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 //config:config GREP
  22 //config:       bool "grep (8.5 kb)"
  23 //config:       default y
  24 //config:       help
  25 //config:       grep is used to search files for a specified pattern.
  26 //config:
  27 //config:config EGREP
  28 //config:       bool "egrep (7.6 kb)"
  29 //config:       default y
  30 //config:       help
  31 //config:       Alias to "grep -E".
  32 //config:
  33 //config:config FGREP
  34 //config:       bool "fgrep (7.6 kb)"
  35 //config:       default y
  36 //config:       help
  37 //config:       Alias to "grep -F".
  38 //config:
  39 //config:config FEATURE_GREP_CONTEXT
  40 //config:       bool "Enable before and after context flags (-A, -B and -C)"
  41 //config:       default y
  42 //config:       depends on GREP || EGREP || FGREP
  43 //config:       help
  44 //config:       Print the specified number of leading (-B) and/or trailing (-A)
  45 //config:       context surrounding our matching lines.
  46 //config:       Print the specified number of context lines (-C).
  47
  48 //applet:IF_GREP(APPLET(grep, BB_DIR_BIN, BB_SUID_DROP))
  49 //                APPLET_ODDNAME:name   main  location    suid_type     help
  50 //applet:IF_EGREP(APPLET_ODDNAME(egrep, grep, BB_DIR_BIN, BB_SUID_DROP, egrep))
  51 //applet:IF_FGREP(APPLET_ODDNAME(fgrep, grep, BB_DIR_BIN, BB_SUID_DROP, fgrep))
  52
  53 //kbuild:lib-$(CONFIG_GREP) += grep.o
  54 //kbuild:lib-$(CONFIG_EGREP) += grep.o
  55 //kbuild:lib-$(CONFIG_FGREP) += grep.o
  56
  57 #include "libbb.h"
  58 #include "common_bufsiz.h"
  59 #include "xregex.h"
  60
  61
  62 /* options */
  63 //usage:#define grep_trivial_usage
  64 //usage:       "[-HhnlLoqvsriwFE"
  65 //usage:        IF_EXTRA_COMPAT("z")
  66 //usage:       "] [-m N] "
  67 //usage:        IF_FEATURE_GREP_CONTEXT("[-A/B/C N] ")
  68 //usage:       "PATTERN/-e PATTERN.../-f FILE [FILE]..."
  69 //usage:#define grep_full_usage "\n\n"
  70 //usage:       "Search for PATTERN in FILEs (or stdin)\n"
  71 //usage:     "\n        -H      Add 'filename:' prefix"
  72 //usage:     "\n        -h      Do not add 'filename:' prefix"
  73 //usage:     "\n        -n      Add 'line_no:' prefix"
  74 //usage:     "\n        -l      Show only names of files that match"
  75 //usage:     "\n        -L      Show only names of files that don't match"
  76 //usage:     "\n        -c      Show only count of matching lines"
  77 //usage:     "\n        -o      Show only the matching part of line"
  78 //usage:     "\n        -q      Quiet. Return 0 if PATTERN is found, 1 otherwise"
  79 //usage:     "\n        -v      Select non-matching lines"
  80 //usage:     "\n        -s      Suppress open and read errors"
  81 //usage:     "\n        -r      Recurse"
  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:*Lorm:+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_m, /* -m MAX_MATCHES */
 136         OPTBIT_w, /* -w whole word match */
 137         OPTBIT_x, /* -x whole line match */
 138         IF_FEATURE_GREP_CONTEXT(    OPTBIT_A ,) /* -A NUM: after-match context */
 139         IF_FEATURE_GREP_CONTEXT(    OPTBIT_B ,) /* -B NUM: before-match context */
 140         IF_FEATURE_GREP_CONTEXT(    OPTBIT_C ,) /* -C NUM: -A and -B combined */
 141         OPTBIT_E, /* extended regexp */
 142         IF_EXTRA_COMPAT(            OPTBIT_z ,) /* input is NUL terminated */
 143         OPT_l = 1 << OPTBIT_l,
 144         OPT_n = 1 << OPTBIT_n,
 145         OPT_q = 1 << OPTBIT_q,
 146         OPT_v = 1 << OPTBIT_v,
 147         OPT_s = 1 << OPTBIT_s,
 148         OPT_c = 1 << OPTBIT_c,
 149         OPT_F = 1 << OPTBIT_F,
 150         OPT_i = 1 << OPTBIT_i,
 151         OPT_H = 1 << OPTBIT_H,
 152         OPT_h = 1 << OPTBIT_h,
 153         OPT_e = 1 << OPTBIT_e,
 154         OPT_f = 1 << OPTBIT_f,
 155         OPT_L = 1 << OPTBIT_L,
 156         OPT_o = 1 << OPTBIT_o,
 157         OPT_r = 1 << OPTBIT_r,
 158         OPT_m = 1 << OPTBIT_m,
 159         OPT_w = 1 << OPTBIT_w,
 160         OPT_x = 1 << OPTBIT_x,
 161         OPT_A = IF_FEATURE_GREP_CONTEXT(    (1 << OPTBIT_A)) + 0,
 162         OPT_B = IF_FEATURE_GREP_CONTEXT(    (1 << OPTBIT_B)) + 0,
