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. */
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
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" */
51 USE_FEATURE_GREP_CONTEXT(OPTBIT_A ,)
52 USE_FEATURE_GREP_CONTEXT(OPTBIT_B ,)
53 USE_FEATURE_GREP_CONTEXT(OPTBIT_C ,)
54 USE_FEATURE_GREP_CONTEXT(OPTBIT_E ,)
55 USE_DESKTOP( OPTBIT_w ,)
56 OPT_l = 1 << OPTBIT_l,
57 OPT_n = 1 << OPTBIT_n,
58 OPT_q = 1 << OPTBIT_q,
59 OPT_v = 1 << OPTBIT_v,
60 OPT_s = 1 << OPTBIT_s,
61 OPT_c = 1 << OPTBIT_c,
62 OPT_F = 1 << OPTBIT_F,
63 OPT_i = 1 << OPTBIT_i,
64 OPT_H = 1 << OPTBIT_H,
65 OPT_h = 1 << OPTBIT_h,
66 OPT_e = 1 << OPTBIT_e,
67 OPT_f = 1 << OPTBIT_f,
68 OPT_L = 1 << OPTBIT_L,
69 OPT_o = 1 << OPTBIT_o,
70 OPT_r = 1 << OPTBIT_r,
71 OPT_A = USE_FEATURE_GREP_CONTEXT((1 << OPTBIT_A)) + 0,
72 OPT_B = USE_FEATURE_GREP_CONTEXT((1 << OPTBIT_B)) + 0,
73 OPT_C = USE_FEATURE_GREP_CONTEXT((1 << OPTBIT_C)) + 0,
74 OPT_E = USE_FEATURE_GREP_CONTEXT((1 << OPTBIT_E)) + 0,
75 OPT_w = USE_DESKTOP( (1 << OPTBIT_w)) + 0,
78 #define PRINT_FILES_WITH_MATCHES (option_mask32 & OPT_l)
79 #define PRINT_LINE_NUM (option_mask32 & OPT_n)
80 #define BE_QUIET (option_mask32 & OPT_q)
81 #define SUPPRESS_ERR_MSGS (option_mask32 & OPT_s)
82 #define PRINT_MATCH_COUNTS (option_mask32 & OPT_c)
83 #define FGREP_FLAG (option_mask32 & OPT_F)
84 #define PRINT_FILES_WITHOUT_MATCHES (option_mask32 & OPT_L)
86 typedef unsigned char byte_t;
89 static byte_t invert_search;
90 static byte_t print_filename;
91 static byte_t open_errors;
93 #if ENABLE_FEATURE_GREP_CONTEXT
94 static int lines_before;
95 static int lines_after;
96 static char **before_buf;
97 static int last_line_printed;
98 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
100 /* globals used internally */
101 static llist_t *pattern_head; /* growable list of patterns to match */
102 static const char *cur_file; /* the current file we are reading */
104 typedef struct grep_list_data_t {
107 #define PATTERN_MEM_A 1
109 int flg_mem_alocated_compiled;
112 static void print_line(const char *line, int linenum, char decoration)
114 #if ENABLE_FEATURE_GREP_CONTEXT
115 /* possibly print the little '--' separator */
116 if ((lines_before || lines_after) && last_line_printed &&
117 last_line_printed < linenum - 1) {
120 last_line_printed = linenum;
123 printf("%s%c", cur_file, decoration);
125 printf("%i%c", linenum, decoration);
126 /* Emulate weird GNU grep behavior with -ov */
127 if ((option_mask32 & (OPT_v+OPT_o)) != (OPT_v+OPT_o))
131 static int grep_file(FILE *file)
138 #if ENABLE_FEATURE_GREP_CONTEXT
139 int print_n_lines_after = 0;
140 int curpos = 0; /* track where we are in the circular 'before' buffer */
141 int idx = 0; /* used for iteration through the circular buffer */
142 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
144 while ((line = xmalloc_getline(file)) != NULL) {
145 llist_t *pattern_ptr = pattern_head;
146 grep_list_data_t * gl;
150 while (pattern_ptr) {
151 gl = (grep_list_data_t *)pattern_ptr->data;
153 ret = strstr(line, gl->pattern) != NULL;
156 * test for a postitive-assertion match (regexec returns success (0)
157 * and the user did not specify invert search), or a negative-assertion
158 * match (regexec returns failure (REG_NOMATCH) and the user specified
161 if (!(gl->flg_mem_alocated_compiled & COMPILED)) {
162 gl->flg_mem_alocated_compiled |= COMPILED;
163 xregcomp(&(gl->preg), gl->pattern, reflags);
167 if (regexec(&(gl->preg), line, 1, ®match, 0) == 0) {
168 if (!(option_mask32 & OPT_w))
173 c = line[regmatch.rm_so - 1];
174 if (!isalnum(c) && c != '_') {
175 c = line[regmatch.rm_eo];
176 if (!c || (!isalnum(c) && c != '_'))
182 pattern_ptr = pattern_ptr->link;
183 } /* while (pattern_ptr) */
