1 /*
2  * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/queue.h>
19 #include <sys/uio.h>
20 #include <sys/socket.h>
21 #include <sys/stat.h>
22 #include <sys/mman.h>
23 
24 #include <ctype.h>
25 #include <endian.h>
26 #include <fcntl.h>
27 #include <fnmatch.h>
28 #include <limits.h>
29 #include <dirent.h>
30 #include <stdlib.h>
31 #include <stdio.h>
32 #include <sha1.h>
33 #include <string.h>
34 #include <time.h>
35 #include <unistd.h>
36 #include <zlib.h>
37 #include <errno.h>
38 #include <libgen.h>
39 #include <stdint.h>
40 #include <imsg.h>
41 #include <uuid.h>
42 
43 #include "got_error.h"
44 #include "got_reference.h"
45 #include "got_repository.h"
46 #include "got_path.h"
47 #include "got_cancel.h"
48 #include "got_worktree.h"
49 #include "got_object.h"
50 
51 #include "got_lib_delta.h"
52 #include "got_lib_inflate.h"
53 #include "got_lib_object.h"
54 #include "got_lib_object_parse.h"
55 #include "got_lib_object_create.h"
56 #include "got_lib_pack.h"
57 #include "got_lib_privsep.h"
58 #include "got_lib_worktree.h"
59 #include "got_lib_sha1.h"
60 #include "got_lib_object_cache.h"
61 #include "got_lib_repository.h"
62 #include "got_lib_gotconfig.h"
63 
64 #ifndef nitems
65 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
66 #endif
67 
68 const char *
69 got_repo_get_path(struct got_repository *repo)
70 {
71 	return repo->path;
72 }
73 
74 const char *
75 got_repo_get_path_git_dir(struct got_repository *repo)
76 {
77 	return repo->path_git_dir;
78 }
79 
80 const char *
81 got_repo_get_gitconfig_author_name(struct got_repository *repo)
82 {
83 	return repo->gitconfig_author_name;
84 }
85 
86 const char *
87 got_repo_get_gitconfig_author_email(struct got_repository *repo)
88 {
89 	return repo->gitconfig_author_email;
90 }
91 
92 const char *
93 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
94 {
95 	return repo->global_gitconfig_author_name;
96 }
97 
98 const char *
99 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
100 {
101 	return repo->global_gitconfig_author_email;
102 }
103 
104 const char *
105 got_repo_get_gitconfig_owner(struct got_repository *repo)
106 {
107 	return repo->gitconfig_owner;
108 }
109 
110 int
111 got_repo_is_bare(struct got_repository *repo)
112 {
113 	return (strcmp(repo->path, repo->path_git_dir) == 0);
114 }
115 
116 static char *
117 get_path_git_child(struct got_repository *repo, const char *basename)
118 {
119 	char *path_child;
120 
121 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
122 	    basename) == -1)
123 		return NULL;
124 
125 	return path_child;
126 }
127 
128 char *
129 got_repo_get_path_objects(struct got_repository *repo)
130 {
131 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
132 }
133 
134 char *
135 got_repo_get_path_objects_pack(struct got_repository *repo)
136 {
137 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
138 }
139 
140 char *
141 got_repo_get_path_refs(struct got_repository *repo)
142 {
143 	return get_path_git_child(repo, GOT_REFS_DIR);
144 }
145 
146 char *
147 got_repo_get_path_packed_refs(struct got_repository *repo)
148 {
149 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
150 }
151 
152 static char *
153 get_path_head(struct got_repository *repo)
154 {
155 	return get_path_git_child(repo, GOT_HEAD_FILE);
156 }
157 
158 char *
159 got_repo_get_path_gitconfig(struct got_repository *repo)
160 {
161 	return get_path_git_child(repo, GOT_GITCONFIG);
162 }
163 
164 char *
165 got_repo_get_path_gotconfig(struct got_repository *repo)
166 {
167 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
168 }
169 
170 const struct got_gotconfig *
171 got_repo_get_gotconfig(struct got_repository *repo)
172 {
173 	return repo->gotconfig;
174 }
175 
176 void
177 got_repo_get_gitconfig_remotes(int *nremotes,
178     const struct got_remote_repo **remotes, struct got_repository *repo)
179 {
180 	*nremotes = repo->ngitconfig_remotes;
181 	*remotes = repo->gitconfig_remotes;
182 }
183 
184 static int
185 is_git_repo(struct got_repository *repo)
186 {
187 	const char *path_git = got_repo_get_path_git_dir(repo);
188 	char *path_objects = got_repo_get_path_objects(repo);
189 	char *path_refs = got_repo_get_path_refs(repo);
190 	char *path_head = get_path_head(repo);
191 	int ret = 0;
192 	struct stat sb;
193 	struct got_reference *head_ref;
194 
195 	if (lstat(path_git, &sb) == -1)
196 		goto done;
197 	if (!S_ISDIR(sb.st_mode))
198 		goto done;
199 
200 	if (lstat(path_objects, &sb) == -1)
201 		goto done;
202 	if (!S_ISDIR(sb.st_mode))
203 		goto done;
204 
205 	if (lstat(path_refs, &sb) == -1)
206 		goto done;
207 	if (!S_ISDIR(sb.st_mode))
208 		goto done;
209 
210 	if (lstat(path_head, &sb) == -1)
211 		goto done;
212 	if (!S_ISREG(sb.st_mode))
213 		goto done;
214 
215 	/* Check if the HEAD reference can be opened. */
216 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
217 		goto done;
218 	got_ref_close(head_ref);
219 
220 	ret = 1;
221 done:
222 	free(path_objects);
223 	free(path_refs);
224 	free(path_head);
225 	return ret;
226 
227 }
228 
229 const struct got_error *
230 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
231     struct got_object *obj)
232 {
233 #ifndef GOT_NO_OBJ_CACHE
234 	const struct got_error *err = NULL;
235 	err = got_object_cache_add(&repo->objcache, id, obj);
236 	if (err) {
237 		if (err->code == GOT_ERR_OBJ_EXISTS ||
238 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
239 			err = NULL;
240 		return err;
241 	}
242 	obj->refcnt++;
243 #endif
244 	return NULL;
245 }
246 
247 struct got_object *
248 got_repo_get_cached_object(struct got_repository *repo,
249     struct got_object_id *id)
250 {
251 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
252 }
253 
254 const struct got_error *
255 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
256     struct got_tree_object *tree)
257 {
258 #ifndef GOT_NO_OBJ_CACHE
259 	const struct got_error *err = NULL;
260 	err = got_object_cache_add(&repo->treecache, id, tree);
261 	if (err) {
262 		if (err->code == GOT_ERR_OBJ_EXISTS ||
263 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
264 			err = NULL;
265 		return err;
266 	}
267 	tree->refcnt++;
268 #endif
269 	return NULL;
