]>
Commit | Line | Data |
---|---|---|
5010cb5f JH |
1 | /* |
2 | * Builtin "git grep" | |
3 | * | |
4 | * Copyright (c) 2006 Junio C Hamano | |
5 | */ | |
6 | #include "cache.h" | |
7 | #include "blob.h" | |
8 | #include "tree.h" | |
9 | #include "commit.h" | |
10 | #include "tag.h" | |
1362671f | 11 | #include "tree-walk.h" |
5010cb5f | 12 | #include "builtin.h" |
3e230fa1 | 13 | #include "parse-options.h" |
60ecac98 | 14 | #include "userdiff.h" |
83b5d2f5 | 15 | #include "grep.h" |
5010cb5f | 16 | |
5f7c643a JK |
17 | #ifndef NO_EXTERNAL_GREP |
18 | #ifdef __unix__ | |
19 | #define NO_EXTERNAL_GREP 0 | |
20 | #else | |
21 | #define NO_EXTERNAL_GREP 1 | |
22 | #endif | |
23 | #endif | |
24 | ||
3e230fa1 RS |
25 | static char const * const grep_usage[] = { |
26 | "git grep [options] [-e] <pattern> [<rev>...] [[--] path...]", | |
27 | NULL | |
28 | }; | |
29 | ||
7e8f59d5 RS |
30 | static int grep_config(const char *var, const char *value, void *cb) |
31 | { | |
32 | struct grep_opt *opt = cb; | |
33 | ||
60ecac98 RS |
34 | switch (userdiff_config(var, value)) { |
35 | case 0: break; | |
36 | case -1: return -1; | |
37 | default: return 0; | |
38 | } | |
39 | ||
fe3420b6 | 40 | if (!strcmp(var, "color.grep")) { |
7e8f59d5 RS |
41 | opt->color = git_config_colorbool(var, value, -1); |
42 | return 0; | |
43 | } | |
fe3420b6 | 44 | if (!strcmp(var, "color.grep.external")) |
a94982ef | 45 | return git_config_string(&(opt->color_external), var, value); |
fe3420b6 | 46 | if (!strcmp(var, "color.grep.match")) { |
7e8f59d5 RS |
47 | if (!value) |
48 | return config_error_nonbool(var); | |
49 | color_parse(value, var, opt->color_match); | |
50 | return 0; | |
51 | } | |
52 | return git_color_default_config(var, value, cb); | |
53 | } | |
54 | ||
e0eb889f JH |
55 | /* |
56 | * git grep pathspecs are somewhat different from diff-tree pathspecs; | |
57 | * pathname wildcards are allowed. | |
58 | */ | |
1362671f | 59 | static int pathspec_matches(const char **paths, const char *name) |
5010cb5f | 60 | { |
e0eb889f | 61 | int namelen, i; |
1362671f | 62 | if (!paths || !*paths) |
5010cb5f JH |
63 | return 1; |
64 | namelen = strlen(name); | |
1362671f JH |
65 | for (i = 0; paths[i]; i++) { |
66 | const char *match = paths[i]; | |
67 | int matchlen = strlen(match); | |
1e3d90e0 | 68 | const char *cp, *meta; |
e0eb889f | 69 | |
bb9e15a8 UKK |
70 | if (!matchlen || |
71 | ((matchlen <= namelen) && | |
72 | !strncmp(name, match, matchlen) && | |
73 | (match[matchlen-1] == '/' || | |
74 | name[matchlen] == '\0' || name[matchlen] == '/'))) | |
e0eb889f JH |
75 | return 1; |
76 | if (!fnmatch(match, name, 0)) | |
77 | return 1; | |
78 | if (name[namelen-1] != '/') | |
5010cb5f | 79 | continue; |
e0eb889f | 80 | |
1e3d90e0 | 81 | /* We are being asked if the directory ("name") is worth |
e0eb889f JH |
82 | * descending into. |
83 | * | |
84 | * Find the longest leading directory name that does | |
85 | * not have metacharacter in the pathspec; the name | |
86 | * we are looking at must overlap with that directory. | |
5010cb5f | 87 | */ |
1e3d90e0 | 88 | for (cp = match, meta = NULL; cp - match < matchlen; cp++) { |
e0eb889f | 89 | char ch = *cp; |
1e3d90e0 JH |
90 | if (ch == '*' || ch == '[' || ch == '?') { |
91 | meta = cp; | |
5010cb5f | 92 | break; |
1e3d90e0 | 93 | } |
e0eb889f | 94 | } |
1e3d90e0 JH |
95 | if (!meta) |
96 | meta = cp; /* fully literal */ | |
97 | ||
98 | if (namelen <= meta - match) { | |
e0eb889f JH |
99 | /* Looking at "Documentation/" and |
100 | * the pattern says "Documentation/howto/", or | |
1e3d90e0 JH |
101 | * "Documentation/diff*.txt". The name we |
102 | * have should match prefix. | |
e0eb889f JH |
103 | */ |
104 | if (!memcmp(match, name, namelen)) | |
105 | return 1; | |
1e3d90e0 | 106 | continue; |
e0eb889f | 107 | } |
1e3d90e0 JH |
108 | |
109 | if (meta - match < namelen) { | |
e0eb889f | 110 | /* Looking at "Documentation/howto/" and |
1e3d90e0 JH |
111 | * the pattern says "Documentation/h*"; |
112 | * match up to "Do.../h"; this avoids descending | |
113 | * into "Documentation/technical/". | |
