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 #include <sys/syslimits.h>
24 
25 #include <ctype.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 <zlib.h>
36 #include <errno.h>
37 #include <libgen.h>
38 #include <stdint.h>
39 #include <imsg.h>
40 #include <uuid.h>
41 
42 #include "got_error.h"
43 #include "got_reference.h"
44 #include "got_repository.h"
45 #include "got_path.h"
46 #include "got_cancel.h"
47 #include "got_worktree.h"
48 #include "got_object.h"
49 
50 #include "got_lib_delta.h"
51 #include "got_lib_inflate.h"
52 #include "got_lib_object.h"
53 #include "got_lib_object_parse.h"
54 #include "got_lib_object_create.h"
55 #include "got_lib_pack.h"
56 #include "got_lib_privsep.h"
57 #include "got_lib_worktree.h"
58 #include "got_lib_sha1.h"
59 #include "got_lib_object_cache.h"
60 #include "got_lib_repository.h"
61 
62 #ifndef nitems
63 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
64 #endif
65 
66 const char *
67 got_repo_get_path(struct got_repository *repo)
68 {
69 	return repo->path;
70 }
71 
72 const char *
73 got_repo_get_path_git_dir(struct got_repository *repo)
74 {
75 	return repo->path_git_dir;
76 }
77 
78 const char *
79 got_repo_get_gitconfig_author_name(struct got_repository *repo)
80 {
81 	return repo->gitconfig_author_name;
82 }
83 
84 const char *
85 got_repo_get_gitconfig_author_email(struct got_repository *repo)
86 {
87 	return repo->gitconfig_author_email;
88 }
89 
90 const char *
91 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
92 {
93 	return repo->global_gitconfig_author_name;
94 }
95 
96 const char *
97 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
98 {
99 	return repo->global_gitconfig_author_email;
100 }
101 
102 const char *
103 got_repo_get_gitconfig_owner(struct got_repository *repo)
104 {
105 	return repo->gitconfig_owner;
106 }
107 
108 int
109 got_repo_is_bare(struct got_repository *repo)
110 {
111 	return (strcmp(repo->path, repo->path_git_dir) == 0);
112 }
113 
114 static char *
115 get_path_git_child(struct got_repository *repo, const char *basename)
116 {
117 	char *path_child;
118 	
119 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
120 	    basename) == -1)
121 		return NULL;
122 
123 	return path_child;
124 }
125 
126 char *
127 got_repo_get_path_objects(struct got_repository *repo)
128 {
129 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
130 }
131 
132 char *
133 got_repo_get_path_objects_pack(struct got_repository *repo)
134 {
135 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
136 }
137 
138 char *
139 got_repo_get_path_refs(struct got_repository *repo)
140 {
141 	return get_path_git_child(repo, GOT_REFS_DIR);
142 }
143 
144 char *
145 got_repo_get_path_packed_refs(struct got_repository *repo)
146 {
147 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
148 }
149 
150 static char *
151 get_path_head(struct got_repository *repo)
152 {
153 	return get_path_git_child(repo, GOT_HEAD_FILE);
154 }
155 
156 char *
157 got_repo_get_path_gitconfig(struct got_repository *repo)
158 {
159 	return get_path_git_child(repo, GOT_GITCONFIG);
160 }
161 
162 void
163 got_repo_get_gitconfig_remotes(int *nremotes, struct got_remote_repo **remotes,
164     struct got_repository *repo)
165 {
166 	*nremotes = repo->ngitconfig_remotes;
167 	*remotes = repo->gitconfig_remotes;
168 }
169 
170 static int
171 is_git_repo(struct got_repository *repo)
172 {
173 	const char *path_git = got_repo_get_path_git_dir(repo);
174 	char *path_objects = got_repo_get_path_objects(repo);
175 	char *path_refs = got_repo_get_path_refs(repo);
176 	char *path_head = get_path_head(repo);
177 	int ret = 0;
178 	struct stat sb;
179 	struct got_reference *head_ref;
180 
181 	if (lstat(path_git, &sb) == -1)
182 		goto done;
183 	if (!S_ISDIR(sb.st_mode))
184 		goto done;
185 
186 	if (lstat(path_objects, &sb) == -1)
187 		goto done;
188 	if (!S_ISDIR(sb.st_mode))
189 		goto done;
190 
191 	if (lstat(path_refs, &sb) == -1)
192 		goto done;
193 	if (!S_ISDIR(sb.st_mode))
194 		goto done;
195 
196 	if (lstat(path_head, &sb) == -1)
197 		goto done;
198 	if (!S_ISREG(sb.st_mode))
199 		goto done;
200 
201 	/* Check if the HEAD reference can be opened. */
202 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
203 		goto done;
204 	got_ref_close(head_ref);
205 
206 	ret = 1;
207 done:
208 	free(path_objects);
209 	free(path_refs);
210 	free(path_head);
211 	return ret;
212 
213 }
214 
215 const struct got_error *
216 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
217     struct got_object *obj)
218 {
219 #ifndef GOT_NO_OBJ_CACHE
220 	const struct got_error *err = NULL;
221 	err = got_object_cache_add(&repo->objcache, id, obj);
222 	if (err) {
223 		if (err->code == GOT_ERR_OBJ_EXISTS ||
224 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
225 			err = NULL;
226 		return err;
227 	}
228 	obj->refcnt++;
229 #endif
230 	return NULL;
231 }
232 
233 struct got_object *
234 got_repo_get_cached_object(struct got_repository *repo,
235     struct got_object_id *id)
236 {
237 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
238 }
239 
240 const struct got_error *
241 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
242     struct got_tree_object *tree)
243 {
244 #ifndef GOT_NO_OBJ_CACHE
245 	const struct got_error *err = NULL;
246 	err = got_object_cache_add(&repo->treecache, id, tree);
247 	if (err) {
248 		if (err->code == GOT_ERR_OBJ_EXISTS ||
249 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
250 			err = NULL;
251 		return err;
252 	}
253 	tree->refcnt++;
254 #endif
255 	return NULL;
256 }
257 
258 struct got_tree_object *
259 got_repo_get_cached_tree(struct got_repository *repo,
260     struct got_object_id *id)
261 {
262 	return (struct got_tree_object *)got_object_cache_get(
263 	    &repo->treecache, id);
264 }
265 
266 const struct got_error *
267 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