 163         OPT_C = IF_FEATURE_GREP_CONTEXT(    (1 << OPTBIT_C)) + 0,
 164         OPT_E = 1 << OPTBIT_E,
 165         OPT_z = IF_EXTRA_COMPAT(            (1 << OPTBIT_z)) + 0,
 166 };
 167
 168 #define PRINT_FILES_WITH_MATCHES    (option_mask32 & OPT_l)
 169 #define PRINT_LINE_NUM              (option_mask32 & OPT_n)
 170 #define BE_QUIET                    (option_mask32 & OPT_q)
 171 #define SUPPRESS_ERR_MSGS           (option_mask32 & OPT_s)
 172 #define PRINT_MATCH_COUNTS          (option_mask32 & OPT_c)
 173 #define FGREP_FLAG                  (option_mask32 & OPT_F)
 174 #define PRINT_FILES_WITHOUT_MATCHES (option_mask32 & OPT_L)
 175 #define NUL_DELIMITED               (option_mask32 & OPT_z)
 176
 177 struct globals {
 178         int max_matches;
 179 #if !ENABLE_EXTRA_COMPAT
 180         int reflags;
 181 #else
 182         RE_TRANSLATE_TYPE case_fold; /* RE_TRANSLATE_TYPE is [[un]signed] char* */
 183 #endif
 184         smalluint invert_search;
 185         smalluint print_filename;
 186         smalluint open_errors;
 187 #if ENABLE_FEATURE_GREP_CONTEXT
 188         smalluint did_print_line;
 189         int lines_before;
 190         int lines_after;
 191         char **before_buf;
 192         IF_EXTRA_COMPAT(size_t *before_buf_size;)
 193         int last_line_printed;
 194 #endif
 195         /* globals used internally */
 196         llist_t *pattern_head;   /* growable list of patterns to match */
 197         const char *cur_file;    /* the current file we are reading */
 198 } FIX_ALIASING;
 199 #define G (*(struct globals*)bb_common_bufsiz1)
 200 #define INIT_G() do { \
 201         setup_common_bufsiz(); \
 202         BUILD_BUG_ON(sizeof(G) > COMMON_BUFSIZE); \
 203 } while (0)
 204 #define max_matches       (G.max_matches         )
 205 #if !ENABLE_EXTRA_COMPAT
 206 # define reflags          (G.reflags             )
 207 #else
 208 # define case_fold        (G.case_fold           )
 209 /* http://www.delorie.com/gnu/docs/regex/regex_46.html */
 210 # define reflags           re_syntax_options
 211 # undef REG_NOSUB
 212 # undef REG_EXTENDED
 213 # undef REG_ICASE
 214 # define REG_NOSUB    bug:is:here /* should not be used */
 215 /* Just RE_SYNTAX_EGREP is not enough, need to enable {n[,[m]]} too */
 216 # define REG_EXTENDED (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES)
 217 # define REG_ICASE    bug:is:here /* should not be used */
 218 #endif
 219 #define invert_search     (G.invert_search       )
 220 #define print_filename    (G.print_filename      )
 221 #define open_errors       (G.open_errors         )
 222 #define did_print_line    (G.did_print_line      )
 223 #define lines_before      (G.lines_before        )
 224 #define lines_after       (G.lines_after         )
 225 #define before_buf        (G.before_buf          )
 226 #define before_buf_size   (G.before_buf_size     )
 227 #define last_line_printed (G.last_line_printed   )
 228 #define pattern_head      (G.pattern_head        )
 229 #define cur_file          (G.cur_file            )
 230
 231
 232 typedef struct grep_list_data_t {
 233         char *pattern;
 234 /* for GNU regex, matched_range must be persistent across grep_file() calls */
 235 #if !ENABLE_EXTRA_COMPAT
 236         regex_t compiled_regex;
 237         regmatch_t matched_range;
 238 #else
 239         struct re_pattern_buffer compiled_regex;
 240         struct re_registers matched_range;
 241 #endif
 242 #define ALLOCATED 1
 243 #define COMPILED 2
 244         int flg_mem_allocated_compiled;
 245 } grep_list_data_t;
 246
 247 #if !ENABLE_EXTRA_COMPAT
 248 #define print_line(line, line_len, linenum, decoration) \
 249         print_line(line, linenum, decoration)
 250 #endif
 251 static void print_line(const char *line, size_t line_len, int linenum, char decoration)
 252 {
 253 #if ENABLE_FEATURE_GREP_CONTEXT
 254         /* Happens when we go to next file, immediately hit match
 255          * and try to print prev context... from prev file! Don't do it */
 256         if (linenum < 1)
 257                 return;
 258         /* possibly print the little '--' separator */
 259         if ((lines_before || lines_after) && did_print_line