e0eb889f | 114 | */ |
1e3d90e0 | 115 | if (!memcmp(match, name, meta - match)) |
e0eb889f | 116 | return 1; |
1e3d90e0 | 117 | continue; |
e0eb889f | 118 | } |
5010cb5f JH |
119 | } |
120 | return 0; | |
121 | } | |
122 | ||
0d042fec | 123 | static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name, int tree_name_len) |
5010cb5f JH |
124 | { |
125 | unsigned long size; | |
126 | char *data; | |
21666f1a | 127 | enum object_type type; |
0d042fec | 128 | char *to_free = NULL; |
5010cb5f | 129 | int hit; |
0d042fec | 130 | |
21666f1a | 131 | data = read_sha1_file(sha1, &type, &size); |
5010cb5f JH |
132 | if (!data) { |
133 | error("'%s': unable to read %s", name, sha1_to_hex(sha1)); | |
134 | return 0; | |
135 | } | |
0d042fec JH |
136 | if (opt->relative && opt->prefix_length) { |
137 | static char name_buf[PATH_MAX]; | |
138 | char *cp; | |
139 | int name_len = strlen(name) - opt->prefix_length + 1; | |
140 | ||
141 | if (!tree_name_len) | |
142 | name += opt->prefix_length; | |
143 | else { | |
144 | if (ARRAY_SIZE(name_buf) <= name_len) | |
145 | cp = to_free = xmalloc(name_len); | |
146 | else | |
147 | cp = name_buf; | |
148 | memcpy(cp, name, tree_name_len); | |
149 | strcpy(cp + tree_name_len, | |
150 | name + tree_name_len + opt->prefix_length); | |
151 | name = cp; | |
152 | } | |
153 | } | |
5010cb5f JH |
154 | hit = grep_buffer(opt, name, data, size); |
155 | free(data); | |
0d042fec | 156 | free(to_free); |
5010cb5f JH |
157 | return hit; |
158 | } | |
159 | ||
160 | static int grep_file(struct grep_opt *opt, const char *filename) | |
161 | { | |
162 | struct stat st; | |
163 | int i; | |
164 | char *data; | |
dc49cd76 SP |
165 | size_t sz; |
166 | ||
5010cb5f JH |
167 | if (lstat(filename, &st) < 0) { |
168 | err_ret: | |
169 | if (errno != ENOENT) | |
170 | error("'%s': %s", filename, strerror(errno)); | |
171 | return 0; | |
172 | } | |
173 | if (!st.st_size) | |
174 | return 0; /* empty file -- no grep hit */ | |
175 | if (!S_ISREG(st.st_mode)) | |
176 | return 0; | |
dc49cd76 | 177 | sz = xsize_t(st.st_size); |
5010cb5f JH |
178 | i = open(filename, O_RDONLY); |
179 | if (i < 0) | |
180 | goto err_ret; | |
dc49cd76 SP |
181 | data = xmalloc(sz + 1); |
182 | if (st.st_size != read_in_full(i, data, sz)) { | |
5010cb5f JH |
183 | error("'%s': short read %s", filename, strerror(errno)); |
184 | close(i); | |
185 | free(data); | |
186 | return 0; | |
187 | } | |
188 | close(i); | |
0d042fec JH |
189 | if (opt->relative && opt->prefix_length) |
190 | filename += opt->prefix_length; | |
dc49cd76 | 191 | i = grep_buffer(opt, filename, data, sz); |
5010cb5f JH |
192 | free(data); |
193 | return i; | |
194 | } | |
195 | ||
5f7c643a | 196 | #if !NO_EXTERNAL_GREP |
1e2398d7 LT |
197 | static int exec_grep(int argc, const char **argv) |
198 | { | |
199 | pid_t pid; | |
200 | int status; | |
201 | ||
202 | argv[argc] = NULL; | |
203 | pid = fork(); | |
204 | if (pid < 0) | |
205 | return pid; | |
206 | if (!pid) { | |
207 | execvp("grep", (char **) argv); | |
208 | exit(255); | |
209 | } | |
210 | while (waitpid(pid, &status, 0) < 0) { | |
211 | if (errno == EINTR) | |
212 | continue; | |
213 | return -1; | |
214 | } | |
215 | if (WIFEXITED(status)) { | |
216 | if (!WEXITSTATUS(status)) | |
217 | return 1; | |
218 | return 0; | |
219 | } | |
220 | return -1; | |
221 | } | |
222 | ||
223 | #define MAXARGS 1000 | |
ffa0a7ab JH |
224 | #define ARGBUF 4096 |
225 | #define push_arg(a) do { \ | |
226 | if (nr < MAXARGS) argv[nr++] = (a); \ | |
227 | else die("maximum number of args exceeded"); \ | |
228 | } while (0) | |
1e2398d7 | 229 | |
d99ebf08 JH |
230 | /* |
231 | * If you send a singleton filename to grep, it does not give | |
232 | * the name of the file. GNU grep has "-H" but we would want | |
233 | * that behaviour in a portable way. | |
234 | * | |
235 | * So we keep two pathnames in argv buffer unsent to grep in | |
236 | * the main loop if we need to do more than one grep. | |
237 | */ | |
238 | static int flush_grep(struct grep_opt *opt, | |
239 | int argc, int arg0, const char **argv, int *kept) | |