268     struct got_commit_object *commit)
269 {
270 #ifndef GOT_NO_OBJ_CACHE
271 	const struct got_error *err = NULL;
272 	err = got_object_cache_add(&repo->commitcache, id, commit);
273 	if (err) {
274 		if (err->code == GOT_ERR_OBJ_EXISTS ||
275 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
276 			err = NULL;
277 		return err;
278 	}
279 	commit->refcnt++;
280 #endif
281 	return NULL;
282 }
283 
284 struct got_commit_object *
285 got_repo_get_cached_commit(struct got_repository *repo,
286     struct got_object_id *id)
287 {
288 	return (struct got_commit_object *)got_object_cache_get(
289 	    &repo->commitcache, id);
290 }
291 
292 const struct got_error *
293 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
294     struct got_tag_object *tag)
295 {
296 #ifndef GOT_NO_OBJ_CACHE
297 	const struct got_error *err = NULL;
298 	err = got_object_cache_add(&repo->tagcache, id, tag);
299 	if (err) {
300 		if (err->code == GOT_ERR_OBJ_EXISTS ||
301 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
302 			err = NULL;
303 		return err;
304 	}
305 	tag->refcnt++;
306 #endif
307 	return NULL;
308 }
309 
310 struct got_tag_object *
311 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
312 {
313 	return (struct got_tag_object *)got_object_cache_get(
314 	    &repo->tagcache, id);
315 }
316 
317 const struct got_error *
318 open_repo(struct got_repository *repo, const char *path)
319 {
320 	const struct got_error *err = NULL;
321 
322 	/* bare git repository? */
323 	repo->path_git_dir = strdup(path);
324 	if (repo->path_git_dir == NULL)
325 		return got_error_from_errno("strdup");
326 	if (is_git_repo(repo)) {
327 		repo->path = strdup(repo->path_git_dir);
328 		if (repo->path == NULL) {
329 			err = got_error_from_errno("strdup");
330 			goto done;
331 		}
332 		return NULL;
333 	}
334 
335 	/* git repository with working tree? */
336 	free(repo->path_git_dir);
337 	repo->path_git_dir = NULL;
338 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
339 		err = got_error_from_errno("asprintf");
340 		goto done;
341 	}
342 	if (is_git_repo(repo)) {
343 		repo->path = strdup(path);
344 		if (repo->path == NULL) {
345 			err = got_error_from_errno("strdup");
346 			goto done;
347 		}
348 		return NULL;
349 	}
350 
351 	err = got_error(GOT_ERR_NOT_GIT_REPO);
352 done:
353 	if (err) {
354 		free(repo->path);
355 		repo->path = NULL;
356 		free(repo->path_git_dir);
357 		repo->path_git_dir = NULL;
358 	}
359 	return err;
360 }
361 
362 static const struct got_error *
363 parse_gitconfig_file(int *gitconfig_repository_format_version,
364     char **gitconfig_author_name, char **gitconfig_author_email,
365     struct got_remote_repo **remotes, int *nremotes,
366     char **gitconfig_owner,
367     const char *gitconfig_path)
368 {
369 	const struct got_error *err = NULL, *child_err = NULL;
370 	int fd = -1;
371 	int imsg_fds[2] = { -1, -1 };
372 	pid_t pid;
373 	struct imsgbuf *ibuf;
374 
375 	*gitconfig_repository_format_version = 0;
376 	*gitconfig_author_name = NULL;
377 	*gitconfig_author_email = NULL;
378 
379 	fd = open(gitconfig_path, O_RDONLY);
380 	if (fd == -1) {
381 		if (errno == ENOENT)
382 			return NULL;
383 		return got_error_from_errno2("open", gitconfig_path);
384 	}
385 
386 	ibuf = calloc(1, sizeof(*ibuf));
387 	if (ibuf == NULL) {
388 		err = got_error_from_errno("calloc");
389 		goto done;
390 	}
391 
392 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
393 		err = got_error_from_errno("socketpair");
394 		goto done;
395 	}
396 
397 	pid = fork();
398 	if (pid == -1) {
399 		err = got_error_from_errno("fork");
400 		goto done;
401 	} else if (pid == 0) {
402 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
403 		    gitconfig_path);
404 		/* not reached */
405 	}
406 
407 	if (close(imsg_fds[1]) == -1) {
408 		err = got_error_from_errno("close");
409 		goto done;
410 	}
411 	imsg_fds[1] = -1;
412 	imsg_init(ibuf, imsg_fds[0]);
413 
414 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
415 	if (err)
416 		goto done;
417 	fd = -1;
418 
419 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
420 	if (err)
421 		goto done;
422 
423 	err = got_privsep_recv_gitconfig_int(
424 	    gitconfig_repository_format_version, ibuf);
425 	if (err)
426 		goto done;
427 
428 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
429 	if (err)
430 		goto done;
431 
432 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
433 	if (err)
434 		goto done;
435 
436 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
437 	if (err)
438 		goto done;
439 
440 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
441 	if (err)
442 		goto done;
443 
444 	if (remotes && nremotes) {
445 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
446 		if (err)
447 			goto done;
448 
449 		err = got_privsep_recv_gitconfig_remotes(remotes,
450 		    nremotes, ibuf);
451 		if (err)
452 			goto done;
453 	}
454 
455 	if (gitconfig_owner) {
456 		err = got_privsep_send_gitconfig_owner_req(ibuf);
457 		if (err)
458 			goto done;
459 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
460 		if (err)
461 			goto done;
462 	}
463 
464 	imsg_clear(ibuf);
465 	err = got_privsep_send_stop(imsg_fds[0]);
466 	child_err = got_privsep_wait_for_child(pid);
467 	if (child_err && err == NULL)
468 		err = child_err;
469 done:
470 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
471 		err = got_error_from_errno("close");
472 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
473 		err = got_error_from_errno("close");
474 	if (fd != -1 && close(fd) == -1 && err == NULL)
475 		err = got_error_from_errno2("close", gitconfig_path);
476 	free(ibuf);
477 	return err;
478 }
479 
480 static const struct got_error *
481 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
482 {
483 	const struct got_error *err = NULL;
484 	char *repo_gitconfig_path = NULL;
