blob: 22101f3f4e9ae692ee82a4b383158d080e659929
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 GPLv2 or later, see file LICENSE in this source tree. |
9 | */ |
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 */ |
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 | * (C) 2006 Jac Goudsmit added -o option |
19 | */ |
20 | |
21 | //applet:IF_GREP(APPLET(grep, BB_DIR_BIN, BB_SUID_DROP)) |
22 | //applet:IF_FEATURE_GREP_EGREP_ALIAS(APPLET_ODDNAME(egrep, grep, BB_DIR_BIN, BB_SUID_DROP, egrep)) |
23 | //applet:IF_FEATURE_GREP_FGREP_ALIAS(APPLET_ODDNAME(fgrep, grep, BB_DIR_BIN, BB_SUID_DROP, fgrep)) |
24 | |
25 | //kbuild:lib-$(CONFIG_GREP) += grep.o |
26 | |
27 | //config:config GREP |
28 | //config: bool "grep" |
29 | //config: default y |
30 | //config: help |
31 | //config: grep is used to search files for a specified pattern. |
32 | //config: |
33 | //config:config FEATURE_GREP_EGREP_ALIAS |
34 | //config: bool "Enable extended regular expressions (egrep & grep -E)" |
35 | //config: default y |
36 | //config: depends on GREP |
37 | //config: help |
38 | //config: Enabled support for extended regular expressions. Extended |
39 | //config: regular expressions allow for alternation (foo|bar), grouping, |
40 | //config: and various repetition operators. |
41 | //config: |
42 | //config:config FEATURE_GREP_FGREP_ALIAS |
43 | //config: bool "Alias fgrep to grep -F" |
44 | //config: default y |
45 | //config: depends on GREP |
46 | //config: help |
47 | //config: fgrep sees the search pattern as a normal string rather than |
48 | //config: regular expressions. |
49 | //config: grep -F always works, this just creates the fgrep alias. |
50 | //config: |
51 | //config:config FEATURE_GREP_CONTEXT |
52 | //config: bool "Enable before and after context flags (-A, -B and -C)" |
53 | //config: default y |
54 | //config: depends on GREP |
55 | //config: help |
56 | //config: Print the specified number of leading (-B) and/or trailing (-A) |
57 | //config: context surrounding our matching lines. |
58 | //config: Print the specified number of context lines (-C). |
59 | |
60 | #include "libbb.h" |
61 | #include "xregex.h" |
62 | |
63 | |
64 | /* options */ |
65 | //usage:#define grep_trivial_usage |
66 | //usage: "[-HhnlLoqvsriw" |
67 | //usage: "F" |
68 | //usage: IF_FEATURE_GREP_EGREP_ALIAS("E") |
69 | //usage: IF_EXTRA_COMPAT("z") |
70 | //usage: "] [-m N] " |
71 | //usage: IF_FEATURE_GREP_CONTEXT("[-A/B/C N] ") |
72 | //usage: "PATTERN/-e PATTERN.../-f FILE [FILE]..." |
73 | //usage:#define grep_full_usage "\n\n" |
74 | //usage: "Search for PATTERN in FILEs (or stdin)\n" |
75 | //usage: "\n -H Add 'filename:' prefix" |
76 | //usage: "\n -h Do not add 'filename:' prefix" |
77 | //usage: "\n -n Add 'line_no:' prefix" |
78 | //usage: "\n -l Show only names of files that match" |
79 | //usage: "\n -L Show only names of files that don't match" |
80 | //usage: "\n -c Show only count of matching lines" |
81 | //usage: "\n -o Show only the matching part of line" |
82 | //usage: "\n -q Quiet. Return 0 if PATTERN is found, 1 otherwise" |
83 | //usage: "\n -v Select non-matching lines" |
84 | //usage: "\n -s Suppress open and read errors" |
85 | //usage: "\n -r Recurse" |
86 | //usage: "\n -i Ignore case" |
87 | //usage: "\n -w Match whole words only" |
88 | //usage: "\n -x Match whole lines only" |
89 | //usage: "\n -F PATTERN is a literal (not regexp)" |
90 | //usage: IF_FEATURE_GREP_EGREP_ALIAS( |
91 | //usage: "\n -E PATTERN is an extended regexp" |
92 | //usage: ) |
93 | //usage: IF_EXTRA_COMPAT( |
94 | //usage: "\n -z Input is NUL terminated" |
95 | //usage: ) |
96 | //usage: "\n -m N Match up to N times per file" |
97 | //usage: IF_FEATURE_GREP_CONTEXT( |
