1 /* vi: set sw=4 ts=4: */
3 * Mini grep implementation for busybox using libc regex.
5 * Copyright (C) 1999,2000,2001 by Lineo, inc. and Mark Whitley
6 * Copyright (C) 1999,2000,2001 by Mark Whitley <markw@codepoet.org>
8 * Licensed under the GPL v2 or later, see the file LICENSE in this tarball.
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 */
14 * 2004,2006 (C) Vladimir Oleynik <dzo@simtreas.ru> -
15 * correction "-e pattern1 -e pattern2" logic and more optimizations.
19 * (C) 2006 Jac Goudsmit added -o option
22 //kbuild:lib-$(CONFIG_GREP) += grep.o
28 //config: grep is used to search files for a specified pattern.
30 //config:config FEATURE_GREP_EGREP_ALIAS
31 //config: bool "Enable extended regular expressions (egrep & grep -E)"
33 //config: depends on GREP
35 //config: Enabled support for extended regular expressions. Extended
36 //config: regular expressions allow for alternation (foo|bar), grouping,
37 //config: and various repetition operators.
39 //config:config FEATURE_GREP_FGREP_ALIAS
40 //config: bool "Alias fgrep to grep -F"
42 //config: depends on GREP
44 //config: fgrep sees the search pattern as a normal string rather than
45 //config: regular expressions.
46 //config: grep -F always works, this just creates the fgrep alias.
48 //config:config FEATURE_GREP_CONTEXT
49 //config: bool "Enable before and after context flags (-A, -B and -C)"
51 //config: depends on GREP
53 //config: Print the specified number of leading (-B) and/or trailing (-A)
54 //config: context surrounding our matching lines.
55 //config: Print the specified number of context lines (-C).
62 "lnqvscFiHhe:f:Lorm:w" \
63 IF_FEATURE_GREP_CONTEXT("A:B:C:") \
64 IF_FEATURE_GREP_EGREP_ALIAS("E") \
65 IF_EXTRA_COMPAT("z") \
68 /* ignored: -a "assume all files to be text" */
69 /* ignored: -I "assume binary files have no matches" */
72 OPTBIT_l, /* list matched file names only */
73 OPTBIT_n, /* print line# */
74 OPTBIT_q, /* quiet - exit(EXIT_SUCCESS) of first match */
75 OPTBIT_v, /* invert the match, to select non-matching lines */
76 OPTBIT_s, /* suppress errors about file open errors */
77 OPTBIT_c, /* count matches per file (suppresses normal output) */
78 OPTBIT_F, /* literal match */
79 OPTBIT_i, /* case-insensitive */
80 OPTBIT_H, /* force filename display */
81 OPTBIT_h, /* inhibit filename display */
82 OPTBIT_e, /* -e PATTERN */
83 OPTBIT_f, /* -f FILE_WITH_PATTERNS */
84 OPTBIT_L, /* list unmatched file names only */
85 OPTBIT_o, /* show only matching parts of lines */
86 OPTBIT_r, /* recurse dirs */
87 OPTBIT_m, /* -m MAX_MATCHES */
88 OPTBIT_w, /* -w whole word match */
89 IF_FEATURE_GREP_CONTEXT( OPTBIT_A ,) /* -A NUM: after-match context */
90 IF_FEATURE_GREP_CONTEXT( OPTBIT_B ,) /* -B NUM: before-match context */
91 IF_FEATURE_GREP_CONTEXT( OPTBIT_C ,) /* -C NUM: -A and -B combined */
92 IF_FEATURE_GREP_EGREP_ALIAS(OPTBIT_E ,) /* extended regexp */
93 IF_EXTRA_COMPAT( OPTBIT_z ,) /* input is NUL terminated */
94 OPT_l = 1 << OPTBIT_l,
95 OPT_n = 1 << OPTBIT_n,
96 OPT_q = 1 << OPTBIT_q,
