]>
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 JH |
12 | #include "builtin.h" |
13 | #include <regex.h> | |
e0eb889f | 14 | #include <fnmatch.h> |
1e2398d7 | 15 | #include <sys/wait.h> |
5010cb5f | 16 | |
e0eb889f JH |
17 | /* |
18 | * git grep pathspecs are somewhat different from diff-tree pathspecs; | |
19 | * pathname wildcards are allowed. | |
20 | */ | |
1362671f | 21 | static int pathspec_matches(const char **paths, const char *name) |
5010cb5f | 22 | { |
e0eb889f | 23 | int namelen, i; |
1362671f | 24 | if (!paths || !*paths) |
5010cb5f JH |
25 | return 1; |
26 | namelen = strlen(name); | |
1362671f JH |
27 | for (i = 0; paths[i]; i++) { |
28 | const char *match = paths[i]; | |
29 | int matchlen = strlen(match); | |
1e3d90e0 | 30 | const char *cp, *meta; |
e0eb889f JH |
31 | |
32 | if ((matchlen <= namelen) && | |
33 | !strncmp(name, match, matchlen) && | |
34 | (match[matchlen-1] == '/' || | |
35 | name[matchlen] == '\0' || name[matchlen] == '/')) | |
36 | return 1; | |
37 | if (!fnmatch(match, name, 0)) | |
38 | return 1; | |
39 | if (name[namelen-1] != '/') | |
5010cb5f | 40 | continue; |
e0eb889f | 41 | |
1e3d90e0 | 42 | /* We are being asked if the directory ("name") is worth |
e0eb889f JH |
43 | * descending into. |
44 | * | |
45 | * Find the longest leading directory name that does | |
46 | * not have metacharacter in the pathspec; the name | |
47 | * we are looking at must overlap with that directory. | |
5010cb5f | 48 | */ |
1e3d90e0 | 49 | for (cp = match, meta = NULL; cp - match < matchlen; cp++) { |
e0eb889f | 50 | char ch = *cp; |
1e3d90e0 JH |
51 | if (ch == '*' || ch == '[' || ch == '?') { |
52 | meta = cp; | |
5010cb5f | 53 | break; |
1e3d90e0 | 54 | } |
e0eb889f | 55 | } |
1e3d90e0 JH |
56 | if (!meta) |
57 | meta = cp; /* fully literal */ | |
58 | ||
59 | if (namelen <= meta - match) { | |
e0eb889f JH |
60 | /* Looking at "Documentation/" and |
61 | * the pattern says "Documentation/howto/", or | |
1e3d90e0 JH |
62 | * "Documentation/diff*.txt". The name we |
63 | * have should match prefix. | |
e0eb889f JH |
64 | */ |
65 | if (!memcmp(match, name, namelen)) | |
66 | return 1; | |
1e3d90e0 | 67 | continue; |
e0eb889f | 68 | } |
1e3d90e0 JH |
69 | |
70 | if (meta - match < namelen) { | |
e0eb889f | 71 | /* Looking at "Documentation/howto/" and |
1e3d90e0 JH |
72 | * the pattern says "Documentation/h*"; |
73 | * match up to "Do.../h"; this avoids descending | |
74 | * into "Documentation/technical/". | |
e0eb889f | 75 | */ |
1e3d90e0 | 76 | if (!memcmp(match, name, meta - match)) |
e0eb889f | 77 | return 1; |
1e3d90e0 | 78 | continue; |
e0eb889f | 79 | } |
5010cb5f JH |
80 | } |
81 | return 0; | |
82 | } | |
83 | ||
f9b9faf6 JH |
84 | struct grep_pat { |
85 | struct grep_pat *next; | |
aa8c79ad JH |
86 | const char *origin; |
87 | int no; | |
5010cb5f JH |
88 | const char *pattern; |
89 | regex_t regexp; | |
f9b9faf6 JH |
90 | }; |
91 | ||
92 | struct grep_opt { | |
93 | struct grep_pat *pattern_list; | |
94 | struct grep_pat **pattern_tail; | |
95 | regex_t regexp; | |
5010cb5f JH |
96 | unsigned linenum:1; |
97 | unsigned invert:1; | |
df0e7aa8 | 98 | unsigned name_only:1; |
e23d2d6b | 99 | unsigned unmatch_name_only:1; |
2c866cf1 | 100 | unsigned count:1; |
7839a25e | 101 | unsigned word_regexp:1; |
07ea91d8 | 102 | unsigned fixed:1; |
b8d0f5a0 JH |
103 | #define GREP_BINARY_DEFAULT 0 |
104 | #define GREP_BINARY_NOMATCH 1 | |
105 | #define GREP_BINARY_TEXT 2 | |
106 | unsigned binary:2; | |
5010cb5f JH |
107 | int regflags; |
108 | unsigned pre_context; | |
109 | unsigned post_context; | |
110 | }; | |
111 | ||
aa8c79ad JH |
112 | static void add_pattern(struct grep_opt *opt, const char *pat, |