240 | { | |
241 | int status; | |
242 | int count = argc - arg0; | |
243 | const char *kept_0 = NULL; | |
244 | ||
245 | if (count <= 2) { | |
246 | /* | |
247 | * Because we keep at least 2 paths in the call from | |
248 | * the main loop (i.e. kept != NULL), and MAXARGS is | |
249 | * far greater than 2, this usually is a call to | |
250 | * conclude the grep. However, the user could attempt | |
251 | * to overflow the argv buffer by giving too many | |
252 | * options to leave very small number of real | |
253 | * arguments even for the call in the main loop. | |
254 | */ | |
255 | if (kept) | |
256 | die("insanely many options to grep"); | |
257 | ||
258 | /* | |
259 | * If we have two or more paths, we do not have to do | |
260 | * anything special, but we need to push /dev/null to | |
261 | * get "-H" behaviour of GNU grep portably but when we | |
262 | * are not doing "-l" nor "-L" nor "-c". | |
263 | */ | |
264 | if (count == 1 && | |
265 | !opt->name_only && | |
266 | !opt->unmatch_name_only && | |
267 | !opt->count) { | |
268 | argv[argc++] = "/dev/null"; | |
269 | argv[argc] = NULL; | |
270 | } | |
271 | } | |
272 | ||
273 | else if (kept) { | |
274 | /* | |
275 | * Called because we found many paths and haven't finished | |
276 | * iterating over the cache yet. We keep two paths | |
277 | * for the concluding call. argv[argc-2] and argv[argc-1] | |
278 | * has the last two paths, so save the first one away, | |
279 | * replace it with NULL while sending the list to grep, | |
280 | * and recover them after we are done. | |
281 | */ | |
282 | *kept = 2; | |
283 | kept_0 = argv[argc-2]; | |
284 | argv[argc-2] = NULL; | |
285 | argc -= 2; | |
286 | } | |
287 | ||
ed24e401 | 288 | if (opt->pre_context || opt->post_context) { |
046802d0 RS |
289 | /* |
290 | * grep handles hunk marks between files, but we need to | |
291 | * do that ourselves between multiple calls. | |
292 | */ | |
293 | if (opt->show_hunk_mark) | |
ed24e401 | 294 | write_or_die(1, "--\n", 3); |
046802d0 RS |
295 | else |
296 | opt->show_hunk_mark = 1; | |
297 | } | |
298 | ||
d99ebf08 JH |
299 | status = exec_grep(argc, argv); |
300 | ||
301 | if (kept_0) { | |
302 | /* | |
303 | * Then recover them. Now the last arg is beyond the | |
304 | * terminating NULL which is at argc, and the second | |
305 | * from the last is what we saved away in kept_0 | |
306 | */ | |
307 | argv[arg0++] = kept_0; | |
308 | argv[arg0] = argv[argc+1]; | |
309 | } | |
310 | return status; | |
311 | } | |
312 | ||
a94982ef RS |
313 | static void grep_add_color(struct strbuf *sb, const char *escape_seq) |
314 | { | |
315 | size_t orig_len = sb->len; | |
316 | ||
317 | while (*escape_seq) { | |
318 | if (*escape_seq == 'm') | |
319 | strbuf_addch(sb, ';'); | |
320 | else if (*escape_seq != '\033' && *escape_seq != '[') | |
321 | strbuf_addch(sb, *escape_seq); | |
322 | escape_seq++; | |
323 | } | |
324 | if (sb->len > orig_len && sb->buf[sb->len - 1] == ';') | |
325 | strbuf_setlen(sb, sb->len - 1); | |
326 | } | |
327 | ||
1e2398d7 LT |
328 | static int external_grep(struct grep_opt *opt, const char **paths, int cached) |
329 | { | |
fcfe34b5 | 330 | int i, nr, argc, hit, len, status; |
1e2398d7 | 331 | const char *argv[MAXARGS+1]; |
ffa0a7ab JH |
332 | char randarg[ARGBUF]; |
333 | char *argptr = randarg; | |
1e2398d7 LT |
334 | struct grep_pat *p; |
335 | ||
0d042fec | 336 | if (opt->extended || (opt->relative && opt->prefix_length)) |
79d3696c | 337 | return -1; |
ffa0a7ab JH |
338 | len = nr = 0; |
339 | push_arg("grep"); | |
ffa0a7ab | 340 | if (opt->fixed) |
f6647519 | 341 | push_arg("-F"); |
ffa0a7ab JH |
342 | if (opt->linenum) |
343 | push_arg("-n"); | |
7977f0ea LT |
344 | if (!opt->pathname) |
345 | push_arg("-h"); | |
ffa0a7ab JH |
346 | if (opt->regflags & REG_EXTENDED) |
347 | push_arg("-E"); | |
3026402c RF |
348 | if (opt->regflags & REG_ICASE) |
349 | push_arg("-i"); | |
bc395643 JH |
350 | if (opt->binary == GREP_BINARY_NOMATCH) |
351 | push_arg("-I"); | |
1e2398d7 | 352 | if (opt->word_regexp) |
ffa0a7ab | 353 | push_arg("-w"); |