97 OPT_v = 1 << OPTBIT_v,
98 OPT_s = 1 << OPTBIT_s,
99 OPT_c = 1 << OPTBIT_c,
100 OPT_F = 1 << OPTBIT_F,
101 OPT_i = 1 << OPTBIT_i,
102 OPT_H = 1 << OPTBIT_H,
103 OPT_h = 1 << OPTBIT_h,
104 OPT_e = 1 << OPTBIT_e,
105 OPT_f = 1 << OPTBIT_f,
106 OPT_L = 1 << OPTBIT_L,
107 OPT_o = 1 << OPTBIT_o,
108 OPT_r = 1 << OPTBIT_r,
109 OPT_m = 1 << OPTBIT_m,
110 OPT_w = 1 << OPTBIT_w,
111 OPT_A = IF_FEATURE_GREP_CONTEXT( (1 << OPTBIT_A)) + 0,
112 OPT_B = IF_FEATURE_GREP_CONTEXT( (1 << OPTBIT_B)) + 0,
113 OPT_C = IF_FEATURE_GREP_CONTEXT( (1 << OPTBIT_C)) + 0,
114 OPT_E = IF_FEATURE_GREP_EGREP_ALIAS((1 << OPTBIT_E)) + 0,
115 OPT_z = IF_EXTRA_COMPAT( (1 << OPTBIT_z)) + 0,
118 #define PRINT_FILES_WITH_MATCHES (option_mask32 & OPT_l)
119 #define PRINT_LINE_NUM (option_mask32 & OPT_n)
120 #define BE_QUIET (option_mask32 & OPT_q)
121 #define SUPPRESS_ERR_MSGS (option_mask32 & OPT_s)
122 #define PRINT_MATCH_COUNTS (option_mask32 & OPT_c)
123 #define FGREP_FLAG (option_mask32 & OPT_F)
124 #define PRINT_FILES_WITHOUT_MATCHES (option_mask32 & OPT_L)
125 #define NUL_DELIMITED (option_mask32 & OPT_z)
129 #if !ENABLE_EXTRA_COMPAT
132 RE_TRANSLATE_TYPE case_fold; /* RE_TRANSLATE_TYPE is [[un]signed] char* */
134 smalluint invert_search;
135 smalluint print_filename;
136 smalluint open_errors;
137 #if ENABLE_FEATURE_GREP_CONTEXT
138 smalluint did_print_line;
142 IF_EXTRA_COMPAT(size_t *before_buf_size;)
143 int last_line_printed;
145 /* globals used internally */
146 llist_t *pattern_head; /* growable list of patterns to match */
147 const char *cur_file; /* the current file we are reading */
149 #define G (*(struct globals*)&bb_common_bufsiz1)
150 #define INIT_G() do { \
151 struct G_sizecheck { \
152 char G_sizecheck[sizeof(G) > COMMON_BUFSIZE ? -1 : 1]; \
155 #define max_matches (G.max_matches )
156 #if !ENABLE_EXTRA_COMPAT
157 # define reflags (G.reflags )
159 # define case_fold (G.case_fold )
160 /* http://www.delorie.com/gnu/docs/regex/regex_46.html */
161 # define reflags re_syntax_options
165 # define REG_NOSUB bug:is:here /* should not be used */
166 /* Just RE_SYNTAX_EGREP is not enough, need to enable {n[,[m]]} too */
167 # define REG_EXTENDED (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES)
168 # define REG_ICASE bug:is:here /* should not be used */
170 #define invert_search (G.invert_search )
171 #define print_filename (G.print_filename )
172 #define open_errors (G.open_errors )
173 #define did_print_line (G.did_print_line )
174 #define lines_before (G.lines_before )
175 #define lines_after (G.lines_after )
176 #define before_buf (G.before_buf )
177 #define before_buf_size (G.before_buf_size )
178 #define last_line_printed (G.last_line_printed )
179 #define pattern_head (G.pattern_head )
180 #define cur_file (G.cur_file )
183 typedef struct grep_list_data_t {
185 /* for GNU regex, matched_range must be persistent across grep_file() calls */
186 #if !ENABLE_EXTRA_COMPAT
187 regex_t compiled_regex;
188 regmatch_t matched_range;
190 struct re_pattern_buffer compiled_regex;
191 struct re_registers matched_range;