 260          && last_line_printed != linenum - 1
 261         ) {
 262                 puts("--");
 263         }
 264         /* guard against printing "--" before first line of first file */
 265         did_print_line = 1;
 266         last_line_printed = linenum;
 267 #endif
 268         if (print_filename)
 269                 printf("%s%c", cur_file, decoration);
 270         if (PRINT_LINE_NUM)
 271                 printf("%i%c", linenum, decoration);
 272         /* Emulate weird GNU grep behavior with -ov */
 273         if ((option_mask32 & (OPT_v|OPT_o)) != (OPT_v|OPT_o)) {
 274 #if !ENABLE_EXTRA_COMPAT
 275                 puts(line);
 276 #else
 277                 fwrite(line, 1, line_len, stdout);
 278                 putchar(NUL_DELIMITED ? '\0' : '\n');
 279 #endif
 280         }
 281 }
 282
 283 #if ENABLE_EXTRA_COMPAT
 284 /* Unlike getline, this one removes trailing '\n' */
 285 static ssize_t FAST_FUNC bb_getline(char **line_ptr, size_t *line_alloc_len, FILE *file)
 286 {
 287         ssize_t res_sz;
 288         char *line;
 289         int delim = (NUL_DELIMITED ? '\0' : '\n');
 290
 291         res_sz = getdelim(line_ptr, line_alloc_len, delim, file);
 292         line = *line_ptr;
 293
 294         if (res_sz > 0) {
 295                 if (line[res_sz - 1] == delim)
 296                         line[--res_sz] = '\0';
 297         } else {
 298                 free(line); /* uclibc allocates a buffer even on EOF. WTF? */
 299         }
 300         return res_sz;
 301 }
 302 #endif
 303
 304 static int grep_file(FILE *file)
 305 {
 306         smalluint found;
 307         int linenum = 0;
 308         int nmatches = 0;
 309 #if !ENABLE_EXTRA_COMPAT
 310         char *line;
 311 #else
 312         char *line = NULL;
 313         ssize_t line_len;
 314         size_t line_alloc_len;
 315 # define rm_so start[0]
 316 # define rm_eo end[0]
 317 #endif
 318 #if ENABLE_FEATURE_GREP_CONTEXT
 319         int print_n_lines_after = 0;
 320         int curpos = 0; /* track where we are in the circular 'before' buffer */
 321         int idx = 0; /* used for iteration through the circular buffer */
 322 #else
 323         enum { print_n_lines_after = 0 };
 324 #endif
 325
 326         while (
 327 #if !ENABLE_EXTRA_COMPAT
 328                 (line = xmalloc_fgetline(file)) != NULL
 329 #else
 330                 (line_len = bb_getline(&line, &line_alloc_len, file)) >= 0
 331 #endif
 332         ) {
 333                 llist_t *pattern_ptr = pattern_head;
 334                 grep_list_data_t *gl = gl; /* for gcc */
 335
 336                 linenum++;
 337                 found = 0;
 338                 while (pattern_ptr) {
 339                         gl = (grep_list_data_t *)pattern_ptr->data;
 340                         if (FGREP_FLAG) {
 341                                 char *match;
 342                                 char *str = line;
 343  opt_f_again:
 344                                 match = ((option_mask32 & OPT_i)
 345                                         ? strcasestr(str, gl->pattern)
 346                                         : strstr(str, gl->pattern)
 347                                         );
 348                                 if (match) {
 349                                         if (option_mask32 & OPT_x) {
 350                                                 if (match != str)
 351                                                         goto opt_f_not_found;
 352                                                 if (str[strlen(gl->pattern)] != '\0')
 353                                                         goto opt_f_not_found;
 354                                         } else
 355                                         if (option_mask32 & OPT_w) {
 356                                                 char c = (match != str) ? match[-1] : ' ';
 357                                                 if (!isalnum(c) && c != '_') {
 358                                                         c = match[strlen(gl->pattern)];
 359                                                         if (!c || (!isalnum(c) && c != '_'))
