aboutsummaryrefslogtreecommitdiff
path: root/findutils/grep.c
diff options
context:
space:
mode:
Diffstat (limited to 'findutils/grep.c')
-rw-r--r--findutils/grep.c483
1 files changed, 483 insertions, 0 deletions
diff --git a/findutils/grep.c b/findutils/grep.c
new file mode 100644
index 000000000..8bb38f95c
--- /dev/null
+++ b/findutils/grep.c
@@ -0,0 +1,483 @@
1/* vi: set sw=4 ts=4: */
2/*
3 * Mini grep implementation for busybox using libc regex.
4 *
5 * Copyright (C) 1999,2000,2001 by Lineo, inc. and Mark Whitley
6 * Copyright (C) 1999,2000,2001 by Mark Whitley <markw@codepoet.org>
7 *
8 * Licensed under the GPL v2 or later, see the file LICENSE in this tarball.
9 */
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 */
13/*
14 * 2004,2006 (C) Vladimir Oleynik <dzo@simtreas.ru> -
15 * correction "-e pattern1 -e pattern2" logic and more optimizations.
16 * precompiled regex
17 */
18/*
19 * (C) 2006 Jac Goudsmit added -o option
20 */
21
22#include "busybox.h"
23#include "xregex.h"
24
25
26/* options */
27#define GREP_OPTS "lnqvscFiHhe:f:Lor"
28#define GREP_OPT_l (1<<0)
29#define PRINT_FILES_WITH_MATCHES (option_mask32 & GREP_OPT_l)
30#define GREP_OPT_n (1<<1)
31#define PRINT_LINE_NUM (option_mask32 & GREP_OPT_n)
32#define GREP_OPT_q (1<<2)
33#define BE_QUIET (option_mask32 & GREP_OPT_q)
34#define GREP_OPT_v (1<<3)
35#define GREP_OPT_s (1<<4)
36#define SUPPRESS_ERR_MSGS (option_mask32 & GREP_OPT_s)
37#define GREP_OPT_c (1<<5)
38#define PRINT_MATCH_COUNTS (option_mask32 & GREP_OPT_c)
39#define GREP_OPT_F (1<<6)
40#define FGREP_FLAG (option_mask32 & GREP_OPT_F)
41#define GREP_OPT_i (1<<7)
42#define GREP_OPT_H (1<<8)
43#define GREP_OPT_h (1<<9)
44#define GREP_OPT_e (1<<10)
45#define GREP_OPT_f (1<<11)
46#define GREP_OPT_L (1<<12)
47#define PRINT_FILES_WITHOUT_MATCHES (option_mask32 & GREP_OPT_L)
48#define GREP_OPT_o (1<<13)
49#define GREP_OPT_r (1<<14)
50#if ENABLE_FEATURE_GREP_CONTEXT
51# define GREP_OPT_CONTEXT "A:B:C:"
52# define GREP_OPT_A (1<<15)
53# define GREP_OPT_B (1<<16)
54# define GREP_OPT_C (1<<17)
55# define GREP_OPT_E (1<<18)
56#else
57# define GREP_OPT_CONTEXT ""
58# define GREP_OPT_A 0
59# define GREP_OPT_B 0
60# define GREP_OPT_C 0
61# define GREP_OPT_E (1<<15)
62#endif
63#if ENABLE_FEATURE_GREP_EGREP_ALIAS
64# define OPT_EGREP "E"
65#else
66# define OPT_EGREP ""
67#endif
68
69typedef unsigned char byte_t;
70
71static int reflags;
72static byte_t invert_search;
73static byte_t print_filename;
74static byte_t open_errors;
75
76#if ENABLE_FEATURE_GREP_CONTEXT
77static int lines_before;
78static int lines_after;
79static char **before_buf;
80static int last_line_printed;
81#endif /* ENABLE_FEATURE_GREP_CONTEXT */
82
83/* globals used internally */
84static llist_t *pattern_head; /* growable list of patterns to match */
85static const char *cur_file; /* the current file we are reading */
86
87typedef struct GREP_LIST_DATA {
88 char *pattern;
89 regex_t preg;
90#define PATTERN_MEM_A 1
91#define COMPILED 2
92 int flg_mem_alocated_compiled;
93} grep_list_data_t;
94
95static void print_line(const char *line, int linenum, char decoration)
96{
97#if ENABLE_FEATURE_GREP_CONTEXT
98 /* possibly print the little '--' separator */
99 if ((lines_before || lines_after) && last_line_printed &&
100 last_line_printed < linenum - 1) {
101 puts("--");
102 }
103 last_line_printed = linenum;
104#endif
105 if (print_filename)
106 printf("%s%c", cur_file, decoration);
107 if (PRINT_LINE_NUM)
108 printf("%i%c", linenum, decoration);
109 /* Emulate weird GNU grep behavior with -ov */
110 if ((option_mask32 & (GREP_OPT_v+GREP_OPT_o)) != (GREP_OPT_v+GREP_OPT_o))
