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:     "\n        -H      Add 'filename:' prefix"
  76 //usage:     "\n        -h      Do not add 'filename:' prefix"
  77 //usage:     "\n        -n      Add 'line_no:' prefix"
  78 //usage:     "\n        -l      Show only names of files that match"
  79 //usage:     "\n        -L      Show only names of files that don't match"
  80 //usage:     "\n        -c      Show only count of matching lines"
  81 //usage:     "\n        -o      Show only the matching part of line"
  82 //usage:     "\n        -q      Quiet. Return 0 if PATTERN is found, 1 otherwise"
  83 //usage:     "\n        -v      Select non-matching lines"
  84 //usage:     "\n        -s      Suppress open and read errors"
  85 //usage:     "\n        -r      Recurse"
  86 //usage:     "\n        -i      Ignore case"
  87 //usage:     "\n        -w      Match whole words only"
  88 //usage:     "\n        -x      Match whole lines 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:wx" \
 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         OPTBIT_x, /* -x whole line match */
 143         IF_FEATURE_GREP_CONTEXT(    OPTBIT_A ,) /* -A NUM: after-match context */
 144         IF_FEATURE_GREP_CONTEXT(    OPTBIT_B ,) /* -B NUM: before-match context */
 145         IF_FEATURE_GREP_CONTEXT(    OPTBIT_C ,) /* -C NUM: -A and -B combined */
 146         IF_FEATURE_GREP_EGREP_ALIAS(OPTBIT_E ,) /* extended regexp */
 147         IF_EXTRA_COMPAT(            OPTBIT_z ,) /* input is NUL terminated */
 148         OPT_l = 1 << OPTBIT_l,
 149         OPT_n = 1 << OPTBIT_n,
 150         OPT_q = 1 << OPTBIT_q,
 151         OPT_v = 1 << OPTBIT_v,
 152         OPT_s = 1 << OPTBIT_s,
 153         OPT_c = 1 << OPTBIT_c,
 154         OPT_F = 1 << OPTBIT_F,
 155         OPT_i = 1 << OPTBIT_i,
 156         OPT_H = 1 << OPTBIT_H,
 157         OPT_h = 1 << OPTBIT_h,
 158         OPT_e = 1 << OPTBIT_e,
 159         OPT_f = 1 << OPTBIT_f,
 160         OPT_L = 1 << OPTBIT_L,
 161         OPT_o = 1 << OPTBIT_o,
 162         OPT_r = 1 << OPTBIT_r,
 163         OPT_m = 1 << OPTBIT_m,
 164         OPT_w = 1 << OPTBIT_w,
 165         OPT_x = 1 << OPTBIT_x,
 166         OPT_A = IF_FEATURE_GREP_CONTEXT(    (1 << OPTBIT_A)) + 0,
 167         OPT_B = IF_FEATURE_GREP_CONTEXT(    (1 << OPTBIT_B)) + 0,
 168         OPT_C = IF_FEATURE_GREP_CONTEXT(    (1 << OPTBIT_C)) + 0,
 169         OPT_E = IF_FEATURE_GREP_EGREP_ALIAS((1 << OPTBIT_E)) + 0,
 170         OPT_z = IF_EXTRA_COMPAT(            (1 << OPTBIT_z)) + 0,
 171 };
 172
 173 #define PRINT_FILES_WITH_MATCHES    (option_mask32 & OPT_l)
 174 #define PRINT_LINE_NUM              (option_mask32 & OPT_n)
 175 #define BE_QUIET                    (option_mask32 & OPT_q)
 176 #define SUPPRESS_ERR_MSGS           (option_mask32 & OPT_s)
 177 #define PRINT_MATCH_COUNTS          (option_mask32 & OPT_c)
 178 #define FGREP_FLAG                  (option_mask32 & OPT_F)
 179 #define PRINT_FILES_WITHOUT_MATCHES (option_mask32 & OPT_L)
 180 #define NUL_DELIMITED               (option_mask32 & OPT_z)
 181
 182 struct globals {
 183         int max_matches;
 184 #if !ENABLE_EXTRA_COMPAT
 185         int reflags;
 186 #else
 187         RE_TRANSLATE_TYPE case_fold; /* RE_TRANSLATE_TYPE is [[un]signed] char* */
 188 #endif
 189         smalluint invert_search;
 190         smalluint print_filename;
 191         smalluint open_errors;
 192 #if ENABLE_FEATURE_GREP_CONTEXT
 193         smalluint did_print_line;
 194         int lines_before;
 195         int lines_after;
 196         char **before_buf;
 197         IF_EXTRA_COMPAT(size_t *before_buf_size;)
 198         int last_line_printed;
 199 #endif
 200         /* globals used internally */
 201         llist_t *pattern_head;   /* growable list of patterns to match */
 202         const char *cur_file;    /* the current file we are reading */
 203 } FIX_ALIASING;
 204 #define G (*(struct globals*)&bb_common_bufsiz1)
 205 #define INIT_G() do { \
 206         BUILD_BUG_ON(sizeof(G) > COMMON_BUFSIZE); \
 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                                 char *match;
 346                                 char *str = line;
 347  opt_f_again:
 348                                 match = ((option_mask32 & OPT_i)
 349                                         ? strcasestr(str, gl->pattern)
 350                                         : strstr(str, gl->pattern)
 351                                         );
 352                                 if (match) {
 353                                         if (option_mask32 & OPT_x) {
 354                                                 if (match != str)