485 
486 	if (global_gitconfig_path) {
487 		/* Read settings from ~/.gitconfig. */
488 		int dummy_repo_version;
489 		err = parse_gitconfig_file(&dummy_repo_version,
490 		    &repo->global_gitconfig_author_name,
491 		    &repo->global_gitconfig_author_email,
492 		    NULL, NULL, NULL, global_gitconfig_path);
493 		if (err)
494 			return err;
495 	}
496 
497 	/* Read repository's .git/config file. */
498 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
499 	if (repo_gitconfig_path == NULL)
500 		return got_error_from_errno("got_repo_get_path_gitconfig");
501 
502 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
503 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
504 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
505 	    &repo->gitconfig_owner, repo_gitconfig_path);
506 	if (err)
507 		goto done;
508 done:
509 	free(repo_gitconfig_path);
510 	return err;
511 }
512 
513 const struct got_error *
514 got_repo_open(struct got_repository **repop, const char *path,
515     const char *global_gitconfig_path)
516 {
517 	struct got_repository *repo = NULL;
518 	const struct got_error *err = NULL;
519 	char *abspath;
520 	int i, tried_root = 0;
521 
522 	*repop = NULL;
523 
524 	if (got_path_is_absolute(path))
525 		abspath = strdup(path);
526 	else
527 		abspath = got_path_get_absolute(path);
528 	if (abspath == NULL)
529 		return got_error_path(path, GOT_ERR_BAD_PATH);
530 
531 	repo = calloc(1, sizeof(*repo));
532 	if (repo == NULL) {
533 		err = got_error_from_errno("calloc");
534 		goto done;
535 	}
536 
537 	for (i = 0; i < nitems(repo->privsep_children); i++) {
538 		memset(&repo->privsep_children[i], 0,
539 		    sizeof(repo->privsep_children[0]));
540 		repo->privsep_children[i].imsg_fd = -1;
541 	}
542 
543 	err = got_object_cache_init(&repo->objcache,
544 	    GOT_OBJECT_CACHE_TYPE_OBJ);
545 	if (err)
546 		goto done;
547 	err = got_object_cache_init(&repo->treecache,
548 	    GOT_OBJECT_CACHE_TYPE_TREE);
549 	if (err)
550 		goto done;
551 	err = got_object_cache_init(&repo->commitcache,
552 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
553 	if (err)
554 		goto done;
555 	err = got_object_cache_init(&repo->tagcache,
556 	    GOT_OBJECT_CACHE_TYPE_TAG);
557 	if (err)
558 		goto done;
559 
560 	path = realpath(abspath, NULL);
561 	if (path == NULL) {
562 		err = got_error_from_errno2("realpath", abspath);
563 		goto done;
564 	}
565 
566 	do {
567 		err = open_repo(repo, path);
568 		if (err == NULL)
569 			break;
570 		if (err->code != GOT_ERR_NOT_GIT_REPO)
571 			break;
572 		if (path[0] == '/' && path[1] == '\0') {
573 			if (tried_root) {
574 				err = got_error(GOT_ERR_NOT_GIT_REPO);
575 				goto done;
576 			}
577 			tried_root = 1;
578 		}
579 		path = dirname(path);
580 		if (path == NULL) {
581 			err = got_error_from_errno2("dirname", path);
582 			goto done;
583 		}
584 	} while (path);
585 
586 	err = read_gitconfig(repo, global_gitconfig_path);
587 	if (err)
588 		goto done;
589 	if (repo->gitconfig_repository_format_version != 0)
590 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
591 done:
592 	if (err)
593 		got_repo_close(repo);
594 	else
595 		*repop = repo;
596 	free(abspath);
597 	return err;
598 }
599 
600 const struct got_error *
601 got_repo_close(struct got_repository *repo)
602 {
603 	const struct got_error *err = NULL, *child_err;
604 	int i;
605 
606 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
607 		if (repo->packidx_cache[i] == NULL)
608 			break;
609 		got_packidx_close(repo->packidx_cache[i]);
610 	}
611 
612 	for (i = 0; i < nitems(repo->packs); i++) {
613 		if (repo->packs[i].path_packfile == NULL)
614 			break;
615 		got_pack_close(&repo->packs[i]);
616 	}
617 
618 	free(repo->path);
619 	free(repo->path_git_dir);
620 
621 	got_object_cache_close(&repo->objcache);
622 	got_object_cache_close(&repo->treecache);
623 	got_object_cache_close(&repo->commitcache);
624 	got_object_cache_close(&repo->tagcache);
625 
626 	for (i = 0; i < nitems(repo->privsep_children); i++) {
627 		if (repo->privsep_children[i].imsg_fd == -1)
628 			continue;
629 		imsg_clear(repo->privsep_children[i].ibuf);
630 		free(repo->privsep_children[i].ibuf);
631 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
632 		child_err = got_privsep_wait_for_child(
633 		    repo->privsep_children[i].pid);
634 		if (child_err && err == NULL)
635 			err = child_err;
636 		if (close(repo->privsep_children[i].imsg_fd) != 0 &&
637 		    err == NULL)
638 			err = got_error_from_errno("close");
639 	}
640 
641 	free(repo->gitconfig_author_name);
642 	free(repo->gitconfig_author_email);
643 	for (i = 0; i < repo->ngitconfig_remotes; i++) {
644 		free(repo->gitconfig_remotes[i].name);
645 		free(repo->gitconfig_remotes[i].url);
646 	}
647 	free(repo->gitconfig_remotes);
648 	free(repo);
649 
650 	return err;
651 }
652 
653 const struct got_error *
654 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
655     const char *input_path, int check_disk)
656 {
657 	const struct got_error *err = NULL;
658 	const char *repo_abspath = NULL;
659 	size_t repolen, len;
660 	char *canonpath, *path = NULL;
661 
662 	*in_repo_path = NULL;
663 
664 	canonpath = strdup(input_path);
665 	if (canonpath == NULL) {
666 		err = got_error_from_errno("strdup");
667 		goto done;
668 	}
669 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
670 	if (err)
671 		goto done;
672 
673 	repo_abspath = got_repo_get_path(repo);
674 
675 	if (!check_disk || canonpath[0] == '\0') {
676 		path = strdup(canonpath);
677 		if (path == NULL) {
678 			err = got_error_from_errno("strdup");
679 			goto done;
680 		}
681 	} else {
682 		path = realpath(canonpath, NULL);
683 		if (path == NULL) {
684 			if (errno != ENOENT) {
685 				err = got_error_from_errno2("realpath",
686 				    canonpath);
687 				goto done;
688 			}
689 			/*
690 			 * Path is not on disk.