98 | //usage: "\n -A N Print N lines of trailing context" |
99 | //usage: "\n -B N Print N lines of leading context" |
100 | //usage: "\n -C N Same as '-A N -B N'" |
101 | //usage: ) |
102 | //usage: "\n -e PTRN Pattern to match" |
103 | //usage: "\n -f FILE Read pattern from file" |
104 | //usage: |
105 | //usage:#define grep_example_usage |
106 | //usage: "$ grep root /etc/passwd\n" |
107 | //usage: "root:x:0:0:root:/root:/bin/bash\n" |
108 | //usage: "$ grep ^[rR]oo. /etc/passwd\n" |
109 | //usage: "root:x:0:0:root:/root:/bin/bash\n" |
110 | //usage: |
111 | //usage:#define egrep_trivial_usage NOUSAGE_STR |
112 | //usage:#define egrep_full_usage "" |
113 | //usage:#define fgrep_trivial_usage NOUSAGE_STR |
114 | //usage:#define fgrep_full_usage "" |
115 | |
116 | #define OPTSTR_GREP \ |
117 | "lnqvscFiHhe:f:Lorm:wx" \ |
118 | IF_FEATURE_GREP_CONTEXT("A:B:C:") \ |
119 | IF_FEATURE_GREP_EGREP_ALIAS("E") \ |
120 | IF_EXTRA_COMPAT("z") \ |
121 | "aI" |
122 | /* ignored: -a "assume all files to be text" */ |
123 | /* ignored: -I "assume binary files have no matches" */ |
124 | enum { |
125 | OPTBIT_l, /* list matched file names only */ |
126 | OPTBIT_n, /* print line# */ |
127 | OPTBIT_q, /* quiet - exit(EXIT_SUCCESS) of first match */ |
128 | OPTBIT_v, /* invert the match, to select non-matching lines */ |
129 | OPTBIT_s, /* suppress errors about file open errors */ |
130 | OPTBIT_c, /* count matches per file (suppresses normal output) */ |
131 | OPTBIT_F, /* literal match */ |
132 | OPTBIT_i, /* case-insensitive */ |
133 | OPTBIT_H, /* force filename display */ |
134 | OPTBIT_h, /* inhibit filename display */ |
135 | OPTBIT_e, /* -e PATTERN */ |
136 | OPTBIT_f, /* -f FILE_WITH_PATTERNS */ |
137 | OPTBIT_L, /* list unmatched file names only */ |
138 | OPTBIT_o, /* show only matching parts of lines */ |
139 | OPTBIT_r, /* recurse dirs */ |
140 | OPTBIT_m, /* -m MAX_MATCHES */ |
141 | OPTBIT_w, /* -w whole word match */ |
142 | OPTBIT_x, /* -x whole line match */ |
143 | IF_FEATURE_GREP_CONTEXT( OPTBIT_A ,) /* -A NUM: after-match context */ |
144 | IF_FEATURE_GREP_CONTEXT( OPTBIT_B ,) /* -B NUM: before-match context */ |
145 | IF_FEATURE_GREP_CONTEXT( OPTBIT_C ,) /* -C NUM: -A and -B combined */ |
146 | IF_FEATURE_GREP_EGREP_ALIAS(OPTBIT_E ,) /* extended regexp */ |
147 | IF_EXTRA_COMPAT( OPTBIT_z ,) /* input is NUL terminated */ |
148 | OPT_l = 1 << OPTBIT_l, |
149 | OPT_n = 1 << OPTBIT_n, |
150 | OPT_q = 1 << OPTBIT_q, |
151 | OPT_v = 1 << OPTBIT_v, |
152 | OPT_s = 1 << OPTBIT_s, |
153 | OPT_c = 1 << OPTBIT_c, |
154 | OPT_F = 1 << OPTBIT_F, |
155 | OPT_i = 1 << OPTBIT_i, |
156 | OPT_H = 1 << OPTBIT_H, |
157 | OPT_h = 1 << OPTBIT_h, |
158 | OPT_e = 1 << OPTBIT_e, |
159 | OPT_f = 1 << OPTBIT_f, |
160 | OPT_L = 1 << OPTBIT_L, |
161 | OPT_o = 1 << OPTBIT_o, |
162 | OPT_r = 1 << OPTBIT_r, |
163 | OPT_m = 1 << OPTBIT_m, |
164 | OPT_w = 1 << OPTBIT_w, |
165 | OPT_x = 1 << OPTBIT_x, |
166 | OPT_A = IF_FEATURE_GREP_CONTEXT( (1 << OPTBIT_A)) + 0, |
167 | OPT_B = IF_FEATURE_GREP_CONTEXT( (1 << OPTBIT_B)) + 0, |
168 | OPT_C = IF_FEATURE_GREP_CONTEXT( (1 << OPTBIT_C)) + 0, |
169 | OPT_E = IF_FEATURE_GREP_EGREP_ALIAS((1 << OPTBIT_E)) + 0, |
170 | OPT_z = IF_EXTRA_COMPAT( (1 << OPTBIT_z)) + 0, |
171 | }; |
172 | |
173 | #define PRINT_FILES_WITH_MATCHES (option_mask32 & OPT_l) |
174 | #define PRINT_LINE_NUM (option_mask32 & OPT_n) |
175 | #define BE_QUIET (option_mask32 & OPT_q) |
176 | #define SUPPRESS_ERR_MSGS (option_mask32 & OPT_s) |
177 | #define PRINT_MATCH_COUNTS (option_mask32 & OPT_c) |
178 | #define FGREP_FLAG (option_mask32 & OPT_F) |
179 | #define PRINT_FILES_WITHOUT_MATCHES (option_mask32 & OPT_L) |
