, '-');
 516                                                 free(before_buf[idx]);
 517                                                 before_buf[idx] = NULL;
 518                                                 idx = (idx + 1) % lines_before;
 519                                                 first_buf_entry_line_num++;
 520                                         }
 521                                 }
 522
 523                                 /* make a note that we need to print 'after' lines */
 524                                 print_n_lines_after = lines_after;
 525 #endif
 526                                 if (option_mask32 & OPT_o) {
 527                                         if (FGREP_FLAG) {
 528                                                 /* -Fo just prints the pattern
 529                                                  * (unless -v: -Fov doesn't print anything at all) */
 530                                                 if (found)
 531                                                         print_line(gl->pattern, strlen(gl->pattern), linenum, ':');
 532                                         } else while (1) {
 533                                                 unsigned start = gl->matched_range.rm_so;
 534                                                 unsigned end = gl->matched_range.rm_eo;
 535                                                 unsigned len = end - start;
 536                                                 char old = line[end];
 537                                                 line[end] = '\0';
 538                                                 /* Empty match is not printed: try "echo test | grep -o ''" */
 539                                                 if (len != 0)
 540                                                         print_line(line + start, len, linenum, ':');
 541                                                 if (old == '\0')
 542                                                         break;
 543                                                 line[end] = old;
 544                                                 if (len == 0)
 545                                                         end++;
 546 #if !ENABLE_EXTRA_COMPAT
 547                                                 if (regexec(&gl->compiled_regex, line + end,
 548                                                                 1, &gl->matched_range, REG_NOTBOL) != 0)
 549                                                         break;
 550                                                 gl->matched_range.rm_so += end;
 551                                                 gl->matched_range.rm_eo += end;
 552 #else
 553                                                 if (re_search(&gl->compiled_regex, line, line_len,
 554                                                                 end, line_len - end,
 555                                                                 &gl->matched_range) < 0)
 556                                                         break;
 557 #endif
 558                                         }
 559                                 } else {
 560                                         print_line(line, line_len, linenum, ':');
 561                                 }
 562                         }
 563                 }
 564 #if ENABLE_FEATURE_GREP_CONTEXT
 565                 else { /* no match */
 566  //do_not_found:
 567                         /* if we need to print some context lines after the last match, do so */
 568                         if (print_n_lines_after) {
 569                                 print_line(line, strlen(line), linenum, '-');
 570                                 print_n_lines_after--;
 571                         } else if (lines_before) {
 572                                 /* Add the line to the circular 'before' buffer */
 573                                 free(before_buf[curpos]);
 574                                 before_buf[curpos] = line;
 575                                 IF_EXTRA_COMPAT(before_buf_size[curpos] = line_len;)
 576                                 curpos = (curpos + 1) % lines_before;
 577                                 /* avoid free(line) - we took the line */
 578                                 line = NULL;
 579                         }
 580                 }
 581
 582 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
 583 #if !ENABLE_EXTRA_COMPAT
 584                 free(line);
 585 #endif
 586                 /* Did we print all context after last requested match? */
 587                 if ((option_mask32 & OPT_m)
 588                  && !print_n_lines_after
 589                  && nmatches == max_matches
 590                 ) {
 591                         break;
 592                 }
 593         } /* while (read line) */
 594
 595         /* special-case file post-processing for options where we don't print line
 596          * matches, just filenames and possibly match counts */
 597
 598         /* grep -c: print [filename:]count, even if count is zero */
 599         if (PRINT_MATCH_COUNTS) {
 600                 if (print_filename)
 601                         printf("%s:", cur_file);
 602                 printf("%d\n", nmatches);
 603         }
 604
 605         /* grep -L: print just the filename */
 606         if (PRINT_FILES_WITHOUT_MATCHES) {
 607                 /* nmatches is zero, no need to check it:
 608                  * we return 1 early if we detected a match
 609                  * and PRINT_FILES_WITHOUT_MATCHES is set */
 610                 puts(cur_file);
 611         }
 612
 613         return nmatches;
 614 }
 615
