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 	int 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 	int 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, int check_disk)
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 (!check_disk || 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 as repository-relative.
828 				 */
829 			} else {
830 				char *child;
831 				/* Strip common prefix with repository path. */
832 				err = got_path_skip_common_ancestor(&child,
833 				    repo_abspath, path);
834 				if (err)
835 					goto done;
836 				free(path);
837 				path = child;
838 			}
839 		} else {
840 			/*
841 			 * Matched unrelated on-disk path.
842 			 * Treat it as repository-relative.
843 			 */
844 		}
845 	}
846 
847 	/* Make in-repository path absolute */
848 	if (path[0] != '/') {
849 		char *abspath;
850 		if (asprintf(&abspath, "/%s", path) == -1) {
851 			err = got_error_from_errno("asprintf");
852 			goto done;
853 		}
854 		free(path);
855 		path = abspath;
856 	}
857 
858 done:
859 	free(canonpath);
860 	if (err)
861 		free(path);
862 	else
863 		*in_repo_path = path;
864 	return err;
865 }
866 
867 static const struct got_error *
868 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
869     const char *path_packidx)
870 {
871 	const struct got_error *err = NULL;
872 	int i;
873 
874 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
875 		if (repo->packidx_cache[i] == NULL)
876 			break;
877 		if (strcmp(repo->packidx_cache[i]->path_packidx,
878 		    path_packidx) == 0) {
879 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
880 		}
881 	}
882 	if (i == nitems(repo->packidx_cache)) {
883 		err = got_packidx_close(repo->packidx_cache[i - 1]);
884 		if (err)
885 			return err;
886 	}
887 
888 	/*
889 	 * Insert the new pack index at the front so it will
890 	 * be searched first in the future.
891 	 */
892 	memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
893 	    sizeof(repo->packidx_cache) -
894 	    sizeof(repo->packidx_cache[0]));
895 	repo->packidx_cache[0] = packidx;
896 
897 	return NULL;
898 }
899 
900 static int
901 is_packidx_filename(const char *name, size_t len)
902 {
903 	if (len != GOT_PACKIDX_NAMELEN)
904 		return 0;
905 
906 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
907 		return 0;
908 
909 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
910 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
911 		return 0;
912 
913 	return 1;
914 }
915 
916 const struct got_error *
917 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
918     struct got_repository *repo, struct got_object_id *id)
919 {
920 	const struct got_error *err;
921 	char *path_packdir;
922 	DIR *packdir;
923 	struct dirent *dent;
924 	char *path_packidx;
925 	int i;
926 
927 	/* Search pack index cache. */
928 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
929 		if (repo->packidx_cache[i] == NULL)
930 			break;
931 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
932 		if (*idx != -1) {
933 			*packidx = repo->packidx_cache[i];
934 			/*
935 			 * Move this cache entry to the front. Repeatedly
936 			 * searching a wrong pack index can be expensive.
937 			 */
938 			if (i > 0) {
939 				struct got_packidx *p;
940 				p = repo->packidx_cache[0];
941 				repo->packidx_cache[0] = *packidx;
942 				repo->packidx_cache[i] = p;
943 			}
944 			return NULL;
945 		}
946 	}
947 	/* No luck. Search the filesystem. */
948 
949 	path_packdir = got_repo_get_path_objects_pack(repo);
950 	if (path_packdir == NULL)
951 		return got_error_from_errno("got_repo_get_path_objects_pack");
952 
953 	packdir = opendir(path_packdir);
954 	if (packdir == NULL) {
955 		if (errno == ENOENT)
956 			err = got_error_no_obj(id);
957 		else
958 			err = got_error_from_errno2("opendir", path_packdir);
959 		goto done;
960 	}
961 