270 }
271 
272 struct got_tree_object *
273 got_repo_get_cached_tree(struct got_repository *repo,
274     struct got_object_id *id)
275 {
276 	return (struct got_tree_object *)got_object_cache_get(
277 	    &repo->treecache, id);
278 }
279 
280 const struct got_error *
281 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
282     struct got_commit_object *commit)
283 {
284 #ifndef GOT_NO_OBJ_CACHE
285 	const struct got_error *err = NULL;
286 	err = got_object_cache_add(&repo->commitcache, id, commit);
287 	if (err) {
288 		if (err->code == GOT_ERR_OBJ_EXISTS ||
289 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
290 			err = NULL;
291 		return err;
292 	}
293 	commit->refcnt++;
294 #endif
295 	return NULL;
296 }
297 
298 struct got_commit_object *
299 got_repo_get_cached_commit(struct got_repository *repo,
300     struct got_object_id *id)
301 {
302 	return (struct got_commit_object *)got_object_cache_get(
303 	    &repo->commitcache, id);
304 }
305 
306 const struct got_error *
307 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
308     struct got_tag_object *tag)
309 {
310 #ifndef GOT_NO_OBJ_CACHE
311 	const struct got_error *err = NULL;
312 	err = got_object_cache_add(&repo->tagcache, id, tag);
313 	if (err) {
314 		if (err->code == GOT_ERR_OBJ_EXISTS ||
315 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
316 			err = NULL;
317 		return err;
318 	}
319 	tag->refcnt++;
320 #endif
321 	return NULL;
322 }
323 
324 struct got_tag_object *
325 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
326 {
327 	return (struct got_tag_object *)got_object_cache_get(
328 	    &repo->tagcache, id);
329 }
330 
331 const struct got_error *
332 open_repo(struct got_repository *repo, const char *path)
333 {
334 	const struct got_error *err = NULL;
335 
336 	/* bare git repository? */
337 	repo->path_git_dir = strdup(path);
338 	if (repo->path_git_dir == NULL)
339 		return got_error_from_errno("strdup");
340 	if (is_git_repo(repo)) {
341 		repo->path = strdup(repo->path_git_dir);
342 		if (repo->path == NULL) {
343 			err = got_error_from_errno("strdup");
344 			goto done;
345 		}
346 		return NULL;
347 	}
348 
349 	/* git repository with working tree? */
350 	free(repo->path_git_dir);
351 	repo->path_git_dir = NULL;
352 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
353 		err = got_error_from_errno("asprintf");
354 		goto done;
355 	}
356 	if (is_git_repo(repo)) {
357 		repo->path = strdup(path);
358 		if (repo->path == NULL) {
359 			err = got_error_from_errno("strdup");
360 			goto done;
361 		}
362 		return NULL;
363 	}
364 
365 	err = got_error(GOT_ERR_NOT_GIT_REPO);
366 done:
367 	if (err) {
368 		free(repo->path);
369 		repo->path = NULL;
370 		free(repo->path_git_dir);
371 		repo->path_git_dir = NULL;
372 	}
373 	return err;
374 }
375 
376 static const struct got_error *
377 parse_gitconfig_file(int *gitconfig_repository_format_version,
378     char **gitconfig_author_name, char **gitconfig_author_email,
379     struct got_remote_repo **remotes, int *nremotes,
380     char **gitconfig_owner, char ***extensions, int *nextensions,
381     const char *gitconfig_path)
382 {
383 	const struct got_error *err = NULL, *child_err = NULL;
384 	int fd = -1;
385 	int imsg_fds[2] = { -1, -1 };
386 	pid_t pid;
387 	struct imsgbuf *ibuf;
388 
389 	*gitconfig_repository_format_version = 0;
390 	if (extensions)
391 		*extensions = NULL;
392 	if (nextensions)
393 		*nextensions = 0;
394 	*gitconfig_author_name = NULL;
395 	*gitconfig_author_email = NULL;
396 	if (remotes)
397 		*remotes = NULL;
398 	if (nremotes)
399 		*nremotes = 0;
400 	if (gitconfig_owner)
401 		*gitconfig_owner = NULL;
402 
403 	fd = open(gitconfig_path, O_RDONLY);
404 	if (fd == -1) {
405 		if (errno == ENOENT)
406 			return NULL;
407 		return got_error_from_errno2("open", gitconfig_path);
408 	}
409 
410 	ibuf = calloc(1, sizeof(*ibuf));
411 	if (ibuf == NULL) {
412 		err = got_error_from_errno("calloc");
413 		goto done;
414 	}
415 
416 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
417 		err = got_error_from_errno("socketpair");
418 		goto done;
419 	}
420 
421 	pid = fork();
422 	if (pid == -1) {
423 		err = got_error_from_errno("fork");
424 		goto done;
425 	} else if (pid == 0) {
426 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
427 		    gitconfig_path);
428 		/* not reached */
429 	}
430 
431 	if (close(imsg_fds[1]) == -1) {
432 		err = got_error_from_errno("close");
433 		goto done;
434 	}
435 	imsg_fds[1] = -1;
436 	imsg_init(ibuf, imsg_fds[0]);
437 
438 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
439 	if (err)
440 		goto done;
441 	fd = -1;
442 
443 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
444 	if (err)
445 		goto done;
446 
447 	err = got_privsep_recv_gitconfig_int(
448 	    gitconfig_repository_format_version, ibuf);
449 	if (err)
450 		goto done;
451 
452 	if (extensions && nextensions) {
453 		err = got_privsep_send_gitconfig_repository_extensions_req(
454 		    ibuf);
455 		if (err)
456 			goto done;
457 		err = got_privsep_recv_gitconfig_int(nextensions, ibuf);
458 		if (err)
459 			goto done;
460 		if (*nextensions > 0) {
461 			int i;
462 			*extensions = calloc(*nextensions, sizeof(char *));
463 			if (*extensions == NULL) {
464 				err = got_error_from_errno("calloc");
465 				goto done;
466 			}
467 			for (i = 0; i < *nextensions; i++) {
468 				char *ext;
469 				err = got_privsep_recv_gitconfig_str(&ext,
470 				    ibuf);
471 				if (err)
472 					goto done;
473 				(*extensions)[i] = ext;
474 			}
475 		}
476 	}
477 
478 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
479 	if (err)
480 		goto done;
481 
482 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
483 	if (err)
484 		goto done;
485 
486 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
487 	if (err)
488 		goto done;
489 
490 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
491 	if (err)
492 		goto done;
493 
494 	if (remotes && nremotes) {
495 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
496 		if (err)
497 			goto done;
498 
499 		err = got_privsep_recv_gitconfig_remotes(remotes,
500 		    nremotes, ibuf);
501 		if (err)
