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
27 "lnqvscFiHhe:f:Lorm:" \
28 USE_FEATURE_GREP_CONTEXT("A:B:C:") \
29 USE_FEATURE_GREP_EGREP_ALIAS("E") \
32 /* ignored: -a "assume all files to be text" */
33 /* ignored: -I "assume binary files have no matches" */
36 OPTBIT_l, /* list matched file names only */
37 OPTBIT_n, /* print line# */
38 OPTBIT_q, /* quiet - exit(EXIT_SUCCESS) of first match */
39 OPTBIT_v, /* invert the match, to select non-matching lines */
40 OPTBIT_s, /* suppress errors about file open errors */
41 OPTBIT_c, /* count matches per file (suppresses normal output) */
42 OPTBIT_F, /* literal match */
43 OPTBIT_i, /* case-insensitive */
44 OPTBIT_H, /* force filename display */
45 OPTBIT_h, /* inhibit filename display */
46 OPTBIT_e, /* -e PATTERN */
47 OPTBIT_f, /* -f FILE_WITH_PATTERNS */
48 OPTBIT_L, /* list unmatched file names only */
49 OPTBIT_o, /* show only matching parts of lines */
50 OPTBIT_r, /* recurse dirs */
51 OPTBIT_m, /* -m MAX_MATCHES */
52 USE_FEATURE_GREP_CONTEXT( OPTBIT_A ,) /* -A NUM: after-match context */
53 USE_FEATURE_GREP_CONTEXT( OPTBIT_B ,) /* -B NUM: before-match context */
54 USE_FEATURE_GREP_CONTEXT( OPTBIT_C ,) /* -C NUM: -A and -B combined */
55 USE_FEATURE_GREP_EGREP_ALIAS(OPTBIT_E ,) /* extended regexp */
56 USE_DESKTOP( OPTBIT_w ,) /* whole word match */
57 OPT_l = 1 << OPTBIT_l,
58 OPT_n = 1 << OPTBIT_n,
59 OPT_q = 1 << OPTBIT_q,
60 OPT_v = 1 << OPTBIT_v,
61 OPT_s = 1 << OPTBIT_s,
62 OPT_c = 1 << OPTBIT_c,
63 OPT_F = 1 << OPTBIT_F,
64 OPT_i = 1 << OPTBIT_i,
65 OPT_H = 1 << OPTBIT_H,
66 OPT_h = 1 << OPTBIT_h,
67 OPT_e = 1 << OPTBIT_e,
68 OPT_f = 1 << OPTBIT_f,
69 OPT_L = 1 << OPTBIT_L,
70 OPT_o = 1 << OPTBIT_o,
71 OPT_r = 1 << OPTBIT_r,
72 OPT_m = 1 << OPTBIT_m,
73 OPT_A = USE_FEATURE_GREP_CONTEXT( (1 << OPTBIT_A)) + 0,
74 OPT_B = USE_FEATURE_GREP_CONTEXT( (1 << OPTBIT_B)) + 0,
75 OPT_C = USE_FEATURE_GREP_CONTEXT( (1 << OPTBIT_C)) + 0,
76 OPT_E = USE_FEATURE_GREP_EGREP_ALIAS((1 << OPTBIT_E)) + 0,
77 OPT_w = USE_DESKTOP( (1 << OPTBIT_w)) + 0,
80 #define PRINT_FILES_WITH_MATCHES (option_mask32 & OPT_l)
81 #define PRINT_LINE_NUM (option_mask32 & OPT_n)
82 #define BE_QUIET (option_mask32 & OPT_q)
83 #define SUPPRESS_ERR_MSGS (option_mask32 & OPT_s)
84 #define PRINT_MATCH_COUNTS (option_mask32 & OPT_c)
85 #define FGREP_FLAG (option_mask32 & OPT_F)
86 #define PRINT_FILES_WITHOUT_MATCHES (option_mask32 & OPT_L)
90 #if !ENABLE_EXTRA_COMPAT
93 RE_TRANSLATE_TYPE case_fold; /* RE_TRANSLATE_TYPE is [[un]signed] char* */
95 smalluint invert_search;
96 smalluint print_filename;
97 smalluint open_errors;
98 #if ENABLE_FEATURE_GREP_CONTEXT
99 smalluint did_print_line;
103 USE_EXTRA_COMPAT(size_t *before_buf_size;)
104 int last_line_printed;
106 /* globals used internally */
107 llist_t *pattern_head; /* growable list of patterns to match */
108 const char *cur_file; /* the current file we are reading */
110 #define G (*(struct globals*)&bb_common_bufsiz1)
111 #define INIT_G() do { \
112 struct G_sizecheck { \
113 char G_sizecheck[sizeof(G) > COMMON_BUFSIZE ? -1 : 1]; \
116 #define max_matches (G.max_matches )
117 #if !ENABLE_EXTRA_COMPAT
118 #define reflags (G.reflags )
120 #define case_fold (G.case_fold )
121 /* http://www.delorie.com/gnu/docs/regex/regex_46.html */