691 			 * Assume it is already relative to repository root.
692 			 */
693 			path = strdup(canonpath);
694 			if (path == NULL) {
695 				err = got_error_from_errno("strdup");
696 				goto done;
697 			}
698 		}
699 
700 		repolen = strlen(repo_abspath);
701 		len = strlen(path);
702 
703 
704 		if (strcmp(path, repo_abspath) == 0) {
705 			free(path);
706 			path = strdup("");
707 			if (path == NULL) {
708 				err = got_error_from_errno("strdup");
709 				goto done;
710 			}
711 		} else if (len > repolen &&
712 		    got_path_is_child(path, repo_abspath, repolen)) {
713 			/* Matched an on-disk path inside repository. */
714 			if (got_repo_is_bare(repo)) {
715 				/*
716 				 * Matched an on-disk path inside repository
717 				 * database. Treat as repository-relative.
718 				 */
719 			} else {
720 				char *child;
721 				/* Strip common prefix with repository path. */
722 				err = got_path_skip_common_ancestor(&child,
723 				    repo_abspath, path);
724 				if (err)
725 					goto done;
726 				free(path);
727 				path = child;
728 			}
729 		} else {
730 			/*
731 			 * Matched unrelated on-disk path.
732 			 * Treat it as repository-relative.
733 			 */
734 		}
735 	}
736 
737 	/* Make in-repository path absolute */
738 	if (path[0] != '/') {
739 		char *abspath;
740 		if (asprintf(&abspath, "/%s", path) == -1) {
741 			err = got_error_from_errno("asprintf");
742 			goto done;
743 		}
744 		free(path);
745 		path = abspath;
746 	}
747 
748 done:
749 	free(canonpath);
750 	if (err)
751 		free(path);
752 	else
753 		*in_repo_path = path;
754 	return err;
755 }
756 
757 static const struct got_error *
758 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
759     const char *path_packidx)
760 {
761 	const struct got_error *err = NULL;
762 	int i;
763 
764 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
765 		if (repo->packidx_cache[i] == NULL)
766 			break;
767 		if (strcmp(repo->packidx_cache[i]->path_packidx,
768 		    path_packidx) == 0) {
769 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
770 		}
771 	}
772 	if (i == nitems(repo->packidx_cache)) {
773 		err = got_packidx_close(repo->packidx_cache[i - 1]);
774 		if (err)
775 			return err;
776 	}
777 
778 	/*
779 	 * Insert the new pack index at the front so it will
780 	 * be searched first in the future.
781 	 */
782 	memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
783 	    sizeof(repo->packidx_cache) -
784 	    sizeof(repo->packidx_cache[0]));
785 	repo->packidx_cache[0] = packidx;
786 
787 	return NULL;
788 }
789 
790 static int
791 is_packidx_filename(const char *name, size_t len)
792 {
793 	if (len != GOT_PACKIDX_NAMELEN)
794 		return 0;
795 
796 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
797 		return 0;
798 
799 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
800 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
801 		return 0;
802 
803 	return 1;
804 }
805 
806 const struct got_error *
807 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
808     struct got_repository *repo, struct got_object_id *id)
809 {
810 	const struct got_error *err;
811 	char *path_packdir;
812 	DIR *packdir;
813 	struct dirent *dent;
814 	char *path_packidx;
815 	int i;
816 
817 	/* Search pack index cache. */
818 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
819 		if (repo->packidx_cache[i] == NULL)
820 			break;
821 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
822 		if (*idx != -1) {
823 			*packidx = repo->packidx_cache[i];
824 			/*
825 			 * Move this cache entry to the front. Repeatedly
826 			 * searching a wrong pack index can be expensive.