 360                                                                 goto opt_f_found;
 361                                                 }
 362                                                 str = match + 1;
 363                                                 goto opt_f_again;
 364                                         }
 365  opt_f_found:
 366                                         found = 1;
 367  opt_f_not_found: ;
 368                                 }
 369                         } else {
 370 #if ENABLE_EXTRA_COMPAT
 371                                 unsigned start_pos;
 372 #else
 373                                 int match_flg;
 374 #endif
 375                                 char *match_at;
 376
 377                                 if (!(gl->flg_mem_allocated_compiled & COMPILED)) {
 378                                         gl->flg_mem_allocated_compiled |= COMPILED;
 379 #if !ENABLE_EXTRA_COMPAT
 380                                         xregcomp(&gl->compiled_regex, gl->pattern, reflags);
 381 #else
 382                                         memset(&gl->compiled_regex, 0, sizeof(gl->compiled_regex));
 383                                         gl->compiled_regex.translate = case_fold; /* for -i */
 384                                         if (re_compile_pattern(gl->pattern, strlen(gl->pattern), &gl->compiled_regex))
 385                                                 bb_error_msg_and_die("bad regex '%s'", gl->pattern);
 386 #endif
 387                                 }
 388 #if !ENABLE_EXTRA_COMPAT
 389                                 gl->matched_range.rm_so = 0;
 390                                 gl->matched_range.rm_eo = 0;
 391                                 match_flg = 0;
 392 #else
 393                                 start_pos = 0;
 394 #endif
 395                                 match_at = line;
 396  opt_w_again:
 397 //bb_error_msg("'%s' start_pos:%d line_len:%d", match_at, start_pos, line_len);
 398                                 if (
 399 #if !ENABLE_EXTRA_COMPAT
 400                                         regexec(&gl->compiled_regex, match_at, 1, &gl->matched_range, match_flg) == 0
 401 #else
 402                                         re_search(&gl->compiled_regex, match_at, line_len,
 403                                                         start_pos, /*range:*/ line_len,
 404                                                         &gl->matched_range) >= 0
 405 #endif
 406                                 ) {
 407                                         if (option_mask32 & OPT_x) {
 408                                                 found = (gl->matched_range.rm_so == 0
 409                                                          && match_at[gl->matched_range.rm_eo] == '\0');
 410                                         } else
 411                                         if (!(option_mask32 & OPT_w)) {
 412                                                 found = 1;
 413                                         } else {
 414                                                 char c = ' ';
 415                                                 if (match_at > line || gl->matched_range.rm_so != 0) {
 416                                                         c = match_at[gl->matched_range.rm_so - 1];
 417                                                 }
 418                                                 if (!isalnum(c) && c != '_') {
 419                                                         c = match_at[gl->matched_range.rm_eo];
 420                                                 }
 421                                                 if (!isalnum(c) && c != '_') {
 422                                                         found = 1;
 423                                                 } else {
 424                         /*
 425                          * Why check gl->matched_range.rm_eo?
 426                          * Zero-length match makes -w skip the line:
 427                          * "echo foo | grep ^" prints "foo",
 428                          * "echo foo | grep -w ^" prints nothing.
 429                          * Without such check, we can loop forever.
