built-in "git grep"
[git/git.git] / builtin-grep.c
CommitLineData
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"
11#include "diff.h"
12#include "revision.h"
13#include "builtin.h"
14#include <regex.h>
15
16static int pathspec_matches(struct diff_options *opt, const char *name)
17{
18 int i, j;
19 int namelen;
20 if (!opt->nr_paths)
21 return 1;
22 namelen = strlen(name);
23 for (i = 0; i < opt->nr_paths; i++) {
24 const char *match = opt->paths[i];
25 int matchlen = opt->pathlens[i];
26 if (matchlen <= namelen) {
27 if (!strncmp(name, match, matchlen))
28 return 1;
29 continue;
30 }
31 /* If name is "Documentation" and pathspec is
32 * "Documentation/", they should match. Maybe
33 * we would want to strip it in get_pathspec()???
34 */
35 if (strncmp(name, match, namelen))
36 continue;
37 for (j = namelen; j < matchlen; j++)
38 if (match[j] != '/')
39 break;
40 if (matchlen <= j)
41 return 1;
42 }
43 return 0;
44}
45
46struct grep_opt {
47 const char *pattern;
48 regex_t regexp;
49 unsigned linenum:1;
50 unsigned invert:1;
51 int regflags;
52 unsigned pre_context;
53 unsigned post_context;
54};
55
56static char *end_of_line(char *cp, unsigned long *left)
57{
58 unsigned long l = *left;
59 while (l && *cp != '\n') {
60 l--;
61 cp++;
62 }
63 *left = l;
64 return cp;
65}
66
67static void show_line(struct grep_opt *opt, const char *bol, const char *eol,
68 const char *name, unsigned lno, char sign)
69{
70 printf("%s%c", name, sign);
71 if (opt->linenum)
72 printf("%d%c", lno, sign);
73 printf("%.*s\n", eol-bol, bol);
74}
75
76static int grep_buffer(struct grep_opt *opt, const char *name,
77 char *buf, unsigned long size)
78{
79 char *bol = buf;
80 unsigned long left = size;
81 unsigned lno = 1;
82 struct pre_context_line {
83 char *bol;
84 char *eol;
85 } *prev = NULL, *pcl;
86 unsigned last_hit = 0;
87 unsigned last_shown = 0;
88 const char *hunk_mark = "";
89
90 if (opt->pre_context)
91 prev = xcalloc(opt->pre_context, sizeof(*prev));
92 if (opt->pre_context || opt->post_context)
93 hunk_mark = "--\n";
94
95 while (left) {
96 regmatch_t pmatch[10];
97 char *eol, ch;
98 int hit;
99
100 eol = end_of_line(bol, &left);
101 ch = *eol;
102 *eol = 0;
103
104 hit = !regexec(&opt->regexp, bol, ARRAY_SIZE(pmatch),
105 pmatch, 0);
106 if (opt->invert)
107 hit = !hit;
108 if (hit) {
109 /* Hit at this line. If we haven't shown the
110 * pre-context lines, we would need to show them.
111 */
112 if (opt->pre_context) {
113 unsigned from;
114 if (opt->pre_context < lno)
115 from = lno - opt->pre_context;
116 else
117 from = 1;
118 if (from <= last_shown)
119 from = last_shown + 1;
120 if (last_shown && from != last_shown + 1)
121 printf(hunk_mark);
122 while (from < lno) {
123 pcl = &prev[lno-from-1];
124 show_line(opt, pcl->bol, pcl->eol,
125 name, from, '-');
126 from++;
127 }
128 last_shown = lno-1;
129 }
130 if (last_shown && lno != last_shown + 1)
131 printf(hunk_mark);
132 show_line(opt, bol, eol, name, lno, ':');
133 last_shown = last_hit = lno;
134 }
135 else if (last_hit &&
136 lno <= last_hit + opt->post_context) {
137 /* If the last hit is within the post context,
138 * we need to show this line.