962 	while ((dent = readdir(packdir)) != NULL) {
963 		int is_cached = 0;
964 
965 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
966 			continue;
967 
968 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
969 		    dent->d_name) == -1) {
970 			err = got_error_from_errno("asprintf");
971 			goto done;
972 		}
973 
974 		for (i = 0; i < nitems(repo->packidx_cache); i++) {
975 			if (repo->packidx_cache[i] == NULL)
976 				break;
977 			if (strcmp(repo->packidx_cache[i]->path_packidx,
978 			    path_packidx) == 0) {
979 				is_cached = 1;
980 				break;
981 			}
982 		}
983 		if (is_cached) {
984 			free(path_packidx);
985 			continue; /* already searched */
986 		}
987 
988 		err = got_packidx_open(packidx, path_packidx, 0);
989 		if (err) {
990 			free(path_packidx);
991 			goto done;
992 		}
993 
994 		err = cache_packidx(repo, *packidx, path_packidx);
995 		free(path_packidx);
996 		if (err)
997 			goto done;
998 
999 		*idx = got_packidx_get_object_idx(*packidx, id);
1000 		if (*idx != -1) {
1001 			err = NULL; /* found the object */
1002 			goto done;
1003 		}
1004 	}
1005 
1006 	err = got_error_no_obj(id);
1007 done:
1008 	free(path_packdir);
1009 	if (packdir && closedir(packdir) != 0 && err == NULL)
1010 		err = got_error_from_errno("closedir");
1011 	return err;
1012 }
1013 
1014 static const struct got_error *
1015 read_packfile_hdr(int fd, struct got_packidx *packidx)
1016 {
1017 	const struct got_error *err = NULL;
1018 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1019 	struct got_packfile_hdr hdr;
1020 	ssize_t n;
1021 
1022 	n = read(fd, &hdr, sizeof(hdr));
1023 	if (n < 0)
1024 		return got_error_from_errno("read");
1025 	if (n != sizeof(hdr))
1026 		return got_error(GOT_ERR_BAD_PACKFILE);
1027 
1028 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1029 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1030 	    be32toh(hdr.nobjects) != totobj)
1031 		err = got_error(GOT_ERR_BAD_PACKFILE);
1032 
1033 	return err;
1034 }
1035 
1036 static const struct got_error *
1037 open_packfile(int *fd, const char *path_packfile, struct got_packidx *packidx)
1038 {
1039 	const struct got_error *err = NULL;
1040 
1041 	*fd = open(path_packfile, O_RDONLY | O_NOFOLLOW);
1042 	if (*fd == -1)
1043 		return got_error_from_errno2("open", path_packfile);
1044 
1045 	if (packidx) {
1046 		err = read_packfile_hdr(*fd, packidx);
1047 		if (err) {
1048 			close(*fd);
1049 			*fd = -1;
1050 		}
1051 	}
1052 
1053 	return err;
1054 }
1055 
1056 const struct got_error *
1057 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1058     const char *path_packfile, struct got_packidx *packidx)
1059 {
1060 	const struct got_error *err = NULL;
1061 	struct got_pack *pack = NULL;
1062 	struct stat sb;
1063 	int i;
1064 
1065 	if (packp)
1066 		*packp = NULL;
1067 
1068 	for (i = 0; i < nitems(repo->packs); i++) {
1069 		pack = &repo->packs[i];
1070 		if (pack->path_packfile == NULL)
1071 			break;
1072 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1073 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1074 	}
1075 
1076 	if (i == nitems(repo->packs) - 1) {
1077 		err = got_pack_close(&repo->packs[i - 1]);
1078 		if (err)
1079 			return err;
1080 		memmove(&repo->packs[1], &repo->packs[0],
1081 		    sizeof(repo->packs) - sizeof(repo->packs[0]));
1082 		i = 0;
1083 	}
1084 
1085 	pack = &repo->packs[i];
1086 
1087 	pack->path_packfile = strdup(path_packfile);
1088 	if (pack->path_packfile == NULL) {
1089 		err = got_error_from_errno("strdup");
1090 		goto done;
1091 	}
1092 
1093 	err = open_packfile(&pack->fd, path_packfile, packidx);
1094 	if (err)
1095 		goto done;
1096 
1097 	if (fstat(pack->fd, &sb) != 0) {
1098 		err = got_error_from_errno("fstat");
1099 		goto done;
1100 	}
1101 	pack->filesize = sb.st_size;
1102 
1103 	pack->privsep_child = NULL;
1104 
