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)
91 smalluint invert_search;
92 smalluint print_filename;
93 smalluint open_errors;
94 #if ENABLE_FEATURE_GREP_CONTEXT
95 smalluint did_print_line;
99 int last_line_printed;
101 /* globals used internally */
102 llist_t *pattern_head; /* growable list of patterns to match */
103 const char *cur_file; /* the current file we are reading */
105 #define G (*(struct globals*)&bb_common_bufsiz1)
106 #define INIT_G() do { \
107 struct G_sizecheck { \
108 char G_sizecheck[sizeof(G) > COMMON_BUFSIZE ? -1 : 1]; \
111 #define max_matches (G.max_matches )
112 #define reflags (G.reflags )
113 #define invert_search (G.invert_search )
114 #define print_filename (G.print_filename )
115 #define open_errors (G.open_errors )
116 #define did_print_line (G.did_print_line )
117 #define lines_before (G.lines_before )
118 #define lines_after (G.lines_after )
119 #define before_buf (G.before_buf )
120 #define last_line_printed (G.last_line_printed )
121 #define pattern_head (G.pattern_head )
122 #define cur_file (G.cur_file )
125 typedef struct grep_list_data_t {
130 int flg_mem_alocated_compiled;
134 static void print_line(const char *line, int linenum, char decoration)
136 #if ENABLE_FEATURE_GREP_CONTEXT
137 /* Happens when we go to next file, immediately hit match
138 * and try to print prev context... from prev file! Don't do it */
141 /* possibly print the little '--' separator */
142 if ((lines_before || lines_after) && did_print_line &&
143 last_line_printed != linenum - 1) {
146 /* guard against printing "--" before first line of first file */
148 last_line_printed = linenum;
151 printf("%s%c", cur_file, decoration);
153 printf("%i%c", linenum, decoration);
154 /* Emulate weird GNU grep behavior with -ov */
155 if ((option_mask32 & (OPT_v|OPT_o)) != (OPT_v|OPT_o))
159 static int grep_file(FILE *file)
166 #if ENABLE_FEATURE_GREP_CONTEXT
167 int print_n_lines_after = 0;
168 int curpos = 0; /* track where we are in the circular 'before' buffer */
169 int idx = 0; /* used for iteration through the circular buffer */
171 enum { print_n_lines_after = 0 };
172 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
174 while ((line = xmalloc_fgetline(file)) != NULL) {
175 llist_t *pattern_ptr = pattern_head;
176 grep_list_data_t *gl = gl; /* for gcc */
180 while (pattern_ptr) {
181 gl = (grep_list_data_t *)pattern_ptr->data;
183 found |= (strstr(line, gl->pattern) != NULL);
185 if (!(gl->flg_mem_alocated_compiled & COMPILED)) {
186 gl->flg_mem_alocated_compiled |= COMPILED;
187 xregcomp(&(gl->preg), gl->pattern, reflags);
191 if (regexec(&(gl->preg), line, 1, ®match, 0) == 0) {
192 if (!(option_mask32 & OPT_w))
197 c = line[regmatch.rm_so - 1];
198 if (!isalnum(c) && c != '_') {
199 c = line[regmatch.rm_eo];
200 if (!c || (!isalnum(c) && c != '_'))
206 /* If it's non-inverted search, we can stop
208 if (found && !invert_search)
210 pattern_ptr = pattern_ptr->link;
211 } /* while (pattern_ptr) */
213 if (found ^ invert_search) {
215 /* keep track of matches */
218 /* quiet/print (non)matching file names only? */
219 if (option_mask32 & (OPT_q|OPT_l|OPT_L)) {
220 free(line); /* we don't need line anymore */
222 /* manpage says about -q:
223 * "exit immediately with zero status
224 * if any match is found,
225 * even if errors were detected" */
228 /* if we're just printing filenames, we stop after the first match */
229 if (PRINT_FILES_WITH_MATCHES) {
231 /* fall through to "return 1" */
233 /* OPT_L aka PRINT_FILES_WITHOUT_MATCHES: return early */
234 return 1; /* one match */
237 #if ENABLE_FEATURE_GREP_CONTEXT