195 int flg_mem_alocated_compiled;
198 #if !ENABLE_EXTRA_COMPAT
199 #define print_line(line, line_len, linenum, decoration) \
200 print_line(line, linenum, decoration)
202 static void print_line(const char *line, size_t line_len, int linenum, char decoration)
204 #if ENABLE_FEATURE_GREP_CONTEXT
205 /* Happens when we go to next file, immediately hit match
206 * and try to print prev context... from prev file! Don't do it */
209 /* possibly print the little '--' separator */
210 if ((lines_before || lines_after) && did_print_line
211 && last_line_printed != linenum - 1
215 /* guard against printing "--" before first line of first file */
217 last_line_printed = linenum;
220 printf("%s%c", cur_file, decoration);
222 printf("%i%c", linenum, decoration);
223 /* Emulate weird GNU grep behavior with -ov */
224 if ((option_mask32 & (OPT_v|OPT_o)) != (OPT_v|OPT_o)) {
225 #if !ENABLE_EXTRA_COMPAT
228 fwrite(line, 1, line_len, stdout);
229 putchar(NUL_DELIMITED ? '\0' : '\n');
234 #if ENABLE_EXTRA_COMPAT
235 /* Unlike getline, this one removes trailing '\n' */
236 static ssize_t FAST_FUNC bb_getline(char **line_ptr, size_t *line_alloc_len, FILE *file)
240 int delim = (NUL_DELIMITED ? '\0' : '\n');
242 res_sz = getdelim(line_ptr, line_alloc_len, delim, file);
246 if (line[res_sz - 1] == delim)
247 line[--res_sz] = '\0';
249 free(line); /* uclibc allocates a buffer even on EOF. WTF? */
255 static int grep_file(FILE *file)
260 #if !ENABLE_EXTRA_COMPAT
265 size_t line_alloc_len;
266 # define rm_so start[0]
267 # define rm_eo end[0]
269 #if ENABLE_FEATURE_GREP_CONTEXT
270 int print_n_lines_after = 0;
271 int curpos = 0; /* track where we are in the circular 'before' buffer */
272 int idx = 0; /* used for iteration through the circular buffer */
274 enum { print_n_lines_after = 0 };
278 #if !ENABLE_EXTRA_COMPAT
279 (line = xmalloc_fgetline(file)) != NULL
281 (line_len = bb_getline(&line, &line_alloc_len, file)) >= 0
284 llist_t *pattern_ptr = pattern_head;
285 grep_list_data_t *gl = gl; /* for gcc */
289 while (pattern_ptr) {
290 gl = (grep_list_data_t *)pattern_ptr->data;
292 found |= (((option_mask32 & OPT_i)
293 ? strcasestr(line, gl->pattern)
294 : strstr(line, gl->pattern)
297 if (!(gl->flg_mem_alocated_compiled & COMPILED)) {
298 gl->flg_mem_alocated_compiled |= COMPILED;
299 #if !ENABLE_EXTRA_COMPAT
300 xregcomp(&gl->compiled_regex, gl->pattern, reflags);
302 memset(&gl->compiled_regex, 0, sizeof(gl->compiled_regex));
303 gl->compiled_regex.translate = case_fold; /* for -i */
304 if (re_compile_pattern(gl->pattern, strlen(gl->pattern), &gl->compiled_regex))
305 bb_error_msg_and_die("bad regex '%s'", gl->pattern);
308 #if !ENABLE_EXTRA_COMPAT
309 gl->matched_range.rm_so = 0;
310 gl->matched_range.rm_eo = 0;
313 #if !ENABLE_EXTRA_COMPAT
314 regexec(&gl->compiled_regex, line, 1, &gl->matched_range, 0) == 0
316 re_search(&gl->compiled_regex, line, line_len,
317 /*start:*/ 0, /*range:*/ line_len,
318 &gl->matched_range) >= 0
321 if (!(option_mask32 & OPT_w))
325 if (gl->matched_range.rm_so)