1e2398d7 | 354 | if (opt->name_only) |
ffa0a7ab JH |
355 | push_arg("-l"); |
356 | if (opt->unmatch_name_only) | |
357 | push_arg("-L"); | |
83caecca RZ |
358 | if (opt->null_following_name) |
359 | /* in GNU grep git's "-z" translates to "-Z" */ | |
360 | push_arg("-Z"); | |
ffa0a7ab JH |
361 | if (opt->count) |
362 | push_arg("-c"); | |
363 | if (opt->post_context || opt->pre_context) { | |
364 | if (opt->post_context != opt->pre_context) { | |
365 | if (opt->pre_context) { | |
366 | push_arg("-B"); | |
367 | len += snprintf(argptr, sizeof(randarg)-len, | |
4b87474b | 368 | "%u", opt->pre_context) + 1; |
ffa0a7ab JH |
369 | if (sizeof(randarg) <= len) |
370 | die("maximum length of args exceeded"); | |
371 | push_arg(argptr); | |
372 | argptr += len; | |
373 | } | |
374 | if (opt->post_context) { | |
375 | push_arg("-A"); | |
376 | len += snprintf(argptr, sizeof(randarg)-len, | |
4b87474b | 377 | "%u", opt->post_context) + 1; |
ffa0a7ab JH |
378 | if (sizeof(randarg) <= len) |
379 | die("maximum length of args exceeded"); | |
380 | push_arg(argptr); | |
381 | argptr += len; | |
382 | } | |
383 | } | |
384 | else { | |
385 | push_arg("-C"); | |
386 | len += snprintf(argptr, sizeof(randarg)-len, | |
4b87474b | 387 | "%u", opt->post_context) + 1; |
ffa0a7ab JH |
388 | if (sizeof(randarg) <= len) |
389 | die("maximum length of args exceeded"); | |
390 | push_arg(argptr); | |
391 | argptr += len; | |
392 | } | |
393 | } | |
1e2398d7 | 394 | for (p = opt->pattern_list; p; p = p->next) { |
ffa0a7ab JH |
395 | push_arg("-e"); |
396 | push_arg(p->pattern); | |
1e2398d7 | 397 | } |
a94982ef RS |
398 | if (opt->color) { |
399 | struct strbuf sb = STRBUF_INIT; | |
400 | ||
401 | grep_add_color(&sb, opt->color_match); | |
402 | setenv("GREP_COLOR", sb.buf, 1); | |
403 | ||
404 | strbuf_reset(&sb); | |
405 | strbuf_addstr(&sb, "mt="); | |
406 | grep_add_color(&sb, opt->color_match); | |
407 | strbuf_addstr(&sb, ":sl=:cx=:fn=:ln=:bn=:se="); | |
408 | setenv("GREP_COLORS", sb.buf, 1); | |
409 | ||
410 | strbuf_release(&sb); | |
411 | ||
412 | if (opt->color_external && strlen(opt->color_external) > 0) | |
413 | push_arg(opt->color_external); | |
414 | } | |
bbb66c60 | 415 | |
1e2398d7 LT |
416 | hit = 0; |
417 | argc = nr; | |
418 | for (i = 0; i < active_nr; i++) { | |
419 | struct cache_entry *ce = active_cache[i]; | |
fbd01abf | 420 | char *name; |
d99ebf08 | 421 | int kept; |
7a51ed66 | 422 | if (!S_ISREG(ce->ce_mode)) |
1e2398d7 LT |
423 | continue; |
424 | if (!pathspec_matches(paths, ce->name)) | |
425 | continue; | |
bbb66c60 LT |
426 | name = ce->name; |
427 | if (name[0] == '-') { | |
428 | int len = ce_namelen(ce); | |
429 | name = xmalloc(len + 3); | |
430 | memcpy(name, "./", 2); | |
431 | memcpy(name + 2, ce->name, len + 1); | |
432 | } | |
433 | argv[argc++] = name; | |
6326cee5 JH |
434 | if (MAXARGS <= argc) { |
435 | status = flush_grep(opt, argc, nr, argv, &kept); | |
436 | if (0 < status) | |
437 | hit = 1; | |
438 | argc = nr + kept; | |
439 | } | |
36f2587f JH |
440 | if (ce_stage(ce)) { |
441 | do { | |
442 | i++; | |
443 | } while (i < active_nr && | |
444 | !strcmp(ce->name, active_cache[i]->name)); | |
445 | i--; /* compensate for loop control */ | |
446 | } | |
1e2398d7 | 447 | } |
fcfe34b5 | 448 | if (argc > nr) { |
d99ebf08 | 449 | status = flush_grep(opt, argc, nr, argv, NULL); |
fcfe34b5 JH |
450 | if (0 < status) |
451 | hit = 1; | |
452 | } | |
453 | return hit; | |
1e2398d7 | 454 | } |
ff1f9945 | 455 | #endif |
1e2398d7 | 456 | |
1b5fb44a RS |
457 | static int grep_cache(struct grep_opt *opt, const char **paths, int cached, |
458 | int external_grep_allowed) | |
5010cb5f JH |
459 | { |
460 | int hit = 0; | |
461 | int nr; | |
462 | read_cache(); | |
463 | ||
5f7c643a | 464 | #if !NO_EXTERNAL_GREP |
1e2398d7 LT |
465 | /* |
466 | * Use the external "grep" command for the case where | |
467 | * we grep through the checked-out files. It tends to | |
468 | * be a lot more optimized | |
469 | */ | |
1b5fb44a | 470 | if (!cached && external_grep_allowed) { |
1e2398d7 LT |
471 | hit = external_grep(opt, paths, cached); |
