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)
108 struct G_sizecheck { \
109 char G_sizecheck[sizeof(G) > COMMON_BUFSIZE ? -1 : 1]; \
112 #define max_matches (G.max_matches )
113 #define reflags (G.reflags )
114 #define invert_search (G.invert_search )
115 #define print_filename (G.print_filename )
116 #define open_errors (G.open_errors )
117 #define did_print_line (G.did_print_line )
118 #define lines_before (G.lines_before )
119 #define lines_after (G.lines_after )
120 #define before_buf (G.before_buf )
121 #define last_line_printed (G.last_line_printed )
122 #define pattern_head (G.pattern_head )
123 #define cur_file (G.cur_file )
126 typedef struct grep_list_data_t {
131 int flg_mem_alocated_compiled;
135 static void print_line(const char *line, int linenum, char decoration)
137 #if ENABLE_FEATURE_GREP_CONTEXT
138 /* Happens when we go to next file, immediately hit match
139 * and try to print prev context... from prev file! Don't do it */
142 /* possibly print the little '--' separator */
143 if ((lines_before || lines_after) && did_print_line &&
144 last_line_printed != linenum - 1) {
147 /* guard against printing "--" before first line of first file */
149 last_line_printed = linenum;
152 printf("%s%c", cur_file, decoration);
154 printf("%i%c", linenum, decoration);
155 /* Emulate weird GNU grep behavior with -ov */
156 if ((option_mask32 & (OPT_v|OPT_o)) != (OPT_v|OPT_o))
160 static int grep_file(FILE *file)
167 #if ENABLE_FEATURE_GREP_CONTEXT
168 int print_n_lines_after = 0;
169 int curpos = 0; /* track where we are in the circular 'before' buffer */
170 int idx = 0; /* used for iteration through the circular buffer */
172 enum { print_n_lines_after = 0 };
173 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
175 while ((line = xmalloc_fgetline(file)) != NULL) {
176 llist_t *pattern_ptr = pattern_head;
177 grep_list_data_t *gl = gl; /* for gcc */
181 while (pattern_ptr) {
182 gl = (grep_list_data_t *)pattern_ptr->data;
184 found |= (strstr(line, gl->pattern) != NULL);
186 if (!(gl->flg_mem_alocated_compiled & COMPILED)) {
187 gl->flg_mem_alocated_compiled |= COMPILED;
188 xregcomp(&(gl->preg), gl->pattern, reflags);
192 if (regexec(&(gl->preg), line, 1, ®match, 0) == 0) {
193 if (!(option_mask32 & OPT_w))
198 c = line[regmatch.rm_so - 1];
199 if (!isalnum(c) && c != '_') {
200 c = line[regmatch.rm_eo];
201 if (!c || (!isalnum(c) && c != '_'))
207 /* If it's non-inverted search, we can stop
209 if (found && !invert_search)
211 pattern_ptr = pattern_ptr->link;
212 } /* while (pattern_ptr) */
214 if (found ^ invert_search) {
216 /* keep track of matches */
219 /* quiet/print (non)matching file names only? */
220 if (option_mask32 & (OPT_q|OPT_l|OPT_L)) {
221 free(line); /* we don't need line anymore */
223 /* manpage says about -q:
224 * "exit immediately with zero status
225 * if any match is found,
226 * even if errors were detected" */
229 /* if we're just printing filenames, we stop after the first match */
230 if (PRINT_FILES_WITH_MATCHES) {
232 /* fall through to "return 1" */
234 /* OPT_L aka PRINT_FILES_WITHOUT_MATCHES: return early */
235 return 1; /* one match */
238 #if ENABLE_FEATURE_GREP_CONTEXT
239 /* Were we printing context and saw next (unwanted) match? */
240 if ((option_mask32 & OPT_m) && nmatches > max_matches)
244 /* print the matched line */
245 if (PRINT_MATCH_COUNTS == 0) {
246 #if ENABLE_FEATURE_GREP_CONTEXT
247 int prevpos = (curpos == 0) ? lines_before - 1 : curpos - 1;