 430                          */
 431 #if !ENABLE_EXTRA_COMPAT
 432                                                         if (gl->matched_range.rm_eo != 0) {
 433                                                                 match_at += gl->matched_range.rm_eo;
 434                                                                 match_flg |= REG_NOTBOL;
 435                                                                 goto opt_w_again;
 436                                                         }
 437 #else
 438                                                         if (gl->matched_range.rm_eo > start_pos) {
 439                                                                 start_pos = gl->matched_range.rm_eo;
 440                                                                 goto opt_w_again;
 441                                                         }
 442 #endif
 443                                                 }
 444                                         }
 445                                 }
 446                         }
 447                         /* If it's non-inverted search, we can stop
 448                          * at first match */
 449                         if (found && !invert_search)
 450                                 goto do_found;
 451                         pattern_ptr = pattern_ptr->link;
 452                 } /* while (pattern_ptr) */
 453
 454                 if (found ^ invert_search) {
 455  do_found:
 456                         /* keep track of matches */
 457                         nmatches++;
 458
 459                         /* quiet/print (non)matching file names only? */
 460                         if (option_mask32 & (OPT_q|OPT_l|OPT_L)) {
 461                                 free(line); /* we don't need line anymore */
 462                                 if (BE_QUIET) {
 463                                         /* manpage says about -q:
 464                                          * "exit immediately with zero status
 465                                          * if any match is found,
 466                                          * even if errors were detected" */
 467                                         exit(EXIT_SUCCESS);
 468                                 }
 469                                 /* if we're just printing filenames, we stop after the first match */
 470                                 if (PRINT_FILES_WITH_MATCHES) {
 471                                         puts(cur_file);
 472                                         /* fall through to "return 1" */
 473                                 }
 474                                 /* OPT_L aka PRINT_FILES_WITHOUT_MATCHES: return early */
 475                                 return 1; /* one match */
 476                         }
 477
 478 #if ENABLE_FEATURE_GREP_CONTEXT
 479                         /* Were we printing context and saw next (unwanted) match? */
 480                         if ((option_mask32 & OPT_m) && nmatches > max_matches)
 481                                 break;
 482 #endif
 483
 484                         /* print the matched line */
 485                         if (PRINT_MATCH_COUNTS == 0) {
 486 #if ENABLE_FEATURE_GREP_CONTEXT
 487                                 int prevpos = (curpos == 0) ? lines_before - 1 : curpos - 1;
 488
 489                                 /* if we were told to print 'before' lines and there is at least
 490                                  * one line in the circular buffer, print them */
 491                                 if (lines_before && before_buf[prevpos] != NULL) {
 492                                         int first_buf_entry_line_num = linenum - lines_before;
 493
 494                                         /* advance to the first entry in the circular buffer, and
 495                                          * figure out the line number is of the first line in the
 496                                          * buffer */
 497                                         idx = curpos;
 498                                         while (before_buf[idx] == NULL) {
 499                                                 idx = (idx + 1) % lines_before;
 500                                                 first_buf_entry_line_num++;
 501                                         }
 502
 503                                         /* now print each line in the buffer, clearing them as we go */
 504                                         while (before_buf[idx] != NULL) {
 505                                                 print_line(before_buf[idx], before_buf_size[idx], first_buf_entry_line_num, '-');
 506                                                 free(before_buf[idx]);
 507                                                 before_buf[idx] = NULL;
 508                                                 idx = (idx + 1) % lines_before;