113 | const char *origin, int no) | |
f9b9faf6 JH |
114 | { |
115 | struct grep_pat *p = xcalloc(1, sizeof(*p)); | |
116 | p->pattern = pat; | |
aa8c79ad JH |
117 | p->origin = origin; |
118 | p->no = no; | |
f9b9faf6 JH |
119 | *opt->pattern_tail = p; |
120 | opt->pattern_tail = &p->next; | |
121 | p->next = NULL; | |
122 | } | |
123 | ||
124 | static void compile_patterns(struct grep_opt *opt) | |
125 | { | |
126 | struct grep_pat *p; | |
127 | for (p = opt->pattern_list; p; p = p->next) { | |
128 | int err = regcomp(&p->regexp, p->pattern, opt->regflags); | |
129 | if (err) { | |
130 | char errbuf[1024]; | |
aa8c79ad JH |
131 | char where[1024]; |
132 | if (p->no) | |
133 | sprintf(where, "In '%s' at %d, ", | |
134 | p->origin, p->no); | |
135 | else if (p->origin) | |
136 | sprintf(where, "%s, ", p->origin); | |
137 | else | |
138 | where[0] = 0; | |
f9b9faf6 JH |
139 | regerror(err, &p->regexp, errbuf, 1024); |
140 | regfree(&p->regexp); | |
aa8c79ad | 141 | die("%s'%s': %s", where, p->pattern, errbuf); |
f9b9faf6 JH |
142 | } |
143 | } | |
144 | } | |
145 | ||
5010cb5f JH |
146 | static char *end_of_line(char *cp, unsigned long *left) |
147 | { | |
148 | unsigned long l = *left; | |
149 | while (l && *cp != '\n') { | |
150 | l--; | |
151 | cp++; | |
152 | } | |
153 | *left = l; | |
154 | return cp; | |
155 | } | |
156 | ||
7839a25e JH |
157 | static int word_char(char ch) |
158 | { | |
159 | return isalnum(ch) || ch == '_'; | |
160 | } | |
161 | ||
5010cb5f JH |
162 | static void show_line(struct grep_opt *opt, const char *bol, const char *eol, |
163 | const char *name, unsigned lno, char sign) | |
164 | { | |
165 | printf("%s%c", name, sign); | |
166 | if (opt->linenum) | |
167 | printf("%d%c", lno, sign); | |
a24f1e25 | 168 | printf("%.*s\n", (int)(eol-bol), bol); |
5010cb5f JH |
169 | } |
170 | ||
b8d0f5a0 JH |
171 | /* |
172 | * NEEDSWORK: share code with diff.c | |
173 | */ | |
174 | #define FIRST_FEW_BYTES 8000 | |
175 | static int buffer_is_binary(const char *ptr, unsigned long size) | |
176 | { | |
177 | if (FIRST_FEW_BYTES < size) | |
178 | size = FIRST_FEW_BYTES; | |
179 | if (memchr(ptr, 0, size)) | |
180 | return 1; | |
181 | return 0; | |
182 | } | |
183 | ||
07ea91d8 JH |
184 | static int fixmatch(const char *pattern, char *line, regmatch_t *match) |
185 | { | |
186 | char *hit = strstr(line, pattern); | |
187 | if (!hit) { | |
188 | match->rm_so = match->rm_eo = -1; | |
189 | return REG_NOMATCH; | |
190 | } | |
191 | else { | |
192 | match->rm_so = hit - line; | |
193 | match->rm_eo = match->rm_so + strlen(pattern); | |
194 | return 0; | |
195 | } | |
196 | } | |
197 | ||
5010cb5f JH |
198 | static int grep_buffer(struct grep_opt *opt, const char *name, |
199 | char *buf, unsigned long size) | |
200 | { | |
201 | char *bol = buf; | |
202 | unsigned long left = size; | |
203 | unsigned lno = 1; | |
204 | struct pre_context_line { | |
205 | char *bol; | |
206 | char *eol; | |
207 | } *prev = NULL, *pcl; | |
208 | unsigned last_hit = 0; | |
209 | unsigned last_shown = 0; | |
b8d0f5a0 | 210 | int binary_match_only = 0; |
5010cb5f | 211 | const char *hunk_mark = ""; |
2c866cf1 | 212 | unsigned count = 0; |
5010cb5f | 213 | |
b8d0f5a0 JH |
214 | if (buffer_is_binary(buf, size)) { |
215 | switch (opt->binary) { | |
216 | case GREP_BINARY_DEFAULT: | |
217 | binary_match_only = 1; | |
218 | break; | |
219 | case GREP_BINARY_NOMATCH: | |
220 | return 0; /* Assume unmatch */ | |
221 | break; | |
222 | default: | |
223 | break; | |
224 | } | |
225 | } | |
226 | ||
5010cb5f JH |
227 | if (opt->pre_context) |
228 | prev = xcalloc(opt->pre_context, sizeof(*prev)); | |
229 | if (opt->pre_context || opt->post_context) | |