502 			goto done;
503 	}
504 
505 	if (gitconfig_owner) {
506 		err = got_privsep_send_gitconfig_owner_req(ibuf);
507 		if (err)
508 			goto done;
509 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
510 		if (err)
511 			goto done;
512 	}
513 
514 	imsg_clear(ibuf);
515 	err = got_privsep_send_stop(imsg_fds[0]);
516 	child_err = got_privsep_wait_for_child(pid);
517 	if (child_err && err == NULL)
518 		err = child_err;
519 done:
520 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
521 		err = got_error_from_errno("close");
522 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
523 		err = got_error_from_errno("close");
524 	if (fd != -1 && close(fd) == -1 && err == NULL)
525 		err = got_error_from_errno2("close", gitconfig_path);
526 	free(ibuf);
527 	return err;
528 }
529 
530 static const struct got_error *
531 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
532 {
533 	const struct got_error *err = NULL;
534 	char *repo_gitconfig_path = NULL;
535 
536 	if (global_gitconfig_path) {
537 		/* Read settings from ~/.gitconfig. */
538 		int dummy_repo_version;
539 		err = parse_gitconfig_file(&dummy_repo_version,
540 		    &repo->global_gitconfig_author_name,
541 		    &repo->global_gitconfig_author_email,
542 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
543 		if (err)
544 			return err;
545 	}
546 
547 	/* Read repository's .git/config file. */
548 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
549 	if (repo_gitconfig_path == NULL)
550 		return got_error_from_errno("got_repo_get_path_gitconfig");
551 
552 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
553 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
554 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
555 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
556 	    repo_gitconfig_path);
557 	if (err)
558 		goto done;
559 done:
560 	free(repo_gitconfig_path);
561 	return err;
562 }
563 
564 static const struct got_error *
565 read_gotconfig(struct got_repository *repo)
566 {
567 	const struct got_error *err = NULL;
568 	char *gotconfig_path;
569 
570 	gotconfig_path = got_repo_get_path_gotconfig(repo);
571 	if (gotconfig_path == NULL)
572 		return got_error_from_errno("got_repo_get_path_gotconfig");
573 
574 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
575 	free(gotconfig_path);
576 	return err;
577 }
578 
579 /* Supported repository format extensions. */
580 static const char *repo_extensions[] = {
581 	"noop",			/* Got supports repository format version 1. */
582 	"preciousObjects",	/* Got has no garbage collection yet. */
583 	"worktreeConfig",	/* Got does not care about Git work trees. */
584 };
585 
586 const struct got_error *
587 got_repo_open(struct got_repository **repop, const char *path,
588     const char *global_gitconfig_path)
589 {
590 	struct got_repository *repo = NULL;
591 	const struct got_error *err = NULL;
592 	char *abspath, *repo_path = NULL;
593 	size_t i;
594 
595 	*repop = NULL;
596 
597 	if (got_path_is_absolute(path))
598 		abspath = strdup(path);
599 	else
600 		abspath = got_path_get_absolute(path);
601 	if (abspath == NULL)
602 		return got_error_path(path, GOT_ERR_BAD_PATH);
603 
604 	repo = calloc(1, sizeof(*repo));
605 	if (repo == NULL) {
606 		err = got_error_from_errno("calloc");
607 		goto done;
608 	}
609 
610 	for (i = 0; i < nitems(repo->privsep_children); i++) {
611 		memset(&repo->privsep_children[i], 0,
612 		    sizeof(repo->privsep_children[0]));
613 		repo->privsep_children[i].imsg_fd = -1;
614 	}
615 
616 	err = got_object_cache_init(&repo->objcache,
617 	    GOT_OBJECT_CACHE_TYPE_OBJ);
618 	if (err)
619 		goto done;
620 	err = got_object_cache_init(&repo->treecache,
621 	    GOT_OBJECT_CACHE_TYPE_TREE);
622 	if (err)
623 		goto done;
624 	err = got_object_cache_init(&repo->commitcache,
625 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
626 	if (err)
627 		goto done;
628 	err = got_object_cache_init(&repo->tagcache,
629 	    GOT_OBJECT_CACHE_TYPE_TAG);
630 	if (err)
631 		goto done;
632 
633 	repo_path = realpath(abspath, NULL);
634 	if (repo_path == NULL) {
635 		err = got_error_from_errno2("realpath", abspath);
636 		goto done;
637 	}
638 
639 	for (;;) {
640 		char *parent_path;
641 
642 		err = open_repo(repo, repo_path);
643 		if (err == NULL)
644 			break;
645 		if (err->code != GOT_ERR_NOT_GIT_REPO)
646 			goto done;
647 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
648 			err = got_error(GOT_ERR_NOT_GIT_REPO);
649 			goto done;
650 		}
651 		err = got_path_dirname(&parent_path, repo_path);
652 		if (err)
653 			goto done;
654 		free(repo_path);
655 		repo_path = parent_path;
656 	}
657 
658 	err = read_gotconfig(repo);
659 	if (err)
660 		goto done;
661 
662 	err = read_gitconfig(repo, global_gitconfig_path);
663 	if (err)
664 		goto done;
665 	if (repo->gitconfig_repository_format_version != 0)
666 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
667 	for (i = 0; i < repo->nextensions; i++) {
668 		char *ext = repo->extensions[i];
669 		int j, supported = 0;
670 		for (j = 0; j < nitems(repo_extensions); j++) {
671 			if (strcmp(ext, repo_extensions[j]) == 0) {
672 				supported = 1;
673 				break;
674 			}
675 		}
676 		if (!supported) {
677 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
678 			goto done;
679 		}
680 	}
681 done:
682 	if (err)
683 		got_repo_close(repo);
684 	else
685 		*repop = repo;
686 	free(abspath);
687 	free(repo_path);
688 	return err;
689 }
690 
691 const struct got_error *
692 got_repo_close(struct got_repository *repo)
693 {
694 	const struct got_error *err = NULL, *child_err;
695 	size_t i;
696 
697 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
698 		if (repo->packidx_cache[i] == NULL)
699 			break;
700 		got_packidx_close(repo->packidx_cache[i]);
701 	}
702 
703 	for (i = 0; i < nitems(repo->packs); i++) {
704 		if (repo->packs[i].path_packfile == NULL)
705 			break;
706 		got_pack_close(&repo->packs[i]);
707 	}
708 
709 	free(repo->path);
710 	free(repo->path_git_dir);
711 
712 	got_object_cache_close(&repo->objcache);
713 	got_object_cache_close(&repo->treecache);
714 	got_object_cache_close(&repo->commitcache);