827 			 */
828 			if (i > 0) {
829 				struct got_packidx *p;
830 				p = repo->packidx_cache[0];
831 				repo->packidx_cache[0] = *packidx;
832 				repo->packidx_cache[i] = p;
833 			}
834 			return NULL;
835 		}
836 	}
837 	/* No luck. Search the filesystem. */
838 
839 	path_packdir = got_repo_get_path_objects_pack(repo);
840 	if (path_packdir == NULL)
841 		return got_error_from_errno("got_repo_get_path_objects_pack");
842 
843 	packdir = opendir(path_packdir);
844 	if (packdir == NULL) {
845 		if (errno == ENOENT)
846 			err = got_error_no_obj(id);
847 		else
848 			err = got_error_from_errno2("opendir", path_packdir);
849 		goto done;
850 	}
851 
852 	while ((dent = readdir(packdir)) != NULL) {
853 		int is_cached = 0;
854 
855 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
856 			continue;
857 
858 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
859 		    dent->d_name) == -1) {
860 			err = got_error_from_errno("asprintf");
861 			goto done;
862 		}
863 
864 		for (i = 0; i < nitems(repo->packidx_cache); i++) {
865 			if (repo->packidx_cache[i] == NULL)
866 				break;
867 			if (strcmp(repo->packidx_cache[i]->path_packidx,
868 			    path_packidx) == 0) {
869 				is_cached = 1;
870 				break;
871 			}
872 		}
873 		if (is_cached) {
874 			free(path_packidx);
875 			continue; /* already searched */
876 		}
877 
878 		err = got_packidx_open(packidx, path_packidx, 0);
879 		if (err) {
880 			free(path_packidx);
881 			goto done;
882 		}
883 
884 		err = cache_packidx(repo, *packidx, path_packidx);
885 		free(path_packidx);
886 		if (err)
887 			goto done;
888 
889 		*idx = got_packidx_get_object_idx(*packidx, id);
890 		if (*idx != -1) {
891 			err = NULL; /* found the object */
892 			goto done;
893 		}
894 	}
895 
896 	err = got_error_no_obj(id);
897 done:
898 	free(path_packdir);
899 	if (packdir && closedir(packdir) != 0 && err == NULL)
900 		err = got_error_from_errno("closedir");
901 	return err;
902 }
903 
904 static const struct got_error *
905 read_packfile_hdr(int fd, struct got_packidx *packidx)
906 {
907 	const struct got_error *err = NULL;
908 	uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
909 	struct got_packfile_hdr hdr;
910 	ssize_t n;
911 
912 	n = read(fd, &hdr, sizeof(hdr));
913 	if (n < 0)
914 		return got_error_from_errno("read");
915 	if (n != sizeof(hdr))
916 		return got_error(GOT_ERR_BAD_PACKFILE);
917 
918 	if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
919 	    betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
920 	    betoh32(hdr.nobjects) != totobj)
921 		err = got_error(GOT_ERR_BAD_PACKFILE);
922 
923 	return err;
924 }
925 
926 static const struct got_error *
927 open_packfile(int *fd, const char *path_packfile, struct got_packidx *packidx)
928 {
929 	const struct got_error *err = NULL;
930 
931 	*fd = open(path_packfile, O_RDONLY | O_NOFOLLOW);
932 	if (*fd == -1)
933 		return got_error_from_errno2("open", path_packfile);
934 
935 	if (packidx) {
936 		err = read_packfile_hdr(*fd, packidx);
937 		if (err) {
938 			close(*fd);
939 			*fd = -1;
940 		}
941 	}
942 
943 	return err;
944 }
945 
946 const struct got_error *
947 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
948     const char *path_packfile, struct got_packidx *packidx)
949 {
950 	const struct got_error *err = NULL;
951 	struct got_pack *pack = NULL;
952 	struct stat sb;
953 	int i;
954 
955 	if (packp)
956 		*packp = NULL;
957 
958 	for (i = 0; i < nitems(repo->packs); i++) {
959 		pack = &repo->packs[i];
960 		if (pack->path_packfile == NULL)
961 			break;
962 		if (strcmp(pack->path_packfile, path_packfile) == 0)
963 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
964 	}
965 
966 	if (i == nitems(repo->packs) - 1) {
967 		err = got_pack_close(&repo->packs[i - 1]);
968 		if (err)
969 			return err;
970 		memmove(&repo->packs[1], &repo->packs[0],
971 		    sizeof(repo->packs) - sizeof(repo->packs[0]));
972 		i = 0;
973 	}
974 
975 	pack = &repo->packs[i];
976 
977 	pack->path_packfile = strdup(path_packfile);
978 	if (pack->path_packfile == NULL) {
979 		err = got_error_from_errno("strdup");
980 		goto done;
981 	}
982 
983 	err = open_packfile(&pack->fd, path_packfile, packidx);
984 	if (err)
985 		goto done;
986 
987 	if (fstat(pack->fd, &sb) != 0) {
988 		err = got_error_from_errno("fstat");
989 		goto done;
990 	}
991 	pack->filesize = sb.st_size;
992 
993 	pack->privsep_child = NULL;
994 
995 #ifndef GOT_PACK_NO_MMAP
996 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
997 	    pack->fd, 0);
998 	if (pack->map == MAP_FAILED) {
999 		if (errno != ENOMEM) {
1000 			err = got_error_from_errno("mmap");
1001 			goto done;
1002 		}
1003 		pack->map = NULL; /* fall back to read(2) */
1004 	}
1005 #endif
1006 done:
1007 	if (err) {
1008 		if (pack) {
1009 			free(pack->path_packfile);
1010 			memset(pack, 0, sizeof(*pack));
1011 		}
1012 	} else if (packp)
1013 		*packp = pack;
1014 	return err;
1015 }
1016 
1017 struct got_pack *
1018 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1019 {
1020 	struct got_pack *pack = NULL;
1021 	int i;
1022 
1023 	for (i = 0; i < nitems(repo->packs); i++) {
1024 		pack = &repo->packs[i];
1025 		if (pack->path_packfile == NULL)
1026 			break;
1027 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1028 			return pack;
1029 	}
1030 
1031 	return NULL;
1032 }
1033 
1034 const struct got_error *
1035 got_repo_init(const char *repo_path)