1105 #ifndef GOT_PACK_NO_MMAP
1106 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1107 	    pack->fd, 0);
1108 	if (pack->map == MAP_FAILED) {
1109 		if (errno != ENOMEM) {
1110 			err = got_error_from_errno("mmap");
1111 			goto done;
1112 		}
1113 		pack->map = NULL; /* fall back to read(2) */
1114 	}
1115 #endif
1116 done:
1117 	if (err) {
1118 		if (pack) {
1119 			free(pack->path_packfile);
1120 			memset(pack, 0, sizeof(*pack));
1121 		}
1122 	} else if (packp)
1123 		*packp = pack;
1124 	return err;
1125 }
1126 
1127 struct got_pack *
1128 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1129 {
1130 	struct got_pack *pack = NULL;
1131 	int i;
1132 
1133 	for (i = 0; i < nitems(repo->packs); i++) {
1134 		pack = &repo->packs[i];
1135 		if (pack->path_packfile == NULL)
1136 			break;
1137 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1138 			return pack;
1139 	}
1140 
1141 	return NULL;
1142 }
1143 
1144 const struct got_error *
1145 got_repo_init(const char *repo_path)
1146 {
1147 	const struct got_error *err = NULL;
1148 	const char *dirnames[] = {
1149 		GOT_OBJECTS_DIR,
1150 		GOT_OBJECTS_PACK_DIR,
1151 		GOT_REFS_DIR,
1152 	};
1153 	const char *description_str = "Unnamed repository; "
1154 	    "edit this file 'description' to name the repository.";
1155 	const char *headref_str = "ref: refs/heads/main";
1156 	const char *gitconfig_str = "[core]\n"
1157 	    "\trepositoryformatversion = 0\n"
1158 	    "\tfilemode = true\n"
1159 	    "\tbare = true\n";
1160 	char *path;
1161 	int i;
1162 
1163 	if (!got_path_dir_is_empty(repo_path))
1164 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1165 
1166 	for (i = 0; i < nitems(dirnames); i++) {
1167 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1168 			return got_error_from_errno("asprintf");
1169 		}
1170 		err = got_path_mkdir(path);
1171 		free(path);
1172 		if (err)
1173 			return err;
1174 	}
1175 
1176 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1177 		return got_error_from_errno("asprintf");
1178 	err = got_path_create_file(path, description_str);
1179 	free(path);
1180 	if (err)
1181 		return err;
1182 
1183 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1184 		return got_error_from_errno("asprintf");
1185 	err = got_path_create_file(path, headref_str);
1186 	free(path);
1187 	if (err)
1188 		return err;
1189 
1190 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1191 		return got_error_from_errno("asprintf");
1192 	err = got_path_create_file(path, gitconfig_str);
1193 	free(path);
1194 	if (err)
1195 		return err;
1196 
1197 	return NULL;
1198 }
1199 
1200 static const struct got_error *
1201 match_packed_object(struct got_object_id **unique_id,
1202     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1203 {
1204 	const struct got_error *err = NULL;
1205 	char *path_packdir;
1206 	DIR *packdir;
1207 	struct dirent *dent;
1208 	char *path_packidx;
1209 	struct got_object_id_queue matched_ids;
1210 
1211 	SIMPLEQ_INIT(&matched_ids);
1212 
1213 	path_packdir = got_repo_get_path_objects_pack(repo);
1214 	if (path_packdir == NULL)
1215 		return got_error_from_errno("got_repo_get_path_objects_pack");
1216 
1217 	packdir = opendir(path_packdir);
1218 	if (packdir == NULL) {
1219 		if (errno != ENOENT)
1220 			err = got_error_from_errno2("opendir", path_packdir);
1221 		goto done;
1222 	}
1223 
1224 	while ((dent = readdir(packdir)) != NULL) {
1225 		struct got_packidx *packidx;
1226 		struct got_object_qid *qid;
1227 
1228 
1229 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
1230 			continue;
1231 
1232 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
1233 		    dent->d_name) == -1) {
1234 			err = got_error_from_errno("asprintf");
1235 			break;
1236 		}
1237 
1238 		err = got_packidx_open(&packidx, path_packidx, 0);
1239 		free(path_packidx);
1240 		if (err)
