git-grep: don't use sscanf
[git/vmiklos.git] / builtin-grep.c
blob4510d353247355a28979fbf63b15523c2e705591
1 /*
2 * Builtin "git grep"
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 "tree-walk.h"
12 #include "builtin.h"
13 #include "grep.h"
16 * git grep pathspecs are somewhat different from diff-tree pathspecs;
17 * pathname wildcards are allowed.
19 static int pathspec_matches(const char **paths, const char *name)
21 int namelen, i;
22 if (!paths || !*paths)
23 return 1;
24 namelen = strlen(name);
25 for (i = 0; paths[i]; i++) {
26 const char *match = paths[i];
27 int matchlen = strlen(match);
28 const char *cp, *meta;
30 if (!matchlen ||
31 ((matchlen <= namelen) &&
32 !strncmp(name, match, matchlen) &&
33 (match[matchlen-1] == '/' ||
34 name[matchlen] == '\0' || name[matchlen] == '/')))
35 return 1;
36 if (!fnmatch(match, name, 0))
37 return 1;
38 if (name[namelen-1] != '/')
39 continue;
41 /* We are being asked if the directory ("name") is worth
42 * descending into.
44 * Find the longest leading directory name that does
45 * not have metacharacter in the pathspec; the name
46 * we are looking at must overlap with that directory.
48 for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
49 char ch = *cp;
50 if (ch == '*' || ch == '[' || ch == '?') {
51 meta = cp;
52 break;
55 if (!meta)
56 meta = cp; /* fully literal */
58 if (namelen <= meta - match) {
59 /* Looking at "Documentation/" and
60 * the pattern says "Documentation/howto/", or
61 * "Documentation/diff*.txt". The name we
62 * have should match prefix.
64 if (!memcmp(match, name, namelen))
65 return 1;
66 continue;
69 if (meta - match < namelen) {
70 /* Looking at "Documentation/howto/" and
71 * the pattern says "Documentation/h*";
72 * match up to "Do.../h"; this avoids descending
73 * into "Documentation/technical/".
75 if (!memcmp(match, name, meta - match))
76 return 1;
77 continue;
80 return 0;
83 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name, int tree_name_len)
85 unsigned long size;
86 char *data;
87 enum object_type type;
88 char *to_free = NULL;
89 int hit;
91 data = read_sha1_file(sha1, &type, &size);
92 if (!data) {
93 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
94 return 0;
96 if (opt->relative && opt->prefix_length) {
97 static char name_buf[PATH_MAX];
98 char *cp;
99 int name_len = strlen(name) - opt->prefix_length + 1;
101 if (!tree_name_len)
102 name += opt->prefix_length;
103 else {
104 if (ARRAY_SIZE(name_buf) <= name_len)
105 cp = to_free = xmalloc(name_len);
106 else
107 cp = name_buf;
108 memcpy(cp, name, tree_name_len);
109 strcpy(cp + tree_name_len,
110 name + tree_name_len + opt->prefix_length);
111 name = cp;
114 hit = grep_buffer(opt, name, data, size);
115 free(data);
116 free(to_free);
117 return hit;
120 static int grep_file(struct grep_opt *opt, const char *filename)
122 struct stat st;
123 int i;
124 char *data;
125 size_t sz;
127 if (lstat(filename, &st) < 0) {
128 err_ret:
129 if (errno != ENOENT)
130 error("'%s': %s", filename, strerror(errno));
131 return 0;
133 if (!st.st_size)
134 return 0; /* empty file -- no grep hit */