1036 {
1037 	const struct got_error *err = NULL;
1038 	const char *dirnames[] = {
1039 		GOT_OBJECTS_DIR,
1040 		GOT_OBJECTS_PACK_DIR,
1041 		GOT_REFS_DIR,
1042 	};
1043 	const char *description_str = "Unnamed repository; "
1044 	    "edit this file 'description' to name the repository.";
1045 	const char *headref_str = "ref: refs/heads/main";
1046 	const char *gitconfig_str = "[core]\n"
1047 	    "\trepositoryformatversion = 0\n"
1048 	    "\tfilemode = true\n"
1049 	    "\tbare = true\n";
1050 	char *path;
1051 	int i;
1052 
1053 	if (!got_path_dir_is_empty(repo_path))
1054 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1055 
1056 	for (i = 0; i < nitems(dirnames); i++) {
1057 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1058 			return got_error_from_errno("asprintf");
1059 		}
1060 		err = got_path_mkdir(path);
1061 		free(path);
1062 		if (err)
1063 			return err;
1064 	}
1065 
1066 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1067 		return got_error_from_errno("asprintf");
1068 	err = got_path_create_file(path, description_str);
1069 	free(path);
1070 	if (err)
1071 		return err;
1072 
1073 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1074 		return got_error_from_errno("asprintf");
1075 	err = got_path_create_file(path, headref_str);
1076 	free(path);
1077 	if (err)
1078 		return err;
1079 
1080 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1081 		return got_error_from_errno("asprintf");
1082 	err = got_path_create_file(path, gitconfig_str);
1083 	free(path);
1084 	if (err)
1085 		return err;
1086 
1087 	return NULL;
1088 }
1089 
1090 static const struct got_error *
1091 match_packed_object(struct got_object_id **unique_id,
1092     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1093 {
1094 	const struct got_error *err = NULL;
1095 	char *path_packdir;
1096 	DIR *packdir;
1097 	struct dirent *dent;
1098 	char *path_packidx;
1099 	struct got_object_id_queue matched_ids;
1100 
1101 	SIMPLEQ_INIT(&matched_ids);
1102 
1103 	path_packdir = got_repo_get_path_objects_pack(repo);
1104 	if (path_packdir == NULL)
1105 		return got_error_from_errno("got_repo_get_path_objects_pack");
1106 
1107 	packdir = opendir(path_packdir);
1108 	if (packdir == NULL) {
1109 		if (errno != ENOENT)
1110 			err = got_error_from_errno2("opendir", path_packdir);
1111 		goto done;
1112 	}
1113 
1114 	while ((dent = readdir(packdir)) != NULL) {
1115 		struct got_packidx *packidx;
1116 		struct got_object_qid *qid;
1117 
1118 
1119 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
1120 			continue;
1121 
1122 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
1123 		    dent->d_name) == -1) {
1124 			err = got_error_from_errno("asprintf");
1125 			break;
1126 		}
1127 
1128 		err = got_packidx_open(&packidx, path_packidx, 0);
1129 		free(path_packidx);
1130 		if (err)
1131 			break;
1132 
1133 		err = got_packidx_match_id_str_prefix(&matched_ids,
1134 		    packidx, id_str_prefix);
1135 		if (err) {
1136 			got_packidx_close(packidx);
1137 			break;
1138 		}
1139 		err = got_packidx_close(packidx);
1140 		if (err)
1141 			break;
1142 
1143 		SIMPLEQ_FOREACH(qid, &matched_ids, entry) {
1144 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1145 				int matched_type;
1146 				err = got_object_get_type(&matched_type, repo,
1147 				    qid->id);
1148 				if (err)
1149 					goto done;
1150 				if (matched_type != obj_type)
1151 					continue;
1152 			}
1153 			if (*unique_id == NULL) {
1154 				*unique_id = got_object_id_dup(qid->id);
1155 				if (*unique_id == NULL) {
1156 					err = got_error_from_errno("malloc");
1157 					goto done;
1158 				}
1159 			} else {
1160 				if (got_object_id_cmp(*unique_id, qid->id) == 0)
1161 					continue; /* packed multiple times */
1162 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1163 				goto done;
1164 			}
1165 		}
1166 	}
1167 done:
1168 	got_object_id_queue_free(&matched_ids);
1169 	free(path_packdir);
1170 	if (packdir && closedir(packdir) != 0 && err == NULL)
1171 		err = got_error_from_errno("closedir");
1172 	if (err) {
1173 		free(*unique_id);
1174 		*unique_id = NULL;
1175 	}
1176 	return err;
1177 }
1178 
1179 static const struct got_error *
1180 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1181     const char *object_dir, const char *id_str_prefix, int obj_type,
1182     struct got_repository *repo)
1183 {
1184 	const struct got_error *err = NULL;
1185 	char *path;
1186 	DIR *dir = NULL;
1187 	struct dirent *dent;
1188 	struct got_object_id id;
1189 
1190 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1191 		err = got_error_from_errno("asprintf");
1192 		goto done;
1193 	}
1194 
1195 	dir = opendir(path);
1196 	if (dir == NULL) {
1197 		if (errno == ENOENT) {
1198 			err = NULL;
1199 			goto done;
1200 		}
1201 		err = got_error_from_errno2("opendir", path);
1202 		goto done;
1203 	}
1204 	while ((dent = readdir(dir)) != NULL) {
1205 		char *id_str;
1206 		int cmp;
1207 
1208 		if (strcmp(dent->d_name, ".") == 0 ||
1209 		    strcmp(dent->d_name, "..") == 0)
1210 			continue;
1211 
1212 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1213 			err = got_error_from_errno("asprintf");
1214 			goto done;
1215 		}
1216 
1217 		if (!got_parse_sha1_digest(id.sha1, id_str))
1218 			continue;
1219 
1220 		/*
1221 		 * Directory entries do not necessarily appear in
1222 		 * sorted order, so we must iterate over all of them.