1241 			break;
1242 
1243 		err = got_packidx_match_id_str_prefix(&matched_ids,
1244 		    packidx, id_str_prefix);
1245 		if (err) {
1246 			got_packidx_close(packidx);
1247 			break;
1248 		}
1249 		err = got_packidx_close(packidx);
1250 		if (err)
1251 			break;
1252 
1253 		SIMPLEQ_FOREACH(qid, &matched_ids, entry) {
1254 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1255 				int matched_type;
1256 				err = got_object_get_type(&matched_type, repo,
1257 				    qid->id);
1258 				if (err)
1259 					goto done;
1260 				if (matched_type != obj_type)
1261 					continue;
1262 			}
1263 			if (*unique_id == NULL) {
1264 				*unique_id = got_object_id_dup(qid->id);
1265 				if (*unique_id == NULL) {
1266 					err = got_error_from_errno("malloc");
1267 					goto done;
1268 				}
1269 			} else {
1270 				if (got_object_id_cmp(*unique_id, qid->id) == 0)
1271 					continue; /* packed multiple times */
1272 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1273 				goto done;
1274 			}
1275 		}
1276 	}
1277 done:
1278 	got_object_id_queue_free(&matched_ids);
1279 	free(path_packdir);
1280 	if (packdir && closedir(packdir) != 0 && err == NULL)
1281 		err = got_error_from_errno("closedir");
1282 	if (err) {
1283 		free(*unique_id);
1284 		*unique_id = NULL;
1285 	}
1286 	return err;
1287 }
1288 
1289 static const struct got_error *
1290 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1291     const char *object_dir, const char *id_str_prefix, int obj_type,
1292     struct got_repository *repo)
1293 {
1294 	const struct got_error *err = NULL;
1295 	char *path;
1296 	DIR *dir = NULL;
1297 	struct dirent *dent;
1298 	struct got_object_id id;
1299 
1300 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1301 		err = got_error_from_errno("asprintf");
1302 		goto done;
1303 	}
1304 
1305 	dir = opendir(path);
1306 	if (dir == NULL) {
1307 		if (errno == ENOENT) {
1308 			err = NULL;
1309 			goto done;
1310 		}
1311 		err = got_error_from_errno2("opendir", path);
1312 		goto done;
1313 	}
1314 	while ((dent = readdir(dir)) != NULL) {
1315 		char *id_str;
1316 		int cmp;
1317 
1318 		if (strcmp(dent->d_name, ".") == 0 ||
1319 		    strcmp(dent->d_name, "..") == 0)
1320 			continue;
1321 
1322 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1323 			err = got_error_from_errno("asprintf");
1324 			goto done;
1325 		}
1326 
1327 		if (!got_parse_sha1_digest(id.sha1, id_str))
1328 			continue;
1329 
1330 		/*
1331 		 * Directory entries do not necessarily appear in
1332 		 * sorted order, so we must iterate over all of them.
1333 		 */
1334 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1335 		if (cmp != 0) {
1336 			free(id_str);
1337 			continue;
1338 		}
1339 
1340 		if (*unique_id == NULL) {
1341 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1342 				int matched_type;
1343 				err = got_object_get_type(&matched_type, repo,
1344 				    &id);
1345 				if (err)
1346 					goto done;
1347 				if (matched_type != obj_type)
1348 					continue;
1349 			}
1350 			*unique_id = got_object_id_dup(&id);
1351 			if (*unique_id == NULL) {
1352 				err = got_error_from_errno("got_object_id_dup");
1353 				free(id_str);
1354 				goto done;
1355 			}
1356 		} else {
1357 			if (got_object_id_cmp(*unique_id, &id) == 0)
1358 				continue; /* both packed and loose */
1359 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1360 			free(id_str);
1361 			goto done;
1362 		}
1363 	}
1364 done:
1365 	if (dir && closedir(dir) != 0 && err == NULL)
1366 		err = got_error_from_errno("closedir");
1367 	if (err) {
1368 		free(*unique_id);
1369 		*unique_id = NULL;
1370 	}
1371 	free(path);
1372 	return err;
1373 }
1374 
1375 const struct got_error *
1376 got_repo_match_object_id_prefix(struct got_object_id **id,
1377     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1378 {