135 if (!S_ISREG(st.st_mode))
136 return 0;
137 sz = xsize_t(st.st_size);
138 i = open(filename, O_RDONLY);
139 if (i < 0)
140 goto err_ret;
141 data = xmalloc(sz + 1);
142 if (st.st_size != read_in_full(i, data, sz)) {
143 error("'%s': short read %s", filename, strerror(errno));
144 close(i);
145 free(data);
146 return 0;
148 close(i);
149 if (opt->relative && opt->prefix_length)
150 filename += opt->prefix_length;
151 i = grep_buffer(opt, filename, data, sz);
152 free(data);
153 return i;
156 #ifdef __unix__
157 static int exec_grep(int argc, const char **argv)
159 pid_t pid;
160 int status;
162 argv[argc] = NULL;
163 pid = fork();
164 if (pid < 0)
165 return pid;
166 if (!pid) {
167 execvp("grep", (char **) argv);
168 exit(255);
170 while (waitpid(pid, &status, 0) < 0) {
171 if (errno == EINTR)
172 continue;
173 return -1;
175 if (WIFEXITED(status)) {
176 if (!WEXITSTATUS(status))
177 return 1;
178 return 0;
180 return -1;
183 #define MAXARGS 1000
184 #define ARGBUF 4096
185 #define push_arg(a) do { \
186 if (nr < MAXARGS) argv[nr++] = (a); \
187 else die("maximum number of args exceeded"); \
188 } while (0)
190 static int external_grep(struct grep_opt *opt, const char **paths, int cached)
192 int i, nr, argc, hit, len, status;
193 const char *argv[MAXARGS+1];
194 char randarg[ARGBUF];
195 char *argptr = randarg;
196 struct grep_pat *p;
198 if (opt->extended || (opt->relative && opt->prefix_length))
199 return -1;
200 len = nr = 0;
201 push_arg("grep");
202 if (opt->fixed)
203 push_arg("-F");
204 if (opt->linenum)
205 push_arg("-n");
206 if (!opt->pathname)
207 push_arg("-h");
208 if (opt->regflags & REG_EXTENDED)
209 push_arg("-E");
210 if (opt->regflags & REG_ICASE)
211 push_arg("-i");
212 if (opt->word_regexp)
213 push_arg("-w");
214 if (opt->name_only)
215 push_arg("-l");
216 if (opt->unmatch_name_only)
217 push_arg("-L");
218 if (opt->count)
219 push_arg("-c");
220 if (opt->post_context || opt->pre_context) {
221 if (opt->post_context != opt->pre_context) {
222 if (opt->pre_context) {
223 push_arg("-B");
224 len += snprintf(argptr, sizeof(randarg)-len,
225 "%u", opt->pre_context);
226 if (sizeof(randarg) <= len)
227 die("maximum length of args exceeded");
228 push_arg(argptr);
229 argptr += len;
231 if (opt->post_context) {
232 push_arg("-A");
233 len += snprintf(argptr, sizeof(randarg)-len,
234 "%u", opt->post_context);
235 if (sizeof(randarg) <= len)
236 die("maximum length of args exceeded");
237 push_arg(argptr);
238 argptr += len;
241 else {
242 push_arg("-C");
243 len += snprintf(argptr, sizeof(randarg)-len,
244 "%u", opt->post_context);
245 if (sizeof(randarg) <= len)
246 die("maximum length of args exceeded");
247 push_arg(argptr);
248 argptr += len;
251 for (p = opt->pattern_list; p; p = p->next) {
252 push_arg("-e");
253 push_arg(p->pattern);
257 * To make sure we get the header printed out when we want it,
258 * add /dev/null to the paths to grep. This is unnecessary
259 * (and wrong) with "-l" or "-L", which always print out the
260 * name anyway.
262 * GNU grep has "-H", but this is portable.