238 /* Were we printing context and saw next (unwanted) match? */
239 if ((option_mask32 & OPT_m) && nmatches > max_matches)
243 /* print the matched line */
244 if (PRINT_MATCH_COUNTS == 0) {
245 #if ENABLE_FEATURE_GREP_CONTEXT
246 int prevpos = (curpos == 0) ? lines_before - 1 : curpos - 1;
248 /* if we were told to print 'before' lines and there is at least
249 * one line in the circular buffer, print them */
250 if (lines_before && before_buf[prevpos] != NULL) {
251 int first_buf_entry_line_num = linenum - lines_before;
253 /* advance to the first entry in the circular buffer, and
254 * figure out the line number is of the first line in the
257 while (before_buf[idx] == NULL) {
258 idx = (idx + 1) % lines_before;
259 first_buf_entry_line_num++;
262 /* now print each line in the buffer, clearing them as we go */
263 while (before_buf[idx] != NULL) {
264 print_line(before_buf[idx], first_buf_entry_line_num, '-');
265 free(before_buf[idx]);
266 before_buf[idx] = NULL;
267 idx = (idx + 1) % lines_before;
268 first_buf_entry_line_num++;
272 /* make a note that we need to print 'after' lines */
273 print_n_lines_after = lines_after;
275 if (option_mask32 & OPT_o) {
277 /* -Fo just prints the pattern
278 * (unless -v: -Fov doesnt print anything at all) */
280 print_line(gl->pattern, linenum, ':');
282 line[regmatch.rm_eo] = '\0';
283 print_line(line + regmatch.rm_so, linenum, ':');
286 print_line(line, linenum, ':');
290 #if ENABLE_FEATURE_GREP_CONTEXT
291 else { /* no match */
292 /* if we need to print some context lines after the last match, do so */
293 if (print_n_lines_after) {
294 print_line(line, linenum, '-');
295 print_n_lines_after--;
296 } else if (lines_before) {
297 /* Add the line to the circular 'before' buffer */
298 free(before_buf[curpos]);
299 before_buf[curpos] = line;
300 curpos = (curpos + 1) % lines_before;
301 /* avoid free(line) - we took the line */
306 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
309 /* Did we print all context after last requested match? */
310 if ((option_mask32 & OPT_m)
311 && !print_n_lines_after && nmatches == max_matches)
315 /* special-case file post-processing for options where we don't print line
316 * matches, just filenames and possibly match counts */
318 /* grep -c: print [filename:]count, even if count is zero */
319 if (PRINT_MATCH_COUNTS) {
321 printf("%s:", cur_file);
322 printf("%d\n", nmatches);
325 /* grep -L: print just the filename */
326 if (PRINT_FILES_WITHOUT_MATCHES) {
327 /* nmatches is zero, no need to check it:
328 * we return 1 early if we detected a match
329 * and PRINT_FILES_WITHOUT_MATCHES is set */
336 #if ENABLE_FEATURE_CLEAN_UP
337 #define new_grep_list_data(p, m) add_grep_list_data(p, m)
338 static char *add_grep_list_data(char *pattern, int flg_used_mem)
340 #define new_grep_list_data(p, m) add_grep_list_data(p)
341 static char *add_grep_list_data(char *pattern)
344 grep_list_data_t *gl = xzalloc(sizeof(*gl));
345 gl->pattern = pattern;
346 #if ENABLE_FEATURE_CLEAN_UP
347 gl->flg_mem_alocated_compiled = flg_used_mem;
349 /*gl->flg_mem_alocated_compiled = 0;*/
354 static void load_regexes_from_file(llist_t *fopt)
361 char *ffile = cur->data;
365 f = xfopen_stdin(ffile);
366 while ((line = xmalloc_fgetline(f)) != NULL) {
367 llist_add_to(&pattern_head,
368 new_grep_list_data(line, ALLOCATED));
373 static int FAST_FUNC file_action_grep(const char *filename,
374 struct stat *statbuf UNUSED_PARAM,
376 int depth UNUSED_PARAM)
378 FILE *file = fopen_for_read(filename);
380 if (!SUPPRESS_ERR_MSGS)
381 bb_simple_perror_msg(filename);
386 *(int*)matched += grep_file(file);
391 static int grep_dir(const char *dir)