 355                                                         goto opt_f_not_found;
 356                                                 if (str[strlen(gl->pattern)] != '\0')
 357                                                         goto opt_f_not_found;
 358                                         } else
 359                                         if (option_mask32 & OPT_w) {
 360                                                 char c = (match != str) ? match[-1] : ' ';
 361                                                 if (!isalnum(c) && c != '_') {
 362                                                         c = match[strlen(gl->pattern)];
 363                                                         if (!c || (!isalnum(c) && c != '_'))
 364                                                                 goto opt_f_found;
 365                                                 }
 366                                                 str = match + 1;
 367                                                 goto opt_f_again;
 368                                         }
 369  opt_f_found:
 370                                         found = 1;
 371  opt_f_not_found: ;
 372                                 }
 373                         } else {
 374 #if ENABLE_EXTRA_COMPAT
 375                                 unsigned start_pos;
 376 #else
 377                                 int match_flg;
 378 #endif
 379                                 char *match_at;
 380
 381                                 if (!(gl->flg_mem_alocated_compiled & COMPILED)) {
 382                                         gl->flg_mem_alocated_compiled |= COMPILED;
 383 #if !ENABLE_EXTRA_COMPAT
 384                                         xregcomp(&gl->compiled_regex, gl->pattern, reflags);
 385 #else
 386                                         memset(&gl->compiled_regex, 0, sizeof(gl->compiled_regex));
 387                                         gl->compiled_regex.translate = case_fold; /* for -i */
 388                                         if (re_compile_pattern(gl->pattern, strlen(gl->pattern), &gl->compiled_regex))
 389                                                 bb_error_msg_and_die("bad regex '%s'", gl->pattern);
 390 #endif
 391                                 }
 392 #if !ENABLE_EXTRA_COMPAT
 393                                 gl->matched_range.rm_so = 0;
 394                                 gl->matched_range.rm_eo = 0;
 395                                 match_flg = 0;
 396 #else
 397                                 start_pos = 0;
 398 #endif
 399                                 match_at = line;
 400  opt_w_again:
 401 //bb_error_msg("'%s' start_pos:%d line_len:%d", match_at, start_pos, line_len);
 402                                 if (
 403 #if !ENABLE_EXTRA_COMPAT
 404                                         regexec(&gl->compiled_regex, match_at, 1, &gl->matched_range, match_flg) == 0
 405 #else
 406                                         re_search(&gl->compiled_regex, match_at, line_len,
 407                                                         start_pos, /*range:*/ line_len,
 408                                                         &gl->matched_range) >= 0
 409 #endif
 410                                 ) {
 411                                         if (option_mask32 & OPT_x) {
 412                                                 found = (gl->matched_range.rm_so == 0
 413                                                          && match_at[gl->matched_range.rm_eo] == '\0');
 414                                         } else
 415                                         if (!(option_mask32 & OPT_w)) {
 416                                                 found = 1;
 417                                         } else {
 418                                                 char c = ' ';
 419                                                 if (match_at > line || gl->matched_range.rm_so != 0) {
 420                                                         c = match_at[gl->matched_range.rm_so - 1];
 421                                                 }
 422                                                 if (!isalnum(c) && c != '_') {
 423                                                         c = match_at[gl->matched_range.rm_eo];
 424                                                 }
 425                                                 if (!isalnum(c) && c != '_') {
 426                                                         found = 1;
 427                                                 } else {
 428                         /*
 429                          * Why check gl->matched_range.rm_eo?