230 | hunk_mark = "--\n"; | |
231 | ||
232 | while (left) { | |
233 | regmatch_t pmatch[10]; | |
234 | char *eol, ch; | |
f9b9faf6 JH |
235 | int hit = 0; |
236 | struct grep_pat *p; | |
5010cb5f JH |
237 | |
238 | eol = end_of_line(bol, &left); | |
239 | ch = *eol; | |
240 | *eol = 0; | |
241 | ||
f9b9faf6 | 242 | for (p = opt->pattern_list; p; p = p->next) { |
07ea91d8 JH |
243 | if (!opt->fixed) { |
244 | regex_t *exp = &p->regexp; | |
245 | hit = !regexec(exp, bol, ARRAY_SIZE(pmatch), | |
246 | pmatch, 0); | |
247 | } | |
248 | else { | |
249 | hit = !fixmatch(p->pattern, bol, pmatch); | |
250 | } | |
7839a25e JH |
251 | |
252 | if (hit && opt->word_regexp) { | |
253 | /* Match beginning must be either | |
254 | * beginning of the line, or at word | |
255 | * boundary (i.e. the last char must | |
256 | * not be alnum or underscore). | |
257 | */ | |
258 | if ((pmatch[0].rm_so < 0) || | |
259 | (eol - bol) <= pmatch[0].rm_so || | |
260 | (pmatch[0].rm_eo < 0) || | |
261 | (eol - bol) < pmatch[0].rm_eo) | |
262 | die("regexp returned nonsense"); | |
263 | if (pmatch[0].rm_so != 0 && | |
264 | word_char(bol[pmatch[0].rm_so-1])) | |
02ab1c49 JH |
265 | hit = 0; |
266 | if (pmatch[0].rm_eo != (eol-bol) && | |
7839a25e | 267 | word_char(bol[pmatch[0].rm_eo])) |
02ab1c49 | 268 | hit = 0; |
7839a25e | 269 | } |
f9b9faf6 JH |
270 | if (hit) |
271 | break; | |
272 | } | |
273 | /* "grep -v -e foo -e bla" should list lines | |
274 | * that do not have either, so inversion should | |
275 | * be done outside. | |
276 | */ | |
5010cb5f JH |
277 | if (opt->invert) |
278 | hit = !hit; | |
e23d2d6b JH |
279 | if (opt->unmatch_name_only) { |
280 | if (hit) | |
281 | return 0; | |
282 | goto next_line; | |
283 | } | |
5010cb5f | 284 | if (hit) { |
2c866cf1 | 285 | count++; |
b8d0f5a0 JH |
286 | if (binary_match_only) { |
287 | printf("Binary file %s matches\n", name); | |
288 | return 1; | |
289 | } | |
df0e7aa8 JH |
290 | if (opt->name_only) { |
291 | printf("%s\n", name); | |
292 | return 1; | |
293 | } | |
5010cb5f JH |
294 | /* Hit at this line. If we haven't shown the |
295 | * pre-context lines, we would need to show them. | |
2c866cf1 JH |
296 | * When asked to do "count", this still show |
297 | * the context which is nonsense, but the user | |
298 | * deserves to get that ;-). | |
5010cb5f JH |
299 | */ |
300 | if (opt->pre_context) { | |
301 | unsigned from; | |
302 | if (opt->pre_context < lno) | |
303 | from = lno - opt->pre_context; | |
304 | else | |
305 | from = 1; | |
306 | if (from <= last_shown) | |
307 | from = last_shown + 1; | |
308 | if (last_shown && from != last_shown + 1) | |
309 | printf(hunk_mark); | |
310 | while (from < lno) { | |
311 | pcl = &prev[lno-from-1]; | |
312 | show_line(opt, pcl->bol, pcl->eol, | |
313 | name, from, '-'); | |
314 | from++; | |
315 | } | |
316 | last_shown = lno-1; | |
317 | } | |
318 | if (last_shown && lno != last_shown + 1) | |
319 | printf(hunk_mark); | |
2c866cf1 JH |
320 | if (!opt->count) |
321 | show_line(opt, bol, eol, name, lno, ':'); | |
5010cb5f JH |
322 | last_shown = last_hit = lno; |
323 | } | |
324 | else if (last_hit && | |
325 | lno <= last_hit + opt->post_context) { | |
326 | /* If the last hit is within the post context, | |
327 | * we need to show this line. | |
328 | */ | |
329 | if (last_shown && lno != last_shown + 1) | |
330 | printf(hunk_mark); | |
331 | show_line(opt, bol, eol, name, lno, '-'); | |
332 | last_shown = lno; | |
333 | } | |
334 | if (opt->pre_context) { | |
335 | memmove(prev+1, prev, | |
336 | (opt->pre_context-1) * sizeof(*prev)); | |
337 | prev->bol = bol; | |
338 | prev->eol = eol; | |
339 | } | |
e23d2d6b JH |