 616 #if ENABLE_FEATURE_CLEAN_UP
 617 #define new_grep_list_data(p, m) add_grep_list_data(p, m)
 618 static char *add_grep_list_data(char *pattern, int flg_used_mem)
 619 #else
 620 #define new_grep_list_data(p, m) add_grep_list_data(p)
 621 static char *add_grep_list_data(char *pattern)
 622 #endif
 623 {
 624         grep_list_data_t *gl = xzalloc(sizeof(*gl));
 625         gl->pattern = pattern;
 626 #if ENABLE_FEATURE_CLEAN_UP
 627         gl->flg_mem_allocated_compiled = flg_used_mem;
 628 #else
 629         /*gl->flg_mem_allocated_compiled = 0;*/
 630 #endif
 631         return (char *)gl;
 632 }
 633
 634 static void load_regexes_from_file(llist_t *fopt)
 635 {
 636         while (fopt) {
 637                 char *line;
 638                 FILE *fp;
 639                 llist_t *cur = fopt;
 640                 char *ffile = cur->data;
 641
 642                 fopt = cur->link;
 643                 free(cur);
 644                 fp = xfopen_stdin(ffile);
 645                 while ((line = xmalloc_fgetline(fp)) != NULL) {
 646                         llist_add_to(&pattern_head,
 647                                 new_grep_list_data(line, ALLOCATED));
 648                 }
 649                 fclose_if_not_stdin(fp);
 650         }
 651 }
 652
 653 static void load_pattern_list(llist_t **lst, char *pattern)
 654 {
 655         char *p;
 656         while ((p = strsep(&pattern, "\n")) != NULL)
 657                 llist_add_to(lst, new_grep_list_data(p, 0));
 658 }
 659
 660 static int FAST_FUNC file_action_grep(const char *filename,
 661                         struct stat *statbuf,
 662                         void* matched,
 663                         int depth UNUSED_PARAM)
 664 {
 665         FILE *file;
 666
 667         /* If we are given a link to a directory, we should bail out now, rather
 668          * than trying to open the "file" and hoping getline gives us nothing,
 669          * since that is not portable across operating systems (FreeBSD for
 670          * example will return the raw directory contents). */
 671         if (S_ISLNK(statbuf->st_mode)) {
 672                 struct stat sb;
 673                 if (stat(filename, &sb) != 0) {
 674                         if (!SUPPRESS_ERR_MSGS)
 675                                 bb_simple_perror_msg(filename);
 676                         return 0;
 677                 }
 678                 if (S_ISDIR(sb.st_mode))
 679                         return 1;
 680         }
 681
 682         file = fopen_for_read(filename);
 683         if (file == NULL) {
 684                 if (!SUPPRESS_ERR_MSGS)
 685                         bb_simple_perror_msg(filename);
 686                 open_errors = 1;
 687                 return 0;
 688         }
 689         cur_file = filename;
 690         *(int*)matched += grep_file(file);
 691         fclose(file);
 692         return 1;
 693 }
 694
 695 static int grep_dir(const char *dir)
 696 {
 697         int matched = 0;
 698         recursive_action(dir,
 699                 /* recurse=yes */ ACTION_RECURSE |
 700                 /* followLinks=always */ ((option_mask32 & OPT_R) ? ACTION_FOLLOWLINKS : 0) |
 701                 /* followLinks=command line only */ ACTION_FOLLOWLINKS_L0 |
 702                 /* depthFirst=yes */ ACTION_DEPTHFIRST,
 703                 /* fileAction= */ file_action_grep,
 704                 /* dirAction= */ NULL,
 705                 /* userData= */ &matched,
 706                 /* depth= */ 0);
 707         return matched;
 708 }
 709
 710 int grep_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 711 int grep_main(int argc UNUSED_PARAM, char **argv)
 712 {
 713         FILE *file;
 714         int matched;
 715         llist_t *fopt = NULL;
 716 #if ENABLE_FEATURE_GREP_CONTEXT
 717         int Copt, opts;
 718 #endif
 719         INIT_G();
 720
 721         /* For grep, exitcode of 1 is "not found". Other errors are 2: */
 722         xfunc_error_retval = 2;
 723
 724         /* do normal option parsing */
 725 #if ENABLE_FEATURE_GREP_CONTEXT
 726         /* -H unsets -h; -C unsets -A,-B */
 727         opts = getopt32long(argv, "^"
 728                 OPTSTR_GREP
 729                         "\0"
 730                         "H-h:C-AB",
 731                 "color\0" Optional_argument "\xff",
 732                 &pattern_head, &fopt, &max_matches,
 733                 &lines_after, &lines_before, &Copt
 734                 , NULL
 735         );
 736
 737         if (opts & OPT_C) {
 738                 /* -C unsets prev -A and -B, but following -A or -B
 739                  * may override it */
 740                 if (!(opts & OPT_A)) /* not overridden */
 741                         lines_after = Copt;
 742                 if (!(opts & OPT_B)) /* not overridden */
 743                         lines_before = Copt;
 744         }
 745         /* sanity checks */
 746         if (opts & (OPT_c|OPT_q|OPT_l|OPT_L)) {
 747                 option_mask32 &= ~OPT_n;
 748                 lines_before = 0;
 749                 lines_after = 0;
 750         } else if (lines_before > 0) {
 751                 if (lines_before > INT_MAX / sizeof(long long))
 752                         lines_before = INT_MAX / sizeof(long long);
 753                 /* overflow in (lines_before * sizeof(x)) is prevented (above) */
 754                 before_buf = xzalloc(lines_before * sizeof(before_buf[0]));
 755                 IF_EXTRA_COMPAT(before_buf_size = xzalloc(lines_before * sizeof(before_buf_size[0]));)
 756         }
 757 #else
 758         /* with auto sanity checks */
 759         getopt32(argv, "^" OPTSTR_GREP "\0" "H-h:c-n:q-n:l-n:", // why trailing ":"?