 430                          * Zero-length match makes -w skip the line:
 431                          * "echo foo | grep ^" prints "foo",
 432                          * "echo foo | grep -w ^" prints nothing.
 433                          * Without such check, we can loop forever.
 434                          */
 435 #if !ENABLE_EXTRA_COMPAT
 436                                                         if (gl->matched_range.rm_eo != 0) {
 437                                                                 match_at += gl->matched_range.rm_eo;
 438                                                                 match_flg |= REG_NOTBOL;
 439                                                                 goto opt_w_again;
 440                                                         }
 441 #else
 442                                                         if (gl->matched_range.rm_eo > start_pos) {
 443                                                                 start_pos = gl->matched_range.rm_eo;
 444                                                                 goto opt_w_again;
 445                                                         }
 446 #endif
 447                                                 }
 448                                         }
 449                                 }
 450                         }
 451                         /* If it's non-inverted search, we can stop
 452                          * at first match */
 453                         if (found && !invert_search)
 454                                 goto do_found;
 455                         pattern_ptr = pattern_ptr->link;
 456                 } /* while (pattern_ptr) */
 457
 458                 if (found ^ invert_search) {
 459  do_found:
 460                         /* keep track of matches */
 461                         nmatches++;
 462
 463                         /* quiet/print (non)matching file names only? */
 464                         if (option_mask32 & (OPT_q|OPT_l|OPT_L)) {
 465                                 free(line); /* we don't need line anymore */
 466                                 if (BE_QUIET) {
 467                                         /* manpage says about -q:
 468                                          * "exit immediately with zero status
 469                                          * if any match is found,
 470                                          * even if errors were detected" */
 471                                         exit(EXIT_SUCCESS);
 472                                 }
 473                                 /* if we're just printing filenames, we stop after the first match */
 474                                 if (PRINT_FILES_WITH_MATCHES) {
 475                                         puts(cur_file);
 476                                         /* fall through to "return 1" */
 477                                 }
 478                                 /* OPT_L aka PRINT_FILES_WITHOUT_MATCHES: return early */
 479                                 return 1; /* one match */
 480                         }
 481
 482 #if ENABLE_FEATURE_GREP_CONTEXT
 483                         /* Were we printing context and saw next (unwanted) match? */
 484                         if ((option_mask32 & OPT_m) && nmatches > max_matches)
 485                                 break;
 486 #endif
 487
 488                         /* print the matched line */
 489                         if (PRINT_MATCH_COUNTS == 0) {
 490 #if ENABLE_FEATURE_GREP_CONTEXT
 491                                 int prevpos = (curpos == 0) ? lines_before - 1 : curpos - 1;
 492
 493                                 /* if we were told to print 'before' lines and there is at least
 494                                  * one line in the circular buffer, print them */
 495                                 if (lines_before && before_buf[prevpos] != NULL) {
 496                                         int first_buf_entry_line_num = linenum - lines_before;
 497
 498                                         /* advance to the first entry in the circular buffer, and
 499                                          * figure out the line number is of the first line in the
 500                                          * buffer */
 501                                         idx = curpos;
 502                                         while (before_buf[idx] == NULL) {
 503                                                 idx = (idx + 1) % lines_before;
 504                                                 first_buf_entry_line_num++;
 505                                         }
 506
 507                                         /* now print each line in the buffer, clearing them as we go */
 508                                         while (before_buf[idx] != NULL) {
 509                                                 print_line(before_buf[idx], before_buf_size[idx], first_buf_entry_line_num, '-');
 510                                                 free(before_buf[idx]);
 511                                                 before_buf[idx] = NULL;