326 c = line[gl->matched_range.rm_so - 1];
327 if (!isalnum(c) && c != '_') {
328 c = line[gl->matched_range.rm_eo];
329 if (!c || (!isalnum(c) && c != '_'))
335 /* If it's non-inverted search, we can stop
337 if (found && !invert_search)
339 pattern_ptr = pattern_ptr->link;
340 } /* while (pattern_ptr) */
342 if (found ^ invert_search) {
344 /* keep track of matches */
347 /* quiet/print (non)matching file names only? */
348 if (option_mask32 & (OPT_q|OPT_l|OPT_L)) {
349 free(line); /* we don't need line anymore */
351 /* manpage says about -q:
352 * "exit immediately with zero status
353 * if any match is found,
354 * even if errors were detected" */
357 /* if we're just printing filenames, we stop after the first match */
358 if (PRINT_FILES_WITH_MATCHES) {
360 /* fall through to "return 1" */
362 /* OPT_L aka PRINT_FILES_WITHOUT_MATCHES: return early */
363 return 1; /* one match */
366 #if ENABLE_FEATURE_GREP_CONTEXT
367 /* Were we printing context and saw next (unwanted) match? */
368 if ((option_mask32 & OPT_m) && nmatches > max_matches)
372 /* print the matched line */
373 if (PRINT_MATCH_COUNTS == 0) {
374 #if ENABLE_FEATURE_GREP_CONTEXT
375 int prevpos = (curpos == 0) ? lines_before - 1 : curpos - 1;
377 /* if we were told to print 'before' lines and there is at least
378 * one line in the circular buffer, print them */
379 if (lines_before && before_buf[prevpos] != NULL) {
380 int first_buf_entry_line_num = linenum - lines_before;
382 /* advance to the first entry in the circular buffer, and
383 * figure out the line number is of the first line in the
386 while (before_buf[idx] == NULL) {
387 idx = (idx + 1) % lines_before;
388 first_buf_entry_line_num++;
391 /* now print each line in the buffer, clearing them as we go */
392 while (before_buf[idx] != NULL) {
393 print_line(before_buf[idx], before_buf_size[idx], first_buf_entry_line_num, '-');
394 free(before_buf[idx]);
395 before_buf[idx] = NULL;
396 idx = (idx + 1) % lines_before;
397 first_buf_entry_line_num++;
401 /* make a note that we need to print 'after' lines */
402 print_n_lines_after = lines_after;
404 if (option_mask32 & OPT_o) {
406 /* -Fo just prints the pattern
407 * (unless -v: -Fov doesnt print anything at all) */
409 print_line(gl->pattern, strlen(gl->pattern), linenum, ':');
411 unsigned end = gl->matched_range.rm_eo;
412 char old = line[end];
414 print_line(line + gl->matched_range.rm_so,
415 end - gl->matched_range.rm_so,
420 #if !ENABLE_EXTRA_COMPAT
421 if (regexec(&gl->compiled_regex, line + end,
422 1, &gl->matched_range, REG_NOTBOL) != 0)
424 gl->matched_range.rm_so += end;
425 gl->matched_range.rm_eo += end;
427 if (re_search(&gl->compiled_regex, line, line_len,
429 &gl->matched_range) < 0)
434 print_line(line, line_len, linenum, ':');
438 #if ENABLE_FEATURE_GREP_CONTEXT
439 else { /* no match */
440 /* if we need to print some context lines after the last match, do so */
441 if (print_n_lines_after) {
442 print_line(line, strlen(line), linenum, '-');
443 print_n_lines_after--;
444 } else if (lines_before) {