715 	got_object_cache_close(&repo->tagcache);
716 
717 	for (i = 0; i < nitems(repo->privsep_children); i++) {
718 		if (repo->privsep_children[i].imsg_fd == -1)
719 			continue;
720 		imsg_clear(repo->privsep_children[i].ibuf);
721 		free(repo->privsep_children[i].ibuf);
722 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
723 		child_err = got_privsep_wait_for_child(
724 		    repo->privsep_children[i].pid);
725 		if (child_err && err == NULL)
726 			err = child_err;
727 		if (close(repo->privsep_children[i].imsg_fd) != 0 &&
728 		    err == NULL)
729 			err = got_error_from_errno("close");
730 	}
731 
732 	if (repo->gotconfig)
733 		got_gotconfig_free(repo->gotconfig);
734 	free(repo->gitconfig_author_name);
735 	free(repo->gitconfig_author_email);
736 	for (i = 0; i < repo->ngitconfig_remotes; i++)
737 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
738 	free(repo->gitconfig_remotes);
739 	for (i = 0; i < repo->nextensions; i++)
740 		free(repo->extensions[i]);
741 	free(repo->extensions);
742 	free(repo);
743 
744 	return err;
745 }
746 
747 void
748 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
749 {
750 	int i;
751 
752 	free(repo->name);
753 	repo->name = NULL;
754 	free(repo->url);
755 	repo->url = NULL;
756 	for (i = 0; i < repo->nbranches; i++)
757 		free(repo->branches[i]);
758 	free(repo->branches);
759 	repo->branches = NULL;
760 	repo->nbranches = 0;
761 }
762 
763 const struct got_error *
764 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
765     const char *input_path)
766 {
767 	const struct got_error *err = NULL;
768 	const char *repo_abspath = NULL;
769 	size_t repolen, len;
770 	char *canonpath, *path = NULL;
771 
772 	*in_repo_path = NULL;
773 
774 	canonpath = strdup(input_path);
775 	if (canonpath == NULL) {
776 		err = got_error_from_errno("strdup");
777 		goto done;
778 	}
779 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
780 	if (err)
781 		goto done;
782 
783 	repo_abspath = got_repo_get_path(repo);
784 
785 	if (canonpath[0] == '\0') {
786 		path = strdup(canonpath);
787 		if (path == NULL) {
788 			err = got_error_from_errno("strdup");
789 			goto done;
790 		}
791 	} else {
792 		path = realpath(canonpath, NULL);
793 		if (path == NULL) {
794 			if (errno != ENOENT) {
795 				err = got_error_from_errno2("realpath",
796 				    canonpath);
797 				goto done;
798 			}
799 			/*
800 			 * Path is not on disk.
801 			 * Assume it is already relative to repository root.
802 			 */
803 			path = strdup(canonpath);
804 			if (path == NULL) {
805 				err = got_error_from_errno("strdup");
806 				goto done;
807 			}
808 		}
809 
810 		repolen = strlen(repo_abspath);
811 		len = strlen(path);
812 
813 
814 		if (strcmp(path, repo_abspath) == 0) {
815 			free(path);
816 			path = strdup("");
817 			if (path == NULL) {
818 				err = got_error_from_errno("strdup");
819 				goto done;
820 			}
821 		} else if (len > repolen &&
822 		    got_path_is_child(path, repo_abspath, repolen)) {
823 			/* Matched an on-disk path inside repository. */
824 			if (got_repo_is_bare(repo)) {
825 				/*
826 				 * Matched an on-disk path inside repository
827 				 * database. Treat input as repository-relative.
828 				 */
829 				 free(path);
830 				 path = canonpath;
831 				 canonpath = NULL;
832 			} else {
833 				char *child;
834 				/* Strip common prefix with repository path. */
835 				err = got_path_skip_common_ancestor(&child,
836 				    repo_abspath, path);
837 				if (err)
838 					goto done;
839 				free(path);
840 				path = child;
841 			}
842 		} else {
843 			/*
844 			 * Matched unrelated on-disk path.
845 			 * Treat input as repository-relative.
846 			 */
847 			 free(path);
848 			 path = canonpath;
849 			 canonpath = NULL;
850 		}
851 	}
852 
853 	/* Make in-repository path absolute */
854 	if (path[0] != '/') {
855 		char *abspath;
856 		if (asprintf(&abspath, "/%s", path) == -1) {
857 			err = got_error_from_errno("asprintf");
858 			goto done;
859 		}
860 		free(path);
861 		path = abspath;
862 	}
863 
864 done:
865 	free(canonpath);
866 	if (err)
867 		free(path);
868 	else
869 		*in_repo_path = path;
870 	return err;
871 }
872 
873 static const struct got_error *
874 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
875     const char *path_packidx)
876 {
877 	const struct got_error *err = NULL;
878 	size_t i;
879 
880 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
881 		if (repo->packidx_cache[i] == NULL)
882 			break;
883 		if (strcmp(repo->packidx_cache[i]->path_packidx,
884 		    path_packidx) == 0) {
885 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
886 		}
887 	}
888 	if (i == nitems(repo->packidx_cache)) {
889 		err = got_packidx_close(repo->packidx_cache[i - 1]);
890 		if (err)
891 			return err;
892 	}
893 
894 	/*
895 	 * Insert the new pack index at the front so it will
896 	 * be searched first in the future.
897 	 */
898 	memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
899 	    sizeof(repo->packidx_cache) -
900 	    sizeof(repo->packidx_cache[0]));
901 	repo->packidx_cache[0] = packidx;
902 
903 	return NULL;
904 }
905 
906 static int
907 is_packidx_filename(const char *name, size_t len)
908 {
909 	if (len != GOT_PACKIDX_NAMELEN)
910 		return 0;
911 
912 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
913 		return 0;
914 
915 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
916 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
917 		return 0;
918 
919 	return 1;
920 }
921 
922 const struct got_error *
923 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
924     struct got_repository *repo, struct got_object_id *id)
925 {
926 	const struct got_error *err;
927 	char *path_packdir;
928 	DIR *packdir;
929 	struct dirent *dent;
930 	char *path_packidx;
931 	size_t i;
932 
933 	/* Search pack index cache. */
934 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
935 		if (repo->packidx_cache[i] == NULL)
936 			break;
937 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
938 		if (*idx != -1) {
939 			*packidx = repo->packidx_cache[i];
940 			/*
941 			 * Move this cache entry to the front. Repeatedly
942 			 * searching a wrong pack index can be expensive.