1223 		 */
1224 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1225 		if (cmp != 0) {
1226 			free(id_str);
1227 			continue;
1228 		}
1229 
1230 		if (*unique_id == NULL) {
1231 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1232 				int matched_type;
1233 				err = got_object_get_type(&matched_type, repo,
1234 				    &id);
1235 				if (err)
1236 					goto done;
1237 				if (matched_type != obj_type)
1238 					continue;
1239 			}
1240 			*unique_id = got_object_id_dup(&id);
1241 			if (*unique_id == NULL) {
1242 				err = got_error_from_errno("got_object_id_dup");
1243 				free(id_str);
1244 				goto done;
1245 			}
1246 		} else {
1247 			if (got_object_id_cmp(*unique_id, &id) == 0)
1248 				continue; /* both packed and loose */
1249 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1250 			free(id_str);
1251 			goto done;
1252 		}
1253 	}
1254 done:
1255 	if (dir && closedir(dir) != 0 && err == NULL)
1256 		err = got_error_from_errno("closedir");
1257 	if (err) {
1258 		free(*unique_id);
1259 		*unique_id = NULL;
1260 	}
1261 	free(path);
1262 	return err;
1263 }
1264 
1265 const struct got_error *
1266 got_repo_match_object_id_prefix(struct got_object_id **id,
1267     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1268 {
1269 	const struct got_error *err = NULL;
1270 	char *path_objects = got_repo_get_path_objects(repo);
1271 	char *object_dir = NULL;
1272 	size_t len;
1273 	int i;
1274 
1275 	*id = NULL;
1276 
1277 	for (i = 0; i < strlen(id_str_prefix); i++) {
1278 		if (isxdigit((unsigned char)id_str_prefix[i]))
1279 			continue;
1280 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1281 	}
1282 
1283 	len = strlen(id_str_prefix);
1284 	if (len >= 2) {
1285 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1286 		if (err)
1287 			goto done;
1288 		object_dir = strndup(id_str_prefix, 2);
1289 		if (object_dir == NULL) {
1290 			err = got_error_from_errno("strdup");
1291 			goto done;
1292 		}
1293 		err = match_loose_object(id, path_objects, object_dir,
1294 		    id_str_prefix, obj_type, repo);
1295 	} else if (len == 1) {
1296 		int i;
1297 		for (i = 0; i < 0xf; i++) {
1298 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1299 			    == -1) {
1300 				err = got_error_from_errno("asprintf");
1301 				goto done;
1302 			}
1303 			err = match_packed_object(id, repo, object_dir,
1304 			    obj_type);
1305 			if (err)
1306 				goto done;
1307 			err = match_loose_object(id, path_objects, object_dir,
1308 			    id_str_prefix, obj_type, repo);
1309 			if (err)
1310 				goto done;
1311 		}
1312 	} else {
1313 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1314 		goto done;
1315 	}
1316 done:
1317 	free(object_dir);
1318 	if (err) {
1319 		free(*id);
1320 		*id = NULL;
1321 	} else if (*id == NULL)
1322 		err = got_error(GOT_ERR_NO_OBJ);
1323 
1324 	return err;
1325 }
1326 
1327 const struct got_error *
1328 got_repo_match_object_id(struct got_object_id **id, char **label,
1329     const char *id_str, int obj_type, int resolve_tags,
1330     struct got_repository *repo)
1331 {
1332 	const struct got_error *err;
1333 	struct got_tag_object *tag;
1334 	struct got_reference *ref = NULL;
1335 
1336 	*id = NULL;
1337 	if (label)
1338 		*label = NULL;
1339 
1340 	if (resolve_tags) {
1341 		err = got_repo_object_match_tag(&tag, id_str, GOT_OBJ_TYPE_ANY,
1342 		    repo);
1343 		if (err == NULL) {
1344 			*id = got_object_id_dup(
1345 			    got_object_tag_get_object_id(tag));
1346 			if (*id == NULL)
1347 				err = got_error_from_errno("got_object_id_dup");
1348 			else if (label && asprintf(label, "refs/tags/%s",
1349 			    got_object_tag_get_name(tag)) == -1) {
1350 				err = got_error_from_errno("asprintf");
1351 				free(*id);
1352 				*id = NULL;
1353 			}
1354 			got_object_tag_close(tag);
1355 			return err;
1356 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1357 		    err->code != GOT_ERR_NO_OBJ)
1358 			return err;
1359 	}
1360 
1361 	err = got_repo_match_object_id_prefix(id, id_str, obj_type, repo);
1362 	if (err) {
1363 		if (err->code != GOT_ERR_BAD_OBJ_ID_STR)
1364 			return err;
1365 		err = got_ref_open(&ref, repo, id_str, 0);
1366 		if (err != NULL)
1367 			goto done;
1368 		if (label) {
1369 			*label = strdup(got_ref_get_name(ref));
1370 			if (*label == NULL) {
1371 				err = got_error_from_errno("strdup");
1372 				goto done;
1373 			}
1374 		}
1375 		err = got_ref_resolve(id, repo, ref);
1376 	} else if (label) {
1377 		err = got_object_id_str(label, *id);
1378 		if (*label == NULL) {
1379 			err = got_error_from_errno("strdup");
1380 			goto done;
1381 		}
1382 	}
1383 done:
1384 	if (ref)
1385 		got_ref_close(ref);
1386 	return err;
1387 }
1388 
1389 const struct got_error *
1390 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1391     int obj_type, struct got_repository *repo)
1392 {
1393 	const struct got_error *err;
1394 	struct got_reflist_head refs;
1395 	struct got_reflist_entry *re;
1396 	struct got_object_id *tag_id;
1397 
1398 	SIMPLEQ_INIT(&refs);
1399 	*tag = NULL;
1400 
1401 	err = got_ref_list(&refs, repo, "refs/tags", got_ref_cmp_by_name, NULL);
1402 	if (err)
1403 		return err;
1404 
1405 	SIMPLEQ_FOREACH(re, &refs, entry) {
1406 		const char *refname;
1407 		refname = got_ref_get_name(re->ref);
1408 		if (got_ref_is_symbolic(re->ref))
1409 			continue;
1410 		refname += strlen("refs/tags/");
1411 		if (strcmp(refname, name) != 0)
1412 			continue;
1413 		err = got_ref_resolve(&tag_id, repo, re->ref);
1414 		if (err)
1415 			break;
1416 		err = got_object_open_as_tag(tag, repo, tag_id);
1417 		free(tag_id);
1418 		if (err)
1419 			break;
1420 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1421 		    got_object_tag_get_object_type(*tag) == obj_type)
1422 			break;
1423 		got_object_tag_close(*tag);
1424 		*tag = NULL;
1425 	}
1426 
1427 	got_ref_list_free(&refs);
1428 	if (err == NULL && *tag == NULL)
1429 		err = got_error(GOT_ERR_NO_OBJ);
1430 	return err;
1431 }
1432 
1433 static const struct got_error *
1434 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1435     const char *name, mode_t mode, struct got_object_id *blob_id)
1436 {
1437 	const struct got_error *err = NULL;
1438 
1439 	 *new_te = NULL;
1440 