472 | if (hit >= 0) | |
473 | return hit; | |
474 | } | |
475 | #endif | |
476 | ||
5010cb5f JH |
477 | for (nr = 0; nr < active_nr; nr++) { |
478 | struct cache_entry *ce = active_cache[nr]; | |
7a51ed66 | 479 | if (!S_ISREG(ce->ce_mode)) |
5010cb5f | 480 | continue; |
1362671f | 481 | if (!pathspec_matches(paths, ce->name)) |
5010cb5f | 482 | continue; |
57d43466 NTND |
483 | /* |
484 | * If CE_VALID is on, we assume worktree file and its cache entry | |
485 | * are identical, even if worktree file has been modified, so use | |
486 | * cache version instead | |
487 | */ | |
488 | if (cached || (ce->ce_flags & CE_VALID)) { | |
36f2587f JH |
489 | if (ce_stage(ce)) |
490 | continue; | |
0d042fec | 491 | hit |= grep_sha1(opt, ce->sha1, ce->name, 0); |
36f2587f | 492 | } |
5010cb5f JH |
493 | else |
494 | hit |= grep_file(opt, ce->name); | |
36f2587f JH |
495 | if (ce_stage(ce)) { |
496 | do { | |
497 | nr++; | |
498 | } while (nr < active_nr && | |
499 | !strcmp(ce->name, active_cache[nr]->name)); | |
500 | nr--; /* compensate for loop control */ | |
501 | } | |
5010cb5f | 502 | } |
b48fb5b6 | 503 | free_grep_patterns(opt); |
5010cb5f JH |
504 | return hit; |
505 | } | |
506 | ||
1362671f | 507 | static int grep_tree(struct grep_opt *opt, const char **paths, |
5010cb5f JH |
508 | struct tree_desc *tree, |
509 | const char *tree_name, const char *base) | |
510 | { | |
5010cb5f JH |
511 | int len; |
512 | int hit = 0; | |
4c068a98 | 513 | struct name_entry entry; |
e0eb889f | 514 | char *down; |
0d042fec | 515 | int tn_len = strlen(tree_name); |
620e2bb9 DP |
516 | struct strbuf pathbuf; |
517 | ||
518 | strbuf_init(&pathbuf, PATH_MAX + tn_len); | |
5010cb5f | 519 | |
0d042fec | 520 | if (tn_len) { |
620e2bb9 DP |
521 | strbuf_add(&pathbuf, tree_name, tn_len); |
522 | strbuf_addch(&pathbuf, ':'); | |
523 | tn_len = pathbuf.len; | |
5010cb5f | 524 | } |
620e2bb9 DP |
525 | strbuf_addstr(&pathbuf, base); |
526 | len = pathbuf.len; | |
5010cb5f | 527 | |
4c068a98 | 528 | while (tree_entry(tree, &entry)) { |
620e2bb9 DP |
529 | int te_len = tree_entry_len(entry.path, entry.sha1); |
530 | pathbuf.len = len; | |
531 | strbuf_add(&pathbuf, entry.path, te_len); | |
5010cb5f | 532 | |
4c068a98 | 533 | if (S_ISDIR(entry.mode)) |
e0eb889f JH |
534 | /* Match "abc/" against pathspec to |
535 | * decide if we want to descend into "abc" | |
536 | * directory. | |
537 | */ | |
620e2bb9 | 538 | strbuf_addch(&pathbuf, '/'); |
e0eb889f | 539 | |
620e2bb9 | 540 | down = pathbuf.buf + tn_len; |
1362671f | 541 | if (!pathspec_matches(paths, down)) |
5010cb5f | 542 | ; |
4c068a98 | 543 | else if (S_ISREG(entry.mode)) |
620e2bb9 | 544 | hit |= grep_sha1(opt, entry.sha1, pathbuf.buf, tn_len); |
4c068a98 | 545 | else if (S_ISDIR(entry.mode)) { |
21666f1a | 546 | enum object_type type; |
5010cb5f JH |
547 | struct tree_desc sub; |
548 | void *data; | |
6fda5e51 LT |
549 | unsigned long size; |
550 | ||
551 | data = read_sha1_file(entry.sha1, &type, &size); | |
5010cb5f JH |
552 | if (!data) |
553 | die("unable to read tree (%s)", | |
4c068a98 | 554 | sha1_to_hex(entry.sha1)); |
6fda5e51 | 555 | init_tree_desc(&sub, data, size); |
1362671f | 556 | hit |= grep_tree(opt, paths, &sub, tree_name, down); |
5010cb5f JH |
557 | free(data); |
558 | } | |
5010cb5f | 559 | } |
620e2bb9 | 560 | strbuf_release(&pathbuf); |
5010cb5f JH |
561 | return hit; |
562 | } | |
563 | ||
1362671f | 564 | static int grep_object(struct grep_opt *opt, const char **paths, |
5010cb5f JH |
565 | struct object *obj, const char *name) |
566 | { | |
1974632c | 567 | if (obj->type == OBJ_BLOB) |
0d042fec | 568 | return grep_sha1(opt, obj->sha1, name, 0); |
1974632c | 569 | if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) { |
5010cb5f JH |
570 | struct tree_desc tree; |
571 | void *data; | |
6fda5e51 | 572 | unsigned long size; |
5010cb5f JH |
573 | int hit; |
574 | data = read_object_with_reference(obj->sha1, tree_type, | |
6fda5e51 | 575 | &size, NULL); |
5010cb5f JH |
576 | if (!data) |
577 | die("unable to read tree (%s)", sha1_to_hex(obj->sha1)); | |