943 			 */
944 			if (i > 0) {
945 				struct got_packidx *p;
946 				p = repo->packidx_cache[0];
947 				repo->packidx_cache[0] = *packidx;
948 				repo->packidx_cache[i] = p;
949 			}
950 			return NULL;
951 		}
952 	}
953 	/* No luck. Search the filesystem. */
954 
955 	path_packdir = got_repo_get_path_objects_pack(repo);
956 	if (path_packdir == NULL)
957 		return got_error_from_errno("got_repo_get_path_objects_pack");
958 
959 	packdir = opendir(path_packdir);
960 	if (packdir == NULL) {
961 		if (errno == ENOENT)
962 			err = got_error_no_obj(id);
963 		else
964 			err = got_error_from_errno2("opendir", path_packdir);
965 		goto done;
966 	}
967 
968 	while ((dent = readdir(packdir)) != NULL) {
969 		int is_cached = 0;
970 
971 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
972 			continue;
973 
974 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
975 		    dent->d_name) == -1) {
976 			err = got_error_from_errno("asprintf");
977 			goto done;
978 		}
979 
980 		for (i = 0; i < nitems(repo->packidx_cache); i++) {
981 			if (repo->packidx_cache[i] == NULL)
982 				break;
983 			if (strcmp(repo->packidx_cache[i]->path_packidx,
984 			    path_packidx) == 0) {
985 				is_cached = 1;
986 				break;
987 			}
988 		}
989 		if (is_cached) {
990 			free(path_packidx);
991 			continue; /* already searched */
992 		}
993 
994 		err = got_packidx_open(packidx, path_packidx, 0);
995 		if (err) {
996 			free(path_packidx);
997 			goto done;
998 		}
999 
1000 		err = cache_packidx(repo, *packidx, path_packidx);
1001 		free(path_packidx);
1002 		if (err)
1003 			goto done;
1004 
1005 		*idx = got_packidx_get_object_idx(*packidx, id);
1006 		if (*idx != -1) {
1007 			err = NULL; /* found the object */
1008 			goto done;
1009 		}
1010 	}
1011 
1012 	err = got_error_no_obj(id);
1013 done:
1014 	free(path_packdir);
1015 	if (packdir && closedir(packdir) != 0 && err == NULL)
1016 		err = got_error_from_errno("closedir");
1017 	return err;
1018 }
1019 
1020 static const struct got_error *
1021 read_packfile_hdr(int fd, struct got_packidx *packidx)
1022 {
1023 	const struct got_error *err = NULL;
1024 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1025 	struct got_packfile_hdr hdr;
1026 	ssize_t n;
1027 
1028 	n = read(fd, &hdr, sizeof(hdr));
1029 	if (n < 0)
1030 		return got_error_from_errno("read");
1031 	if (n != sizeof(hdr))
1032 		return got_error(GOT_ERR_BAD_PACKFILE);
1033 
1034 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1035 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1036 	    be32toh(hdr.nobjects) != totobj)
1037 		err = got_error(GOT_ERR_BAD_PACKFILE);
1038 
1039 	return err;
1040 }
1041 
1042 static const struct got_error *
1043 open_packfile(int *fd, const char *path_packfile, struct got_packidx *packidx)
1044 {
1045 	const struct got_error *err = NULL;
1046 
1047 	*fd = open(path_packfile, O_RDONLY | O_NOFOLLOW);
1048 	if (*fd == -1)
1049 		return got_error_from_errno2("open", path_packfile);
1050 
1051 	if (packidx) {
1052 		err = read_packfile_hdr(*fd, packidx);
1053 		if (err) {
1054 			close(*fd);
1055 			*fd = -1;
1056 		}
1057 	}
1058 
1059 	return err;
1060 }
1061 
1062 const struct got_error *
1063 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1064     const char *path_packfile, struct got_packidx *packidx)
1065 {
1066 	const struct got_error *err = NULL;
1067 	struct got_pack *pack = NULL;
1068 	struct stat sb;
1069 	size_t i;
1070 
1071 	if (packp)
1072 		*packp = NULL;
1073 
1074 	for (i = 0; i < nitems(repo->packs); i++) {
1075 		pack = &repo->packs[i];
1076 		if (pack->path_packfile == NULL)
1077 			break;
1078 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1079 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1080 	}
1081 
1082 	if (i == nitems(repo->packs) - 1) {
1083 		err = got_pack_close(&repo->packs[i - 1]);
1084 		if (err)
1085 			return err;
1086 		memmove(&repo->packs[1], &repo->packs[0],
1087 		    sizeof(repo->packs) - sizeof(repo->packs[0]));
1088 		i = 0;
1089 	}
1090 
1091 	pack = &repo->packs[i];
1092 
1093 	pack->path_packfile = strdup(path_packfile);
1094 	if (pack->path_packfile == NULL) {
1095 		err = got_error_from_errno("strdup");
1096 		goto done;
1097 	}
1098 
1099 	err = open_packfile(&pack->fd, path_packfile, packidx);
1100 	if (err)
1101 		goto done;
1102 
1103 	if (fstat(pack->fd, &sb) != 0) {
1104 		err = got_error_from_errno("fstat");
1105 		goto done;
1106 	}
1107 	pack->filesize = sb.st_size;
1108 
1109 	pack->privsep_child = NULL;
1110 
1111 #ifndef GOT_PACK_NO_MMAP
1112 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1113 	    pack->fd, 0);
1114 	if (pack->map == MAP_FAILED) {
1115 		if (errno != ENOMEM) {
1116 			err = got_error_from_errno("mmap");
1117 			goto done;
1118 		}
1119 		pack->map = NULL; /* fall back to read(2) */
1120 	}
1121 #endif
1122 done:
1123 	if (err) {
1124 		if (pack) {
1125 			free(pack->path_packfile);
1126 			memset(pack, 0, sizeof(*pack));
1127 		}
1128 	} else if (packp)
1129 		*packp = pack;
1130 	return err;
1131 }
1132 
1133 struct got_pack *
1134 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1135 {
1136 	struct got_pack *pack = NULL;
1137 	size_t i;
1138 
1139 	for (i = 0; i < nitems(repo->packs); i++) {
1140 		pack = &repo->packs[i];
1141 		if (pack->path_packfile == NULL)
1142 			break;
1143 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1144 			return pack;
1145 	}
1146 
1147 	return NULL;
1148 }
1149 
1150 const struct got_error *