1441 	*new_te = calloc(1, sizeof(**new_te));
1442 	if (*new_te == NULL)
1443 		return got_error_from_errno("calloc");
1444 
1445 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1446 	    sizeof((*new_te)->name)) {
1447 		err = got_error(GOT_ERR_NO_SPACE);
1448 		goto done;
1449 	}
1450 
1451 	(*new_te)->mode = S_IFREG | (mode & ((S_IRWXU | S_IRWXG | S_IRWXO)));
1452 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1453 done:
1454 	if (err && *new_te) {
1455 		free(*new_te);
1456 		*new_te = NULL;
1457 	}
1458 	return err;
1459 }
1460 
1461 static const struct got_error *
1462 import_file(struct got_tree_entry **new_te, struct dirent *de,
1463     const char *path, struct got_repository *repo)
1464 {
1465 	const struct got_error *err;
1466 	struct got_object_id *blob_id = NULL;
1467 	char *filepath;
1468 	struct stat sb;
1469 
1470 	if (asprintf(&filepath, "%s%s%s", path,
1471 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1472 		return got_error_from_errno("asprintf");
1473 
1474 	if (lstat(filepath, &sb) != 0) {
1475 		err = got_error_from_errno2("lstat", path);
1476 		goto done;
1477 	}
1478 
1479 	err = got_object_blob_create(&blob_id, filepath, repo);
1480 	if (err)
1481 		goto done;
1482 
1483 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1484 	    blob_id);
1485 done:
1486 	free(filepath);
1487 	if (err)
1488 		free(blob_id);
1489 	return err;
1490 }
1491 
1492 static const struct got_error *
1493 insert_tree_entry(struct got_tree_entry *new_te,
1494     struct got_pathlist_head *paths)
1495 {
1496 	const struct got_error *err = NULL;
1497 	struct got_pathlist_entry *new_pe;
1498 
1499 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1500 	if (err)
1501 		return err;
1502 	if (new_pe == NULL)
1503 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1504 	return NULL;
1505 }
1506 
1507 static const struct got_error *write_tree(struct got_object_id **,
1508     const char *, struct got_pathlist_head *, struct got_repository *,
1509     got_repo_import_cb progress_cb, void *progress_arg);
1510 
1511 static const struct got_error *
1512 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1513     const char *path, struct got_pathlist_head *ignores,
1514     struct got_repository *repo,
1515     got_repo_import_cb progress_cb, void *progress_arg)
1516 {
1517 	const struct got_error *err;
1518 	struct got_object_id *id = NULL;
1519 	char *subdirpath;
1520 
1521 	if (asprintf(&subdirpath, "%s%s%s", path,
1522 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1523 		return got_error_from_errno("asprintf");
1524 
1525 	(*new_te) = calloc(1, sizeof(**new_te));
1526 	if (*new_te == NULL)
1527 		return got_error_from_errno("calloc");
1528 	(*new_te)->mode = S_IFDIR;
1529 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
1530 	    sizeof((*new_te)->name)) {
1531 		err = got_error(GOT_ERR_NO_SPACE);
1532 		goto done;
1533 	}
1534 	err = write_tree(&id, subdirpath, ignores,  repo,
1535 	    progress_cb, progress_arg);
1536 	if (err)
1537 		goto done;
1538 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
1539 	
1540 done:
1541 	free(id);
1542 	free(subdirpath);
1543 	if (err) {
1544 		free(*new_te);
1545 		*new_te = NULL;
1546 	}
1547 	return err;
1548 }
1549 
1550 static const struct got_error *
1551 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1552     struct got_pathlist_head *ignores, struct got_repository *repo,
1553     got_repo_import_cb progress_cb, void *progress_arg)
1554 {
1555 	const struct got_error *err = NULL;
1556 	DIR *dir;
1557 	struct dirent *de;
1558 	int nentries;
1559 	struct got_tree_entry *new_te = NULL;
1560 	struct got_pathlist_head paths;
1561 	struct got_pathlist_entry *pe;
1562 
1563 	*new_tree_id = NULL;
1564 
1565 	TAILQ_INIT(&paths);
1566 
1567 	dir = opendir(path_dir);
1568 	if (dir == NULL) {
1569 		err = got_error_from_errno2("opendir", path_dir);
1570 		goto done;
1571 	}
1572 
1573 	nentries = 0;
1574 	while ((de = readdir(dir)) != NULL) {
1575 		int ignore = 0;
1576 
1577 		if (strcmp(de->d_name, ".") == 0 ||
1578 		    strcmp(de->d_name, "..") == 0)
1579 			continue;
1580 
1581 		TAILQ_FOREACH(pe, ignores, entry) {
1582 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
1583 				ignore = 1;
1584 				break;
1585 			}
1586 		}
1587 		if (ignore)
1588 			continue;
1589 		if (de->d_type == DT_DIR) {
1590 			err = import_subdir(&new_te, de, path_dir,
1591 			    ignores, repo, progress_cb, progress_arg);
1592 			if (err) {
1593 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
1594 					goto done;
1595 				err = NULL;
1596 				continue;
1597 			}
1598 		} else if (de->d_type == DT_REG) {
1599 			err = import_file(&new_te, de, path_dir, repo);
1600 			if (err)
1601 				goto done;
1602 		} else
1603 			continue;
1604 
1605 		err = insert_tree_entry(new_te, &paths);
1606 		if (err)
1607 			goto done;
1608 		nentries++;
1609 	}
1610 
1611 	if (TAILQ_EMPTY(&paths)) {
1612 		err = got_error(GOT_ERR_NO_TREE_ENTRY);
1613 		goto done;
1614 	}
1615 
1616 	TAILQ_FOREACH(pe, &paths, entry) {
1617 		struct got_tree_entry *te = pe->data;
1618 		char *path;
1619 		if (!S_ISREG(te->mode))
1620 			continue;
1621 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
1622 			err = got_error_from_errno("asprintf");
1623 			goto done;
1624 		}
1625 		err = (*progress_cb)(progress_arg, path);
1626 		free(path);
1627 		if (err)
1628 			goto done;
1629 	}
1630 
1631 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
1632 done:
1633 	if (dir)
1634 		closedir(dir);
1635 	got_pathlist_free(&paths);
1636 	return err;
1637 }
1638 
1639 const struct got_error *
1640 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
1641     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
1642     struct got_repository *repo, got_repo_import_cb progress_cb,
1643     void *progress_arg)
1644 {
1645 	const struct got_error *err;
1646 	struct got_object_id *new_tree_id;
1647 
1648 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
1649 	    progress_cb, progress_arg);
1650 	if (err)
1651 		return err;
1652 
1653 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
1654 	    author, time(NULL), author, time(NULL), logmsg, repo);
1655 	free(new_tree_id);
1656 	return err;
1657 }