249 /* if we were told to print 'before' lines and there is at least
250 * one line in the circular buffer, print them */
251 if (lines_before && before_buf[prevpos] != NULL) {
252 int first_buf_entry_line_num = linenum - lines_before;
254 /* advance to the first entry in the circular buffer, and
255 * figure out the line number is of the first line in the
258 while (before_buf[idx] == NULL) {
259 idx = (idx + 1) % lines_before;
260 first_buf_entry_line_num++;
263 /* now print each line in the buffer, clearing them as we go */
264 while (before_buf[idx] != NULL) {
265 print_line(before_buf[idx], first_buf_entry_line_num, '-');
266 free(before_buf[idx]);
267 before_buf[idx] = NULL;
268 idx = (idx + 1) % lines_before;
269 first_buf_entry_line_num++;
273 /* make a note that we need to print 'after' lines */
274 print_n_lines_after = lines_after;
276 if (option_mask32 & OPT_o) {
278 /* -Fo just prints the pattern
279 * (unless -v: -Fov doesnt print anything at all) */
281 print_line(gl->pattern, linenum, ':');
283 line[regmatch.rm_eo] = '\0';
284 print_line(line + regmatch.rm_so, linenum, ':');
287 print_line(line, linenum, ':');
291 #if ENABLE_FEATURE_GREP_CONTEXT
292 else { /* no match */
293 /* if we need to print some context lines after the last match, do so */
294 if (print_n_lines_after) {
295 print_line(line, linenum, '-');
296 print_n_lines_after--;
297 } else if (lines_before) {
298 /* Add the line to the circular 'before' buffer */
299 free(before_buf[curpos]);
300 before_buf[curpos] = line;
301 curpos = (curpos + 1) % lines_before;
302 /* avoid free(line) - we took the line */
307 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
310 /* Did we print all context after last requested match? */
311 if ((option_mask32 & OPT_m)
312 && !print_n_lines_after && nmatches == max_matches)
316 /* special-case file post-processing for options where we don't print line
317 * matches, just filenames and possibly match counts */
319 /* grep -c: print [filename:]count, even if count is zero */
320 if (PRINT_MATCH_COUNTS) {
322 printf("%s:", cur_file);
323 printf("%d\n", nmatches);
326 /* grep -L: print just the filename */
327 if (PRINT_FILES_WITHOUT_MATCHES) {
328 /* nmatches is zero, no need to check it:
329 * we return 1 early if we detected a match
330 * and PRINT_FILES_WITHOUT_MATCHES is set */
337 #if ENABLE_FEATURE_CLEAN_UP
338 #define new_grep_list_data(p, m) add_grep_list_data(p, m)
339 static char *add_grep_list_data(char *pattern, int flg_used_mem)
341 #define new_grep_list_data(p, m) add_grep_list_data(p)
342 static char *add_grep_list_data(char *pattern)
345 grep_list_data_t *gl = xzalloc(sizeof(*gl));
346 gl->pattern = pattern;
347 #if ENABLE_FEATURE_CLEAN_UP
348 gl->flg_mem_alocated_compiled = flg_used_mem;
350 /*gl->flg_mem_alocated_compiled = 0;*/
355 static void load_regexes_from_file(llist_t *fopt)
362 char *ffile = cur->data;
366 f = xfopen_stdin(ffile);
367 while ((line = xmalloc_fgetline(f)) != NULL) {
368 llist_add_to(&pattern_head,
369 new_grep_list_data(line, ALLOCATED));
374 static int file_action_grep(const char *filename,
375 struct stat *statbuf ATTRIBUTE_UNUSED,
377 int depth ATTRIBUTE_UNUSED)
379 FILE *file = fopen(filename, "r");
381 if (!SUPPRESS_ERR_MSGS)
382 bb_simple_perror_msg(filename);
387 *(int*)matched += grep_file(file);
392 static int grep_dir(const char *dir)
395 recursive_action(dir,
396 /* recurse=yes */ ACTION_RECURSE |
398 /* depthFirst=yes */ ACTION_DEPTHFIRST,