445 /* Add the line to the circular 'before' buffer */
446 free(before_buf[curpos]);
447 before_buf[curpos] = line;
448 IF_EXTRA_COMPAT(before_buf_size[curpos] = line_len;)
449 curpos = (curpos + 1) % lines_before;
450 /* avoid free(line) - we took the line */
455 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
456 #if !ENABLE_EXTRA_COMPAT
459 /* Did we print all context after last requested match? */
460 if ((option_mask32 & OPT_m)
461 && !print_n_lines_after
462 && nmatches == max_matches
466 } /* while (read line) */
468 /* special-case file post-processing for options where we don't print line
469 * matches, just filenames and possibly match counts */
471 /* grep -c: print [filename:]count, even if count is zero */
472 if (PRINT_MATCH_COUNTS) {
474 printf("%s:", cur_file);
475 printf("%d\n", nmatches);
478 /* grep -L: print just the filename */
479 if (PRINT_FILES_WITHOUT_MATCHES) {
480 /* nmatches is zero, no need to check it:
481 * we return 1 early if we detected a match
482 * and PRINT_FILES_WITHOUT_MATCHES is set */
489 #if ENABLE_FEATURE_CLEAN_UP
490 #define new_grep_list_data(p, m) add_grep_list_data(p, m)
491 static char *add_grep_list_data(char *pattern, int flg_used_mem)
493 #define new_grep_list_data(p, m) add_grep_list_data(p)
494 static char *add_grep_list_data(char *pattern)
497 grep_list_data_t *gl = xzalloc(sizeof(*gl));
498 gl->pattern = pattern;
499 #if ENABLE_FEATURE_CLEAN_UP
500 gl->flg_mem_alocated_compiled = flg_used_mem;
502 /*gl->flg_mem_alocated_compiled = 0;*/
507 static void load_regexes_from_file(llist_t *fopt)
514 char *ffile = cur->data;
518 f = xfopen_stdin(ffile);
519 while ((line = xmalloc_fgetline(f)) != NULL) {
520 llist_add_to(&pattern_head,
521 new_grep_list_data(line, ALLOCATED));
526 static int FAST_FUNC file_action_grep(const char *filename,
527 struct stat *statbuf UNUSED_PARAM,
529 int depth UNUSED_PARAM)
531 FILE *file = fopen_for_read(filename);
533 if (!SUPPRESS_ERR_MSGS)
534 bb_simple_perror_msg(filename);
539 *(int*)matched += grep_file(file);
544 static int grep_dir(const char *dir)
547 recursive_action(dir,
548 /* recurse=yes */ ACTION_RECURSE |
550 /* depthFirst=yes */ ACTION_DEPTHFIRST,
551 /* fileAction= */ file_action_grep,
552 /* dirAction= */ NULL,
553 /* userData= */ &matched,
558 int grep_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
559 int grep_main(int argc UNUSED_PARAM, char **argv)
563 llist_t *fopt = NULL;
565 /* do normal option parsing */
566 #if ENABLE_FEATURE_GREP_CONTEXT
569 /* -H unsets -h; -C unsets -A,-B; -e,-f are lists;
570 * -m,-A,-B,-C have numeric param */
571 opt_complementary = "H-h:C-AB:e::f::m+:A+:B+:C+";
574 &pattern_head, &fopt, &max_matches,
575 &lines_after, &lines_before, &Copt);
577 if (option_mask32 & OPT_C) {
578 /* -C unsets prev -A and -B, but following -A or -B
580 if (!(option_mask32 & OPT_A)) /* not overridden */
582 if (!(option_mask32 & OPT_B)) /* not overridden */
586 if (option_mask32 & (OPT_c|OPT_q|OPT_l|OPT_L)) {
587 option_mask32 &= ~OPT_n;
590 } else if (lines_before > 0) {