180 | #define NUL_DELIMITED (option_mask32 & OPT_z) |
181 | |
182 | struct globals { |
183 | int max_matches; |
184 | #if !ENABLE_EXTRA_COMPAT |
185 | int reflags; |
186 | #else |
187 | RE_TRANSLATE_TYPE case_fold; /* RE_TRANSLATE_TYPE is [[un]signed] char* */ |
188 | #endif |
189 | smalluint invert_search; |
190 | smalluint print_filename; |
191 | smalluint open_errors; |
192 | #if ENABLE_FEATURE_GREP_CONTEXT |
193 | smalluint did_print_line; |
194 | int lines_before; |
195 | int lines_after; |
196 | char **before_buf; |
197 | IF_EXTRA_COMPAT(size_t *before_buf_size;) |
198 | int last_line_printed; |
199 | #endif |
200 | /* globals used internally */ |
201 | llist_t *pattern_head; /* growable list of patterns to match */ |
202 | const char *cur_file; /* the current file we are reading */ |
203 | } FIX_ALIASING; |
204 | #define G (*(struct globals*)&bb_common_bufsiz1) |
205 | #define INIT_G() do { \ |
206 | struct G_sizecheck { \ |
207 | char G_sizecheck[sizeof(G) > COMMON_BUFSIZE ? -1 : 1]; \ |
208 | }; \ |
209 | } while (0) |
210 | #define max_matches (G.max_matches ) |
211 | #if !ENABLE_EXTRA_COMPAT |
212 | # define reflags (G.reflags ) |
213 | #else |
214 | # define case_fold (G.case_fold ) |
215 | /* http://www.delorie.com/gnu/docs/regex/regex_46.html */ |
216 | # define reflags re_syntax_options |
217 | # undef REG_NOSUB |
218 | # undef REG_EXTENDED |
219 | # undef REG_ICASE |
220 | # define REG_NOSUB bug:is:here /* should not be used */ |
221 | /* Just RE_SYNTAX_EGREP is not enough, need to enable {n[,[m]]} too */ |
222 | # define REG_EXTENDED (RE_SYNTAX_EGREP | RE_INTERVALS | RE_NO_BK_BRACES) |
223 | # define REG_ICASE bug:is:here /* should not be used */ |
224 | #endif |
225 | #define invert_search (G.invert_search ) |
226 | #define print_filename (G.print_filename ) |
227 | #define open_errors (G.open_errors ) |
228 | #define did_print_line (G.did_print_line ) |
229 | #define lines_before (G.lines_before ) |
230 | #define lines_after (G.lines_after ) |
231 | #define before_buf (G.before_buf ) |
232 | #define before_buf_size (G.before_buf_size ) |
233 | #define last_line_printed (G.last_line_printed ) |
234 | #define pattern_head (G.pattern_head ) |
235 | #define cur_file (G.cur_file ) |
236 | |
237 | |
238 | typedef struct grep_list_data_t { |
239 | char *pattern; |
240 | /* for GNU regex, matched_range must be persistent across grep_file() calls */ |
241 | #if !ENABLE_EXTRA_COMPAT |
242 | regex_t compiled_regex; |
243 | regmatch_t matched_range; |
244 | #else |
245 | struct re_pattern_buffer compiled_regex; |
246 | struct re_registers matched_range; |
247 | #endif |
248 | #define ALLOCATED 1 |
249 | #define COMPILED 2 |
250 | int flg_mem_alocated_compiled; |
251 | } grep_list_data_t; |
252 | |
253 | #if !ENABLE_EXTRA_COMPAT |
254 | #define print_line(line, line_len, linenum, decoration) \ |
255 | print_line(line, linenum, decoration) |
256 | #endif |
257 | static void print_line(const char *line, size_t line_len, int linenum, char decoration) |
258 | { |
259 | #if ENABLE_FEATURE_GREP_CONTEXT |
260 | /* Happens when we go to next file, immediately hit match |
261 | * and try to print prev context... from prev file! Don't do it */ |
262 | if (linenum < 1) |
263 | return; |
264 | /* possibly print the little '--' separator */ |
265 | if ((lines_before || lines_after) && did_print_line |
266 | && last_line_printed != linenum - 1 |
267 | ) { |
268 | puts("--"); |
269 | } |
270 | /* guard against printing "--" before first line of first file */ |
271 | did_print_line = 1; |
272 | last_line_printed = linenum; |
273 | #endif |
274 | if (print_filename) |
275 | printf("%s%c", cur_file, decoration); |
276 | if (PRINT_LINE_NUM) |