 512                                                 idx = (idx + 1) % lines_before;
 513                                                 first_buf_entry_line_num++;
 514                                         }
 515                                 }
 516
 517                                 /* make a note that we need to print 'after' lines */
 518                                 print_n_lines_after = lines_after;
 519 #endif
 520                                 if (option_mask32 & OPT_o) {
 521                                         if (FGREP_FLAG) {
 522                                                 /* -Fo just prints the pattern
 523                                                  * (unless -v: -Fov doesnt print anything at all) */
 524                                                 if (found)
 525                                                         print_line(gl->pattern, strlen(gl->pattern), linenum, ':');
 526                                         } else while (1) {
 527                                                 unsigned start = gl->matched_range.rm_so;
 528                                                 unsigned end = gl->matched_range.rm_eo;
 529                                                 unsigned len = end - start;
 530                                                 char old = line[end];
 531                                                 line[end] = '\0';
 532                                                 /* Empty match is not printed: try "echo test | grep -o ''" */
 533                                                 if (len != 0)
 534                                                         print_line(line + start, len, linenum, ':');
 535                                                 if (old == '\0')
 536                                                         break;
 537                                                 line[end] = old;
 538                                                 if (len == 0)
 539                                                         end++;
 540 #if !ENABLE_EXTRA_COMPAT
 541                                                 if (regexec(&gl->compiled_regex, line + end,
 542                                                                 1, &gl->matched_range, REG_NOTBOL) != 0)
 543                                                         break;
 544                                                 gl->matched_range.rm_so += end;
 545                                                 gl->matched_range.rm_eo += end;
 546 #else
 547                                                 if (re_search(&gl->compiled_regex, line, line_len,
 548                                                                 end, line_len - end,
 549                                                                 &gl->matched_range) < 0)
 550                                                         break;
 551 #endif
 552                                         }
 553                                 } else {
 554                                         print_line(line, line_len, linenum, ':');
 555                                 }
 556                         }
 557                 }
 558 #if ENABLE_FEATURE_GREP_CONTEXT
 559                 else { /* no match */
 560                         /* if we need to print some context lines after the last match, do so */
 561                         if (print_n_lines_after) {
 562                                 print_line(line, strlen(line), linenum, '-');
 563                                 print_n_lines_after--;
 564                         } else if (lines_before) {
 565                                 /* Add the line to the circular 'before' buffer */
 566                                 free(before_buf[curpos]);
 567                                 before_buf[curpos] = line;
 568                                 IF_EXTRA_COMPAT(before_buf_size[curpos] = line_len;)
 569                                 curpos = (curpos + 1) % lines_before;
 570                                 /* avoid free(line) - we took the line */
 571                                 line = NULL;
 572                         }
 573                 }
 574
 575 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
 576 #if !ENABLE_EXTRA_COMPAT
 577                 free(line);
 578 #endif
 579                 /* Did we print all context after last requested match? */
 580                 if ((option_mask32 & OPT_m)
 581                  && !print_n_lines_after
 582                  && nmatches == max_matches
 583                 ) {
 584                         break;
 585                 }
 586         } /* while (read line) */
 587
 588         /* special-case file post-processing for options where we don't print line
 589          * matches, just filenames and possibly match counts */
 590
 591         /* grep -c: print [filename:]count, even if count is zero */
 592         if (PRINT_MATCH_COUNTS) {
 593                 if (print_filename)
 594                         printf("%s:", cur_file);
 595                 printf("%d\n", nmatches);
 596         }
 597
 598         /* grep -L: print just the filename */