1379 	const struct got_error *err = NULL;
1380 	char *path_objects = got_repo_get_path_objects(repo);
1381 	char *object_dir = NULL;
1382 	size_t len;
1383 	int i;
1384 
1385 	*id = NULL;
1386 
1387 	for (i = 0; i < strlen(id_str_prefix); i++) {
1388 		if (isxdigit((unsigned char)id_str_prefix[i]))
1389 			continue;
1390 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1391 	}
1392 
1393 	len = strlen(id_str_prefix);
1394 	if (len >= 2) {
1395 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1396 		if (err)
1397 			goto done;
1398 		object_dir = strndup(id_str_prefix, 2);
1399 		if (object_dir == NULL) {
1400 			err = got_error_from_errno("strdup");
1401 			goto done;
1402 		}
1403 		err = match_loose_object(id, path_objects, object_dir,
1404 		    id_str_prefix, obj_type, repo);
1405 	} else if (len == 1) {
1406 		int i;
1407 		for (i = 0; i < 0xf; i++) {
1408 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1409 			    == -1) {
1410 				err = got_error_from_errno("asprintf");
1411 				goto done;
1412 			}
1413 			err = match_packed_object(id, repo, object_dir,
1414 			    obj_type);
1415 			if (err)
1416 				goto done;
1417 			err = match_loose_object(id, path_objects, object_dir,
1418 			    id_str_prefix, obj_type, repo);
1419 			if (err)
1420 				goto done;
1421 		}
1422 	} else {
1423 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1424 		goto done;
1425 	}
1426 done:
1427 	free(object_dir);
1428 	if (err) {
1429 		free(*id);
1430 		*id = NULL;
1431 	} else if (*id == NULL)
1432 		err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1433 
1434 	return err;
1435 }
1436 
1437 const struct got_error *
1438 got_repo_match_object_id(struct got_object_id **id, char **label,
1439     const char *id_str, int obj_type, int resolve_tags,
1440     struct got_repository *repo)
1441 {
1442 	const struct got_error *err;
1443 	struct got_tag_object *tag;
1444 	struct got_reference *ref = NULL;
1445 
1446 	*id = NULL;
1447 	if (label)
1448 		*label = NULL;
1449 
1450 	if (resolve_tags) {
1451 		err = got_repo_object_match_tag(&tag, id_str, GOT_OBJ_TYPE_ANY,
1452 		    repo);
1453 		if (err == NULL) {
1454 			*id = got_object_id_dup(
1455 			    got_object_tag_get_object_id(tag));
1456 			if (*id == NULL)
1457 				err = got_error_from_errno("got_object_id_dup");
1458 			else if (label && asprintf(label, "refs/tags/%s",
1459 			    got_object_tag_get_name(tag)) == -1) {
1460 				err = got_error_from_errno("asprintf");
1461 				free(*id);
1462 				*id = NULL;
1463 			}
1464 			got_object_tag_close(tag);
1465 			return err;
1466 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1467 		    err->code != GOT_ERR_NO_OBJ)
1468 			return err;
1469 	}
1470 
1471 	err = got_repo_match_object_id_prefix(id, id_str, obj_type, repo);
1472 	if (err) {
1473 		if (err->code != GOT_ERR_BAD_OBJ_ID_STR)
1474 			return err;
1475 		err = got_ref_open(&ref, repo, id_str, 0);
1476 		if (err != NULL)
1477 			goto done;
1478 		if (label) {
1479 			*label = strdup(got_ref_get_name(ref));
1480 			if (*label == NULL) {
1481 				err = got_error_from_errno("strdup");
1482 				goto done;
1483 			}
1484 		}
1485 		err = got_ref_resolve(id, repo, ref);
1486 	} else if (label) {
1487 		err = got_object_id_str(label, *id);
1488 		if (*label == NULL) {
1489 			err = got_error_from_errno("strdup");
1490 			goto done;
1491 		}
1492 	}
1493 done:
1494 	if (ref)
1495 		got_ref_close(ref);
1496 	return err;
1497 }
1498 
1499 const struct got_error *
1500 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1501     int obj_type, struct got_repository *repo)
1502 {
1503 	const struct got_error *err;
1504 	struct got_reflist_head refs;
1505 	struct got_reflist_entry *re;
1506 	struct got_object_id *tag_id;
1507 
1508 	SIMPLEQ_INIT(&refs);
1509 	*tag = NULL;
1510 
1511 	err = got_ref_list(&refs, repo, "refs/tags", got_ref_cmp_by_name, NULL);
