b9621384ec78e5054d4b04dc4d5addf123f1af0a
[oweals/busybox.git] / findutils / grep.c
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                 grep_list_data_t *gl = gl; /* for gcc */
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 #else
379                                 int match_flg;
380 #endif
381                                 char *match_at;
382
383                                 if (!(gl->flg_mem_alocated_compiled & COMPILED)) {
384                                         gl->flg_mem_alocated_compiled |= COMPILED;
385 #if !ENABLE_EXTRA_COMPAT
386                                         xregcomp(&gl->compiled_regex, gl->pattern, reflags);
387 #else
388                                         memset(&gl->compiled_regex, 0, sizeof(gl->compiled_regex));
389                                         gl->compiled_regex.translate = case_fold; /* for -i */
390                                         if (re_compile_pattern(gl->pattern, strlen(gl->pattern), &gl->compiled_regex))
391                                                 bb_error_msg_and_die("bad regex '%s'", gl->pattern);
392 #endif
393                                 }
394 #if !ENABLE_EXTRA_COMPAT
395                                 gl->matched_range.rm_so = 0;
396                                 gl->matched_range.rm_eo = 0;
397                                 match_flg = 0;
398 #else
399                                 start_pos = 0;
400 #endif
401                                 match_at = line;
402  opt_w_again:
403 //bb_error_msg("'%s' start_pos:%d line_len:%d", match_at, start_pos, line_len);
404                                 if (
405 #if !ENABLE_EXTRA_COMPAT
406                                         regexec(&gl->compiled_regex, match_at, 1, &gl->matched_range, match_flg) == 0
407 #else
408                                         re_search(&gl->compiled_regex, match_at, line_len,
409                                                         start_pos, /*range:*/ line_len,
410                                                         &gl->matched_range) >= 0
411 #endif
412                                 ) {
413                                         if (option_mask32 & OPT_x) {
414                                                 found = (gl->matched_range.rm_so == 0
415                                                          && match_at[gl->matched_range.rm_eo] == '\0');
416                                         } else
417                                         if (!(option_mask32 & OPT_w)) {
418                                                 found = 1;
419                                         } else {
420                                                 char c = ' ';
421                                                 if (match_at > line || gl->matched_range.rm_so != 0) {
422                                                         c = match_at[gl->matched_range.rm_so - 1];
423                                                 }
424                                                 if (!isalnum(c) && c != '_') {
425                                                         c = match_at[gl->matched_range.rm_eo];
426                                                 }
427                                                 if (!isalnum(c) && c != '_') {
428                                                         found = 1;
429                                                 } else {
430                         /*
431                          * Why check gl->matched_range.rm_eo?
432                          * Zero-length match makes -w skip the line:
433                          * "echo foo | grep ^" prints "foo",
434                          * "echo foo | grep -w ^" prints nothing.
435                          * Without such check, we can loop forever.
436                          */
437 #if !ENABLE_EXTRA_COMPAT
438                                                         if (gl->matched_range.rm_eo != 0) {
439                                                                 match_at += gl->matched_range.rm_eo;
440                                                                 match_flg |= REG_NOTBOL;
441                                                                 goto opt_w_again;
442                                                         }
443 #else
444                                                         if (gl->matched_range.rm_eo > start_pos) {
445                                                                 start_pos = gl->matched_range.rm_eo;
446                                                                 goto opt_w_again;
447                                                         }
448 #endif
449                                                 }
450                                         }
451                                 }
452                         }
453                         /* If it's non-inverted search, we can stop
454                          * at first match */
455                         if (found && !invert_search)
456                                 goto do_found;
457                         pattern_ptr = pattern_ptr->link;
458                 } /* while (pattern_ptr) */
459
460                 if (found ^ invert_search) {
461  do_found:
462                         /* keep track of matches */
463                         nmatches++;
464
465                         /* quiet/print (non)matching file names only? */
466                         if (option_mask32 & (OPT_q|OPT_l|OPT_L)) {
467                                 free(line); /* we don't need line anymore */
468                                 if (BE_QUIET) {
469                                         /* manpage says about -q:
470                                          * "exit immediately with zero status
471                                          * if any match is found,
472                                          * even if errors were detected" */