 599         if (PRINT_FILES_WITHOUT_MATCHES) {
 600                 /* nmatches is zero, no need to check it:
 601                  * we return 1 early if we detected a match
 602                  * and PRINT_FILES_WITHOUT_MATCHES is set */
 603                 puts(cur_file);
 604         }
 605
 606         return nmatches;
 607 }
 608
 609 #if ENABLE_FEATURE_CLEAN_UP
 610 #define new_grep_list_data(p, m) add_grep_list_data(p, m)
 611 static char *add_grep_list_data(char *pattern, int flg_used_mem)
 612 #else
 613 #define new_grep_list_data(p, m) add_grep_list_data(p)
 614 static char *add_grep_list_data(char *pattern)
 615 #endif
 616 {
 617         grep_list_data_t *gl = xzalloc(sizeof(*gl));
 618         gl->pattern = pattern;
 619 #if ENABLE_FEATURE_CLEAN_UP
 620         gl->flg_mem_alocated_compiled = flg_used_mem;
 621 #else
 622         /*gl->flg_mem_alocated_compiled = 0;*/
 623 #endif
 624         return (char *)gl;
 625 }
 626
 627 static void load_regexes_from_file(llist_t *fopt)
 628 {
 629         while (fopt) {
 630                 char *line;
 631                 FILE *fp;
 632                 llist_t *cur = fopt;
 633                 char *ffile = cur->data;
 634
 635                 fopt = cur->link;
 636                 free(cur);
 637                 fp = xfopen_stdin(ffile);
 638                 while ((line = xmalloc_fgetline(fp)) != NULL) {
 639                         llist_add_to(&pattern_head,
 640                                 new_grep_list_data(line, ALLOCATED));
 641                 }
 642                 fclose_if_not_stdin(fp);
 643         }
 644 }
 645
 646 static int FAST_FUNC file_action_grep(const char *filename,
 647                         struct stat *statbuf UNUSED_PARAM,
 648                         void* matched,
 649                         int depth UNUSED_PARAM)
 650 {
 651         FILE *file = fopen_for_read(filename);
 652         if (file == NULL) {
 653                 if (!SUPPRESS_ERR_MSGS)
 654                         bb_simple_perror_msg(filename);
 655                 open_errors = 1;
 656                 return 0;
 657         }
 658         cur_file = filename;
 659         *(int*)matched += grep_file(file);
 660         fclose(file);
 661         return 1;
 662 }
 663
 664 static int grep_dir(const char *dir)
 665 {
 666         int matched = 0;
 667         recursive_action(dir,
 668                 /* recurse=yes */ ACTION_RECURSE |
 669                 /* followLinks=command line only */ ACTION_FOLLOWLINKS_L0 |
 670                 /* depthFirst=yes */ ACTION_DEPTHFIRST,
 671                 /* fileAction= */ file_action_grep,
 672                 /* dirAction= */ NULL,
 673                 /* userData= */ &matched,
 674                 /* depth= */ 0);
 675         return matched;
 676 }
 677
 678 int grep_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 679 int grep_main(int argc UNUSED_PARAM, char **argv)
 680 {
 681         FILE *file;
 682         int matched;
 683         llist_t *fopt = NULL;
 684
 685         /* do normal option parsing */
 686 #if ENABLE_FEATURE_GREP_CONTEXT
 687         int Copt, opts;
 688
 689         /* -H unsets -h; -C unsets -A,-B; -e,-f are lists;
 690          * -m,-A,-B,-C have numeric param */
 691         opt_complementary = "H-h:C-AB:e::f::m+:A+:B+:C+";
 692         opts = getopt32(argv,
 693                 OPTSTR_GREP,
 694                 &pattern_head, &fopt, &max_matches,
 695                 &lines_after, &lines_before, &Copt);
 696
 697         if (opts & OPT_C) {
 698                 /* -C unsets prev -A and -B, but following -A or -B
 699                  * may override it */
 700                 if (!(opts & OPT_A)) /* not overridden */
 701                         lines_after = Copt;
 702                 if (!(opts & OPT_B)) /* not overridden */
 703                         lines_before = Copt;
 704         }
 705         /* sanity checks */
 706         if (opts & (OPT_c|OPT_q|OPT_l|OPT_L)) {
 707                 option_mask32 &= ~OPT_n;
 708                 lines_before = 0;
 709                 lines_after = 0;
 710         } else if (lines_before > 0) {
 711                 if (lines_before > INT_MAX / sizeof(long long))
 712                         lines_before = INT_MAX / sizeof(long long);
 713                 /* overflow in (lines_before * sizeof(x)) is prevented (above) */
 714                 before_buf = xzalloc(lines_before * sizeof(before_buf[0]));
 715                 IF_EXTRA_COMPAT(before_buf_size = xzalloc(lines_before * sizeof(before_buf_size[0]));)
 716         }
 717 #else
 718         /* with auto sanity checks */
 719         /* -H unsets -h; -c,-q or -l unset -n; -e,-f are lists; -m N */
 720         opt_complementary = "H-h:c-n:q-n:l-n:e::f::m+";
 721         getopt32(argv, OPTSTR_GREP,
 722                 &pattern_head, &fopt, &max_matches);
 723 #endif
 724         invert_search = ((option_mask32 & OPT_v) != 0); /* 0 | 1 */
 725
 726         {       /* convert char **argv to grep_list_data_t */
 727                 llist_t *cur;
 728                 for (cur = pattern_head; cur; cur = cur->link)