185 if (ret ^ invert_search) {
186 if (PRINT_FILES_WITH_MATCHES || BE_QUIET)
189 /* if we found a match but were told to be quiet, stop here */
190 if (BE_QUIET || PRINT_FILES_WITHOUT_MATCHES)
193 /* keep track of matches */
196 /* if we're just printing filenames, we stop after the first match */
197 if (PRINT_FILES_WITH_MATCHES)
200 /* print the matched line */
201 if (PRINT_MATCH_COUNTS == 0) {
202 #if ENABLE_FEATURE_GREP_CONTEXT
203 int prevpos = (curpos == 0) ? lines_before - 1 : curpos - 1;
205 /* if we were told to print 'before' lines and there is at least
206 * one line in the circular buffer, print them */
207 if (lines_before && before_buf[prevpos] != NULL) {
208 int first_buf_entry_line_num = linenum - lines_before;
210 /* advance to the first entry in the circular buffer, and
211 * figure out the line number is of the first line in the
214 while (before_buf[idx] == NULL) {
215 idx = (idx + 1) % lines_before;
216 first_buf_entry_line_num++;
219 /* now print each line in the buffer, clearing them as we go */
220 while (before_buf[idx] != NULL) {
221 print_line(before_buf[idx], first_buf_entry_line_num, '-');
222 free(before_buf[idx]);
223 before_buf[idx] = NULL;
224 idx = (idx + 1) % lines_before;
225 first_buf_entry_line_num++;
229 /* make a note that we need to print 'after' lines */
230 print_n_lines_after = lines_after;
232 if (option_mask32 & OPT_o) {
233 line[regmatch.rm_eo] = '\0';
234 print_line(line + regmatch.rm_so, linenum, ':');
236 print_line(line, linenum, ':');
240 #if ENABLE_FEATURE_GREP_CONTEXT
241 else { /* no match */
242 /* Add the line to the circular 'before' buffer */
244 free(before_buf[curpos]);
245 before_buf[curpos] = xstrdup(line);
246 curpos = (curpos + 1) % lines_before;
250 /* if we need to print some context lines after the last match, do so */
251 if (print_n_lines_after && (last_line_printed != linenum)) {
252 print_line(line, linenum, '-');
253 print_n_lines_after--;
255 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
259 /* special-case file post-processing for options where we don't print line
260 * matches, just filenames and possibly match counts */
262 /* grep -c: print [filename:]count, even if count is zero */
263 if (PRINT_MATCH_COUNTS) {
265 printf("%s:", cur_file);
266 printf("%d\n", nmatches);
269 /* grep -l: print just the filename, but only if we grepped the line in the file */
270 if (PRINT_FILES_WITH_MATCHES && nmatches > 0) {
274 /* grep -L: print just the filename, but only if we didn't grep the line in the file */
275 if (PRINT_FILES_WITHOUT_MATCHES && nmatches == 0) {
282 #if ENABLE_FEATURE_CLEAN_UP
283 #define new_grep_list_data(p, m) add_grep_list_data(p, m)
284 static char * add_grep_list_data(char *pattern, int flg_used_mem)
286 #define new_grep_list_data(p, m) add_grep_list_data(p)
287 static char * add_grep_list_data(char *pattern)
290 grep_list_data_t *gl = xmalloc(sizeof(grep_list_data_t));
291 gl->pattern = pattern;
292 #if ENABLE_FEATURE_CLEAN_UP
293 gl->flg_mem_alocated_compiled = flg_used_mem;
295 gl->flg_mem_alocated_compiled = 0;
301 static void load_regexes_from_file(llist_t *fopt)
308 char *ffile = cur->data;
312 f = xfopen(ffile, "r");
313 while ((line = xmalloc_getline(f)) != NULL) {
314 llist_add_to(&pattern_head,
315 new_grep_list_data(line, PATTERN_MEM_A));
320 static int file_action_grep(const char *filename, struct stat *statbuf, void* matched, int depth)
322 FILE *file = fopen(filename, "r");
324 if (!SUPPRESS_ERR_MSGS)
325 bb_perror_msg("%s", cur_file);
330 *(int*)matched += grep_file(file);
335 static int grep_dir(const char *dir)
338 recursive_action(dir,
339 /* recurse= */ action_recurse |
340 /* followLinks= */ /* no. 0 | */