473                                         exit(EXIT_SUCCESS);
474                                 }
475                                 /* if we're just printing filenames, we stop after the first match */
476                                 if (PRINT_FILES_WITH_MATCHES) {
477                                         puts(cur_file);
478                                         /* fall through to "return 1" */
479                                 }
480                                 /* OPT_L aka PRINT_FILES_WITHOUT_MATCHES: return early */
481                                 return 1; /* one match */
482                         }
483
484 #if ENABLE_FEATURE_GREP_CONTEXT
485                         /* Were we printing context and saw next (unwanted) match? */
486                         if ((option_mask32 & OPT_m) && nmatches > max_matches)
487                                 break;
488 #endif
489
490                         /* print the matched line */
491                         if (PRINT_MATCH_COUNTS == 0) {
492 #if ENABLE_FEATURE_GREP_CONTEXT
493                                 int prevpos = (curpos == 0) ? lines_before - 1 : curpos - 1;
494
495                                 /* if we were told to print 'before' lines and there is at least
496                                  * one line in the circular buffer, print them */
497                                 if (lines_before && before_buf[prevpos] != NULL) {
498                                         int first_buf_entry_line_num = linenum - lines_before;
499
500                                         /* advance to the first entry in the circular buffer, and
501                                          * figure out the line number is of the first line in the
502                                          * buffer */
503                                         idx = curpos;
504                                         while (before_buf[idx] == NULL) {
505                                                 idx = (idx + 1) % lines_before;
506                                                 first_buf_entry_line_num++;
507                                         }
508
509                                         /* now print each line in the buffer, clearing them as we go */
510                                         while (before_buf[idx] != NULL) {
511                                                 print_line(before_buf[idx], before_buf_size[idx], first_buf_entry_line_num, '-');
512                                                 free(before_buf[idx]);
513                                                 before_buf[idx] = NULL;
514                                                 idx = (idx + 1) % lines_before;
515                                                 first_buf_entry_line_num++;
516                                         }
517                                 }
518
519                                 /* make a note that we need to print 'after' lines */
520                                 print_n_lines_after = lines_after;
521 #endif
522                                 if (option_mask32 & OPT_o) {
523                                         if (FGREP_FLAG) {
524                                                 /* -Fo just prints the pattern
525                                                  * (unless -v: -Fov doesnt print anything at all) */
526                                                 if (found)
527                                                         print_line(gl->pattern, strlen(gl->pattern), linenum, ':');
528                                         } else while (1) {
529                                                 unsigned start = gl->matched_range.rm_so;
530                                                 unsigned end = gl->matched_range.rm_eo;
531                                                 unsigned len = end - start;
532                                                 char old = line[end];
533                                                 line[end] = '\0';
534                                                 /* Empty match is not printed: try "echo test | grep -o ''" */
535                                                 if (len != 0)
536                                                         print_line(line + start, len, linenum, ':');
537                                                 if (old == '\0')
538                                                         break;
539                                                 line[end] = old;
540                                                 if (len == 0)
541                                                         end++;
542 #if !ENABLE_EXTRA_COMPAT
543                                                 if (regexec(&gl->compiled_regex, line + end,
544                                                                 1, &gl->matched_range, REG_NOTBOL) != 0)
545                                                         break;
546                                                 gl->matched_range.rm_so += end;
547                                                 gl->matched_range.rm_eo += end;
548 #else
549                                                 if (re_search(&gl->compiled_regex, line, line_len,
550                                                                 end, line_len - end,
551                                                                 &gl->matched_range) < 0)
552                                                         break;
553 #endif
554                                         }
555                                 } else {
556                                         print_line(line, line_len, linenum, ':');
557                                 }
558                         }
559                 }
560 #if ENABLE_FEATURE_GREP_CONTEXT
561                 else { /* no match */
562                         /* if we need to print some context lines after the last match, do so */