399 /* fileAction= */ file_action_grep,
400 /* dirAction= */ NULL,
401 /* userData= */ &matched,
406 int grep_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
407 int grep_main(int argc, char **argv)
411 llist_t *fopt = NULL;
413 /* do normal option parsing */
414 #if ENABLE_FEATURE_GREP_CONTEXT
417 /* -H unsets -h; -C unsets -A,-B; -e,-f are lists;
418 * -m,-A,-B,-C have numeric param */
419 opt_complementary = "H-h:C-AB:e::f::m+:A+:B+:C+";
422 &pattern_head, &fopt, &max_matches,
423 &lines_after, &lines_before, &Copt);
425 if (option_mask32 & OPT_C) {
426 /* -C unsets prev -A and -B, but following -A or -B
428 if (!(option_mask32 & OPT_A)) /* not overridden */
430 if (!(option_mask32 & OPT_B)) /* not overridden */
432 //option_mask32 |= OPT_A|OPT_B; /* for parser */
435 if (option_mask32 & (OPT_c|OPT_q|OPT_l|OPT_L)) {
436 option_mask32 &= ~OPT_n;
439 } else if (lines_before > 0)
440 before_buf = xzalloc(lines_before * sizeof(char *));
442 /* with auto sanity checks */
443 /* -H unsets -h; -c,-q or -l unset -n; -e,-f are lists; -m N */
444 opt_complementary = "H-h:c-n:q-n:l-n:e::f::m+";
445 getopt32(argv, OPTSTR_GREP,
446 &pattern_head, &fopt, &max_matches);
448 invert_search = ((option_mask32 & OPT_v) != 0); /* 0 | 1 */
450 if (pattern_head != NULL) {
451 /* convert char **argv to grep_list_data_t */
454 for (cur = pattern_head; cur; cur = cur->link)
455 cur->data = new_grep_list_data(cur->data, 0);
457 if (option_mask32 & OPT_f)
458 load_regexes_from_file(fopt);
460 if (ENABLE_FEATURE_GREP_FGREP_ALIAS && applet_name[0] == 'f')
461 option_mask32 |= OPT_F;
463 if (!(option_mask32 & (OPT_o | OPT_w)))
466 if (ENABLE_FEATURE_GREP_EGREP_ALIAS
467 && (applet_name[0] == 'e' || (option_mask32 & OPT_E))
469 reflags |= REG_EXTENDED;
472 if (option_mask32 & OPT_i)
473 reflags |= REG_ICASE;
478 /* if we didn't get a pattern from -e and no command file was specified,
479 * first parameter should be the pattern. no pattern, no worky */
480 if (pattern_head == NULL) {
484 pattern = new_grep_list_data(*argv++, 0);
485 llist_add_to(&pattern_head, pattern);
489 /* argv[0..(argc-1)] should be names of file to grep through. If
490 * there is more than one file to grep, we will print the filenames. */
493 /* -H / -h of course override */
494 if (option_mask32 & OPT_H)
496 if (option_mask32 & OPT_h)
499 /* If no files were specified, or '-' was specified, take input from
500 * stdin. Otherwise, we grep through all the files specified. */
505 if (!cur_file || LONE_DASH(cur_file)) {
506 cur_file = "(standard input)";
508 if (option_mask32 & OPT_r) {
510 if (stat(cur_file, &st) == 0 && S_ISDIR(st.st_mode)) {
511 if (!(option_mask32 & OPT_h))
513 matched += grep_dir(cur_file);
517 /* else: fopen(dir) will succeed, but reading won't */
518 file = fopen(cur_file, "r");
520 if (!SUPPRESS_ERR_MSGS)
521 bb_simple_perror_msg(cur_file);
526 matched += grep_file(file);
527 fclose_if_not_stdin(file);
529 } while (--argc > 0);
531 /* destroy all the elments in the pattern list */
532 if (ENABLE_FEATURE_CLEAN_UP) {
533 while (pattern_head) {
534 llist_t *pattern_head_ptr = pattern_head;
535 grep_list_data_t *gl = (grep_list_data_t *)pattern_head_ptr->data;
537 pattern_head = pattern_head->link;
538 if (gl->flg_mem_alocated_compiled & ALLOCATED)
540 if (gl->flg_mem_alocated_compiled & COMPILED)
541 regfree(&(gl->preg));
543 free(pattern_head_ptr);
546 /* 0 = success, 1 = failed, 2 = error */
549 return !matched; /* invert return value: 0 = success, 1 = failed */