341 /* depthFirst= */ action_depthFirst,
342 /* fileAction= */ file_action_grep,
343 /* dirAction= */ NULL,
344 /* userData= */ &matched,
349 int grep_main(int argc, char **argv);
350 int grep_main(int argc, char **argv)
354 llist_t *fopt = NULL;
356 /* do normal option parsing */
357 #if ENABLE_FEATURE_GREP_CONTEXT
362 opt_complementary = "H-h:e::f::C-AB";
365 &pattern_head, &fopt,
366 &slines_after, &slines_before, &Copt);
368 if (option_mask32 & OPT_C) {
369 /* -C unsets prev -A and -B, but following -A or -B
371 if (!(option_mask32 & OPT_A)) /* not overridden */
373 if (!(option_mask32 & OPT_B)) /* not overridden */
374 slines_before = Copt;
375 option_mask32 |= OPT_A|OPT_B; /* for parser */
377 if (option_mask32 & OPT_A) {
378 lines_after = xatoi_u(slines_after);
380 if (option_mask32 & OPT_B) {
381 lines_before = xatoi_u(slines_before);
384 if (option_mask32 & (OPT_c|OPT_q|OPT_l|OPT_L)) {
385 option_mask32 &= ~OPT_n;
388 } else if (lines_before > 0)
389 before_buf = xzalloc(lines_before * sizeof(char *));
391 /* with auto sanity checks */
392 opt_complementary = "H-h:e::f::c-n:q-n:l-n";
393 getopt32(argc, argv, OPTSTR_GREP,
394 &pattern_head, &fopt);
396 invert_search = ((option_mask32 & OPT_v) != 0); /* 0 | 1 */
398 if (pattern_head != NULL) {
399 /* convert char **argv to grep_list_data_t */
402 for (cur = pattern_head; cur; cur = cur->link)
403 cur->data = new_grep_list_data(cur->data, 0);
405 if (option_mask32 & OPT_f)
406 load_regexes_from_file(fopt);
408 if (ENABLE_FEATURE_GREP_FGREP_ALIAS && applet_name[0] == 'f')
409 option_mask32 |= OPT_F;
411 if (!(option_mask32 & (OPT_o | OPT_w)))
414 if (ENABLE_FEATURE_GREP_EGREP_ALIAS &&
415 (applet_name[0] == 'e' || (option_mask32 & OPT_E)))
416 reflags |= REG_EXTENDED;
418 if (option_mask32 & OPT_i)
419 reflags |= REG_ICASE;
424 /* if we didn't get a pattern from a -e and no command file was specified,
425 * argv[optind] should be the pattern. no pattern, no worky */
426 if (pattern_head == NULL) {
430 pattern = new_grep_list_data(*argv++, 0);
431 llist_add_to(&pattern_head, pattern);
435 /* argv[(optind)..(argc-1)] should be names of file to grep through. If
436 * there is more than one file to grep, we will print the filenames. */
439 /* -H / -h of course override */
440 if (option_mask32 & OPT_H)
442 if (option_mask32 & OPT_h)
445 /* If no files were specified, or '-' was specified, take input from
446 * stdin. Otherwise, we grep through all the files specified. */
453 if (!cur_file || (*cur_file == '-' && !cur_file[1])) {
454 cur_file = "(standard input)";
456 if (option_mask32 & OPT_r) {
458 if (stat(cur_file, &st) == 0 && S_ISDIR(st.st_mode)) {
459 if (!(option_mask32 & OPT_h))
461 matched += grep_dir(cur_file);
465 /* else: fopen(dir) will succeed, but reading won't */
466 file = fopen(cur_file, "r");
468 if (!SUPPRESS_ERR_MSGS)
469 bb_perror_msg("%s", cur_file);
474 matched += grep_file(file);
475 fclose_if_not_stdin(file);
478 /* we found a match but were told to be quiet, stop here and
484 /* destroy all the elments in the pattern list */
485 if (ENABLE_FEATURE_CLEAN_UP) {
486 while (pattern_head) {
487 llist_t *pattern_head_ptr = pattern_head;
488 grep_list_data_t *gl =
489 (grep_list_data_t *)pattern_head_ptr->data;
491 pattern_head = pattern_head->link;
492 if ((gl->flg_mem_alocated_compiled & PATTERN_MEM_A))
494 if ((gl->flg_mem_alocated_compiled & COMPILED))
495 regfree(&(gl->preg));
496 free(pattern_head_ptr);
499 /* 0 = success, 1 = failed, 2 = error */
500 /* If the -q option is specified, the exit status shall be zero
501 * if an input line is selected, even if an error was detected. */
502 if (BE_QUIET && matched)
506 return !matched; /* invert return value 0 = success, 1 = failed */