340 | |
341 | next_line: | |
5010cb5f JH |
342 | *eol = ch; |
343 | bol = eol + 1; | |
7ed36f56 JH |
344 | if (!left) |
345 | break; | |
5010cb5f JH |
346 | left--; |
347 | lno++; | |
348 | } | |
e23d2d6b JH |
349 | |
350 | if (opt->unmatch_name_only) { | |
351 | /* We did not see any hit, so we want to show this */ | |
352 | printf("%s\n", name); | |
353 | return 1; | |
354 | } | |
355 | ||
2c866cf1 JH |
356 | /* NEEDSWORK: |
357 | * The real "grep -c foo *.c" gives many "bar.c:0" lines, | |
358 | * which feels mostly useless but sometimes useful. Maybe | |
359 | * make it another option? For now suppress them. | |
360 | */ | |
361 | if (opt->count && count) | |
362 | printf("%s:%u\n", name, count); | |
5010cb5f JH |
363 | return !!last_hit; |
364 | } | |
365 | ||
366 | static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name) | |
367 | { | |
368 | unsigned long size; | |
369 | char *data; | |
370 | char type[20]; | |
371 | int hit; | |
372 | data = read_sha1_file(sha1, type, &size); | |
373 | if (!data) { | |
374 | error("'%s': unable to read %s", name, sha1_to_hex(sha1)); | |
375 | return 0; | |
376 | } | |
377 | hit = grep_buffer(opt, name, data, size); | |
378 | free(data); | |
379 | return hit; | |
380 | } | |
381 | ||
382 | static int grep_file(struct grep_opt *opt, const char *filename) | |
383 | { | |
384 | struct stat st; | |
385 | int i; | |
386 | char *data; | |
387 | if (lstat(filename, &st) < 0) { | |
388 | err_ret: | |
389 | if (errno != ENOENT) | |
390 | error("'%s': %s", filename, strerror(errno)); | |
391 | return 0; | |
392 | } | |
393 | if (!st.st_size) | |
394 | return 0; /* empty file -- no grep hit */ | |
395 | if (!S_ISREG(st.st_mode)) | |
396 | return 0; | |
397 | i = open(filename, O_RDONLY); | |
398 | if (i < 0) | |
399 | goto err_ret; | |
400 | data = xmalloc(st.st_size + 1); | |
401 | if (st.st_size != xread(i, data, st.st_size)) { | |
402 | error("'%s': short read %s", filename, strerror(errno)); | |
403 | close(i); | |
404 | free(data); | |
405 | return 0; | |
406 | } | |
407 | close(i); | |
408 | i = grep_buffer(opt, filename, data, st.st_size); | |
409 | free(data); | |
410 | return i; | |
411 | } | |
412 | ||
1e2398d7 LT |
413 | static int exec_grep(int argc, const char **argv) |
414 | { | |
415 | pid_t pid; | |
416 | int status; | |
417 | ||
418 | argv[argc] = NULL; | |
419 | pid = fork(); | |
420 | if (pid < 0) | |
421 | return pid; | |
422 | if (!pid) { | |
423 | execvp("grep", (char **) argv); | |
424 | exit(255); | |
425 | } | |
426 | while (waitpid(pid, &status, 0) < 0) { | |
427 | if (errno == EINTR) | |
428 | continue; | |
429 | return -1; | |
430 | } | |
431 | if (WIFEXITED(status)) { | |
432 | if (!WEXITSTATUS(status)) | |
433 | return 1; | |
434 | return 0; | |
435 | } | |
436 | return -1; | |
437 | } | |
438 | ||
439 | #define MAXARGS 1000 | |
ffa0a7ab JH |
440 | #define ARGBUF 4096 |
441 | #define push_arg(a) do { \ | |
442 | if (nr < MAXARGS) argv[nr++] = (a); \ | |
443 | else die("maximum number of args exceeded"); \ | |
444 | } while (0) | |
1e2398d7 LT |
445 | |
446 | static int external_grep(struct grep_opt *opt, const char **paths, int cached) | |
447 | { | |
ffa0a7ab | 448 | int i, nr, argc, hit, len; |
1e2398d7 | 449 | const char *argv[MAXARGS+1]; |
ffa0a7ab JH |
450 | char randarg[ARGBUF]; |
451 | char *argptr = randarg; | |
1e2398d7 LT |
452 | struct grep_pat *p; |
453 | ||
ffa0a7ab JH |
454 | len = nr = 0; |
455 | push_arg("grep"); | |
ffa0a7ab | 456 | if (opt->fixed) |
f6647519 | 457 | push_arg("-F"); |
ffa0a7ab JH |
458 | if (opt->linenum) |
459 | push_arg("-n"); | |
460 | if (opt->regflags & REG_EXTENDED) | |
461 | push_arg("-E"); | |
3026402c RF |