122 #define reflags re_syntax_options
126 #define REG_NOSUB bug:is:here /* should not be used */
127 #define REG_EXTENDED RE_SYNTAX_EGREP
128 #define REG_ICASE bug:is:here /* should not be used */
130 #define invert_search (G.invert_search )
131 #define print_filename (G.print_filename )
132 #define open_errors (G.open_errors )
133 #define did_print_line (G.did_print_line )
134 #define lines_before (G.lines_before )
135 #define lines_after (G.lines_after )
136 #define before_buf (G.before_buf )
137 #define before_buf_size (G.before_buf_size )
138 #define last_line_printed (G.last_line_printed )
139 #define pattern_head (G.pattern_head )
140 #define cur_file (G.cur_file )
143 typedef struct grep_list_data_t {
145 /* for GNU regex, matched_range must be persistent across grep_file() calls */
146 #if !ENABLE_EXTRA_COMPAT
147 regex_t compiled_regex;
148 regmatch_t matched_range;
150 struct re_pattern_buffer compiled_regex;
151 struct re_registers matched_range;
155 int flg_mem_alocated_compiled;
158 #if !ENABLE_EXTRA_COMPAT
159 #define print_line(line, line_len, linenum, decoration) \
160 print_line(line, linenum, decoration)
162 static void print_line(const char *line, size_t line_len, int linenum, char decoration)
164 #if ENABLE_FEATURE_GREP_CONTEXT
165 /* Happens when we go to next file, immediately hit match
166 * and try to print prev context... from prev file! Don't do it */
169 /* possibly print the little '--' separator */
170 if ((lines_before || lines_after) && did_print_line
171 && last_line_printed != linenum - 1
175 /* guard against printing "--" before first line of first file */
177 last_line_printed = linenum;
180 printf("%s%c", cur_file, decoration);
182 printf("%i%c", linenum, decoration);
183 /* Emulate weird GNU grep behavior with -ov */
184 if ((option_mask32 & (OPT_v|OPT_o)) != (OPT_v|OPT_o)) {
185 #if !ENABLE_EXTRA_COMPAT
188 fwrite(line, 1, line_len, stdout);
194 #if ENABLE_EXTRA_COMPAT
195 /* Unlike getline, this one removes trailing '\n' */
196 static ssize_t FAST_FUNC bb_getline(char **line_ptr, size_t *line_alloc_len, FILE *file)
201 res_sz = getline(line_ptr, line_alloc_len, file);
205 if (line[res_sz - 1] == '\n')
206 line[--res_sz] = '\0';
208 free(line); /* uclibc allocates a buffer even on EOF. WTF? */
214 static int grep_file(FILE *file)
219 #if !ENABLE_EXTRA_COMPAT
224 size_t line_alloc_len;
225 #define rm_so start[0]
228 #if ENABLE_FEATURE_GREP_CONTEXT
229 int print_n_lines_after = 0;
230 int curpos = 0; /* track where we are in the circular 'before' buffer */
231 int idx = 0; /* used for iteration through the circular buffer */
233 enum { print_n_lines_after = 0 };
234 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
237 #if !ENABLE_EXTRA_COMPAT
238 (line = xmalloc_fgetline(file)) != NULL
240 (line_len = bb_getline(&line, &line_alloc_len, file)) >= 0
243 llist_t *pattern_ptr = pattern_head;
244 grep_list_data_t *gl = gl; /* for gcc */
248 while (pattern_ptr) {
249 gl = (grep_list_data_t *)pattern_ptr->data;
251 found |= (strstr(line, gl->pattern) != NULL);
253 if (!(gl->flg_mem_alocated_compiled & COMPILED)) {
254 gl->flg_mem_alocated_compiled |= COMPILED;
255 #if !ENABLE_EXTRA_COMPAT
256 xregcomp(&gl->compiled_regex, gl->pattern, reflags);