111 puts(line);
112}
113
114
115static int grep_file(FILE *file)
116{
117 char *line;
118 byte_t ret;
119 int linenum = 0;
120 int nmatches = 0;
121 regmatch_t regmatch;
122#if ENABLE_FEATURE_GREP_CONTEXT
123 int print_n_lines_after = 0;
124 int curpos = 0; /* track where we are in the circular 'before' buffer */
125 int idx = 0; /* used for iteration through the circular buffer */
126#endif /* ENABLE_FEATURE_GREP_CONTEXT */
127
128 while ((line = xmalloc_getline(file)) != NULL) {
129 llist_t *pattern_ptr = pattern_head;
130 grep_list_data_t * gl;
131
132 linenum++;
133 ret = 0;
134 while (pattern_ptr) {
135 gl = (grep_list_data_t *)pattern_ptr->data;
136 if (FGREP_FLAG) {
137 ret = strstr(line, gl->pattern) != NULL;
138 } else {
139 /*
140 * test for a postitive-assertion match (regexec returns success (0)
141 * and the user did not specify invert search), or a negative-assertion
142 * match (regexec returns failure (REG_NOMATCH) and the user specified
143 * invert search)
144 */
145 if (!(gl->flg_mem_alocated_compiled & COMPILED)) {
146 gl->flg_mem_alocated_compiled |= COMPILED;
147 xregcomp(&(gl->preg), gl->pattern, reflags);
148 }
149 regmatch.rm_so = 0;
150 regmatch.rm_eo = 0;
151 ret |= regexec(&(gl->preg), line, 1, &regmatch, 0) == 0;
152 }
153 pattern_ptr = pattern_ptr->link;
154 } /* while (pattern_ptr) */
155
156 if (ret ^ invert_search) {
157
158 if (PRINT_FILES_WITH_MATCHES || BE_QUIET)
159 free(line);
160
161 /* if we found a match but were told to be quiet, stop here */
162 if (BE_QUIET || PRINT_FILES_WITHOUT_MATCHES)
163 return -1;
164
165 /* keep track of matches */
166 nmatches++;
167
168 /* if we're just printing filenames, we stop after the first match */
169 if (PRINT_FILES_WITH_MATCHES)
170 break;
171
172 /* print the matched line */
173 if (PRINT_MATCH_COUNTS == 0) {
174#if ENABLE_FEATURE_GREP_CONTEXT
175 int prevpos = (curpos == 0) ? lines_before - 1 : curpos - 1;
176
177 /* if we were told to print 'before' lines and there is at least
178 * one line in the circular buffer, print them */
179 if (lines_before && before_buf[prevpos] != NULL) {
180 int first_buf_entry_line_num = linenum - lines_before;
181
182 /* advance to the first entry in the circular buffer, and
183 * figure out the line number is of the first line in the
184 * buffer */
185 idx = curpos;
186 while (before_buf[idx] == NULL) {
187 idx = (idx + 1) % lines_before;
188 first_buf_entry_line_num++;
189 }
190
191 /* now print each line in the buffer, clearing them as we go */
192 while (before_buf[idx] != NULL) {
193 print_line(before_buf[idx], first_buf_entry_line_num, '-');
194 free(before_buf[idx]);
195 before_buf[idx] = NULL;
196 idx = (idx + 1) % lines_before;
197 first_buf_entry_line_num++;
198 }
199 }
200
201 /* make a note that we need to print 'after' lines */
202 print_n_lines_after = lines_after;
203#endif
204 if (option_mask32 & GREP_OPT_o) {
205 line[regmatch.rm_eo] = '\0';
206 print_line(line + regmatch.rm_so, linenum, ':');
207 } else {
208 print_line(line, linenum, ':');
209 }
210 }
211 }
212#if ENABLE_FEATURE_GREP_CONTEXT
213 else { /* no match */
214 /* Add the line to the circular 'before' buffer */
215 if (lines_before) {
216 free(before_buf[curpos]);
217 before_buf[curpos] = xstrdup(line);
218 curpos = (curpos + 1) % lines_before;
219 }
220 }
221
222 /* if we need to print some context lines after the last match, do so */
223 if (print_n_lines_after && (last_line_printed != linenum)) {
224 print_line(line, linenum, '-');
225 print_n_lines_after--;
226 }