264 if (!opt->name_only && !opt->unmatch_name_only)
265 push_arg("/dev/null");
267 hit = 0;
268 argc = nr;
269 for (i = 0; i < active_nr; i++) {
270 struct cache_entry *ce = active_cache[i];
271 char *name;
272 if (!S_ISREG(ntohl(ce->ce_mode)))
273 continue;
274 if (!pathspec_matches(paths, ce->name))
275 continue;
276 name = ce->name;
277 if (name[0] == '-') {
278 int len = ce_namelen(ce);
279 name = xmalloc(len + 3);
280 memcpy(name, "./", 2);
281 memcpy(name + 2, ce->name, len + 1);
283 argv[argc++] = name;
284 if (argc < MAXARGS && !ce_stage(ce))
285 continue;
286 status = exec_grep(argc, argv);
287 if (0 < status)
288 hit = 1;
289 argc = nr;
290 if (ce_stage(ce)) {
291 do {
292 i++;
293 } while (i < active_nr &&
294 !strcmp(ce->name, active_cache[i]->name));
295 i--; /* compensate for loop control */
298 if (argc > nr) {
299 status = exec_grep(argc, argv);
300 if (0 < status)
301 hit = 1;
303 return hit;
305 #endif
307 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
309 int hit = 0;
310 int nr;
311 read_cache();
313 #ifdef __unix__
315 * Use the external "grep" command for the case where
316 * we grep through the checked-out files. It tends to
317 * be a lot more optimized
319 if (!cached) {
320 hit = external_grep(opt, paths, cached);
321 if (hit >= 0)
322 return hit;
324 #endif
326 for (nr = 0; nr < active_nr; nr++) {
327 struct cache_entry *ce = active_cache[nr];
328 if (!S_ISREG(ntohl(ce->ce_mode)))
329 continue;
330 if (!pathspec_matches(paths, ce->name))
331 continue;
332 if (cached) {
333 if (ce_stage(ce))
334 continue;
335 hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
337 else
338 hit |= grep_file(opt, ce->name);
339 if (ce_stage(ce)) {
340 do {
341 nr++;
342 } while (nr < active_nr &&
343 !strcmp(ce->name, active_cache[nr]->name));
344 nr--; /* compensate for loop control */
347 free_grep_patterns(opt);
348 return hit;
351 static int grep_tree(struct grep_opt *opt, const char **paths,
352 struct tree_desc *tree,
353 const char *tree_name, const char *base)
355 int len;
356 int hit = 0;
357 struct name_entry entry;
358 char *down;
359 int tn_len = strlen(tree_name);
360 char *path_buf = xmalloc(PATH_MAX + tn_len + 100);
362 if (tn_len) {
363 tn_len = sprintf(path_buf, "%s:", tree_name);
364 down = path_buf + tn_len;
365 strcat(down, base);
367 else {
368 down = path_buf;
369 strcpy(down, base);
371 len = strlen(path_buf);
373 while (tree_entry(tree, &entry)) {
374 strcpy(path_buf + len, entry.path);
376 if (S_ISDIR(entry.mode))
377 /* Match "abc/" against pathspec to
378 * decide if we want to descend into "abc"
379 * directory.
381 strcpy(path_buf + len + entry.pathlen, "/");
383 if (!pathspec_matches(paths, down))
385 else if (S_ISREG(entry.mode))
386 hit |= grep_sha1(opt, entry.sha1, path_buf, tn_len);
387 else if (S_ISDIR(entry.mode)) {
388 enum object_type type;
389 struct tree_desc sub;
390 void *data;
391 data = read_sha1_file(entry.sha1, &type, &sub.size);
392 if (!data)
393 die("unable to read tree (%s)",
394 sha1_to_hex(entry.sha1));
395 sub.buf = data;
396 hit |= grep_tree(opt, paths, &sub, tree_name, down);
397 free(data);
400 return hit;
403 static int grep_object(struct grep_opt *opt, const char **paths,
404 struct object *obj, const char *name)
406 if (obj->type == OBJ_BLOB)
407 return grep_sha1(opt, obj->sha1, name, 0);
408 if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
409 struct tree_desc tree;
410 void *data;
411 int hit;
412 data = read_object_with_reference(obj->sha1, tree_type,
413 &tree.size, NULL);
414 if (!data)
415 die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
416 tree.buf = data;
417 hit = grep_tree(opt, paths, &tree, name, "");
418 free(data);
419 return hit;
421 die("unable to grep from object of type %s", typename(obj->type));
424 static const char builtin_grep_usage[] =
425 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
427 static const char emsg_invalid_context_len[] =
428 "%s: invalid context length argument";
429 static const char emsg_missing_context_len[] =
430 "missing context length argument";
431 static const char emsg_missing_argument[] =
432 "option requires an argument -%s";
434 static int strtoul_ui(char const *s, unsigned int *result)
436 unsigned long ul;
437 char *p;
439 errno = 0;
440 ul = strtoul(s, &p, 10);
441 if (errno || *p || p == s || (unsigned int) ul != ul)
442 return -1;
443 *result = ul;
444 return 0;
447 int cmd_grep(int argc, const char **argv, const char *prefix)
449 int hit = 0;
450 int cached = 0;
451 int seen_dashdash = 0;
452 struct grep_opt opt;
453 struct object_array list = { 0, 0, NULL };
454 const char **paths = NULL;
455 int i;
457 memset(&opt, 0, sizeof(opt));
458 opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
459 opt.relative = 1;
460 opt.pathname = 1;
461 opt.pattern_tail = &opt.pattern_list;
462 opt.regflags = REG_NEWLINE;
465 * If there is no -- then the paths must exist in the working
466 * tree. If there is no explicit pattern specified with -e or
467 * -f, we take the first unrecognized non option to be the
468 * pattern, but then what follows it must be zero or more
469 * valid refs up to the -- (if exists), and then existing
470 * paths. If there is an explicit pattern, then the first
471 * unrecognized non option is the beginning of the refs list
472 * that continues up to the -- (if exists), and then paths.