462 | if (opt->regflags & REG_ICASE) |
463 | push_arg("-i"); | |
1e2398d7 | 464 | if (opt->word_regexp) |
ffa0a7ab | 465 | push_arg("-w"); |
1e2398d7 | 466 | if (opt->name_only) |
ffa0a7ab JH |
467 | push_arg("-l"); |
468 | if (opt->unmatch_name_only) | |
469 | push_arg("-L"); | |
470 | if (opt->count) | |
471 | push_arg("-c"); | |
472 | if (opt->post_context || opt->pre_context) { | |
473 | if (opt->post_context != opt->pre_context) { | |
474 | if (opt->pre_context) { | |
475 | push_arg("-B"); | |
476 | len += snprintf(argptr, sizeof(randarg)-len, | |
477 | "%u", opt->pre_context); | |
478 | if (sizeof(randarg) <= len) | |
479 | die("maximum length of args exceeded"); | |
480 | push_arg(argptr); | |
481 | argptr += len; | |
482 | } | |
483 | if (opt->post_context) { | |
484 | push_arg("-A"); | |
485 | len += snprintf(argptr, sizeof(randarg)-len, | |
486 | "%u", opt->post_context); | |
487 | if (sizeof(randarg) <= len) | |
488 | die("maximum length of args exceeded"); | |
489 | push_arg(argptr); | |
490 | argptr += len; | |
491 | } | |
492 | } | |
493 | else { | |
494 | push_arg("-C"); | |
495 | len += snprintf(argptr, sizeof(randarg)-len, | |
496 | "%u", opt->post_context); | |
497 | if (sizeof(randarg) <= len) | |
498 | die("maximum length of args exceeded"); | |
499 | push_arg(argptr); | |
500 | argptr += len; | |
501 | } | |
502 | } | |
1e2398d7 | 503 | for (p = opt->pattern_list; p; p = p->next) { |
ffa0a7ab JH |
504 | push_arg("-e"); |
505 | push_arg(p->pattern); | |
1e2398d7 | 506 | } |
bbb66c60 LT |
507 | |
508 | /* | |
509 | * To make sure we get the header printed out when we want it, | |
510 | * add /dev/null to the paths to grep. This is unnecessary | |
511 | * (and wrong) with "-l" or "-L", which always print out the | |
512 | * name anyway. | |
513 | * | |
514 | * GNU grep has "-H", but this is portable. | |
515 | */ | |
516 | if (!opt->name_only && !opt->unmatch_name_only) | |
517 | push_arg("/dev/null"); | |
1e2398d7 LT |
518 | |
519 | hit = 0; | |
520 | argc = nr; | |
521 | for (i = 0; i < active_nr; i++) { | |
522 | struct cache_entry *ce = active_cache[i]; | |
fbd01abf | 523 | char *name; |
1e2398d7 LT |
524 | if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode))) |
525 | continue; | |
526 | if (!pathspec_matches(paths, ce->name)) | |
527 | continue; | |
bbb66c60 LT |
528 | name = ce->name; |
529 | if (name[0] == '-') { | |
530 | int len = ce_namelen(ce); | |
531 | name = xmalloc(len + 3); | |
532 | memcpy(name, "./", 2); | |
533 | memcpy(name + 2, ce->name, len + 1); | |
534 | } | |
535 | argv[argc++] = name; | |
1e2398d7 LT |
536 | if (argc < MAXARGS) |
537 | continue; | |
538 | hit += exec_grep(argc, argv); | |
539 | argc = nr; | |
540 | } | |
541 | if (argc > nr) | |
542 | hit += exec_grep(argc, argv); | |
543 | return 0; | |
544 | } | |
545 | ||
1362671f | 546 | static int grep_cache(struct grep_opt *opt, const char **paths, int cached) |
5010cb5f JH |
547 | { |
548 | int hit = 0; | |
549 | int nr; | |
550 | read_cache(); | |
551 | ||
1e2398d7 LT |
552 | #ifdef __unix__ |
553 | /* | |
554 | * Use the external "grep" command for the case where | |
555 | * we grep through the checked-out files. It tends to | |
556 | * be a lot more optimized | |
557 | */ | |
558 | if (!cached) { | |
559 | hit = external_grep(opt, paths, cached); | |
560 | if (hit >= 0) | |
561 | return hit; | |
562 | } | |
563 | #endif | |
564 | ||
5010cb5f JH |
565 | for (nr = 0; nr < active_nr; nr++) { |
566 | struct cache_entry *ce = active_cache[nr]; | |
567 | if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode))) | |
568 | continue; | |
1362671f | 569 | if (!pathspec_matches(paths, ce->name)) |
5010cb5f JH |