 509                                                 first_buf_entry_line_num++;
 510                                         }
 511                                 }
 512
 513                                 /* make a note that we need to print 'after' lines */
 514                                 print_n_lines_after = lines_after;
 515 #endif
 516                                 if (option_mask32 & OPT_o) {
 517                                         if (FGREP_FLAG) {
 518                                                 /* -Fo just prints the pattern
 519                                                  * (unless -v: -Fov doesn't print anything at all) */
 520                                                 if (found)
 521                                                         print_line(gl->pattern, strlen(gl->pattern), linenum, ':');
 522                                         } else while (1) {
 523                                                 unsigned start = gl->matched_range.rm_so;
 524                                                 unsigned end = gl->matched_range.rm_eo;
 525                                                 unsigned len = end - start;
 526                                                 char old = line[end];
 527                                                 line[end] = '\0';
 528                                                 /* Empty match is not printed: try "echo test | grep -o ''" */
 529                                                 if (len != 0)
 530                                                         print_line(line + start, len, linenum, ':');
 531                                                 if (old == '\0')
 532                                                         break;
 533                                                 line[end] = old;
 534                                                 if (len == 0)
 535                                                         end++;
 536 #if !ENABLE_EXTRA_COMPAT
 537                                                 if (regexec(&gl->compiled_regex, line + end,
 538                                                                 1, &gl->matched_range, REG_NOTBOL) != 0)
 539                                                         break;
 540                                                 gl->matched_range.rm_so += end;
 541                                                 gl->matched_range.rm_eo += end;
 542 #else
 543                                                 if (re_search(&gl->compiled_regex, line, line_len,
 544                                                                 end, line_len - end,
 545                                                                 &gl->matched_range) < 0)
 546                                                         break;
 547 #endif
 548                                         }
 549                                 } else {
 550                                         print_line(line, line_len, linenum, ':');
 551                                 }
 552                         }
 553                 }
 554 #if ENABLE_FEATURE_GREP_CONTEXT
 555                 else { /* no match */
 556                         /* if we need to print some context lines after the last match, do so */
 557                         if (print_n_lines_after) {
 558                                 print_line(line, strlen(line), linenum, '-');
 559                                 print_n_lines_after--;
 560                         } else if (lines_before) {
 561                                 /* Add the line to the circular 'before' buffer */
 562                                 free(before_buf[curpos]);
 563                                 before_buf[curpos] = line;
 564                                 IF_EXTRA_COMPAT(before_buf_size[curpos] = line_len;)
 565                                 curpos = (curpos + 1) % lines_before;
 566                                 /* avoid free(line) - we took the line */
 567                                 line = NULL;
 568                         }
 569                 }
 570
 571 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
 572 #if !ENABLE_EXTRA_COMPAT
 573                 free(line);
 574 #endif
 575                 /* Did we print all context after last requested match? */
 576                 if ((option_mask32 & OPT_m)
 577                  && !print_n_lines_after
 578                  && nmatches == max_matches
 579                 ) {
 580                         break;
 581                 }
 582         } /* while (read line) */
 583
 584         /* special-case file post-processing for options where we don't print line
 585          * matches, just filenames and possibly match counts */
 586
 587         /* grep -c: print [filename:]count, even if count is zero */
 588         if (PRINT_MATCH_COUNTS) {
 589                 if (print_filename)
 590                         printf("%s:", cur_file);