394 recursive_action(dir,
395 /* recurse=yes */ ACTION_RECURSE |
397 /* depthFirst=yes */ ACTION_DEPTHFIRST,
398 /* fileAction= */ file_action_grep,
399 /* dirAction= */ NULL,
400 /* userData= */ &matched,
405 int grep_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
406 int grep_main(int argc, char **argv)
410 llist_t *fopt = NULL;
412 /* do normal option parsing */
413 #if ENABLE_FEATURE_GREP_CONTEXT
416 /* -H unsets -h; -C unsets -A,-B; -e,-f are lists;
417 * -m,-A,-B,-C have numeric param */
418 opt_complementary = "H-h:C-AB:e::f::m+:A+:B+:C+";
421 &pattern_head, &fopt, &max_matches,
422 &lines_after, &lines_before, &Copt);
424 if (option_mask32 & OPT_C) {
425 /* -C unsets prev -A and -B, but following -A or -B
427 if (!(option_mask32 & OPT_A)) /* not overridden */
429 if (!(option_mask32 & OPT_B)) /* not overridden */
431 //option_mask32 |= OPT_A|OPT_B; /* for parser */
434 if (option_mask32 & (OPT_c|OPT_q|OPT_l|OPT_L)) {
435 option_mask32 &= ~OPT_n;
438 } else if (lines_before > 0)
439 before_buf = xzalloc(lines_before * sizeof(char *));
441 /* with auto sanity checks */
442 /* -H unsets -h; -c,-q or -l unset -n; -e,-f are lists; -m N */
443 opt_complementary = "H-h:c-n:q-n:l-n:e::f::m+";
444 getopt32(argv, OPTSTR_GREP,
445 &pattern_head, &fopt, &max_matches);
447 invert_search = ((option_mask32 & OPT_v) != 0); /* 0 | 1 */
449 if (pattern_head != NULL) {
450 /* convert char **argv to grep_list_data_t */
453 for (cur = pattern_head; cur; cur = cur->link)
454 cur->data = new_grep_list_data(cur->data, 0);
456 if (option_mask32 & OPT_f)
457 load_regexes_from_file(fopt);
459 if (ENABLE_FEATURE_GREP_FGREP_ALIAS && applet_name[0] == 'f')
460 option_mask32 |= OPT_F;
462 if (!(option_mask32 & (OPT_o | OPT_w)))
465 if (ENABLE_FEATURE_GREP_EGREP_ALIAS
466 && (applet_name[0] == 'e' || (option_mask32 & OPT_E))
468 reflags |= REG_EXTENDED;
471 if (option_mask32 & OPT_i)
472 reflags |= REG_ICASE;
477 /* if we didn't get a pattern from -e and no command file was specified,
478 * first parameter should be the pattern. no pattern, no worky */
479 if (pattern_head == NULL) {
483 pattern = new_grep_list_data(*argv++, 0);
484 llist_add_to(&pattern_head, pattern);
488 /* argv[0..(argc-1)] should be names of file to grep through. If
489 * there is more than one file to grep, we will print the filenames. */
492 /* -H / -h of course override */
493 if (option_mask32 & OPT_H)
495 if (option_mask32 & OPT_h)
498 /* If no files were specified, or '-' was specified, take input from
499 * stdin. Otherwise, we grep through all the files specified. */
504 if (!cur_file || LONE_DASH(cur_file)) {
505 cur_file = "(standard input)";
507 if (option_mask32 & OPT_r) {
509 if (stat(cur_file, &st) == 0 && S_ISDIR(st.st_mode)) {
510 if (!(option_mask32 & OPT_h))
512 matched += grep_dir(cur_file);
516 /* else: fopen(dir) will succeed, but reading won't */
517 file = fopen_for_read(cur_file);
519 if (!SUPPRESS_ERR_MSGS)
520 bb_simple_perror_msg(cur_file);
525 matched += grep_file(file);
526 fclose_if_not_stdin(file);
528 } while (--argc > 0);
530 /* destroy all the elments in the pattern list */
531 if (ENABLE_FEATURE_CLEAN_UP) {
532 while (pattern_head) {
533 llist_t *pattern_head_ptr = pattern_head;
534 grep_list_data_t *gl = (grep_list_data_t *)pattern_head_ptr->data;
536 pattern_head = pattern_head->link;
537 if (gl->flg_mem_alocated_compiled & ALLOCATED)
539 if (gl->flg_mem_alocated_compiled & COMPILED)
540 regfree(&(gl->preg));
542 free(pattern_head_ptr);
545 /* 0 = success, 1 = failed, 2 = error */
548 return !matched; /* invert return value: 0 = success, 1 = failed */