570 | continue; |
571 | if (cached) | |
572 | hit |= grep_sha1(opt, ce->sha1, ce->name); | |
573 | else | |
574 | hit |= grep_file(opt, ce->name); | |
575 | } | |
576 | return hit; | |
577 | } | |
578 | ||
1362671f | 579 | static int grep_tree(struct grep_opt *opt, const char **paths, |
5010cb5f JH |
580 | struct tree_desc *tree, |
581 | const char *tree_name, const char *base) | |
582 | { | |
5010cb5f JH |
583 | int len; |
584 | int hit = 0; | |
4c068a98 | 585 | struct name_entry entry; |
e0eb889f | 586 | char *down; |
5010cb5f JH |
587 | char *path_buf = xmalloc(PATH_MAX + strlen(tree_name) + 100); |
588 | ||
589 | if (tree_name[0]) { | |
590 | int offset = sprintf(path_buf, "%s:", tree_name); | |
e0eb889f JH |
591 | down = path_buf + offset; |
592 | strcat(down, base); | |
5010cb5f JH |
593 | } |
594 | else { | |
e0eb889f JH |
595 | down = path_buf; |
596 | strcpy(down, base); | |
5010cb5f JH |
597 | } |
598 | len = strlen(path_buf); | |
599 | ||
4c068a98 LT |
600 | while (tree_entry(tree, &entry)) { |
601 | strcpy(path_buf + len, entry.path); | |
5010cb5f | 602 | |
4c068a98 | 603 | if (S_ISDIR(entry.mode)) |
e0eb889f JH |
604 | /* Match "abc/" against pathspec to |
605 | * decide if we want to descend into "abc" | |
606 | * directory. | |
607 | */ | |
4c068a98 | 608 | strcpy(path_buf + len + entry.pathlen, "/"); |
e0eb889f | 609 | |
1362671f | 610 | if (!pathspec_matches(paths, down)) |
5010cb5f | 611 | ; |
4c068a98 LT |
612 | else if (S_ISREG(entry.mode)) |
613 | hit |= grep_sha1(opt, entry.sha1, path_buf); | |
614 | else if (S_ISDIR(entry.mode)) { | |
5010cb5f JH |
615 | char type[20]; |
616 | struct tree_desc sub; | |
617 | void *data; | |
4c068a98 | 618 | data = read_sha1_file(entry.sha1, type, &sub.size); |
5010cb5f JH |
619 | if (!data) |
620 | die("unable to read tree (%s)", | |
4c068a98 | 621 | sha1_to_hex(entry.sha1)); |
5010cb5f | 622 | sub.buf = data; |
1362671f | 623 | hit |= grep_tree(opt, paths, &sub, tree_name, down); |
5010cb5f JH |
624 | free(data); |
625 | } | |
5010cb5f JH |
626 | } |
627 | return hit; | |
628 | } | |
629 | ||
1362671f | 630 | static int grep_object(struct grep_opt *opt, const char **paths, |
5010cb5f JH |
631 | struct object *obj, const char *name) |
632 | { | |
885a86ab | 633 | if (obj->type == TYPE_BLOB) |
5010cb5f | 634 | return grep_sha1(opt, obj->sha1, name); |
885a86ab | 635 | if (obj->type == TYPE_COMMIT || obj->type == TYPE_TREE) { |
5010cb5f JH |
636 | struct tree_desc tree; |
637 | void *data; | |
638 | int hit; | |
639 | data = read_object_with_reference(obj->sha1, tree_type, | |
640 | &tree.size, NULL); | |
641 | if (!data) | |
642 | die("unable to read tree (%s)", sha1_to_hex(obj->sha1)); | |
643 | tree.buf = data; | |
1362671f | 644 | hit = grep_tree(opt, paths, &tree, name, ""); |
5010cb5f JH |
645 | free(data); |
646 | return hit; | |
647 | } | |
885a86ab | 648 | die("unable to grep from object of type %s", typename(obj->type)); |
5010cb5f JH |
649 | } |
650 | ||
651 | static const char builtin_grep_usage[] = | |
652 | "git-grep <option>* <rev>* [-e] <pattern> [<path>...]"; | |
653 | ||
654 | int cmd_grep(int argc, const char **argv, char **envp) | |
655 | { | |
5010cb5f | 656 | int hit = 0; |
5010cb5f | 657 | int cached = 0; |
5acd64ed | 658 | int seen_dashdash = 0; |
5010cb5f | 659 | struct grep_opt opt; |
1f1e895f | 660 | struct object_array list = { 0, 0, NULL }; |
1362671f JH |
661 | const char *prefix = setup_git_directory(); |
662 | const char **paths = NULL; | |
5acd64ed | 663 | int i; |
5010cb5f JH |
664 | |
665 | memset(&opt, 0, sizeof(opt)); | |
f9b9faf6 | 666 | opt.pattern_tail = &opt.pattern_list; |
5010cb5f JH |
667 | opt.regflags = REG_NEWLINE; |