6fda5e51 | 578 | init_tree_desc(&tree, data, size); |
1362671f | 579 | hit = grep_tree(opt, paths, &tree, name, ""); |
5010cb5f JH |
580 | free(data); |
581 | return hit; | |
582 | } | |
885a86ab | 583 | die("unable to grep from object of type %s", typename(obj->type)); |
5010cb5f JH |
584 | } |
585 | ||
ff3c7f9a RS |
586 | static int context_callback(const struct option *opt, const char *arg, |
587 | int unset) | |
3e230fa1 RS |
588 | { |
589 | struct grep_opt *grep_opt = opt->value; | |
590 | int value; | |
591 | const char *endp; | |
592 | ||
593 | if (unset) { | |
594 | grep_opt->pre_context = grep_opt->post_context = 0; | |
595 | return 0; | |
596 | } | |
597 | value = strtol(arg, (char **)&endp, 10); | |
598 | if (*endp) { | |
599 | return error("switch `%c' expects a numerical value", | |
600 | opt->short_name); | |
601 | } | |
602 | grep_opt->pre_context = grep_opt->post_context = value; | |
603 | return 0; | |
604 | } | |
605 | ||
ff3c7f9a | 606 | static int file_callback(const struct option *opt, const char *arg, int unset) |
3e230fa1 RS |
607 | { |
608 | struct grep_opt *grep_opt = opt->value; | |
609 | FILE *patterns; | |
610 | int lno = 0; | |
611 | struct strbuf sb; | |
612 | ||
613 | patterns = fopen(arg, "r"); | |
614 | if (!patterns) | |
d824cbba | 615 | die_errno("cannot open '%s'", arg); |
3e230fa1 RS |
616 | while (strbuf_getline(&sb, patterns, '\n') == 0) { |
617 | /* ignore empty line like grep does */ | |
618 | if (sb.len == 0) | |
619 | continue; | |
620 | append_grep_pattern(grep_opt, strbuf_detach(&sb, NULL), arg, | |
621 | ++lno, GREP_PATTERN); | |
622 | } | |
623 | fclose(patterns); | |
624 | strbuf_release(&sb); | |
625 | return 0; | |
626 | } | |
627 | ||
ff3c7f9a | 628 | static int not_callback(const struct option *opt, const char *arg, int unset) |
3e230fa1 RS |
629 | { |
630 | struct grep_opt *grep_opt = opt->value; | |
631 | append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT); | |
632 | return 0; | |
633 | } | |
634 | ||
ff3c7f9a | 635 | static int and_callback(const struct option *opt, const char *arg, int unset) |
3e230fa1 RS |
636 | { |
637 | struct grep_opt *grep_opt = opt->value; | |
638 | append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND); | |
639 | return 0; | |
640 | } | |
641 | ||
ff3c7f9a | 642 | static int open_callback(const struct option *opt, const char *arg, int unset) |
3e230fa1 RS |
643 | { |
644 | struct grep_opt *grep_opt = opt->value; | |
645 | append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN); | |
646 | return 0; | |
647 | } | |
648 | ||
ff3c7f9a | 649 | static int close_callback(const struct option *opt, const char *arg, int unset) |
3e230fa1 RS |
650 | { |
651 | struct grep_opt *grep_opt = opt->value; | |
652 | append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN); | |
653 | return 0; | |
654 | } | |
655 | ||
ff3c7f9a RS |
656 | static int pattern_callback(const struct option *opt, const char *arg, |
657 | int unset) | |
3e230fa1 RS |
658 | { |
659 | struct grep_opt *grep_opt = opt->value; | |
660 | append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN); | |
661 | return 0; | |
662 | } | |
5010cb5f | 663 | |
ff3c7f9a | 664 | static int help_callback(const struct option *opt, const char *arg, int unset) |
3e230fa1 RS |
665 | { |
666 | return -1; | |
667 | } | |
088b084b | 668 | |
a633fca0 | 669 | int cmd_grep(int argc, const char **argv, const char *prefix) |
5010cb5f | 670 | { |
5010cb5f | 671 | int hit = 0; |
5010cb5f | 672 | int cached = 0; |
1b5fb44a | 673 | int external_grep_allowed = 1; |
5acd64ed | 674 | int seen_dashdash = 0; |
5010cb5f | 675 | struct grep_opt opt; |
1f1e895f | 676 | struct object_array list = { 0, 0, NULL }; |
1362671f | 677 | const char **paths = NULL; |
5acd64ed | 678 | int i; |
3e230fa1 RS |
679 | int dummy; |
680 | struct option options[] = { | |
681 | OPT_BOOLEAN(0, "cached", &cached, | |
682 | "search in index instead of in the work tree"), | |
683 | OPT_GROUP(""), | |
684 | OPT_BOOLEAN('v', "invert-match", &opt.invert, | |
685 | "show non-matching lines"), | |
686 | OPT_BIT('i', "ignore-case", &opt.regflags, | |