 591                 printf("%d\n", nmatches);
 592         }
 593
 594         /* grep -L: print just the filename */
 595         if (PRINT_FILES_WITHOUT_MATCHES) {
 596                 /* nmatches is zero, no need to check it:
 597                  * we return 1 early if we detected a match
 598                  * and PRINT_FILES_WITHOUT_MATCHES is set */
 599                 puts(cur_file);
 600         }
 601
 602         return nmatches;
 603 }
 604
 605 #if ENABLE_FEATURE_CLEAN_UP
 606 #define new_grep_list_data(p, m) add_grep_list_data(p, m)
 607 static char *add_grep_list_data(char *pattern, int flg_used_mem)
 608 #else
 609 #define new_grep_list_data(p, m) add_grep_list_data(p)
 610 static char *add_grep_list_data(char *pattern)
 611 #endif
 612 {
 613         grep_list_data_t *gl = xzalloc(sizeof(*gl));
 614         gl->pattern = pattern;
 615 #if ENABLE_FEATURE_CLEAN_UP
 616         gl->flg_mem_allocated_compiled = flg_used_mem;
 617 #else
 618         /*gl->flg_mem_allocated_compiled = 0;*/
 619 #endif
 620         return (char *)gl;
 621 }
 622
 623 static void load_regexes_from_file(llist_t *fopt)
 624 {
 625         while (fopt) {
 626                 char *line;
 627                 FILE *fp;
 628                 llist_t *cur = fopt;
 629                 char *ffile = cur->data;
 630
 631                 fopt = cur->link;
 632                 free(cur);
 633                 fp = xfopen_stdin(ffile);
 634                 while ((line = xmalloc_fgetline(fp)) != NULL) {
 635                         llist_add_to(&pattern_head,
 636                                 new_grep_list_data(line, ALLOCATED));
 637                 }
 638                 fclose_if_not_stdin(fp);
 639         }
 640 }
 641
 642 static int FAST_FUNC file_action_grep(const char *filename,
 643                         struct stat *statbuf UNUSED_PARAM,
 644                         void* matched,
 645                         int depth UNUSED_PARAM)
 646 {
 647         FILE *file = fopen_for_read(filename);
 648         if (file == NULL) {
 649                 if (!SUPPRESS_ERR_MSGS)
 650                         bb_simple_perror_msg(filename);
 651                 open_errors = 1;
 652                 return 0;
 653         }
 654         cur_file = filename;
 655         *(int*)matched += grep_file(file);
 656         fclose(file);
 657         return 1;
 658 }
 659
 660 static int grep_dir(const char *dir)
 661 {
 662         int matched = 0;
 663         recursive_action(dir,
 664                 /* recurse=yes */ ACTION_RECURSE |
 665                 /* followLinks=command line only */ ACTION_FOLLOWLINKS_L0 |
 666                 /* depthFirst=yes */ ACTION_DEPTHFIRST,
 667                 /* fileAction= */ file_action_grep,
 668                 /* dirAction= */ NULL,
 669                 /* userData= */ &matched,
 670                 /* depth= */ 0);
 671         return matched;
 672 }
 673
 674 int grep_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 675 int grep_main(int argc UNUSED_PARAM, char **argv)
 676 {
 677         FILE *file;
 678         int matched;
 679         llist_t *fopt = NULL;
 680 #if ENABLE_FEATURE_GREP_CONTEXT
 681         int Copt, opts;
 682 #endif
 683         INIT_G();
 684
 685         /* For grep, exitcode of 1 is "not found". Other errors are 2: */
 686         xfunc_error_retval = 2;
 687
 688         /* do normal option parsing */
 689 #if ENABLE_FEATURE_GREP_CONTEXT
 690         /* -H unsets -h; -C unsets -A,-B */
 691         opts = getopt32(argv,
 692                 "^" OPTSTR_GREP "\0" "H-h:C-AB",
 693                 &pattern_head, &fopt, &max_matches,
 694                 &lines_after, &lines_before, &Copt);
 695
 696         if (opts & OPT_C) {
 697                 /* -C unsets prev -A and -B, but following -A or -B
 698                  * may override it */
 699                 if (!(opts & OPT_A)) /* not overridden */
 700                         lines_after = Copt;
 701                 if (!(opts & OPT_B)) /* not overridden */
 702                         lines_before = Copt;
 703         }
 704         /* sanity checks */
 705         if (opts & (OPT_c|OPT_q|OPT_l|OPT_L)) {
 706                 option_mask32 &= ~OPT_n;
 707                 lines_before = 0;
 708                 lines_after = 0;
 709         } else if (lines_before > 0) {
 710                 if (lines_before > INT_MAX / sizeof(long long))
 711                         lines_before = INT_MAX / sizeof(long long);
 712                 /* overflow in (lines_before * sizeof(x)) is prevented (above) */
 713                 before_buf = xzalloc(lines_before * sizeof(before_buf[0]));
 714                 IF_EXTRA_COMPAT(before_buf_size = xzalloc(lines_before * sizeof(before_buf_size[0]));)
 715         }
 716 #else
 717         /* with auto sanity checks */
 718         getopt32(argv, "^" OPTSTR_GREP "\0" "H-h:c-n:q-n:l-n:", // why trailing ":"?