1512 	if (err)
1513 		return err;
1514 
1515 	SIMPLEQ_FOREACH(re, &refs, entry) {
1516 		const char *refname;
1517 		refname = got_ref_get_name(re->ref);
1518 		if (got_ref_is_symbolic(re->ref))
1519 			continue;
1520 		refname += strlen("refs/tags/");
1521 		if (strcmp(refname, name) != 0)
1522 			continue;
1523 		err = got_ref_resolve(&tag_id, repo, re->ref);
1524 		if (err)
1525 			break;
1526 		err = got_object_open_as_tag(tag, repo, tag_id);
1527 		free(tag_id);
1528 		if (err)
1529 			break;
1530 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1531 		    got_object_tag_get_object_type(*tag) == obj_type)
1532 			break;
1533 		got_object_tag_close(*tag);
1534 		*tag = NULL;
1535 	}
1536 
1537 	got_ref_list_free(&refs);
1538 	if (err == NULL && *tag == NULL)
1539 		err = got_error_path(name, GOT_ERR_NO_OBJ);
1540 	return err;
1541 }
1542 
1543 static const struct got_error *
1544 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1545     const char *name, mode_t mode, struct got_object_id *blob_id)
1546 {
1547 	const struct got_error *err = NULL;
1548 
1549 	 *new_te = NULL;
1550 
1551 	*new_te = calloc(1, sizeof(**new_te));
1552 	if (*new_te == NULL)
1553 		return got_error_from_errno("calloc");
1554 
1555 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1556 	    sizeof((*new_te)->name)) {
1557 		err = got_error(GOT_ERR_NO_SPACE);
1558 		goto done;
1559 	}
1560 
1561 	if (S_ISLNK(mode)) {
1562 		(*new_te)->mode = S_IFLNK;
1563 	} else {
1564 		(*new_te)->mode = S_IFREG;
1565 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1566 	}
1567 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1568 done:
1569 	if (err && *new_te) {
1570 		free(*new_te);
1571 		*new_te = NULL;
1572 	}
1573 	return err;
1574 }
1575 
1576 static const struct got_error *
1577 import_file(struct got_tree_entry **new_te, struct dirent *de,
1578     const char *path, struct got_repository *repo)
1579 {
1580 	const struct got_error *err;
1581 	struct got_object_id *blob_id = NULL;
1582 	char *filepath;
1583 	struct stat sb;
1584 
1585 	if (asprintf(&filepath, "%s%s%s", path,
1586 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1587 		return got_error_from_errno("asprintf");
1588 
1589 	if (lstat(filepath, &sb) != 0) {
1590 		err = got_error_from_errno2("lstat", path);
1591 		goto done;
1592 	}
1593 
1594 	err = got_object_blob_create(&blob_id, filepath, repo);
1595 	if (err)
1596 		goto done;
1597 
1598 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1599 	    blob_id);
1600 done:
1601 	free(filepath);
1602 	if (err)
1603 		free(blob_id);
1604 	return err;
1605 }
1606 
1607 static const struct got_error *
1608 insert_tree_entry(struct got_tree_entry *new_te,
1609     struct got_pathlist_head *paths)
1610 {
1611 	const struct got_error *err = NULL;
1612 	struct got_pathlist_entry *new_pe;
1613 
1614 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1615 	if (err)
1616 		return err;
1617 	if (new_pe == NULL)
1618 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1619 	return NULL;
1620 }
1621 
1622 static const struct got_error *write_tree(struct got_object_id **,
1623     const char *, struct got_pathlist_head *, struct got_repository *,
1624     got_repo_import_cb progress_cb, void *progress_arg);
1625 
1626 static const struct got_error *
1627 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1628     const char *path, struct got_pathlist_head *ignores,
1629     struct got_repository *repo,
1630     got_repo_import_cb progress_cb, void *progress_arg)
1631 {
1632 	const struct got_error *err;
1633 	struct got_object_id *id = NULL;
1634 	char *subdirpath;
1635 
1636 	if (asprintf(&subdirpath, "%s%s%s", path,
1637 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1638 		return got_error_from_errno("asprintf");
1639 
1640 	(*new_te) = calloc(1, sizeof(**new_te));