563                         if (print_n_lines_after) {
564                                 print_line(line, strlen(line), linenum, '-');
565                                 print_n_lines_after--;
566                         } else if (lines_before) {
567                                 /* Add the line to the circular 'before' buffer */
568                                 free(before_buf[curpos]);
569                                 before_buf[curpos] = line;
570                                 IF_EXTRA_COMPAT(before_buf_size[curpos] = line_len;)
571                                 curpos = (curpos + 1) % lines_before;
572                                 /* avoid free(line) - we took the line */
573                                 line = NULL;
574                         }
575                 }
576
577 #endif /* ENABLE_FEATURE_GREP_CONTEXT */
578 #if !ENABLE_EXTRA_COMPAT
579                 free(line);
580 #endif
581                 /* Did we print all context after last requested match? */
582                 if ((option_mask32 & OPT_m)
583                  && !print_n_lines_after
584                  && nmatches == max_matches
585                 ) {
586                         break;
587                 }
588         } /* while (read line) */
589
590         /* special-case file post-processing for options where we don't print line
591          * matches, just filenames and possibly match counts */
592
593         /* grep -c: print [filename:]count, even if count is zero */
594         if (PRINT_MATCH_COUNTS) {
595                 if (print_filename)
596                         printf("%s:", cur_file);
597                 printf("%d\n", nmatches);
598         }
599
600         /* grep -L: print just the filename */
601         if (PRINT_FILES_WITHOUT_MATCHES) {
602                 /* nmatches is zero, no need to check it:
603                  * we return 1 early if we detected a match
604                  * and PRINT_FILES_WITHOUT_MATCHES is set */
605                 puts(cur_file);
606         }
607
608         return nmatches;
609 }
610
611 #if ENABLE_FEATURE_CLEAN_UP
612 #define new_grep_list_data(p, m) add_grep_list_data(p, m)
613 static char *add_grep_list_data(char *pattern, int flg_used_mem)
614 #else
615 #define new_grep_list_data(p, m) add_grep_list_data(p)
616 static char *add_grep_list_data(char *pattern)
617 #endif
618 {
619         grep_list_data_t *gl = xzalloc(sizeof(*gl));
620         gl->pattern = pattern;
621 #if ENABLE_FEATURE_CLEAN_UP
622         gl->flg_mem_alocated_compiled = flg_used_mem;
623 #else
624         /*gl->flg_mem_alocated_compiled = 0;*/
625 #endif
626         return (char *)gl;
627 }
628
629 static void load_regexes_from_file(llist_t *fopt)
630 {
631         while (fopt) {
632                 char *line;
633                 FILE *fp;
634                 llist_t *cur = fopt;
635                 char *ffile = cur->data;
636
637                 fopt = cur->link;
638                 free(cur);
639                 fp = xfopen_stdin(ffile);
640                 while ((line = xmalloc_fgetline(fp)) != NULL) {
641                         llist_add_to(&pattern_head,
642                                 new_grep_list_data(line, ALLOCATED));
643                 }
644                 fclose_if_not_stdin(fp);
645         }
646 }
647
648 static int FAST_FUNC file_action_grep(const char *filename,
649                         struct stat *statbuf UNUSED_PARAM,
650                         void* matched,
651                         int depth UNUSED_PARAM)
652 {
653         FILE *file = fopen_for_read(filename);
654         if (file == NULL) {
655                 if (!SUPPRESS_ERR_MSGS)
656                         bb_simple_perror_msg(filename);
657                 open_errors = 1;
658                 return 0;
659         }
660         cur_file = filename;
661         *(int*)matched += grep_file(file);
662         fclose(file);
663         return 1;
664 }
665
666 static int grep_dir(const char *dir)
667 {
668         int matched = 0;
669         recursive_action(dir,
670                 /* recurse=yes */ ACTION_RECURSE |
671                 /* followLinks=command line only */ ACTION_FOLLOWLINKS_L0 |
672                 /* depthFirst=yes */ ACTION_DEPTHFIRST,
673                 /* fileAction= */ file_action_grep,
674                 /* dirAction= */ NULL,
675                 /* userData= */ &matched,
676                 /* depth= */ 0);
677         return matched;
678 }
679
680 int grep_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
681 int grep_main(int argc UNUSED_PARAM, char **argv)
682 {
683         FILE *file;
684         int matched;
685         llist_t *fopt = NULL;
686
687         /* do normal option parsing */
688 #if ENABLE_FEATURE_GREP_CONTEXT
689         int Copt, opts;
690
691         /* -H unsets -h; -C unsets -A,-B; -e,-f are lists;
692          * -m,-A,-B,-C have numeric param */
693         opt_complementary = "H-h:C-AB:e::f::m+:A+:B+:C+";
694         opts = getopt32(argv,
695                 OPTSTR_GREP,
696                 &pattern_head, &fopt, &max_matches,
697                 &lines_after, &lines_before, &Copt);
698
699         if (opts & OPT_C) {
700                 /* -C unsets prev -A and -B, but following -A or -B
701                  * may override it */
702                 if (!(opts & OPT_A)) /* not overridden */
703                         lines_after = Copt;
704                 if (!(opts & OPT_B)) /* not overridden */
705                         lines_before = Copt;
706         }
707         /* sanity checks */
708         if (opts & (OPT_c|OPT_q|OPT_l|OPT_L)) {
709                 option_mask32 &= ~OPT_n;
710                 lines_before = 0;
711                 lines_after = 0;
712         } else if (lines_before > 0) {
713                 if (lines_before > INT_MAX / sizeof(long long))