277 | printf("%i%c", linenum, decoration); |
278 | /* Emulate weird GNU grep behavior with -ov */ |
279 | if ((option_mask32 & (OPT_v|OPT_o)) != (OPT_v|OPT_o)) { |
280 | #if !ENABLE_EXTRA_COMPAT |
281 | puts(line); |
282 | #else |
283 | fwrite(line, 1, line_len, stdout); |
284 | putchar(NUL_DELIMITED ? '\0' : '\n'); |
285 | #endif |
286 | } |
287 | } |
288 | |
289 | #if ENABLE_EXTRA_COMPAT |
290 | /* Unlike getline, this one removes trailing '\n' */ |
291 | static ssize_t FAST_FUNC bb_getline(char **line_ptr, size_t *line_alloc_len, FILE *file) |
292 | { |
293 | ssize_t res_sz; |
294 | char *line; |
295 | int delim = (NUL_DELIMITED ? '\0' : '\n'); |
296 | |
297 | res_sz = getdelim(line_ptr, line_alloc_len, delim, file); |
298 | line = *line_ptr; |
299 | |
300 | if (res_sz > 0) { |
301 | if (line[res_sz - 1] == delim) |
302 | line[--res_sz] = '\0'; |
303 | } else { |
304 | free(line); /* uclibc allocates a buffer even on EOF. WTF? */ |
305 | } |
306 | return res_sz; |
307 | } |
308 | #endif |
309 | |
310 | static int grep_file(FILE *file) |
311 | { |
312 | smalluint found; |
313 | int linenum = 0; |
314 | int nmatches = 0; |
315 | #if !ENABLE_EXTRA_COMPAT |
316 | char *line; |
317 | #else |
318 | char *line = NULL; |
319 | ssize_t line_len; |
320 | size_t line_alloc_len; |
321 | # define rm_so start[0] |
322 | # define rm_eo end[0] |
323 | #endif |
324 | #if ENABLE_FEATURE_GREP_CONTEXT |
325 | int print_n_lines_after = 0; |
326 | int curpos = 0; /* track where we are in the circular 'before' buffer */ |
327 | int idx = 0; /* used for iteration through the circular buffer */ |
328 | #else |
329 | enum { print_n_lines_after = 0 }; |
330 | #endif |
331 | |
332 | while ( |
333 | #if !ENABLE_EXTRA_COMPAT |
334 | (line = xmalloc_fgetline(file)) != NULL |
335 | #else |
336 | (line_len = bb_getline(&line, &line_alloc_len, file)) >= 0 |
337 | #endif |
338 | ) { |
339 | llist_t *pattern_ptr = pattern_head; |
340 | static grep_list_data_t *gl; |
341 | |
342 | linenum++; |
343 | found = 0; |
344 | while (pattern_ptr) { |
345 | gl = (grep_list_data_t *)pattern_ptr->data; |
346 | if (FGREP_FLAG) { |
347 | char *match; |
348 | char *str = line; |
349 | opt_f_again: |
350 | match = ((option_mask32 & OPT_i) |
351 | ? strcasestr(str, gl->pattern) |
352 | : strstr(str, gl->pattern) |
353 | ); |
354 | if (match) { |
355 | if (option_mask32 & OPT_x) { |
356 | if (match != str) |
357 | goto opt_f_not_found; |
358 | if (str[strlen(gl->pattern)] != '\0') |
359 | goto opt_f_not_found; |
360 | } else |
361 | if (option_mask32 & OPT_w) { |
362 | char c = (match != str) ? match[-1] : ' '; |
363 | if (!isalnum(c) && c != '_') { |
364 | c = match[strlen(gl->pattern)]; |
365 | if (!c || (!isalnum(c) && c != '_')) |
366 | goto opt_f_found; |
367 | } |
368 | str = match + 1; |
369 | goto opt_f_again; |
370 | } |
371 | opt_f_found: |
372 | found = 1; |
373 | opt_f_not_found: ; |
374 | } |
375 | } else { |
376 | #if ENABLE_EXTRA_COMPAT |
377 | unsigned start_pos; |
378 | #endif |
379 | char *match_at; |
380 | |
381 | if (!(gl->flg_mem_alocated_compiled & COMPILED)) { |
382 | gl->flg_mem_alocated_compiled |= COMPILED; |
383 | #if !ENABLE_EXTRA_COMPAT |
384 | xregcomp(&gl->compiled_regex, gl->pattern, reflags); |
385 | #else |
386 | memset(&gl->compiled_regex, 0, sizeof(gl->compiled_regex)); |
387 | gl->compiled_regex.translate = case_fold; /* for -i */ |
388 | if (re_compile_pattern(gl->pattern, strlen(gl->pattern), &gl->compiled_regex)) |
389 | bb_error_msg_and_die("bad regex '%s'", gl->pattern); |
390 | #endif |
391 | } |
392 | #if !ENABLE_EXTRA_COMPAT |
393 | gl->matched_range.rm_so = 0; |
394 | gl->matched_range.rm_eo = 0; |
395 | #else |
396 | start_pos = 0; |
397 | #endif |