 729                         cur->data = new_grep_list_data(cur->data, 0);
 730         }
 731         if (option_mask32 & OPT_f) {
 732                 load_regexes_from_file(fopt);
 733                 if (!pattern_head) { /* -f EMPTY_FILE? */
 734                         /* GNU grep treats it as "nothing matches" */
 735                         llist_add_to(&pattern_head, new_grep_list_data((char*) "", 0));
 736                         invert_search ^= 1;
 737                 }
 738         }
 739
 740         if (ENABLE_FEATURE_GREP_FGREP_ALIAS && applet_name[0] == 'f')
 741                 option_mask32 |= OPT_F;
 742
 743 #if !ENABLE_EXTRA_COMPAT
 744         if (!(option_mask32 & (OPT_o | OPT_w | OPT_x)))
 745                 reflags = REG_NOSUB;
 746 #endif
 747
 748         if (ENABLE_FEATURE_GREP_EGREP_ALIAS
 749          && (applet_name[0] == 'e' || (option_mask32 & OPT_E))
 750         ) {
 751                 reflags |= REG_EXTENDED;
 752         }
 753 #if ENABLE_EXTRA_COMPAT
 754         else {
 755                 reflags = RE_SYNTAX_GREP;
 756         }
 757 #endif
 758
 759         if (option_mask32 & OPT_i) {
 760 #if !ENABLE_EXTRA_COMPAT
 761                 reflags |= REG_ICASE;
 762 #else
 763                 int i;
 764                 case_fold = xmalloc(256);
 765                 for (i = 0; i < 256; i++)
 766                         case_fold[i] = (unsigned char)i;
 767                 for (i = 'a'; i <= 'z'; i++)
 768                         case_fold[i] = (unsigned char)(i - ('a' - 'A'));
 769 #endif
 770         }
 771
 772         argv += optind;
 773
 774         /* if we didn't get a pattern from -e and no command file was specified,
 775          * first parameter should be the pattern. no pattern, no worky */
 776         if (pattern_head == NULL) {
 777                 char *pattern;
 778                 if (*argv == NULL)
 779                         bb_show_usage();
 780                 pattern = new_grep_list_data(*argv++, 0);
 781                 llist_add_to(&pattern_head, pattern);
 782         }
 783
 784         /* argv[0..(argc-1)] should be names of file to grep through. If
 785          * there is more than one file to grep, we will print the filenames. */
 786         if (argv[0] && argv[1])
 787                 print_filename = 1;
 788         /* -H / -h of course override */
 789         if (option_mask32 & OPT_H)
 790                 print_filename = 1;
 791         if (option_mask32 & OPT_h)
 792                 print_filename = 0;
 793
 794         /* If no files were specified, or '-' was specified, take input from
 795          * stdin. Otherwise, we grep through all the files specified. */
 796         matched = 0;
 797         do {
 798                 cur_file = *argv;
 799                 file = stdin;
 800                 if (!cur_file || LONE_DASH(cur_file)) {
 801                         cur_file = "(standard input)";
 802                 } else {
 803                         if (option_mask32 & OPT_r) {
 804                                 struct stat st;
 805                                 if (stat(cur_file, &st) == 0 && S_ISDIR(st.st_mode)) {
 806                                         if (!(option_mask32 & OPT_h))
 807                                                 print_filename = 1;
 808                                         matched += grep_dir(cur_file);
 809                                         goto grep_done;
 810                                 }
 811                         }
 812                         /* else: fopen(dir) will succeed, but reading won't */
 813                         file = fopen_for_read(cur_file);
 814                         if (file == NULL) {
 815                                 if (!SUPPRESS_ERR_MSGS)
 816                                         bb_simple_perror_msg(cur_file);
 817                                 open_errors = 1;
 818                                 continue;
 819                         }
 820                 }
 821                 matched += grep_file(file);
 822                 fclose_if_not_stdin(file);
 823  grep_done: ;
 824         } while (*argv && *++argv);
 825
 826         /* destroy all the elments in the pattern list */
 827         if (ENABLE_FEATURE_CLEAN_UP) {
 828                 while (pattern_head) {
 829                         llist_t *pattern_head_ptr = pattern_head;
 830                         grep_list_data_t *gl = (grep_list_data_t *)pattern_head_ptr->data;
 831
 832                         pattern_head = pattern_head->link;
 833                         if (gl->flg_mem_alocated_compiled & ALLOCATED)
 834                                 free(gl->pattern);
 835                         if (gl->flg_mem_alocated_compiled & COMPILED)
 836                                 regfree(&gl->compiled_regex);
 837                         free(gl);
 838                         free(pattern_head_ptr);
 839                 }
 840         }
 841         /* 0 = success, 1 = failed, 2 = error */
 842         if (open_errors)
 843                 return 2;
 844         return !matched; /* invert return value: 0 = success, 1 = failed */
 845 }