258 memset(&gl->compiled_regex, 0, sizeof(gl->compiled_regex));
259 gl->compiled_regex.translate = case_fold; /* for -i */
260 if (re_compile_pattern(gl->pattern, strlen(gl->pattern), &gl->compiled_regex))
261 bb_error_msg_and_die("bad regex '%s'", gl->pattern);
264 #if !ENABLE_EXTRA_COMPAT
265 gl->matched_range.rm_so = 0;
266 gl->matched_range.rm_eo = 0;
269 #if !ENABLE_EXTRA_COMPAT
270 regexec(&gl->compiled_regex, line, 1, &gl->matched_range, 0) == 0
272 re_search(&gl->compiled_regex, line, line_len,
273 /*start:*/ 0, /*range:*/ line_len,
274 &gl->matched_range) >= 0
277 if (!(option_mask32 & OPT_w))
281 if (gl->matched_range.rm_so)
282 c = line[gl->matched_range.rm_so - 1];
283 if (!isalnum(c) && c != '_') {
284 c = line[gl->matched_range.rm_eo];
285 if (!c || (!isalnum(c) && c != '_'))
291 /* If it's non-inverted search, we can stop
293 if (found && !invert_search)
295 pattern_ptr = pattern_ptr->link;
296 } /* while (pattern_ptr) */
298 if (found ^ invert_search) {
300 /* keep track of matches */
303 /* quiet/print (non)matching file names only? */
304 if (option_mask32 & (OPT_q|OPT_l|OPT_L)) {
305 free(line); /* we don't need line anymore */
307 /* manpage says about -q:
308 * "exit immediately with zero status
309 * if any match is found,
310 * even if errors were detected" */
313 /* if we're just printing filenames, we stop after the first match */
314 if (PRINT_FILES_WITH_MATCHES) {
316 /* fall through to "return 1" */
318 /* OPT_L aka PRINT_FILES_WITHOUT_MATCHES: return early */
319 return 1; /* one match */
322 #if ENABLE_FEATURE_GREP_CONTEXT
323 /* Were we printing context and saw next (unwanted) match? */
324 if ((option_mask32 & OPT_m) && nmatches > max_matches)
328 /* print the matched line */
329 if (PRINT_MATCH_COUNTS == 0) {
330 #if ENABLE_FEATURE_GREP_CONTEXT
331 int prevpos = (curpos == 0) ? lines_before - 1 : curpos - 1;
333 /* if we were told to print 'before' lines and there is at least
334 * one line in the circular buffer, print them */
335 if (lines_before && before_buf[prevpos] != NULL) {
336 int first_buf_entry_line_num = linenum - lines_before;
338 /* advance to the first entry in the circular buffer, and
339 * figure out the line number is of the first line in the
342 while (before_buf[idx] == NULL) {
343 idx = (idx + 1) % lines_before;
344 first_buf_entry_line_num++;
347 /* now print each line in the buffer, clearing them as we go */
348 while (before_buf[idx] != NULL) {
349 print_line(before_buf[idx], before_buf_size[idx], first_buf_entry_line_num, '-');
350 free(before_buf[idx]);
351 before_buf[idx] = NULL;
352 idx = (idx + 1) % lines_before;
353 first_buf_entry_line_num++;
357 /* make a note that we need to print 'after' lines */
358 print_n_lines_after = lines_after;
360 if (option_mask32 & OPT_o) {
362 /* -Fo just prints the pattern
363 * (unless -v: -Fov doesnt print anything at all) */
365 print_line(gl->pattern, strlen(gl->pattern), linenum, ':');
367 char old = line[gl->matched_range.rm_eo];
368 line[gl->matched_range.rm_eo] = '\0';
369 print_line(line + gl->matched_range.rm_so,
370 gl->matched_range.rm_eo - gl->matched_range.rm_so,
372 line[gl->matched_range.rm_eo] = old;
373 #if !ENABLE_EXTRA_COMPAT