475 while (1 < argc) {
476 const char *arg = argv[1];
477 argc--; argv++;
478 if (!strcmp("--cached", arg)) {
479 cached = 1;
480 continue;
482 if (!strcmp("-a", arg) ||
483 !strcmp("--text", arg)) {
484 opt.binary = GREP_BINARY_TEXT;
485 continue;
487 if (!strcmp("-i", arg) ||
488 !strcmp("--ignore-case", arg)) {
489 opt.regflags |= REG_ICASE;
490 continue;
492 if (!strcmp("-I", arg)) {
493 opt.binary = GREP_BINARY_NOMATCH;
494 continue;
496 if (!strcmp("-v", arg) ||
497 !strcmp("--invert-match", arg)) {
498 opt.invert = 1;
499 continue;
501 if (!strcmp("-E", arg) ||
502 !strcmp("--extended-regexp", arg)) {
503 opt.regflags |= REG_EXTENDED;
504 continue;
506 if (!strcmp("-F", arg) ||
507 !strcmp("--fixed-strings", arg)) {
508 opt.fixed = 1;
509 continue;
511 if (!strcmp("-G", arg) ||
512 !strcmp("--basic-regexp", arg)) {
513 opt.regflags &= ~REG_EXTENDED;
514 continue;
516 if (!strcmp("-n", arg)) {
517 opt.linenum = 1;
518 continue;
520 if (!strcmp("-h", arg)) {
521 opt.pathname = 0;
522 continue;
524 if (!strcmp("-H", arg)) {
525 opt.pathname = 1;
526 continue;
528 if (!strcmp("-l", arg) ||
529 !strcmp("--files-with-matches", arg)) {
530 opt.name_only = 1;
531 continue;
533 if (!strcmp("-L", arg) ||
534 !strcmp("--files-without-match", arg)) {
535 opt.unmatch_name_only = 1;
536 continue;
538 if (!strcmp("-c", arg) ||
539 !strcmp("--count", arg)) {
540 opt.count = 1;
541 continue;
543 if (!strcmp("-w", arg) ||
544 !strcmp("--word-regexp", arg)) {
545 opt.word_regexp = 1;
546 continue;
548 if (!prefixcmp(arg, "-A") ||
549 !prefixcmp(arg, "-B") ||
550 !prefixcmp(arg, "-C") ||
551 (arg[0] == '-' && '1' <= arg[1] && arg[1] <= '9')) {
552 unsigned num;
553 const char *scan;
554 switch (arg[1]) {
555 case 'A': case 'B': case 'C':
556 if (!arg[2]) {
557 if (argc <= 1)
558 die(emsg_missing_context_len);
559 scan = *++argv;
560 argc--;
562 else
563 scan = arg + 2;
564 break;
565 default:
566 scan = arg + 1;
567 break;
569 if (strtoul_ui(scan, &num))
570 die(emsg_invalid_context_len, scan);
571 switch (arg[1]) {
572 case 'A':
573 opt.post_context = num;
574 break;
575 default:
576 case 'C':
577 opt.post_context = num;
578 case 'B':
579 opt.pre_context = num;
580 break;
582 continue;
584 if (!strcmp("-f", arg)) {
585 FILE *patterns;
586 int lno = 0;
587 char buf[1024];
588 if (argc <= 1)
589 die(emsg_missing_argument, arg);
590 patterns = fopen(argv[1], "r");
591 if (!patterns)
592 die("'%s': %s", argv[1], strerror(errno));
593 while (fgets(buf, sizeof(buf), patterns)) {
594 int len = strlen(buf);
595 if (buf[len-1] == '\n')
596 buf[len-1] = 0;
597 /* ignore empty line like grep does */
598 if (!buf[0])
599 continue;
600 append_grep_pattern(&opt, xstrdup(buf),
601 argv[1], ++lno,
602 GREP_PATTERN);
604 fclose(patterns);
605 argv++;
606 argc--;
607 continue;
609 if (!strcmp("--not", arg)) {