398 | match_at = line; |
399 | opt_w_again: |
400 | //bb_error_msg("'%s' start_pos:%d line_len:%d", match_at, start_pos, line_len); |
401 | if ( |
402 | #if !ENABLE_EXTRA_COMPAT |
403 | regexec(&gl->compiled_regex, match_at, 1, &gl->matched_range, 0) == 0 |
404 | #else |
405 | re_search(&gl->compiled_regex, match_at, line_len, |
406 | start_pos, /*range:*/ line_len, |
407 | &gl->matched_range) >= 0 |
408 | #endif |
409 | ) { |
410 | if (option_mask32 & OPT_x) { |
411 | found = (gl->matched_range.rm_so == 0 |
412 | && match_at[gl->matched_range.rm_eo] == '\0'); |
413 | } else |
414 | if (!(option_mask32 & OPT_w)) { |
415 | found = 1; |
416 | } else { |
417 | char c = ' '; |
418 | if (gl->matched_range.rm_so) |
419 | c = match_at[gl->matched_range.rm_so - 1]; |
420 | if (!isalnum(c) && c != '_') { |
421 | c = match_at[gl->matched_range.rm_eo]; |
422 | if (!c || (!isalnum(c) && c != '_')) { |
423 | found = 1; |
424 | } else { |
425 | /* |
426 | * Why check gl->matched_range.rm_eo? |
427 | * Zero-length match makes -w skip the line: |
428 | * "echo foo | grep ^" prints "foo", |
429 | * "echo foo | grep -w ^" prints nothing. |
430 | * Without such check, we can loop forever. |
431 | */ |
432 | #if !ENABLE_EXTRA_COMPAT |
433 | if (gl->matched_range.rm_eo != 0) { |
434 | match_at += gl->matched_range.rm_eo; |
435 | goto opt_w_again; |
436 | } |
437 | #else |
438 | if (gl->matched_range.rm_eo > start_pos) { |
439 | start_pos = gl->matched_range.rm_eo; |
440 | goto opt_w_again; |
441 | } |
442 | #endif |
443 | } |
444 | } |
445 | } |
446 | } |
447 | } |
448 | /* If it's non-inverted search, we can stop |
449 | * at first match */ |
450 | if (found && !invert_search) |
451 | goto do_found; |
452 | pattern_ptr = pattern_ptr->link; |
453 | } /* while (pattern_ptr) */ |
454 | |
455 | if (found ^ invert_search) { |
456 | do_found: |
457 | /* keep track of matches */ |
458 | nmatches++; |
459 | |
460 | /* quiet/print (non)matching file names only? */ |
461 | if (option_mask32 & (OPT_q|OPT_l|OPT_L)) { |
462 | free(line); /* we don't need line anymore */ |
463 | if (BE_QUIET) { |
464 | /* manpage says about -q: |
465 | * "exit immediately with zero status |
466 | * if any match is found, |
467 | * even if errors were detected" */ |
468 | exit(EXIT_SUCCESS); |
469 | } |
470 | /* if we're just printing filenames, we stop after the first match */ |
471 | if (PRINT_FILES_WITH_MATCHES) { |
472 | puts(cur_file); |
473 | /* fall through to "return 1" */ |
474 | } |
475 | /* OPT_L aka PRINT_FILES_WITHOUT_MATCHES: return early */ |
476 | return 1; /* one match */ |
477 | } |
478 | |
479 | #if ENABLE_FEATURE_GREP_CONTEXT |
480 | /* Were we printing context and saw next (unwanted) match? */ |
481 | if ((option_mask32 & OPT_m) && nmatches > max_matches) |
482 | break; |
483 | #endif |
484 | |
485 | /* print the matched line */ |
486 | if (PRINT_MATCH_COUNTS == 0) { |
487 | #if ENABLE_FEATURE_GREP_CONTEXT |
488 | int prevpos = (curpos == 0) ? lines_before - 1 : curpos - 1; |
489 | |
490 | /* if we were told to print 'before' lines and there is at least |
491 | * one line in the circular buffer, print them */ |
492 | if (lines_before && before_buf[prevpos] != NULL) { |
493 | int first_buf_entry_line_num = linenum - lines_before; |
494 | |
495 | /* advance to the first entry in the circular buffer, and |
496 | * figure out the line number is of the first line in the |
497 | * buffer */ |
498 | idx = curpos; |
499 | while (before_buf[idx] == NULL) { |
500 | idx = (idx + 1) % lines_before; |
501 | first_buf_entry_line_num++; |
502 | } |
503 | |
504 | /* now print each line in the buffer, clearing them as we go */ |
505 | while (before_buf[idx] != NULL) { |