227#endif /* ENABLE_FEATURE_GREP_CONTEXT */
228 free(line);
229 }
230
231 /* special-case file post-processing for options where we don't print line
232 * matches, just filenames and possibly match counts */
233
234 /* grep -c: print [filename:]count, even if count is zero */
235 if (PRINT_MATCH_COUNTS) {
236 if (print_filename)
237 printf("%s:", cur_file);
238 printf("%d\n", nmatches);
239 }
240
241 /* grep -l: print just the filename, but only if we grepped the line in the file */
242 if (PRINT_FILES_WITH_MATCHES && nmatches > 0) {
243 puts(cur_file);
244 }
245
246 /* grep -L: print just the filename, but only if we didn't grep the line in the file */
247 if (PRINT_FILES_WITHOUT_MATCHES && nmatches == 0) {
248 puts(cur_file);
249 }
250
251 return nmatches;
252}
253
254#if ENABLE_FEATURE_CLEAN_UP
255#define new_grep_list_data(p, m) add_grep_list_data(p, m)
256static char * add_grep_list_data(char *pattern, int flg_used_mem)
257#else
258#define new_grep_list_data(p, m) add_grep_list_data(p)
259static char * add_grep_list_data(char *pattern)
260#endif
261{
262 grep_list_data_t *gl = xmalloc(sizeof(grep_list_data_t));
263 gl->pattern = pattern;
264#if ENABLE_FEATURE_CLEAN_UP
265 gl->flg_mem_alocated_compiled = flg_used_mem;
266#else
267 gl->flg_mem_alocated_compiled = 0;
268#endif
269 return (char *)gl;
270}
271
272
273static void load_regexes_from_file(llist_t *fopt)
274{
275 char *line;
276 FILE *f;
277
278 while (fopt) {
279 llist_t *cur = fopt;
280 char *ffile = cur->data;
281
282 fopt = cur->link;
283 free(cur);
284 f = xfopen(ffile, "r");
285 while ((line = xmalloc_getline(f)) != NULL) {
286 llist_add_to(&pattern_head,
287 new_grep_list_data(line, PATTERN_MEM_A));
288 }
289 }
290}
291
292
293static int file_action_grep(const char *filename, struct stat *statbuf, void* matched, int depth)
294{
295 FILE *file = fopen(filename, "r");
296 if (file == NULL) {
297 if (!SUPPRESS_ERR_MSGS)
298 bb_perror_msg("%s", cur_file);
299 open_errors = 1;
300 return 0;
301 }
302 cur_file = filename;
303 *(int*)matched += grep_file(file);
304 fclose(file);
305 return 1;
306}
307
308
309static int grep_dir(const char *dir)
310{
311 int matched = 0;
312 recursive_action(dir,
313 /* recurse= */ 1,
314 /* followLinks= */ 0,
315 /* depthFirst= */ 1,
316 /* fileAction= */ file_action_grep,
317 /* dirAction= */ NULL,
318 /* userData= */ &matched,
319 /* depth= */ 0);
320 return matched;
321}
322
323
324int grep_main(int argc, char **argv)
325{
326 FILE *file;
327 int matched;
328 llist_t *fopt = NULL;
329
330 /* do normal option parsing */
331#if ENABLE_FEATURE_GREP_CONTEXT
332 char *slines_after;
333 char *slines_before;
334 char *Copt;
335
336 opt_complementary = "H-h:e::f::C-AB";
337 getopt32(argc, argv,
338 GREP_OPTS GREP_OPT_CONTEXT OPT_EGREP,
339 &pattern_head, &fopt,
340 &slines_after, &slines_before, &Copt);
341
342 if (option_mask32 & GREP_OPT_C) {
343 /* -C unsets prev -A and -B, but following -A or -B
344 may override it */
345 if (!(option_mask32 & GREP_OPT_A)) /* not overridden */
346 slines_after = Copt;
347 if (!(option_mask32 & GREP_OPT_B)) /* not overridden */
348 slines_before = Copt;
349 option_mask32 |= GREP_OPT_A|GREP_OPT_B; /* for parser */
350 }
351 if (option_mask32 & GREP_OPT_A) {
352 lines_after = xatoi_u(slines_after);
353 }
354 if (option_mask32 & GREP_OPT_B) {
355 lines_before = xatoi_u(slines_before);
356 }
357 /* sanity checks */
358 if (option_mask32 & (GREP_OPT_c|GREP_OPT_q|GREP_OPT_l|GREP_OPT_L)) {
359 option_mask32 &= ~GREP_OPT_n;
360 lines_before = 0;
361 lines_after = 0;
362 } else if (lines_before > 0)