687 | "case insensitive matching", REG_ICASE), | |
688 | OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp, | |
689 | "match patterns only at word boundaries"), | |
690 | OPT_SET_INT('a', "text", &opt.binary, | |
691 | "process binary files as text", GREP_BINARY_TEXT), | |
692 | OPT_SET_INT('I', NULL, &opt.binary, | |
693 | "don't match patterns in binary files", | |
694 | GREP_BINARY_NOMATCH), | |
695 | OPT_GROUP(""), | |
696 | OPT_BIT('E', "extended-regexp", &opt.regflags, | |
697 | "use extended POSIX regular expressions", REG_EXTENDED), | |
698 | OPT_NEGBIT('G', "basic-regexp", &opt.regflags, | |
699 | "use basic POSIX regular expressions (default)", | |
700 | REG_EXTENDED), | |
701 | OPT_BOOLEAN('F', "fixed-strings", &opt.fixed, | |
702 | "interpret patterns as fixed strings"), | |
703 | OPT_GROUP(""), | |
704 | OPT_BOOLEAN('n', NULL, &opt.linenum, "show line numbers"), | |
705 | OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1), | |
706 | OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1), | |
707 | OPT_NEGBIT(0, "full-name", &opt.relative, | |
708 | "show filenames relative to top directory", 1), | |
709 | OPT_BOOLEAN('l', "files-with-matches", &opt.name_only, | |
710 | "show only filenames instead of matching lines"), | |
711 | OPT_BOOLEAN(0, "name-only", &opt.name_only, | |
712 | "synonym for --files-with-matches"), | |
713 | OPT_BOOLEAN('L', "files-without-match", | |
714 | &opt.unmatch_name_only, | |
715 | "show only the names of files without match"), | |
716 | OPT_BOOLEAN('z', "null", &opt.null_following_name, | |
717 | "print NUL after filenames"), | |
718 | OPT_BOOLEAN('c', "count", &opt.count, | |
719 | "show the number of matches instead of matching lines"), | |
720 | OPT_SET_INT(0, "color", &opt.color, "highlight matches", 1), | |
721 | OPT_GROUP(""), | |
722 | OPT_CALLBACK('C', NULL, &opt, "n", | |
723 | "show <n> context lines before and after matches", | |
724 | context_callback), | |
725 | OPT_INTEGER('B', NULL, &opt.pre_context, | |
726 | "show <n> context lines before matches"), | |
727 | OPT_INTEGER('A', NULL, &opt.post_context, | |
728 | "show <n> context lines after matches"), | |
729 | OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM", | |
730 | context_callback), | |
2944e4e6 RS |
731 | OPT_BOOLEAN('p', "show-function", &opt.funcname, |
732 | "show a line with the function name before matches"), | |
3e230fa1 RS |
733 | OPT_GROUP(""), |
734 | OPT_CALLBACK('f', NULL, &opt, "file", | |
735 | "read patterns from file", file_callback), | |
736 | { OPTION_CALLBACK, 'e', NULL, &opt, "pattern", | |
737 | "match <pattern>", PARSE_OPT_NONEG, pattern_callback }, | |
738 | { OPTION_CALLBACK, 0, "and", &opt, NULL, | |
739 | "combine patterns specified with -e", | |
740 | PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback }, | |
741 | OPT_BOOLEAN(0, "or", &dummy, ""), | |
742 | { OPTION_CALLBACK, 0, "not", &opt, NULL, "", | |
743 | PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback }, | |
744 | { OPTION_CALLBACK, '(', NULL, &opt, NULL, "", | |
745 | PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH, | |
746 | open_callback }, | |
747 | { OPTION_CALLBACK, ')', NULL, &opt, NULL, "", | |
748 | PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH, | |
749 | close_callback }, | |
750 | OPT_BOOLEAN(0, "all-match", &opt.all_match, | |
751 | "show only matches from files that match all patterns"), | |
752 | OPT_GROUP(""), | |
753 | #if NO_EXTERNAL_GREP | |
754 | OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed, | |
755 | "allow calling of grep(1) (ignored by this build)"), | |
756 | #else | |
757 | OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed, | |
758 | "allow calling of grep(1) (default)"), | |
759 | #endif | |
760 | { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage", | |
761 | PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback }, | |
762 | OPT_END() | |
763 | }; | |
5010cb5f JH |
764 | |
765 | memset(&opt, 0, sizeof(opt)); | |
0d042fec JH |
766 | opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0; |