506 | print_line(before_buf[idx], before_buf_size[idx], first_buf_entry_line_num, '-'); |
507 | free(before_buf[idx]); |
508 | before_buf[idx] = NULL; |
509 | idx = (idx + 1) % lines_before; |
510 | first_buf_entry_line_num++; |
511 | } |
512 | } |
513 | |
514 | /* make a note that we need to print 'after' lines */ |
515 | print_n_lines_after = lines_after; |
516 | #endif |
517 | if (option_mask32 & OPT_o) { |
518 | if (FGREP_FLAG) { |
519 | /* -Fo just prints the pattern |
520 | * (unless -v: -Fov doesnt print anything at all) */ |
521 | if (found) |
522 | print_line(gl->pattern, strlen(gl->pattern), linenum, ':'); |
523 | } else while (1) { |
524 | unsigned start = gl->matched_range.rm_so; |
525 | unsigned end = gl->matched_range.rm_eo; |
526 | unsigned len = end - start; |
527 | char old = line[end]; |
528 | line[end] = '\0'; |
529 | /* Empty match is not printed: try "echo test | grep -o ''" */ |
530 | if (len != 0) |
531 | print_line(line + start, len, linenum, ':'); |
532 | if (old == '\0') |
533 | break; |
534 | line[end] = old; |
535 | if (len == 0) |
536 | end++; |
537 | #if !ENABLE_EXTRA_COMPAT |
538 | if (regexec(&gl->compiled_regex, line + end, |
539 | 1, &gl->matched_range, REG_NOTBOL) != 0) |
540 | break; |
541 | gl->matched_range.rm_so += end; |
542 | gl->matched_range.rm_eo += end; |
543 | #else |
544 | if (re_search(&gl->compiled_regex, line, line_len, |
545 | end, line_len - end, |
546 | &gl->matched_range) < 0) |
547 | break; |
548 | #endif |
549 | } |
550 | } else { |
551 | print_line(line, line_len, linenum, ':'); |
552 | } |
553 | } |
554 | } |
555 | #if ENABLE_FEATURE_GREP_CONTEXT |
556 | else { /* no match */ |
557 | /* if we need to print some context lines after the last match, do so */ |
558 | if (print_n_lines_after) { |
559 | print_line(line, strlen(line), linenum, '-'); |
560 | print_n_lines_after--; |
561 | } else if (lines_before) { |
562 | /* Add the line to the circular 'before' buffer */ |
563 | free(before_buf[curpos]); |
564 | before_buf[curpos] = line; |
565 | IF_EXTRA_COMPAT(before_buf_size[curpos] = line_len;) |
566 | curpos = (curpos + 1) % lines_before; |
567 | /* avoid free(line) - we took the line */ |
568 | line = NULL; |
569 | } |
570 | } |
571 | |
572 | #endif /* ENABLE_FEATURE_GREP_CONTEXT */ |
573 | #if !ENABLE_EXTRA_COMPAT |
574 | free(line); |
575 | #endif |
576 | /* Did we print all context after last requested match? */ |
577 | if ((option_mask32 & OPT_m) |
578 | && !print_n_lines_after |
579 | && nmatches == max_matches |
580 | ) { |
581 | break; |
582 | } |
583 | } /* while (read line) */ |
584 | |
585 | /* special-case file post-processing for options where we don't print line |
586 | * matches, just filenames and possibly match counts */ |
587 | |
588 | /* grep -c: print [filename:]count, even if count is zero */ |
589 | if (PRINT_MATCH_COUNTS) { |
590 | if (print_filename) |
591 | printf("%s:", cur_file); |
592 | printf("%d\n", nmatches); |
593 | } |
594 | |
595 | /* grep -L: print just the filename */ |
596 | if (PRINT_FILES_WITHOUT_MATCHES) { |
597 | /* nmatches is zero, no need to check it: |
598 | * we return 1 early if we detected a match |
599 | * and PRINT_FILES_WITHOUT_MATCHES is set */ |
600 | puts(cur_file); |
601 | } |
602 | |
603 | return nmatches; |
604 | } |
605 | |
606 | #if ENABLE_FEATURE_CLEAN_UP |
607 | #define new_grep_list_data(p, m) add_grep_list_data(p, m) |
608 | static char *add_grep_list_data(char *pattern, int flg_used_mem) |
609 | #else |
610 | #define new_grep_list_data(p, m) add_grep_list_data(p) |
611 | static char *add_grep_list_data(char *pattern) |
612 | #endif |
613 | { |
614 | grep_list_data_t *gl = xzalloc(sizeof(*gl)); |
615 | gl->pattern = pattern; |