668 | ||
669 | /* | |
5acd64ed JH |
670 | * If there is no -- then the paths must exist in the working |
671 | * tree. If there is no explicit pattern specified with -e or | |
672 | * -f, we take the first unrecognized non option to be the | |
673 | * pattern, but then what follows it must be zero or more | |
674 | * valid refs up to the -- (if exists), and then existing | |
675 | * paths. If there is an explicit pattern, then the first | |
676 | * unrecocnized non option is the beginning of the refs list | |
677 | * that continues up to the -- (if exists), and then paths. | |
5010cb5f | 678 | */ |
5acd64ed | 679 | |
1362671f JH |
680 | while (1 < argc) { |
681 | const char *arg = argv[1]; | |
682 | argc--; argv++; | |
683 | if (!strcmp("--cached", arg)) { | |
684 | cached = 1; | |
685 | continue; | |
686 | } | |
b8d0f5a0 JH |
687 | if (!strcmp("-a", arg) || |
688 | !strcmp("--text", arg)) { | |
689 | opt.binary = GREP_BINARY_TEXT; | |
690 | continue; | |
691 | } | |
1362671f JH |
692 | if (!strcmp("-i", arg) || |
693 | !strcmp("--ignore-case", arg)) { | |
694 | opt.regflags |= REG_ICASE; | |
695 | continue; | |
696 | } | |
b8d0f5a0 JH |
697 | if (!strcmp("-I", arg)) { |
698 | opt.binary = GREP_BINARY_NOMATCH; | |
699 | continue; | |
700 | } | |
1362671f JH |
701 | if (!strcmp("-v", arg) || |
702 | !strcmp("--invert-match", arg)) { | |
703 | opt.invert = 1; | |
704 | continue; | |
705 | } | |
706 | if (!strcmp("-E", arg) || | |
707 | !strcmp("--extended-regexp", arg)) { | |
708 | opt.regflags |= REG_EXTENDED; | |
709 | continue; | |
710 | } | |
07ea91d8 JH |
711 | if (!strcmp("-F", arg) || |
712 | !strcmp("--fixed-strings", arg)) { | |
713 | opt.fixed = 1; | |
714 | continue; | |
715 | } | |
1362671f JH |
716 | if (!strcmp("-G", arg) || |
717 | !strcmp("--basic-regexp", arg)) { | |
718 | opt.regflags &= ~REG_EXTENDED; | |
719 | continue; | |
720 | } | |
721 | if (!strcmp("-n", arg)) { | |
722 | opt.linenum = 1; | |
723 | continue; | |
724 | } | |
725 | if (!strcmp("-H", arg)) { | |
726 | /* We always show the pathname, so this | |
727 | * is a noop. | |
728 | */ | |
729 | continue; | |
730 | } | |
731 | if (!strcmp("-l", arg) || | |
732 | !strcmp("--files-with-matches", arg)) { | |
733 | opt.name_only = 1; | |
734 | continue; | |
735 | } | |
e23d2d6b JH |
736 | if (!strcmp("-L", arg) || |
737 | !strcmp("--files-without-match", arg)) { | |
738 | opt.unmatch_name_only = 1; | |
739 | continue; | |
740 | } | |
2c866cf1 JH |
741 | if (!strcmp("-c", arg) || |
742 | !strcmp("--count", arg)) { | |
743 | opt.count = 1; | |
744 | continue; | |
745 | } | |
7839a25e JH |
746 | if (!strcmp("-w", arg) || |
747 | !strcmp("--word-regexp", arg)) { | |
748 | opt.word_regexp = 1; | |
749 | continue; | |
750 | } | |
f462ebb4 JH |
751 | if (!strncmp("-A", arg, 2) || |
752 | !strncmp("-B", arg, 2) || | |
753 | !strncmp("-C", arg, 2) || | |
754 | (arg[0] == '-' && '1' <= arg[1] && arg[1] <= '9')) { | |
1362671f | 755 | unsigned num; |
f462ebb4 JH |
756 | const char *scan; |
757 | switch (arg[1]) { | |
758 | case 'A': case 'B': case 'C': | |
759 | if (!arg[2]) { | |
760 | if (argc <= 1) | |
761 | usage(builtin_grep_usage); | |
762 | scan = *++argv; | |
763 | argc--; | |
764 | } | |
765 | else | |
766 | scan = arg + 2; | |
767 | break; | |
768 | default: | |
769 | scan = arg + 1; | |
770 | break; | |
771 | } | |
772 | if (sscanf(scan, "%u", &num) != 1) | |
5010cb5f | 773 | usage(builtin_grep_usage); |
1362671f JH |
774 | switch (arg[1]) { |
775 | case 'A': | |
776 | opt.post_context = num; | |
777 | break; | |
f462ebb4 | 778 | default: |
1362671f JH |
779 | case 'C': |
780 | opt.post_context = num; | |
781 | case 'B': | |