139 */
140 if (last_shown && lno != last_shown + 1)
141 printf(hunk_mark);
142 show_line(opt, bol, eol, name, lno, '-');
143 last_shown = lno;
144 }
145 if (opt->pre_context) {
146 memmove(prev+1, prev,
147 (opt->pre_context-1) * sizeof(*prev));
148 prev->bol = bol;
149 prev->eol = eol;
150 }
151 *eol = ch;
152 bol = eol + 1;
153 left--;
154 lno++;
155 }
156 return !!last_hit;
157}
158
159static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name)
160{
161 unsigned long size;
162 char *data;
163 char type[20];
164 int hit;
165 data = read_sha1_file(sha1, type, &size);
166 if (!data) {
167 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
168 return 0;
169 }
170 hit = grep_buffer(opt, name, data, size);
171 free(data);
172 return hit;
173}
174
175static int grep_file(struct grep_opt *opt, const char *filename)
176{
177 struct stat st;
178 int i;
179 char *data;
180 if (lstat(filename, &st) < 0) {
181 err_ret:
182 if (errno != ENOENT)
183 error("'%s': %s", filename, strerror(errno));
184 return 0;
185 }
186 if (!st.st_size)
187 return 0; /* empty file -- no grep hit */
188 if (!S_ISREG(st.st_mode))
189 return 0;
190 i = open(filename, O_RDONLY);
191 if (i < 0)
192 goto err_ret;
193 data = xmalloc(st.st_size + 1);
194 if (st.st_size != xread(i, data, st.st_size)) {
195 error("'%s': short read %s", filename, strerror(errno));
196 close(i);
197 free(data);
198 return 0;
199 }
200 close(i);
201 i = grep_buffer(opt, filename, data, st.st_size);
202 free(data);
203 return i;
204}
205
206static int grep_cache(struct grep_opt *opt, struct rev_info *revs, int cached)
207{
208 int hit = 0;
209 int nr;
210 read_cache();
211
212 for (nr = 0; nr < active_nr; nr++) {
213 struct cache_entry *ce = active_cache[nr];
214 if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
215 continue;
216 if (!pathspec_matches(&revs->diffopt, ce->name))
217 continue;
218 if (cached)
219 hit |= grep_sha1(opt, ce->sha1, ce->name);
220 else
221 hit |= grep_file(opt, ce->name);
222 }
223 return hit;
224}
225
226static int grep_tree(struct grep_opt *opt, struct rev_info *revs,
227 struct tree_desc *tree,
228 const char *tree_name, const char *base)
229{
230 unsigned mode;
231 int len;
232 int hit = 0;
233 const char *path;
234 const unsigned char *sha1;
235 char *down_base;
236 char *path_buf = xmalloc(PATH_MAX + strlen(tree_name) + 100);
237
238 if (tree_name[0]) {
239 int offset = sprintf(path_buf, "%s:", tree_name);
240 down_base = path_buf + offset;
241 strcat(down_base, base);
242 }
243 else {
244 down_base = path_buf;
245 strcpy(down_base, base);
246 }
247 len = strlen(path_buf);
248
249 while (tree->size) {
250 int pathlen;
251 sha1 = tree_entry_extract(tree, &path, &mode);
252 pathlen = strlen(path);
253 strcpy(path_buf + len, path);
254
255 if (!pathspec_matches(&revs->diffopt, down_base))
256 ;
257 else if (S_ISREG(mode))
258 hit |= grep_sha1(opt, sha1, path_buf);
259 else if (S_ISDIR(mode)) {
260 char type[20];
261 struct tree_desc sub;
262 void *data;
263 data = read_sha1_file(sha1, type, &sub.size);
264 if (!data)
265 die("unable to read tree (%s)",
266 sha1_to_hex(sha1));
267 strcpy(path_buf + len + pathlen, "/");
268 sub.buf = data;
269 hit = grep_tree(opt, revs, &sub, tree_name, down_base);
270 free(data);
271 }
272 update_tree_entry(tree);
273 }
274 return hit;
275}
276
277static int grep_object(struct grep_opt *opt, struct rev_info *revs,
278 struct object *obj, const char *name)
279{
280 if (!strcmp(obj->type, blob_type))
281 return grep_sha1(opt, obj->sha1, name);
282 if (!strcmp(obj->type, commit_type) ||
283 !strcmp(obj->type, tree_type)) {
284 struct tree_desc tree;
285 void *data;
286 int hit;
287 data = read_object_with_reference(obj->sha1, tree_type,
288 &tree.size, NULL);
289 if (!data)
290 die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
291 tree.buf = data;
292 hit = grep_tree(opt, revs, &tree, name, "");
293 free(data);
294 return hit;
295 }
296 die("unable to grep from object of type %s", obj->type);
297}
298
299static const char builtin_grep_usage[] =
300"git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
301
302int cmd_grep(int argc, const char **argv, char **envp)
303{
304 struct rev_info rev;
305 const char **dst, **src;
306 int err;
307 int hit = 0;
308 int no_more_arg = 0;
309 int seen_range = 0;
310 int seen_noncommit = 0;
311 int cached = 0;
312 struct grep_opt opt;
313 struct object_list *list;
314
315 memset(&opt, 0, sizeof(opt));
316 opt.regflags = REG_NEWLINE;
317
318 /*
319 * Interpret and remove the grep options upfront. Sigh...