 760                 &pattern_head, &fopt, &max_matches);
 761 #endif
 762         invert_search = ((option_mask32 & OPT_v) != 0); /* 0 | 1 */
 763
 764         {       /* convert char **argv to pattern_list */
 765                 llist_t *cur, *new = NULL;
 766                 for (cur = pattern_head; cur; cur = cur->link)
 767                         load_pattern_list(&new, cur->data);
 768                 llist_free(pattern_head, NULL);
 769                 pattern_head = new;
 770         }
 771         if (option_mask32 & OPT_f) {
 772                 load_regexes_from_file(fopt);
 773                 if (!pattern_head) { /* -f EMPTY_FILE? */
 774                         /* GNU grep treats it as "nothing matches" except when -x */
 775                         const char *data = (option_mask32 & OPT_x) ? ".*" : "";
 776                         llist_add_to(&pattern_head, new_grep_list_data((char*)data, 0));
 777                         invert_search ^= 1;
 778                 }
 779         }
 780
 781         if (ENABLE_FGREP && applet_name[0] == 'f')
 782                 option_mask32 |= OPT_F;
 783
 784 #if !ENABLE_EXTRA_COMPAT
 785         if (!(option_mask32 & (OPT_o | OPT_w | OPT_x)))
 786                 reflags = REG_NOSUB;
 787 #endif
 788
 789         if ((ENABLE_EGREP && applet_name[0] == 'e')
 790          || (option_mask32 & OPT_E)
 791         ) {
 792                 reflags |= REG_EXTENDED;
 793         }
 794 #if ENABLE_EXTRA_COMPAT
 795         else {
 796                 reflags = RE_SYNTAX_GREP;
 797         }
 798 #endif
 799
 800         if (option_mask32 & OPT_i) {
 801 #if !ENABLE_EXTRA_COMPAT
 802                 reflags |= REG_ICASE;
 803 #else
 804                 int i;
 805                 case_fold = xmalloc(256);
 806                 for (i = 0; i < 256; i++)
 807                         case_fold[i] = (unsigned char)i;
 808                 for (i = 'a'; i <= 'z'; i++)
 809                         case_fold[i] = (unsigned char)(i - ('a' - 'A'));
 810 #endif
 811         }
 812
 813         argv += optind;
 814
 815         /* if we didn't get a pattern from -e and no command file was specified,
 816          * first parameter should be the pattern. no pattern, no worky */
 817         if (pattern_head == NULL) {
 818                 if (*argv == NULL)
 819                         bb_show_usage();
 820                 load_pattern_list(&pattern_head, *argv++);
 821         }
 822
 823         /* argv[0..(argc-1)] should be names of file to grep through. If
 824          * there is more than one file to grep, we will print the filenames. */
 825         if (argv[0] && argv[1])
 826                 print_filename = 1;
 827         /* -H / -h of course override */
 828         if (option_mask32 & OPT_H)
 829                 print_filename = 1;
 830         if (option_mask32 & OPT_h)
 831                 print_filename = 0;
 832
 833         /* If no files were specified, or '-' was specified, take input from
 834          * stdin. Otherwise, we grep through all the files specified. */
 835         matched = 0;
 836         do {
 837                 cur_file = *argv;
 838                 file = stdin;
 839                 if (!cur_file || LONE_DASH(cur_file)) {
 840                         cur_file = "(standard input)";
 841                 } else {
 842                         if (option_mask32 & (OPT_r|OPT_R)) {
 843                                 struct stat st;
 844                                 if (stat(cur_file, &st) == 0 && S_ISDIR(st.st_mode)) {
 845                                         if (!(option_mask32 & OPT_h))
 846                                                 print_filename = 1;
 847                                         matched += grep_dir(cur_file);
 848                                         goto grep_done;
 849                                 }
 850                         }
 851                         /* else: fopen(dir) will succeed, but reading won't */
 852                         file = fopen_for_read(cur_file);
 853                         if (file == NULL) {
 854                                 if (!SUPPRESS_ERR_MSGS)
 855                                         bb_simple_perror_msg(cur_file);
 856                                 open_errors = 1;
 857                                 continue;
 858                         }
 859                 }
 860                 matched += grep_file(file);
 861                 fclose_if_not_stdin(file);
 862  grep_done: ;
 863         } while (*argv && *++argv);
 864
 865         /* destroy all the elements in the pattern list */
 866         if (ENABLE_FEATURE_CLEAN_UP) {
 867                 while (pattern_head) {
 868                         llist_t *pattern_head_ptr = pattern_head;
 869                         grep_list_data_t *gl = (grep_list_data_t *)pattern_head_ptr->data;
 870
 871                         pattern_head = pattern_head->link;
 872                         if (gl->flg_mem_allocated_compiled & ALLOCATED)
 873                                 free(gl->pattern);
 874                         if (gl->flg_mem_allocated_compiled & COMPILED)
 875                                 regfree(&gl->compiled_regex);
 876                         free(gl);
 877                         free(pattern_head_ptr);
 878                 }
 879         }
 880         /* 0 = success, 1 = failed, 2 = error */
 881         if (open_errors)
 882                 return 2;
 883         return !matched; /* invert return value: 0 = success, 1 = failed */
 884 }