1151 got_repo_init(const char *repo_path)
1152 {
1153 	const struct got_error *err = NULL;
1154 	const char *dirnames[] = {
1155 		GOT_OBJECTS_DIR,
1156 		GOT_OBJECTS_PACK_DIR,
1157 		GOT_REFS_DIR,
1158 	};
1159 	const char *description_str = "Unnamed repository; "
1160 	    "edit this file 'description' to name the repository.";
1161 	const char *headref_str = "ref: refs/heads/main";
1162 	const char *gitconfig_str = "[core]\n"
1163 	    "\trepositoryformatversion = 0\n"
1164 	    "\tfilemode = true\n"
1165 	    "\tbare = true\n";
1166 	char *path;
1167 	size_t i;
1168 
1169 	if (!got_path_dir_is_empty(repo_path))
1170 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1171 
1172 	for (i = 0; i < nitems(dirnames); i++) {
1173 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1174 			return got_error_from_errno("asprintf");
1175 		}
1176 		err = got_path_mkdir(path);
1177 		free(path);
1178 		if (err)
1179 			return err;
1180 	}
1181 
1182 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1183 		return got_error_from_errno("asprintf");
1184 	err = got_path_create_file(path, description_str);
1185 	free(path);
1186 	if (err)
1187 		return err;
1188 
1189 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1190 		return got_error_from_errno("asprintf");
1191 	err = got_path_create_file(path, headref_str);
1192 	free(path);
1193 	if (err)
1194 		return err;
1195 
1196 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1197 		return got_error_from_errno("asprintf");
1198 	err = got_path_create_file(path, gitconfig_str);
1199 	free(path);
1200 	if (err)
1201 		return err;
1202 
1203 	return NULL;
1204 }
1205 
1206 static const struct got_error *
1207 match_packed_object(struct got_object_id **unique_id,
1208     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1209 {
1210 	const struct got_error *err = NULL;
1211 	char *path_packdir;
1212 	DIR *packdir;
1213 	struct dirent *dent;
1214 	char *path_packidx;
1215 	struct got_object_id_queue matched_ids;
1216 
1217 	SIMPLEQ_INIT(&matched_ids);
1218 
1219 	path_packdir = got_repo_get_path_objects_pack(repo);
1220 	if (path_packdir == NULL)
1221 		return got_error_from_errno("got_repo_get_path_objects_pack");
1222 
1223 	packdir = opendir(path_packdir);
1224 	if (packdir == NULL) {
1225 		if (errno != ENOENT)
1226 			err = got_error_from_errno2("opendir", path_packdir);
1227 		goto done;
1228 	}
1229 
1230 	while ((dent = readdir(packdir)) != NULL) {
1231 		struct got_packidx *packidx;
1232 		struct got_object_qid *qid;
1233 
1234 
1235 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
1236 			continue;
1237 
1238 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
1239 		    dent->d_name) == -1) {
1240 			err = got_error_from_errno("asprintf");
1241 			break;
1242 		}
1243 
1244 		err = got_packidx_open(&packidx, path_packidx, 0);
1245 		free(path_packidx);
1246 		if (err)
1247 			break;
1248 
1249 		err = got_packidx_match_id_str_prefix(&matched_ids,
1250 		    packidx, id_str_prefix);
1251 		if (err) {
1252 			got_packidx_close(packidx);
1253 			break;
1254 		}
1255 		err = got_packidx_close(packidx);
1256 		if (err)
1257 			break;
1258 
1259 		SIMPLEQ_FOREACH(qid, &matched_ids, entry) {
1260 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1261 				int matched_type;
1262 				err = got_object_get_type(&matched_type, repo,
1263 				    qid->id);
1264 				if (err)
1265 					goto done;
1266 				if (matched_type != obj_type)
1267 					continue;
1268 			}
1269 			if (*unique_id == NULL) {
1270 				*unique_id = got_object_id_dup(qid->id);
1271 				if (*unique_id == NULL) {
1272 					err = got_error_from_errno("malloc");
1273 					goto done;
1274 				}
1275 			} else {
1276 				if (got_object_id_cmp(*unique_id, qid->id) == 0)
1277 					continue; /* packed multiple times */
1278 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1279 				goto done;
1280 			}
1281 		}
1282 	}
1283 done:
1284 	got_object_id_queue_free(&matched_ids);
1285 	free(path_packdir);
1286 	if (packdir && closedir(packdir) != 0 && err == NULL)
1287 		err = got_error_from_errno("closedir");
1288 	if (err) {
1289 		free(*unique_id);
1290 		*unique_id = NULL;
1291 	}
1292 	return err;
1293 }
1294 
1295 static const struct got_error *
1296 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1297     const char *object_dir, const char *id_str_prefix, int obj_type,
1298     struct got_repository *repo)
1299 {
1300 	const struct got_error *err = NULL;
1301 	char *path;
1302 	DIR *dir = NULL;
1303 	struct dirent *dent;
1304 	struct got_object_id id;
1305 
1306 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1307 		err = got_error_from_errno("asprintf");
1308 		goto done;
1309 	}
1310 
1311 	dir = opendir(path);
1312 	if (dir == NULL) {
1313 		if (errno == ENOENT) {
1314 			err = NULL;
1315 			goto done;
1316 		}
1317 		err = got_error_from_errno2("opendir", path);
1318 		goto done;
1319 	}
1320 	while ((dent = readdir(dir)) != NULL) {
1321 		char *id_str;
1322 		int cmp;
1323 
1324 		if (strcmp(dent->d_name, ".") == 0 ||
1325 		    strcmp(dent->d_name, "..") == 0)
1326 			continue;
1327 
1328 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1329 			err = got_error_from_errno("asprintf");
1330 			goto done;
1331 		}
1332 
1333 		if (!got_parse_sha1_digest(id.sha1, id_str))
1334 			continue;
1335 
1336 		/*
1337 		 * Directory entries do not necessarily appear in
1338 		 * sorted order, so we must iterate over all of them.