320 */
321 for (dst = src = &argv[1]; src < argc + argv; ) {
322 const char *arg = *src++;
323 if (!no_more_arg) {
324 if (!strcmp("--", arg)) {
325 no_more_arg = 1;
326 *dst++ = arg;
327 continue;
328 }
329 if (!strcmp("--cached", arg)) {
330 cached = 1;
331 continue;
332 }
333 if (!strcmp("-i", arg) ||
334 !strcmp("--ignore-case", arg)) {
335 opt.regflags |= REG_ICASE;
336 continue;
337 }
338 if (!strcmp("-v", arg) ||
339 !strcmp("--invert-match", arg)) {
340 opt.invert = 1;
341 continue;
342 }
343 if (!strcmp("-E", arg) ||
344 !strcmp("--extended-regexp", arg)) {
345 opt.regflags |= REG_EXTENDED;
346 continue;
347 }
348 if (!strcmp("-G", arg) ||
349 !strcmp("--basic-regexp", arg)) {
350 opt.regflags &= ~REG_EXTENDED;
351 continue;
352 }
353 if (!strcmp("-e", arg)) {
354 if (src < argc + argv) {
355 opt.pattern = *src++;
356 continue;
357 }
358 usage(builtin_grep_usage);
359 }
360 if (!strcmp("-n", arg)) {
361 opt.linenum = 1;
362 continue;
363 }
364 if (!strcmp("-H", arg)) {
365 /* We always show the pathname, so this
366 * is a noop.
367 */
368 continue;
369 }
370 if (!strcmp("-A", arg) ||
371 !strcmp("-B", arg) ||
372 !strcmp("-C", arg)) {
373 unsigned num;
374 if ((argc + argv <= src) ||
375 sscanf(*src++, "%u", &num) != 1)
376 usage(builtin_grep_usage);
377 switch (arg[1]) {
378 case 'A':
379 opt.post_context = num;
380 break;
381 case 'C':
382 opt.post_context = num;
383 case 'B':
384 opt.pre_context = num;
385 break;
386 }
387 continue;
388 }
389 }
390 *dst++ = arg;
391 }
392 if (!opt.pattern)
393 die("no pattern given.");
394
395 err = regcomp(&opt.regexp, opt.pattern, opt.regflags);
396 if (err) {
397 char errbuf[1024];
398 regerror(err, &opt.regexp, errbuf, 1024);
399 regfree(&opt.regexp);
400 die("'%s': %s", opt.pattern, errbuf);
401 }
402
403 init_revisions(&rev);
404 *dst = NULL;
405 argc = setup_revisions(dst - argv, argv, &rev, NULL);
406
407 /*
408 * Do not walk "grep -e foo master next pu -- Documentation/"
409 * but do walk "grep -e foo master..next -- Documentation/".
410 * Ranged request mixed with a blob or tree object, like
411 * "grep -e foo v1.0.0:Documentation/ master..next"
412 * so detect that and complain.
413 */
414 for (list = rev.pending_objects; list; list = list->next) {
415 struct object *real_obj;
416 if (list->item->flags & UNINTERESTING)
417 seen_range = 1;
418 real_obj = deref_tag(list->item, NULL, 0);
419 if (strcmp(real_obj->type, commit_type))
420 seen_noncommit = 1;
421 }
422 if (!rev.pending_objects)
423 return !grep_cache(&opt, &rev, cached);
424 if (cached)
425 die("both --cached and revisions given.");
426
427 if (seen_range && seen_noncommit)
428 die("both A..B and non commit are given.");
429 if (seen_range) {
430 struct commit *commit;
431 prepare_revision_walk(&rev);
432 while ((commit = get_revision(&rev)) != NULL) {
433 unsigned char *sha1 = commit->object.sha1;
434 const char *n = find_unique_abbrev(sha1, rev.abbrev);
435 char rev_name[41];
436 strcpy(rev_name, n);
437 if (grep_object(&opt, &rev, &commit->object, rev_name))
438 hit = 1;
439 commit->buffer = NULL;
440 }
441 return !hit;
442 }
443
444 /* all of them are non-commit; do not walk, and
445 * do not lose their names.
446 */
447 for (list = rev.pending_objects; list; list = list->next) {
448 struct object *real_obj;
449 real_obj = deref_tag(list->item, NULL, 0);
450 if (grep_object(&opt, &rev, real_obj, list->name))
451 hit = 1;
452 }
453 return !hit;
454}