610 append_grep_pattern(&opt, arg, "command line", 0,
611 GREP_NOT);
612 continue;
614 if (!strcmp("--and", arg)) {
615 append_grep_pattern(&opt, arg, "command line", 0,
616 GREP_AND);
617 continue;
619 if (!strcmp("--or", arg))
620 continue; /* no-op */
621 if (!strcmp("(", arg)) {
622 append_grep_pattern(&opt, arg, "command line", 0,
623 GREP_OPEN_PAREN);
624 continue;
626 if (!strcmp(")", arg)) {
627 append_grep_pattern(&opt, arg, "command line", 0,
628 GREP_CLOSE_PAREN);
629 continue;
631 if (!strcmp("--all-match", arg)) {
632 opt.all_match = 1;
633 continue;
635 if (!strcmp("-e", arg)) {
636 if (1 < argc) {
637 append_grep_pattern(&opt, argv[1],
638 "-e option", 0,
639 GREP_PATTERN);
640 argv++;
641 argc--;
642 continue;
644 die(emsg_missing_argument, arg);
646 if (!strcmp("--full-name", arg)) {
647 opt.relative = 0;
648 continue;
650 if (!strcmp("--", arg)) {
651 /* later processing wants to have this at argv[1] */
652 argv--;
653 argc++;
654 break;
656 if (*arg == '-')
657 usage(builtin_grep_usage);
659 /* First unrecognized non-option token */
660 if (!opt.pattern_list) {
661 append_grep_pattern(&opt, arg, "command line", 0,
662 GREP_PATTERN);
663 break;
665 else {
666 /* We are looking at the first path or rev;
667 * it is found at argv[1] after leaving the
668 * loop.
670 argc++; argv--;
671 break;
675 if (!opt.pattern_list)
676 die("no pattern given.");
677 if ((opt.regflags != REG_NEWLINE) && opt.fixed)
678 die("cannot mix --fixed-strings and regexp");
679 compile_grep_patterns(&opt);
681 /* Check revs and then paths */
682 for (i = 1; i < argc; i++) {
683 const char *arg = argv[i];
684 unsigned char sha1[20];
685 /* Is it a rev? */
686 if (!get_sha1(arg, sha1)) {
687 struct object *object = parse_object(sha1);
688 if (!object)
689 die("bad object %s", arg);
690 add_object_array(object, arg, &list);
691 continue;
693 if (!strcmp(arg, "--")) {
694 i++;
695 seen_dashdash = 1;
697 break;
700 /* The rest are paths */
701 if (!seen_dashdash) {
702 int j;
703 for (j = i; j < argc; j++)
704 verify_filename(prefix, argv[j]);
707 if (i < argc) {
708 paths = get_pathspec(prefix, argv + i);
709 if (opt.prefix_length && opt.relative) {
710 /* Make sure we do not get outside of paths */
711 for (i = 0; paths[i]; i++)
712 if (strncmp(prefix, paths[i], opt.prefix_length))
713 die("git-grep: cannot generate relative filenames containing '..'");
716 else if (prefix) {
717 paths = xcalloc(2, sizeof(const char *));
718 paths[0] = prefix;
719 paths[1] = NULL;
722 if (!list.nr)
723 return !grep_cache(&opt, paths, cached);
725 if (cached)
726 die("both --cached and trees are given.");
728 for (i = 0; i < list.nr; i++) {
729 struct object *real_obj;
730 real_obj = deref_tag(list.objects[i].item, NULL, 0);
731 if (grep_object(&opt, paths, real_obj, list.objects[i].name))
732 hit = 1;
734 free_grep_patterns(&opt);
735 return !hit;