1641 	if (*new_te == NULL)
1642 		return got_error_from_errno("calloc");
1643 	(*new_te)->mode = S_IFDIR;
1644 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
1645 	    sizeof((*new_te)->name)) {
1646 		err = got_error(GOT_ERR_NO_SPACE);
1647 		goto done;
1648 	}
1649 	err = write_tree(&id, subdirpath, ignores,  repo,
1650 	    progress_cb, progress_arg);
1651 	if (err)
1652 		goto done;
1653 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
1654 	
1655 done:
1656 	free(id);
1657 	free(subdirpath);
1658 	if (err) {
1659 		free(*new_te);
1660 		*new_te = NULL;
1661 	}
1662 	return err;
1663 }
1664 
1665 static const struct got_error *
1666 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1667     struct got_pathlist_head *ignores, struct got_repository *repo,
1668     got_repo_import_cb progress_cb, void *progress_arg)
1669 {
1670 	const struct got_error *err = NULL;
1671 	DIR *dir;
1672 	struct dirent *de;
1673 	int nentries;
1674 	struct got_tree_entry *new_te = NULL;
1675 	struct got_pathlist_head paths;
1676 	struct got_pathlist_entry *pe;
1677 
1678 	*new_tree_id = NULL;
1679 
1680 	TAILQ_INIT(&paths);
1681 
1682 	dir = opendir(path_dir);
1683 	if (dir == NULL) {
1684 		err = got_error_from_errno2("opendir", path_dir);
1685 		goto done;
1686 	}
1687 
1688 	nentries = 0;
1689 	while ((de = readdir(dir)) != NULL) {
1690 		int ignore = 0;
1691 		int type;
1692 
1693 		if (strcmp(de->d_name, ".") == 0 ||
1694 		    strcmp(de->d_name, "..") == 0)
1695 			continue;
1696 
1697 		TAILQ_FOREACH(pe, ignores, entry) {
1698 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
1699 				ignore = 1;
1700 				break;
1701 			}
1702 		}
1703 		if (ignore)
1704 			continue;
1705 
1706 		err = got_path_dirent_type(&type, path_dir, de);
1707 		if (err)
1708 			goto done;
1709 
1710 		if (type == DT_DIR) {
1711 			err = import_subdir(&new_te, de, path_dir,
1712 			    ignores, repo, progress_cb, progress_arg);
1713 			if (err) {
1714 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
1715 					goto done;
1716 				err = NULL;
1717 				continue;
1718 			}
1719 		} else if (type == DT_REG || type == DT_LNK) {
1720 			err = import_file(&new_te, de, path_dir, repo);
1721 			if (err)
1722 				goto done;
1723 		} else
1724 			continue;
1725 
1726 		err = insert_tree_entry(new_te, &paths);
1727 		if (err)
1728 			goto done;
1729 		nentries++;
1730 	}
1731 
1732 	if (TAILQ_EMPTY(&paths)) {
1733 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
1734 		    "cannot create tree without any entries");
1735 		goto done;
1736 	}
1737 
1738 	TAILQ_FOREACH(pe, &paths, entry) {
1739 		struct got_tree_entry *te = pe->data;
1740 		char *path;
1741 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
1742 			continue;
1743 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
1744 			err = got_error_from_errno("asprintf");
1745 			goto done;
1746 		}
1747 		err = (*progress_cb)(progress_arg, path);
1748 		free(path);
1749 		if (err)
1750 			goto done;
1751 	}
1752 
1753 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
1754 done:
1755 	if (dir)
1756 		closedir(dir);
1757 	got_pathlist_free(&paths);
1758 	return err;
1759 }
1760 
1761 const struct got_error *
1762 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
1763     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
1764     struct got_repository *repo, got_repo_import_cb progress_cb,
1765     void *progress_arg)
1766 {
1767 	const struct got_error *err;
1768 	struct got_object_id *new_tree_id;
1769 
1770 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
1771 	    progress_cb, progress_arg);
1772 	if (err)
1773 		return err;
1774 
1775 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
1776 	    author, time(NULL), author, time(NULL), logmsg, repo);
1777 	free(new_tree_id);
1778 	return err;
1779 }