767 | opt.relative = 1; | |
7977f0ea | 768 | opt.pathname = 1; |
f9b9faf6 | 769 | opt.pattern_tail = &opt.pattern_list; |
5010cb5f JH |
770 | opt.regflags = REG_NEWLINE; |
771 | ||
7e8f59d5 RS |
772 | strcpy(opt.color_match, GIT_COLOR_RED GIT_COLOR_BOLD); |
773 | opt.color = -1; | |
774 | git_config(grep_config, &opt); | |
775 | if (opt.color == -1) | |
776 | opt.color = git_use_color_default; | |
777 | ||
5010cb5f | 778 | /* |
5acd64ed JH |
779 | * If there is no -- then the paths must exist in the working |
780 | * tree. If there is no explicit pattern specified with -e or | |
781 | * -f, we take the first unrecognized non option to be the | |
782 | * pattern, but then what follows it must be zero or more | |
783 | * valid refs up to the -- (if exists), and then existing | |
784 | * paths. If there is an explicit pattern, then the first | |
82e5a82f | 785 | * unrecognized non option is the beginning of the refs list |
5acd64ed | 786 | * that continues up to the -- (if exists), and then paths. |
5010cb5f | 787 | */ |
37782920 | 788 | argc = parse_options(argc, argv, prefix, options, grep_usage, |
3e230fa1 RS |
789 | PARSE_OPT_KEEP_DASHDASH | |
790 | PARSE_OPT_STOP_AT_NON_OPTION | | |
791 | PARSE_OPT_NO_INTERNAL_HELP); | |
792 | ||
793 | /* First unrecognized non-option token */ | |
794 | if (argc > 0 && !opt.pattern_list) { | |
795 | append_grep_pattern(&opt, argv[0], "command line", 0, | |
796 | GREP_PATTERN); | |
797 | argv++; | |
798 | argc--; | |
5010cb5f | 799 | } |
5acd64ed | 800 | |
2944e4e6 | 801 | if ((opt.color && !opt.color_external) || opt.funcname) |
1b5fb44a | 802 | external_grep_allowed = 0; |
f9b9faf6 | 803 | if (!opt.pattern_list) |
5010cb5f | 804 | die("no pattern given."); |
07ea91d8 JH |
805 | if ((opt.regflags != REG_NEWLINE) && opt.fixed) |
806 | die("cannot mix --fixed-strings and regexp"); | |
83b5d2f5 | 807 | compile_grep_patterns(&opt); |
5acd64ed JH |
808 | |
809 | /* Check revs and then paths */ | |
3e230fa1 | 810 | for (i = 0; i < argc; i++) { |
5acd64ed | 811 | const char *arg = argv[i]; |
1362671f | 812 | unsigned char sha1[20]; |
5acd64ed JH |
813 | /* Is it a rev? */ |
814 | if (!get_sha1(arg, sha1)) { | |
815 | struct object *object = parse_object(sha1); | |
5acd64ed JH |
816 | if (!object) |
817 | die("bad object %s", arg); | |
1f1e895f | 818 | add_object_array(object, arg, &list); |
5acd64ed JH |
819 | continue; |
820 | } | |
821 | if (!strcmp(arg, "--")) { | |
822 | i++; | |
823 | seen_dashdash = 1; | |
824 | } | |
825 | break; | |
1362671f | 826 | } |
5acd64ed JH |
827 | |
828 | /* The rest are paths */ | |
829 | if (!seen_dashdash) { | |
830 | int j; | |
c39c4f47 | 831 | for (j = i; j < argc; j++) |
5acd64ed JH |
832 | verify_filename(prefix, argv[j]); |
833 | } | |
834 | ||
0d042fec | 835 | if (i < argc) { |
5acd64ed | 836 | paths = get_pathspec(prefix, argv + i); |
0d042fec JH |
837 | if (opt.prefix_length && opt.relative) { |
838 | /* Make sure we do not get outside of paths */ | |
839 | for (i = 0; paths[i]; i++) | |
840 | if (strncmp(prefix, paths[i], opt.prefix_length)) | |
7e44c935 | 841 | die("git grep: cannot generate relative filenames containing '..'"); |
0d042fec JH |
842 | } |
843 | } | |
1362671f JH |
844 | else if (prefix) { |
845 | paths = xcalloc(2, sizeof(const char *)); | |
846 | paths[0] = prefix; | |
847 | paths[1] = NULL; | |
848 | } | |
5010cb5f | 849 | |
6577f542 NTND |
850 | if (!list.nr) { |
851 | if (!cached) | |
852 | setup_work_tree(); | |
1b5fb44a | 853 | return !grep_cache(&opt, paths, cached, external_grep_allowed); |
6577f542 | 854 | } |
aa8c79ad | 855 | |
5010cb5f | 856 | if (cached) |
aa8c79ad | 857 | die("both --cached and trees are given."); |
5010cb5f | 858 | |
1f1e895f | 859 | for (i = 0; i < list.nr; i++) { |
5010cb5f | 860 | struct object *real_obj; |
1f1e895f LT |
861 | real_obj = deref_tag(list.objects[i].item, NULL, 0); |
862 | if (grep_object(&opt, paths, real_obj, list.objects[i].name)) | |
5010cb5f JH |
863 | hit = 1; |
864 | } | |
b48fb5b6 | 865 | free_grep_patterns(&opt); |
5010cb5f JH |
866 | return !hit; |
867 | } |