714                         lines_before = INT_MAX / sizeof(long long);
715                 /* overflow in (lines_before * sizeof(x)) is prevented (above) */
716                 before_buf = xzalloc(lines_before * sizeof(before_buf[0]));
717                 IF_EXTRA_COMPAT(before_buf_size = xzalloc(lines_before * sizeof(before_buf_size[0]));)
718         }
719 #else
720         /* with auto sanity checks */
721         /* -H unsets -h; -c,-q or -l unset -n; -e,-f are lists; -m N */
722         opt_complementary = "H-h:c-n:q-n:l-n:e::f::m+";
723         getopt32(argv, OPTSTR_GREP,
724                 &pattern_head, &fopt, &max_matches);
725 #endif
726         invert_search = ((option_mask32 & OPT_v) != 0); /* 0 | 1 */
727
728         {       /* convert char **argv to grep_list_data_t */
729                 llist_t *cur;
730                 for (cur = pattern_head; cur; cur = cur->link)
731                         cur->data = new_grep_list_data(cur->data, 0);
732         }
733         if (option_mask32 & OPT_f) {
734                 load_regexes_from_file(fopt);
735                 if (!pattern_head) { /* -f EMPTY_FILE? */
736                         /* GNU grep treats it as "nothing matches" */
737                         llist_add_to(&pattern_head, new_grep_list_data((char*) "", 0));
738                         invert_search ^= 1;
739                 }
740         }
741
742         if (ENABLE_FEATURE_GREP_FGREP_ALIAS && applet_name[0] == 'f')
743                 option_mask32 |= OPT_F;
744
745 #if !ENABLE_EXTRA_COMPAT
746         if (!(option_mask32 & (OPT_o | OPT_w | OPT_x)))
747                 reflags = REG_NOSUB;
748 #endif
749
750         if (ENABLE_FEATURE_GREP_EGREP_ALIAS
751          && (applet_name[0] == 'e' || (option_mask32 & OPT_E))
752         ) {
753                 reflags |= REG_EXTENDED;
754         }
755 #if ENABLE_EXTRA_COMPAT
756         else {
757                 reflags = RE_SYNTAX_GREP;
758         }
759 #endif
760
761         if (option_mask32 & OPT_i) {
762 #if !ENABLE_EXTRA_COMPAT
763                 reflags |= REG_ICASE;
764 #else
765                 int i;
766                 case_fold = xmalloc(256);
767                 for (i = 0; i < 256; i++)
768                         case_fold[i] = (unsigned char)i;
769                 for (i = 'a'; i <= 'z'; i++)
770                         case_fold[i] = (unsigned char)(i - ('a' - 'A'));
771 #endif
772         }
773
774         argv += optind;
775
776         /* if we didn't get a pattern from -e and no command file was specified,
777          * first parameter should be the pattern. no pattern, no worky */
778         if (pattern_head == NULL) {
779                 char *pattern;
780                 if (*argv == NULL)
781                         bb_show_usage();
782                 pattern = new_grep_list_data(*argv++, 0);
783                 llist_add_to(&pattern_head, pattern);
784         }
785
786         /* argv[0..(argc-1)] should be names of file to grep through. If
787          * there is more than one file to grep, we will print the filenames. */
788         if (argv[0] && argv[1])
789                 print_filename = 1;
790         /* -H / -h of course override */
791         if (option_mask32 & OPT_H)
792                 print_filename = 1;
793         if (option_mask32 & OPT_h)
794                 print_filename = 0;
795
796         /* If no files were specified, or '-' was specified, take input from
797          * stdin. Otherwise, we grep through all the files specified. */
798         matched = 0;
799         do {
800                 cur_file = *argv;
801                 file = stdin;
802                 if (!cur_file || LONE_DASH(cur_file)) {
803                         cur_file = "(standard input)";
804                 } else {
805                         if (option_mask32 & OPT_r) {
806                                 struct stat st;
807                                 if (stat(cur_file, &st) == 0 && S_ISDIR(st.st_mode)) {
808                                         if (!(option_mask32 & OPT_h))
809                                                 print_filename = 1;
810                                         matched += grep_dir(cur_file);
811                                         goto grep_done;
812                                 }
813                         }
814                         /* else: fopen(dir) will succeed, but reading won't */
815                         file = fopen_for_read(cur_file);
816                         if (file == NULL) {
817                                 if (!SUPPRESS_ERR_MSGS)
818                                         bb_simple_perror_msg(cur_file);
819                                 open_errors = 1;
820                                 continue;
821                         }
822                 }
823                 matched += grep_file(file);
824                 fclose_if_not_stdin(file);
825  grep_done: ;
826         } while (*argv && *++argv);
827
828         /* destroy all the elments in the pattern list */
829         if (ENABLE_FEATURE_CLEAN_UP) {
830                 while (pattern_head) {
831                         llist_t *pattern_head_ptr = pattern_head;
832                         grep_list_data_t *gl = (grep_list_data_t *)pattern_head_ptr->data;
833
834                         pattern_head = pattern_head->link;
835                         if (gl->flg_mem_alocated_compiled & ALLOCATED)
836                                 free(gl->pattern);
837                         if (gl->flg_mem_alocated_compiled & COMPILED)
838                                 regfree(&gl->compiled_regex);
839                         free(gl);
840                         free(pattern_head_ptr);
841                 }
842         }
843         /* 0 = success, 1 = failed, 2 = error */
844         if (open_errors)
845                 return 2;
846         return !matched; /* invert return value: 0 = success, 1 = failed */
847 }