376 if (re_search(&gl->compiled_regex, line, line_len,
377 gl->matched_range.rm_eo, line_len - gl->matched_range.rm_eo,
378 &gl->matched_range) < 0)
383 print_line(line, line_len, linenum, ':');
387 #if ENABLE_FEATURE_GREP_CONTEXT
388 else { /* no match */
389 /* if we need to print some context lines after the last match, do so */
390 if (print_n_lines_after) {
391 print_line(line, strlen(line), linenum, '-');
392 print_n_lines_after--;
393 } else if (lines_before) {
394 /* Add the line to the circular 'before' buffer */
395 free(before_buf[curpos]);
396 before_buf[curpos] = line;
397 USE_EXTRA_COMPAT(before_buf_size[curpos] = line_len;)
398 curpos = (curpos + 1) % lines_before;
399 /* avoid free(line) - we took the line */
404 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
405 #if !ENABLE_EXTRA_COMPAT
408 /* Did we print all context after last requested match? */
409 if ((option_mask32 & OPT_m)
410 && !print_n_lines_after && nmatches == max_matches)
412 } /* while (read line) */
414 /* special-case file post-processing for options where we don't print line
415 * matches, just filenames and possibly match counts */
417 /* grep -c: print [filename:]count, even if count is zero */
418 if (PRINT_MATCH_COUNTS) {
420 printf("%s:", cur_file);
421 printf("%d\n", nmatches);
424 /* grep -L: print just the filename */
425 if (PRINT_FILES_WITHOUT_MATCHES) {
426 /* nmatches is zero, no need to check it:
427 * we return 1 early if we detected a match
428 * and PRINT_FILES_WITHOUT_MATCHES is set */
435 #if ENABLE_FEATURE_CLEAN_UP
436 #define new_grep_list_data(p, m) add_grep_list_data(p, m)
437 static char *add_grep_list_data(char *pattern, int flg_used_mem)
439 #define new_grep_list_data(p, m) add_grep_list_data(p)
440 static char *add_grep_list_data(char *pattern)
443 grep_list_data_t *gl = xzalloc(sizeof(*gl));
444 gl->pattern = pattern;
445 #if ENABLE_FEATURE_CLEAN_UP
446 gl->flg_mem_alocated_compiled = flg_used_mem;
448 /*gl->flg_mem_alocated_compiled = 0;*/
453 static void load_regexes_from_file(llist_t *fopt)
460 char *ffile = cur->data;
464 f = xfopen_stdin(ffile);
465 while ((line = xmalloc_fgetline(f)) != NULL) {
466 llist_add_to(&pattern_head,
467 new_grep_list_data(line, ALLOCATED));
472 static int FAST_FUNC file_action_grep(const char *filename,
473 struct stat *statbuf UNUSED_PARAM,
475 int depth UNUSED_PARAM)
477 FILE *file = fopen_for_read(filename);
479 if (!SUPPRESS_ERR_MSGS)
480 bb_simple_perror_msg(filename);
485 *(int*)matched += grep_file(file);
490 static int grep_dir(const char *dir)
493 recursive_action(dir,
494 /* recurse=yes */ ACTION_RECURSE |
496 /* depthFirst=yes */ ACTION_DEPTHFIRST,
497 /* fileAction= */ file_action_grep,
498 /* dirAction= */ NULL,
499 /* userData= */ &matched,
504 int grep_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
505 int grep_main(int argc, char **argv)
509 llist_t *fopt = NULL;
511 /* do normal option parsing */
512 #if ENABLE_FEATURE_GREP_CONTEXT
515 /* -H unsets -h; -C unsets -A,-B; -e,-f are lists;
516 * -m,-A,-B,-C have numeric param */
517 opt_complementary = "H-h:C-AB:e::f::m+:A+:B+:C+";
520 &pattern_head, &fopt, &max_matches,
521 &lines_after, &lines_before, &Copt);
523 if (option_mask32 & OPT_C) {
524 /* -C unsets prev -A and -B, but following -A or -B