616 | #if ENABLE_FEATURE_CLEAN_UP |
617 | gl->flg_mem_alocated_compiled = flg_used_mem; |
618 | #else |
619 | /*gl->flg_mem_alocated_compiled = 0;*/ |
620 | #endif |
621 | return (char *)gl; |
622 | } |
623 | |
624 | static void load_regexes_from_file(llist_t *fopt) |
625 | { |
626 | while (fopt) { |
627 | char *line; |
628 | FILE *fp; |
629 | llist_t *cur = fopt; |
630 | char *ffile = cur->data; |
631 | |
632 | fopt = cur->link; |
633 | free(cur); |
634 | fp = xfopen_stdin(ffile); |
635 | while ((line = xmalloc_fgetline(fp)) != NULL) { |
636 | llist_add_to(&pattern_head, |
637 | new_grep_list_data(line, ALLOCATED)); |
638 | } |
639 | fclose_if_not_stdin(fp); |
640 | } |
641 | } |
642 | |
643 | static int FAST_FUNC file_action_grep(const char *filename, |
644 | struct stat *statbuf UNUSED_PARAM, |
645 | void* matched, |
646 | int depth UNUSED_PARAM) |
647 | { |
648 | FILE *file = fopen_for_read(filename); |
649 | if (file == NULL) { |
650 | if (!SUPPRESS_ERR_MSGS) |
651 | bb_simple_perror_msg(filename); |
652 | open_errors = 1; |
653 | return 0; |
654 | } |
655 | cur_file = filename; |
656 | *(int*)matched += grep_file(file); |
657 | fclose(file); |
658 | return 1; |
659 | } |
660 | |
661 | static int grep_dir(const char *dir) |
662 | { |
663 | int matched = 0; |
664 | recursive_action(dir, |
665 | /* recurse=yes */ ACTION_RECURSE | |
666 | /* followLinks=no */ |
667 | /* depthFirst=yes */ ACTION_DEPTHFIRST, |
668 | /* fileAction= */ file_action_grep, |
669 | /* dirAction= */ NULL, |
670 | /* userData= */ &matched, |
671 | /* depth= */ 0); |
672 | return matched; |
673 | } |
674 | |
675 | int grep_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; |
676 | int grep_main(int argc UNUSED_PARAM, char **argv) |
677 | { |
678 | FILE *file; |
679 | int matched; |
680 | llist_t *fopt = NULL; |
681 | |
682 | /* do normal option parsing */ |
683 | #if ENABLE_FEATURE_GREP_CONTEXT |
684 | int Copt, opts; |
685 | |
686 | /* -H unsets -h; -C unsets -A,-B; -e,-f are lists; |
687 | * -m,-A,-B,-C have numeric param */ |
688 | opt_complementary = "H-h:C-AB:e::f::m+:A+:B+:C+"; |
689 | opts = getopt32(argv, |
690 | OPTSTR_GREP, |
691 | &pattern_head, &fopt, &max_matches, |
692 | &lines_after, &lines_before, &Copt); |
693 | |
694 | if (opts & OPT_C) { |
695 | /* -C unsets prev -A and -B, but following -A or -B |
696 | * may override it */ |
697 | if (!(opts & OPT_A)) /* not overridden */ |
698 | lines_after = Copt; |
699 | if (!(opts & OPT_B)) /* not overridden */ |
700 | lines_before = Copt; |
701 | } |
702 | /* sanity checks */ |
703 | if (opts & (OPT_c|OPT_q|OPT_l|OPT_L)) { |
704 | option_mask32 &= ~OPT_n; |
705 | lines_before = 0; |
706 | lines_after = 0; |
707 | } else if (lines_before > 0) { |
708 | if ((unsigned) lines_before > (unsigned) (INT_MAX / sizeof(long long))) |
709 | lines_before = INT_MAX / sizeof(long long); |
710 | /* overflow in (lines_before * sizeof(x)) is prevented (above) */ |
711 | before_buf = xzalloc(lines_before * sizeof(before_buf[0])); |
712 | IF_EXTRA_COMPAT(before_buf_size = xzalloc(lines_before * sizeof(before_buf_size[0]));) |
713 | } |
714 | #else |
715 | /* with auto sanity checks */ |
716 | /* -H unsets -h; -c,-q or -l unset -n; -e,-f are lists; -m N */ |
717 | opt_complementary = "H-h:c-n:q-n:l-n:e::f::m+"; |
718 | getopt32(argv, OPTSTR_GREP, |
719 | &pattern_head, &fopt, &max_matches); |
720 | #endif |
721 | invert_search = ((option_mask32 & OPT_v) != 0); /* 0 | 1 */ |
722 | |
723 | { /* convert char **argv to grep_list_data_t */ |
724 | llist_t *cur; |
725 | for (cur = pattern_head; cur; cur = cur->link) |
726 | cur->data = new_grep_list_data(cur->data, 0); |
727 | } |
728 | if (option_mask32 & OPT_f) { |