1339 		 */
1340 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1341 		if (cmp != 0) {
1342 			free(id_str);
1343 			continue;
1344 		}
1345 
1346 		if (*unique_id == NULL) {
1347 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1348 				int matched_type;
1349 				err = got_object_get_type(&matched_type, repo,
1350 				    &id);
1351 				if (err)
1352 					goto done;
1353 				if (matched_type != obj_type)
1354 					continue;
1355 			}
1356 			*unique_id = got_object_id_dup(&id);
1357 			if (*unique_id == NULL) {
1358 				err = got_error_from_errno("got_object_id_dup");
1359 				free(id_str);
1360 				goto done;
1361 			}
1362 		} else {
1363 			if (got_object_id_cmp(*unique_id, &id) == 0)
1364 				continue; /* both packed and loose */
1365 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1366 			free(id_str);
1367 			goto done;
1368 		}
1369 	}
1370 done:
1371 	if (dir && closedir(dir) != 0 && err == NULL)
1372 		err = got_error_from_errno("closedir");
1373 	if (err) {
1374 		free(*unique_id);
1375 		*unique_id = NULL;
1376 	}
1377 	free(path);
1378 	return err;
1379 }
1380 
1381 const struct got_error *
1382 got_repo_match_object_id_prefix(struct got_object_id **id,
1383     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1384 {
1385 	const struct got_error *err = NULL;
1386 	char *path_objects = got_repo_get_path_objects(repo);
1387 	char *object_dir = NULL;
1388 	size_t len;
1389 	int i;
1390 
1391 	*id = NULL;
1392 
1393 	for (i = 0; i < strlen(id_str_prefix); i++) {
1394 		if (isxdigit((unsigned char)id_str_prefix[i]))
1395 			continue;
1396 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1397 	}
1398 
1399 	len = strlen(id_str_prefix);
1400 	if (len >= 2) {
1401 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1402 		if (err)
1403 			goto done;
1404 		object_dir = strndup(id_str_prefix, 2);
1405 		if (object_dir == NULL) {
1406 			err = got_error_from_errno("strdup");
1407 			goto done;
1408 		}
1409 		err = match_loose_object(id, path_objects, object_dir,
1410 		    id_str_prefix, obj_type, repo);
1411 	} else if (len == 1) {
1412 		int i;
1413 		for (i = 0; i < 0xf; i++) {
1414 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1415 			    == -1) {
1416 				err = got_error_from_errno("asprintf");
1417 				goto done;
1418 			}
1419 			err = match_packed_object(id, repo, object_dir,
1420 			    obj_type);
1421 			if (err)
1422 				goto done;
1423 			err = match_loose_object(id, path_objects, object_dir,
1424 			    id_str_prefix, obj_type, repo);
1425 			if (err)
1426 				goto done;
1427 		}
1428 	} else {
1429 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1430 		goto done;
1431 	}
1432 done:
1433 	free(object_dir);
1434 	if (err) {
1435 		free(*id);
1436 		*id = NULL;
1437 	} else if (*id == NULL)
1438 		err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1439 
1440 	return err;
1441 }
1442 
1443 const struct got_error *
1444 got_repo_match_object_id(struct got_object_id **id, char **label,
1445     const char *id_str, int obj_type, int resolve_tags,
1446     struct got_repository *repo)
1447 {
1448 	const struct got_error *err;
1449 	struct got_tag_object *tag;
1450 	struct got_reference *ref = NULL;
1451 
1452 	*id = NULL;
1453 	if (label)
1454 		*label = NULL;
1455 
1456 	if (resolve_tags) {
1457 		err = got_repo_object_match_tag(&tag, id_str, GOT_OBJ_TYPE_ANY,
1458 		    repo);
1459 		if (err == NULL) {
1460 			*id = got_object_id_dup(
1461 			    got_object_tag_get_object_id(tag));
1462 			if (*id == NULL)
1463 				err = got_error_from_errno("got_object_id_dup");
1464 			else if (label && asprintf(label, "refs/tags/%s",
1465 			    got_object_tag_get_name(tag)) == -1) {
1466 				err = got_error_from_errno("asprintf");
1467 				free(*id);
1468 				*id = NULL;
1469 			}
1470 			got_object_tag_close(tag);
1471 			return err;
1472 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1473 		    err->code != GOT_ERR_NO_OBJ)
1474 			return err;
1475 	}
1476 
1477 	err = got_repo_match_object_id_prefix(id, id_str, obj_type, repo);
1478 	if (err) {
1479 		if (err->code != GOT_ERR_BAD_OBJ_ID_STR)
1480 			return err;
1481 		err = got_ref_open(&ref, repo, id_str, 0);
1482 		if (err != NULL)
1483 			goto done;
1484 		if (label) {
1485 			*label = strdup(got_ref_get_name(ref));
1486 			if (*label == NULL) {
1487 				err = got_error_from_errno("strdup");
1488 				goto done;
1489 			}
1490 		}
1491 		err = got_ref_resolve(id, repo, ref);
1492 	} else if (label) {
1493 		err = got_object_id_str(label, *id);
1494 		if (*label == NULL) {
1495 			err = got_error_from_errno("strdup");
1496 			goto done;
1497 		}
1498 	}
1499 done:
1500 	if (ref)
1501 		got_ref_close(ref);
1502 	return err;
1503 }
1504 
1505 const struct got_error *
1506 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1507     int obj_type, struct got_repository *repo)
1508 {
1509 	const struct got_error *err;
1510 	struct got_reflist_head refs;
1511 	struct got_reflist_entry *re;
1512 	struct got_object_id *tag_id;
1513 
1514 	SIMPLEQ_INIT(&refs);
1515 	*tag = NULL;
1516 
1517 	err = got_ref_list(&refs, repo, "refs/tags", got_ref_cmp_by_name, NULL);
1518 	if (err)
1519 		return err;
1520 
1521 	SIMPLEQ_FOREACH(re, &refs, entry) {
1522 		const char *refname;
1523 		refname = got_ref_get_name(re->ref);
1524 		if (got_ref_is_symbolic(re->ref))
1525 			continue;
1526 		refname += strlen("refs/tags/");
1527 		if (strcmp(refname, name) != 0)
1528 			continue;
1529 		err = got_ref_resolve(&tag_id, repo, re->ref);
1530 		if (err)
1531 			break;
1532 		err = got_object_open_as_tag(tag, repo, tag_id);
1533 		free(tag_id);
1534 		if (err)
1535 			break;
1536 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1537 		    got_object_tag_get_object_type(*tag) == obj_type)
1538 			break;
1539 		got_object_tag_close(*tag);
1540 		*tag = NULL;
1541 	}
1542 
1543 	got_ref_list_free(&refs);
1544 	if (err == NULL && *tag == NULL)
1545 		err = got_error_path(name, GOT_ERR_NO_OBJ);
1546 	return err;
1547 }
1548 
1549 static const struct got_error *
1550 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1551     const char *name, mode_t mode, struct got_object_id *blob_id)
1552 {
1553 	const struct got_error *err = NULL;
1554 
1555 	 *new_te = NULL;
1556 
1557 	*new_te = calloc(1, sizeof(**new_te));
1558 	if (*new_te == NULL)
1559 		return got_error_from_errno("calloc");
1560 