 719                 &pattern_head, &fopt, &max_matches);
 720 #endif
 721         invert_search = ((option_mask32 & OPT_v) != 0); /* 0 | 1 */
 722
 723         {       /* convert char **argv to grep_list_data_t */
 724                 llist_t *cur;
 725                 for (cur = pattern_head; cur; cur = cur->link)
 726                         cur->data = new_grep_list_data(cur->data, 0);
 727         }
 728         if (option_mask32 & OPT_f) {
 729                 load_regexes_from_file(fopt);
 730                 if (!pattern_head) { /* -f EMPTY_FILE? */
 731                         /* GNU grep treats it as "nothing matches" */
 732                         llist_add_to(&pattern_head, new_grep_list_data((char*) "", 0));
 733                         invert_search ^= 1;
 734                 }
 735         }
 736
 737         if (ENABLE_FGREP && applet_name[0] == 'f')
 738                 option_mask32 |= OPT_F;
 739
 740 #if !ENABLE_EXTRA_COMPAT
 741         if (!(option_mask32 & (OPT_o | OPT_w | OPT_x)))
 742                 reflags = REG_NOSUB;
 743 #endif
 744
 745         if ((ENABLE_EGREP && applet_name[0] == 'e')
 746          || (option_mask32 & OPT_E)
 747         ) {
 748                 reflags |= REG_EXTENDED;
 749         }
 750 #if ENABLE_EXTRA_COMPAT
 751         else {
 752                 reflags = RE_SYNTAX_GREP;
 753         }
 754 #endif
 755
 756         if (option_mask32 & OPT_i) {
 757 #if !ENABLE_EXTRA_COMPAT
 758                 reflags |= REG_ICASE;
 759 #else
 760                 int i;
 761                 case_fold = xmalloc(256);
 762                 for (i = 0; i < 256; i++)
 763                         case_fold[i] = (unsigned char)i;
 764                 for (i = 'a'; i <= 'z'; i++)
 765                         case_fold[i] = (unsigned char)(i - ('a' - 'A'));
 766 #endif
 767         }
 768
 769         argv += optind;
 770
 771         /* if we didn't get a pattern from -e and no command file was specified,
 772          * first parameter should be the pattern. no pattern, no worky */
 773         if (pattern_head == NULL) {
 774                 char *pattern;
 775                 if (*argv == NULL)
 776                         bb_show_usage();
 777                 pattern = new_grep_list_data(*argv++, 0);
 778                 llist_add_to(&pattern_head, pattern);
 779         }
 780
 781         /* argv[0..(argc-1)] should be names of file to grep through. If
 782          * there is more than one file to grep, we will print the filenames. */
 783         if (argv[0] && argv[1])
 784                 print_filename = 1;
 785         /* -H / -h of course override */
 786         if (option_mask32 & OPT_H)
 787                 print_filename = 1;
 788         if (option_mask32 & OPT_h)
 789                 print_filename = 0;
 790
 791         /* If no files were specified, or '-' was specified, take input from
 792          * stdin. Otherwise, we grep through all the files specified. */
 793         matched = 0;
 794         do {
 795                 cur_file = *argv;
 796                 file = stdin;
 797                 if (!cur_file || LONE_DASH(cur_file)) {
 798                         cur_file = "(standard input)";
 799                 } else {
 800                         if (option_mask32 & OPT_r) {
 801                                 struct stat st;
 802                                 if (stat(cur_file, &st) == 0 && S_ISDIR(st.st_mode)) {
 803                                         if (!(option_mask32 & OPT_h))
 804                                                 print_filename = 1;
 805                                         matched += grep_dir(cur_file);
 806                                         goto grep_done;
 807                                 }
 808                         }
 809                         /* else: fopen(dir) will succeed, but reading won't */
 810                         file = fopen_for_read(cur_file);
 811                         if (file == NULL) {
 812                                 if (!SUPPRESS_ERR_MSGS)
 813                                         bb_simple_perror_msg(cur_file);
 814                                 open_errors = 1;
 815                                 continue;
 816                         }
 817                 }
 818                 matched += grep_file(file);
 819                 fclose_if_not_stdin(file);
 820  grep_done: ;
 821         } while (*argv && *++argv);
 822
 823         /* destroy all the elements in the pattern list */
 824         if (ENABLE_FEATURE_CLEAN_UP) {
 825                 while (pattern_head) {
 826                         llist_t *pattern_head_ptr = pattern_head;
 827                         grep_list_data_t *gl = (grep_list_data_t *)pattern_head_ptr->data;
 828
 829                         pattern_head = pattern_head->link;
 830                         if (gl->flg_mem_allocated_compiled & ALLOCATED)
 831                                 free(gl->pattern);
 832                         if (gl->flg_mem_allocated_compiled & COMPILED)
 833                                 regfree(&gl->compiled_regex);
 834                         free(gl);
 835                         free(pattern_head_ptr);
 836                 }
 837         }
 838         /* 0 = success, 1 = failed, 2 = error */
 839         if (open_errors)
 840                 return 2;
 841         return !matched; /* invert return value: 0 = success, 1 = failed */
 842 }