591 before_buf = xzalloc(lines_before * sizeof(before_buf[0]));
592 IF_EXTRA_COMPAT(before_buf_size = xzalloc(lines_before * sizeof(before_buf_size[0]));)
595 /* with auto sanity checks */
596 /* -H unsets -h; -c,-q or -l unset -n; -e,-f are lists; -m N */
597 opt_complementary = "H-h:c-n:q-n:l-n:e::f::m+";
598 getopt32(argv, OPTSTR_GREP,
599 &pattern_head, &fopt, &max_matches);
601 invert_search = ((option_mask32 & OPT_v) != 0); /* 0 | 1 */
603 if (pattern_head != NULL) {
604 /* convert char **argv to grep_list_data_t */
607 for (cur = pattern_head; cur; cur = cur->link)
608 cur->data = new_grep_list_data(cur->data, 0);
610 if (option_mask32 & OPT_f)
611 load_regexes_from_file(fopt);
613 if (ENABLE_FEATURE_GREP_FGREP_ALIAS && applet_name[0] == 'f')
614 option_mask32 |= OPT_F;
616 #if !ENABLE_EXTRA_COMPAT
617 if (!(option_mask32 & (OPT_o | OPT_w)))
621 if (ENABLE_FEATURE_GREP_EGREP_ALIAS
622 && (applet_name[0] == 'e' || (option_mask32 & OPT_E))
624 reflags |= REG_EXTENDED;
626 #if ENABLE_EXTRA_COMPAT
628 reflags = RE_SYNTAX_GREP;
632 if (option_mask32 & OPT_i) {
633 #if !ENABLE_EXTRA_COMPAT
634 reflags |= REG_ICASE;
637 case_fold = xmalloc(256);
638 for (i = 0; i < 256; i++)
639 case_fold[i] = (unsigned char)i;
640 for (i = 'a'; i <= 'z'; i++)
641 case_fold[i] = (unsigned char)(i - ('a' - 'A'));
647 /* if we didn't get a pattern from -e and no command file was specified,
648 * first parameter should be the pattern. no pattern, no worky */
649 if (pattern_head == NULL) {
653 pattern = new_grep_list_data(*argv++, 0);
654 llist_add_to(&pattern_head, pattern);
657 /* argv[0..(argc-1)] should be names of file to grep through. If
658 * there is more than one file to grep, we will print the filenames. */
659 if (argv[0] && argv[1])
661 /* -H / -h of course override */
662 if (option_mask32 & OPT_H)
664 if (option_mask32 & OPT_h)
667 /* If no files were specified, or '-' was specified, take input from
668 * stdin. Otherwise, we grep through all the files specified. */
673 if (!cur_file || LONE_DASH(cur_file)) {
674 cur_file = "(standard input)";
676 if (option_mask32 & OPT_r) {
678 if (stat(cur_file, &st) == 0 && S_ISDIR(st.st_mode)) {
679 if (!(option_mask32 & OPT_h))
681 matched += grep_dir(cur_file);
685 /* else: fopen(dir) will succeed, but reading won't */
686 file = fopen_for_read(cur_file);
688 if (!SUPPRESS_ERR_MSGS)
689 bb_simple_perror_msg(cur_file);
694 matched += grep_file(file);
695 fclose_if_not_stdin(file);
697 } while (*argv && *++argv);
699 /* destroy all the elments in the pattern list */
700 if (ENABLE_FEATURE_CLEAN_UP) {
701 while (pattern_head) {
702 llist_t *pattern_head_ptr = pattern_head;
703 grep_list_data_t *gl = (grep_list_data_t *)pattern_head_ptr->data;
705 pattern_head = pattern_head->link;
706 if (gl->flg_mem_alocated_compiled & ALLOCATED)
708 if (gl->flg_mem_alocated_compiled & COMPILED)
709 regfree(&gl->compiled_regex);
711 free(pattern_head_ptr);
714 /* 0 = success, 1 = failed, 2 = error */
717 return !matched; /* invert return value: 0 = success, 1 = failed */