782 | opt.pre_context = num; | |
783 | break; | |
5010cb5f | 784 | } |
1362671f JH |
785 | continue; |
786 | } | |
aa8c79ad JH |
787 | if (!strcmp("-f", arg)) { |
788 | FILE *patterns; | |
789 | int lno = 0; | |
790 | char buf[1024]; | |
791 | if (argc <= 1) | |
792 | usage(builtin_grep_usage); | |
793 | patterns = fopen(argv[1], "r"); | |
794 | if (!patterns) | |
5acd64ed | 795 | die("'%s': %s", argv[1], strerror(errno)); |
aa8c79ad JH |
796 | while (fgets(buf, sizeof(buf), patterns)) { |
797 | int len = strlen(buf); | |
798 | if (buf[len-1] == '\n') | |
799 | buf[len-1] = 0; | |
800 | /* ignore empty line like grep does */ | |
801 | if (!buf[0]) | |
802 | continue; | |
803 | add_pattern(&opt, strdup(buf), argv[1], ++lno); | |
804 | } | |
805 | fclose(patterns); | |
806 | argv++; | |
807 | argc--; | |
808 | continue; | |
809 | } | |
1362671f JH |
810 | if (!strcmp("-e", arg)) { |
811 | if (1 < argc) { | |
aa8c79ad | 812 | add_pattern(&opt, argv[1], "-e option", 0); |
f9b9faf6 | 813 | argv++; |
1362671f | 814 | argc--; |
5010cb5f JH |
815 | continue; |
816 | } | |
1362671f JH |
817 | usage(builtin_grep_usage); |
818 | } | |
5acd64ed JH |
819 | if (!strcmp("--", arg)) |
820 | break; | |
821 | if (*arg == '-') | |
1362671f | 822 | usage(builtin_grep_usage); |
5acd64ed JH |
823 | |
824 | /* First unrecognized non-option token */ | |
f9b9faf6 | 825 | if (!opt.pattern_list) { |
aa8c79ad | 826 | add_pattern(&opt, arg, "command line", 0); |
1362671f JH |
827 | break; |
828 | } | |
829 | else { | |
830 | /* We are looking at the first path or rev; | |
5acd64ed | 831 | * it is found at argv[1] after leaving the |
1362671f JH |
832 | * loop. |
833 | */ | |
834 | argc++; argv--; | |
835 | break; | |
5010cb5f | 836 | } |
5010cb5f | 837 | } |
5acd64ed | 838 | |
f9b9faf6 | 839 | if (!opt.pattern_list) |
5010cb5f | 840 | die("no pattern given."); |
07ea91d8 JH |
841 | if ((opt.regflags != REG_NEWLINE) && opt.fixed) |
842 | die("cannot mix --fixed-strings and regexp"); | |
843 | if (!opt.fixed) | |
844 | compile_patterns(&opt); | |
5acd64ed JH |
845 | |
846 | /* Check revs and then paths */ | |
847 | for (i = 1; i < argc; i++) { | |
848 | const char *arg = argv[i]; | |
1362671f | 849 | unsigned char sha1[20]; |
5acd64ed JH |
850 | /* Is it a rev? */ |
851 | if (!get_sha1(arg, sha1)) { | |
852 | struct object *object = parse_object(sha1); | |
5acd64ed JH |
853 | if (!object) |
854 | die("bad object %s", arg); | |
1f1e895f | 855 | add_object_array(object, arg, &list); |
5acd64ed JH |
856 | continue; |
857 | } | |
858 | if (!strcmp(arg, "--")) { | |
859 | i++; | |
860 | seen_dashdash = 1; | |
861 | } | |
862 | break; | |
1362671f | 863 | } |
5acd64ed JH |
864 | |
865 | /* The rest are paths */ | |
866 | if (!seen_dashdash) { | |
867 | int j; | |
c39c4f47 | 868 | for (j = i; j < argc; j++) |
5acd64ed JH |
869 | verify_filename(prefix, argv[j]); |
870 | } | |
871 | ||
872 | if (i < argc) | |
873 | paths = get_pathspec(prefix, argv + i); | |
1362671f JH |
874 | else if (prefix) { |
875 | paths = xcalloc(2, sizeof(const char *)); | |
876 | paths[0] = prefix; | |
877 | paths[1] = NULL; | |
878 | } | |
5010cb5f | 879 | |
1f1e895f | 880 | if (!list.nr) |
1362671f | 881 | return !grep_cache(&opt, paths, cached); |
aa8c79ad | 882 | |
5010cb5f | 883 | if (cached) |
aa8c79ad | 884 | die("both --cached and trees are given."); |
5010cb5f | 885 | |
1f1e895f | 886 | for (i = 0; i < list.nr; i++) { |
5010cb5f | 887 | struct object *real_obj; |
1f1e895f LT |
888 | real_obj = deref_tag(list.objects[i].item, NULL, 0); |
889 | if (grep_object(&opt, paths, real_obj, list.objects[i].name)) | |
5010cb5f JH |
890 | hit = 1; |
891 | } | |
892 | return !hit; | |
893 | } |