363 before_buf = (char **)xzalloc(lines_before * sizeof(char *));
364#else
365 /* with auto sanity checks */
366 opt_complementary = "H-h:e::f::c-n:q-n:l-n";
367 getopt32(argc, argv, GREP_OPTS OPT_EGREP,
368 &pattern_head, &fopt);
369#endif
370 invert_search = ((option_mask32 & GREP_OPT_v) != 0); /* 0 | 1 */
371
372 if (pattern_head != NULL) {
373 /* convert char *argv[] to grep_list_data_t */
374 llist_t *cur;
375
376 for (cur = pattern_head; cur; cur = cur->link)
377 cur->data = new_grep_list_data(cur->data, 0);
378 }
379 if (option_mask32 & GREP_OPT_f)
380 load_regexes_from_file(fopt);
381
382 if (ENABLE_FEATURE_GREP_FGREP_ALIAS && applet_name[0] == 'f')
383 option_mask32 |= GREP_OPT_F;
384
385 if (!(option_mask32 & GREP_OPT_o))
386 reflags = REG_NOSUB;
387
388 if (ENABLE_FEATURE_GREP_EGREP_ALIAS &&
389 (applet_name[0] == 'e' || (option_mask32 & GREP_OPT_E)))
390 reflags |= REG_EXTENDED;
391
392 if (option_mask32 & GREP_OPT_i)
393 reflags |= REG_ICASE;
394
395 argv += optind;
396 argc -= optind;
397
398 /* if we didn't get a pattern from a -e and no command file was specified,
399 * argv[optind] should be the pattern. no pattern, no worky */
400 if (pattern_head == NULL) {
401 if (*argv == NULL)
402 bb_show_usage();
403 else {
404 char *pattern = new_grep_list_data(*argv++, 0);
405
406 llist_add_to(&pattern_head, pattern);
407 argc--;
408 }
409 }
410
411 /* argv[(optind)..(argc-1)] should be names of file to grep through. If
412 * there is more than one file to grep, we will print the filenames. */
413 if (argc > 1)
414 print_filename = 1;
415 /* -H / -h of course override */
416 if (option_mask32 & GREP_OPT_H)
417 print_filename = 1;
418 if (option_mask32 & GREP_OPT_h)
419 print_filename = 0;
420
421 /* If no files were specified, or '-' was specified, take input from
422 * stdin. Otherwise, we grep through all the files specified. */
423 if (argc == 0)
424 argc++;
425 matched = 0;
426 while (argc--) {
427 cur_file = *argv++;
428 file = stdin;
429 if (!cur_file || (*cur_file == '-' && !cur_file[1])) {
430 cur_file = "(standard input)";
431 } else {
432 if (option_mask32 & GREP_OPT_r) {
433 struct stat st;
434 if (stat(cur_file, &st) == 0 && S_ISDIR(st.st_mode)) {
435 if (!(option_mask32 & GREP_OPT_h))
436 print_filename = 1;
437 matched += grep_dir(cur_file);
438 goto grep_done;
439 }
440 }
441 /* else: fopen(dir) will succeed, but reading won't */
442 file = fopen(cur_file, "r");
443 if (file == NULL) {
444 if (!SUPPRESS_ERR_MSGS)
445 bb_perror_msg("%s", cur_file);
446 open_errors = 1;
447 continue;
448 }
449 }
450 matched += grep_file(file);
451 fclose_if_not_stdin(file);
452 grep_done:
453 if (matched < 0) {
454 /* we found a match but were told to be quiet, stop here and
455 * return success */
456 break;
457 }
458 }
459
460 /* destroy all the elments in the pattern list */
461 if (ENABLE_FEATURE_CLEAN_UP) {
462 while (pattern_head) {
463 llist_t *pattern_head_ptr = pattern_head;
464 grep_list_data_t *gl =
465 (grep_list_data_t *)pattern_head_ptr->data;
466
467 pattern_head = pattern_head->link;
468 if ((gl->flg_mem_alocated_compiled & PATTERN_MEM_A))
469 free(gl->pattern);
470 if ((gl->flg_mem_alocated_compiled & COMPILED))
471 regfree(&(gl->preg));
472 free(pattern_head_ptr);
473 }
474 }
475 /* 0 = success, 1 = failed, 2 = error */
476 /* If the -q option is specified, the exit status shall be zero
477 * if an input line is selected, even if an error was detected. */
478 if (BE_QUIET && matched)
479 return 0;
480 if (open_errors)
481 return 2;
482 return !matched; /* invert return value 0 = success, 1 = failed */
483}