526 if (!(option_mask32 & OPT_A)) /* not overridden */
528 if (!(option_mask32 & OPT_B)) /* not overridden */
532 if (option_mask32 & (OPT_c|OPT_q|OPT_l|OPT_L)) {
533 option_mask32 &= ~OPT_n;
536 } else if (lines_before > 0) {
537 before_buf = xzalloc(lines_before * sizeof(before_buf[0]));
538 USE_EXTRA_COMPAT(before_buf_size = xzalloc(lines_before * sizeof(before_buf_size[0]));)
541 /* with auto sanity checks */
542 /* -H unsets -h; -c,-q or -l unset -n; -e,-f are lists; -m N */
543 opt_complementary = "H-h:c-n:q-n:l-n:e::f::m+";
544 getopt32(argv, OPTSTR_GREP,
545 &pattern_head, &fopt, &max_matches);
547 invert_search = ((option_mask32 & OPT_v) != 0); /* 0 | 1 */
549 if (pattern_head != NULL) {
550 /* convert char **argv to grep_list_data_t */
553 for (cur = pattern_head; cur; cur = cur->link)
554 cur->data = new_grep_list_data(cur->data, 0);
556 if (option_mask32 & OPT_f)
557 load_regexes_from_file(fopt);
559 if (ENABLE_FEATURE_GREP_FGREP_ALIAS && applet_name[0] == 'f')
560 option_mask32 |= OPT_F;
562 #if !ENABLE_EXTRA_COMPAT
563 if (!(option_mask32 & (OPT_o | OPT_w)))
567 if (ENABLE_FEATURE_GREP_EGREP_ALIAS
568 && (applet_name[0] == 'e' || (option_mask32 & OPT_E))
570 reflags |= REG_EXTENDED;
572 #if ENABLE_EXTRA_COMPAT
574 reflags = RE_SYNTAX_GREP;
578 if (option_mask32 & OPT_i) {
579 #if !ENABLE_EXTRA_COMPAT
580 reflags |= REG_ICASE;
583 case_fold = xmalloc(256);
584 for (i = 0; i < 256; i++)
585 case_fold[i] = (unsigned char)i;
586 for (i = 'a'; i <= 'z'; i++)
587 case_fold[i] = (unsigned char)(i - ('a' - 'A'));
594 /* if we didn't get a pattern from -e and no command file was specified,
595 * first parameter should be the pattern. no pattern, no worky */
596 if (pattern_head == NULL) {
600 pattern = new_grep_list_data(*argv++, 0);
601 llist_add_to(&pattern_head, pattern);
605 /* argv[0..(argc-1)] should be names of file to grep through. If
606 * there is more than one file to grep, we will print the filenames. */
609 /* -H / -h of course override */
610 if (option_mask32 & OPT_H)
612 if (option_mask32 & OPT_h)
615 /* If no files were specified, or '-' was specified, take input from
616 * stdin. Otherwise, we grep through all the files specified. */
621 if (!cur_file || LONE_DASH(cur_file)) {
622 cur_file = "(standard input)";
624 if (option_mask32 & OPT_r) {
626 if (stat(cur_file, &st) == 0 && S_ISDIR(st.st_mode)) {
627 if (!(option_mask32 & OPT_h))
629 matched += grep_dir(cur_file);
633 /* else: fopen(dir) will succeed, but reading won't */
634 file = fopen_for_read(cur_file);
636 if (!SUPPRESS_ERR_MSGS)
637 bb_simple_perror_msg(cur_file);
642 matched += grep_file(file);
643 fclose_if_not_stdin(file);
645 } while (--argc > 0);
647 /* destroy all the elments in the pattern list */
648 if (ENABLE_FEATURE_CLEAN_UP) {
649 while (pattern_head) {
650 llist_t *pattern_head_ptr = pattern_head;
651 grep_list_data_t *gl = (grep_list_data_t *)pattern_head_ptr->data;
653 pattern_head = pattern_head->link;
654 if (gl->flg_mem_alocated_compiled & ALLOCATED)
656 if (gl->flg_mem_alocated_compiled & COMPILED)
657 regfree(&gl->compiled_regex);
659 free(pattern_head_ptr);
662 /* 0 = success, 1 = failed, 2 = error */
665 return !matched; /* invert return value: 0 = success, 1 = failed */