729 | load_regexes_from_file(fopt); |
730 | if (!pattern_head) { /* -f EMPTY_FILE? */ |
731 | /* GNU grep treats it as "nothing matches" */ |
732 | llist_add_to(&pattern_head, new_grep_list_data((char*) "", 0)); |
733 | invert_search ^= 1; |
734 | } |
735 | } |
736 | |
737 | if (ENABLE_FEATURE_GREP_FGREP_ALIAS && applet_name[0] == 'f') |
738 | option_mask32 |= OPT_F; |
739 | |
740 | #if !ENABLE_EXTRA_COMPAT |
741 | if (!(option_mask32 & (OPT_o | OPT_w | OPT_x))) |
742 | reflags = REG_NOSUB; |
743 | #endif |
744 | |
745 | if (ENABLE_FEATURE_GREP_EGREP_ALIAS |
746 | && (applet_name[0] == 'e' || (option_mask32 & OPT_E)) |
747 | ) { |
748 | reflags |= REG_EXTENDED; |
749 | } |
750 | #if ENABLE_EXTRA_COMPAT |
751 | else { |
752 | reflags = RE_SYNTAX_GREP; |
753 | } |
754 | #endif |
755 | |
756 | if (option_mask32 & OPT_i) { |
757 | #if !ENABLE_EXTRA_COMPAT |
758 | reflags |= REG_ICASE; |
759 | #else |
760 | int i; |
761 | case_fold = xmalloc(256); |
762 | for (i = 0; i < 256; i++) |
763 | case_fold[i] = (unsigned char)i; |
764 | for (i = 'a'; i <= 'z'; i++) |
765 | case_fold[i] = (unsigned char)(i - ('a' - 'A')); |
766 | #endif |
767 | } |
768 | |
769 | argv += optind; |
770 | |
771 | /* if we didn't get a pattern from -e and no command file was specified, |
772 | * first parameter should be the pattern. no pattern, no worky */ |
773 | if (pattern_head == NULL) { |
774 | char *pattern; |
775 | if (*argv == NULL) |
776 | bb_show_usage(); |
777 | pattern = new_grep_list_data(*argv++, 0); |
778 | llist_add_to(&pattern_head, pattern); |
779 | } |
780 | |
781 | /* argv[0..(argc-1)] should be names of file to grep through. If |
782 | * there is more than one file to grep, we will print the filenames. */ |
783 | if (argv[0] && argv[1]) |
784 | print_filename = 1; |
785 | /* -H / -h of course override */ |
786 | if (option_mask32 & OPT_H) |
787 | print_filename = 1; |
788 | if (option_mask32 & OPT_h) |
789 | print_filename = 0; |
790 | |
791 | /* If no files were specified, or '-' was specified, take input from |
792 | * stdin. Otherwise, we grep through all the files specified. */ |
793 | matched = 0; |
794 | do { |
795 | cur_file = *argv; |
796 | file = stdin; |
797 | if (!cur_file || LONE_DASH(cur_file)) { |
798 | cur_file = "(standard input)"; |
799 | } else { |
800 | if (option_mask32 & OPT_r) { |
801 | struct stat st; |
802 | if (stat(cur_file, &st) == 0 && S_ISDIR(st.st_mode)) { |
803 | if (!(option_mask32 & OPT_h)) |
804 | print_filename = 1; |
805 | matched += grep_dir(cur_file); |
806 | goto grep_done; |
807 | } |
808 | } |
809 | /* else: fopen(dir) will succeed, but reading won't */ |
810 | file = fopen_for_read(cur_file); |
811 | if (file == NULL) { |
812 | if (!SUPPRESS_ERR_MSGS) |
813 | bb_simple_perror_msg(cur_file); |
814 | open_errors = 1; |
815 | continue; |
816 | } |
817 | } |
818 | matched += grep_file(file); |
819 | fclose_if_not_stdin(file); |
820 | grep_done: ; |
821 | } while (*argv && *++argv); |
822 | |
823 | /* destroy all the elments in the pattern list */ |
824 | if (ENABLE_FEATURE_CLEAN_UP) { |
825 | while (pattern_head) { |
826 | llist_t *pattern_head_ptr = pattern_head; |
827 | grep_list_data_t *gl = (grep_list_data_t *)pattern_head_ptr->data; |
828 | |
829 | pattern_head = pattern_head->link; |
830 | if (gl->flg_mem_alocated_compiled & ALLOCATED) |
831 | free(gl->pattern); |
832 | if (gl->flg_mem_alocated_compiled & COMPILED) |
833 | regfree(&gl->compiled_regex); |
834 | free(gl); |
835 | free(pattern_head_ptr); |
836 | } |
837 | } |
838 | /* 0 = success, 1 = failed, 2 = error */ |
839 | if (open_errors) |
840 | return 2; |
841 | return !matched; /* invert return value: 0 = success, 1 = failed */ |
842 | } |
843 |