1561 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1562 	    sizeof((*new_te)->name)) {
1563 		err = got_error(GOT_ERR_NO_SPACE);
1564 		goto done;
1565 	}
1566 
1567 	if (S_ISLNK(mode)) {
1568 		(*new_te)->mode = S_IFLNK;
1569 	} else {
1570 		(*new_te)->mode = S_IFREG;
1571 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1572 	}
1573 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1574 done:
1575 	if (err && *new_te) {
1576 		free(*new_te);
1577 		*new_te = NULL;
1578 	}
1579 	return err;
1580 }
1581 
1582 static const struct got_error *
1583 import_file(struct got_tree_entry **new_te, struct dirent *de,
1584     const char *path, struct got_repository *repo)
1585 {
1586 	const struct got_error *err;
1587 	struct got_object_id *blob_id = NULL;
1588 	char *filepath;
1589 	struct stat sb;
1590 
1591 	if (asprintf(&filepath, "%s%s%s", path,
1592 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1593 		return got_error_from_errno("asprintf");
1594 
1595 	if (lstat(filepath, &sb) != 0) {
1596 		err = got_error_from_errno2("lstat", path);
1597 		goto done;
1598 	}
1599 
1600 	err = got_object_blob_create(&blob_id, filepath, repo);
1601 	if (err)
1602 		goto done;
1603 
1604 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1605 	    blob_id);
1606 done:
1607 	free(filepath);
1608 	if (err)
1609 		free(blob_id);
1610 	return err;
1611 }
1612 
1613 static const struct got_error *
1614 insert_tree_entry(struct got_tree_entry *new_te,
1615     struct got_pathlist_head *paths)
1616 {
1617 	const struct got_error *err = NULL;
1618 	struct got_pathlist_entry *new_pe;
1619 
1620 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1621 	if (err)
1622 		return err;
1623 	if (new_pe == NULL)
1624 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1625 	return NULL;
1626 }
1627 
1628 static const struct got_error *write_tree(struct got_object_id **,
1629     const char *, struct got_pathlist_head *, struct got_repository *,
1630     got_repo_import_cb progress_cb, void *progress_arg);
1631 
1632 static const struct got_error *
1633 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1634     const char *path, struct got_pathlist_head *ignores,
1635     struct got_repository *repo,
1636     got_repo_import_cb progress_cb, void *progress_arg)
1637 {
1638 	const struct got_error *err;
1639 	struct got_object_id *id = NULL;
1640 	char *subdirpath;
1641 
1642 	if (asprintf(&subdirpath, "%s%s%s", path,
1643 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1644 		return got_error_from_errno("asprintf");
1645 
1646 	(*new_te) = calloc(1, sizeof(**new_te));
1647 	if (*new_te == NULL)
1648 		return got_error_from_errno("calloc");
1649 	(*new_te)->mode = S_IFDIR;
1650 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
1651 	    sizeof((*new_te)->name)) {
1652 		err = got_error(GOT_ERR_NO_SPACE);
1653 		goto done;
1654 	}
1655 	err = write_tree(&id, subdirpath, ignores,  repo,
1656 	    progress_cb, progress_arg);
1657 	if (err)
1658 		goto done;
1659 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
1660 	
1661 done:
1662 	free(id);
1663 	free(subdirpath);
1664 	if (err) {
1665 		free(*new_te);
1666 		*new_te = NULL;
1667 	}
1668 	return err;
1669 }
1670 
1671 static const struct got_error *
1672 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1673     struct got_pathlist_head *ignores, struct got_repository *repo,
1674     got_repo_import_cb progress_cb, void *progress_arg)
1675 {
1676 	const struct got_error *err = NULL;
1677 	DIR *dir;
1678 	struct dirent *de;
1679 	int nentries;
1680 	struct got_tree_entry *new_te = NULL;
1681 	struct got_pathlist_head paths;
1682 	struct got_pathlist_entry *pe;
1683 
1684 	*new_tree_id = NULL;
1685 
1686 	TAILQ_INIT(&paths);
1687 
1688 	dir = opendir(path_dir);
1689 	if (dir == NULL) {
1690 		err = got_error_from_errno2("opendir", path_dir);
1691 		goto done;
1692 	}
1693 
1694 	nentries = 0;
1695 	while ((de = readdir(dir)) != NULL) {
1696 		int ignore = 0;
1697 		int type;
1698 
1699 		if (strcmp(de->d_name, ".") == 0 ||
1700 		    strcmp(de->d_name, "..") == 0)
1701 			continue;
1702 
1703 		TAILQ_FOREACH(pe, ignores, entry) {
1704 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
1705 				ignore = 1;
1706 				break;
1707 			}
1708 		}
1709 		if (ignore)
1710 			continue;
1711 
1712 		err = got_path_dirent_type(&type, path_dir, de);
1713 		if (err)
1714 			goto done;
1715 
1716 		if (type == DT_DIR) {
1717 			err = import_subdir(&new_te, de, path_dir,
1718 			    ignores, repo, progress_cb, progress_arg);
1719 			if (err) {
1720 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
1721 					goto done;
1722 				err = NULL;
1723 				continue;
1724 			}
1725 		} else if (type == DT_REG || type == DT_LNK) {
1726 			err = import_file(&new_te, de, path_dir, repo);
1727 			if (err)
1728 				goto done;
1729 		} else
1730 			continue;
1731 
1732 		err = insert_tree_entry(new_te, &paths);
1733 		if (err)
1734 			goto done;
1735 		nentries++;
1736 	}
1737 
1738 	if (TAILQ_EMPTY(&paths)) {
1739 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
1740 		    "cannot create tree without any entries");
1741 		goto done;
1742 	}
1743 
1744 	TAILQ_FOREACH(pe, &paths, entry) {
1745 		struct got_tree_entry *te = pe->data;
1746 		char *path;
1747 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
1748 			continue;
1749 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
1750 			err = got_error_from_errno("asprintf");
1751 			goto done;
1752 		}
1753 		err = (*progress_cb)(progress_arg, path);
1754 		free(path);
1755 		if (err)
1756 			goto done;
1757 	}
1758 
1759 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
1760 done:
1761 	if (dir)
1762 		closedir(dir);
1763 	got_pathlist_free(&paths);
1764 	return err;
1765 }
1766 
1767 const struct got_error *
1768 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
1769     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
1770     struct got_repository *repo, got_repo_import_cb progress_cb,
1771     void *progress_arg)
1772 {
1773 	const struct got_error *err;
1774 	struct got_object_id *new_tree_id;
1775 
1776 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
1777 	    progress_cb, progress_arg);
1778 	if (err)
1779 		return err;
1780 
1781 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
1782 	    author, time(NULL), author, time(NULL), logmsg, repo);
1783 	free(new_tree_id);
1784 	return err;
1785 }