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