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, len;
661 	char *canonpath, *path = NULL;
662 
663 	*in_repo_path = NULL;
664 
665 	canonpath = strdup(input_path);
666 	if (canonpath == NULL) {
667 		err = got_error_from_errno("strdup");
668 		goto done;
669 	}
670 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
671 	if (err)
672 		goto done;
673 
674 	repo_abspath = got_repo_get_path(repo);
675 
676 	if (!check_disk || canonpath[0] == '\0') {
677 		path = strdup(canonpath);
678 		if (path == NULL) {
679 			err = got_error_from_errno("strdup");
680 			goto done;
681 		}
682 	} else {
683 		int is_repo_child = 0;
684 
685 		path = realpath(canonpath, NULL);
686 		if (path == NULL) {
687 			if (errno != ENOENT) {
688 				err = got_error_from_errno2("realpath",
689 				    canonpath);
690 				goto done;
691 			}
692 			/*
693 			 * Path is not on disk.
694 			 * Assume it is already relative to repository root.
695 			 */
696 			path = strdup(canonpath);
697 			if (path == NULL) {
698 				err = got_error_from_errno("strdup");
699 				goto done;
700 			}
701 		}
702 
703 		repolen = strlen(repo_abspath);
704 		len = strlen(path);
705 
706 		if (len > repolen && strncmp(path, repo_abspath, repolen) == 0)
707 			is_repo_child = 1;
708 
709 		if (strcmp(path, repo_abspath) == 0) {
710 			free(path);
711 			path = strdup("");
712 			if (path == NULL) {
713 				err = got_error_from_errno("strdup");
714 				goto done;
715 			}
716 		} else if (is_repo_child) {
717 			/* Matched an on-disk path inside repository. */
718 			if (got_repo_is_bare(repo)) {
719 				/*
720 				 * Matched an on-disk path inside repository
721 				 * database. Treat as repository-relative.
722 				 */
723 			} else {
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 			}
733 		} else {
734 			/*
735 			 * Matched unrelated on-disk path.
736 			 * Treat it as repository-relative.
737 			 */
738 		}
739 	}
740 
741 	/* Make in-repository path absolute */
742 	if (path[0] != '/') {
743 		char *abspath;
744 		if (asprintf(&abspath, "/%s", path) == -1) {
745 			err = got_error_from_errno("asprintf");
746 			goto done;
747 		}
748 		free(path);
749 		path = abspath;
750 	}
751 
752 done:
753 	free(canonpath);
754 	if (err)
755 		free(path);
756 	else
757 		*in_repo_path = path;
758 	return err;
759 }
760 
761 static const struct got_error *
762 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
763     const char *path_packidx)
764 {
765 	const struct got_error *err = NULL;
766 	int i;
767 
768 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
769 		if (repo->packidx_cache[i] == NULL)
770 			break;
771 		if (strcmp(repo->packidx_cache[i]->path_packidx,
772 		    path_packidx) == 0) {
773 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
774 		}
775 	}
776 	if (i == nitems(repo->packidx_cache)) {
777 		err = got_packidx_close(repo->packidx_cache[i - 1]);
778 		if (err)
779 			return err;
780 	}
781 
782 	/*
783 	 * Insert the new pack index at the front so it will
784 	 * be searched first in the future.
785 	 */
786 	memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
787 	    sizeof(repo->packidx_cache) -
788 	    sizeof(repo->packidx_cache[0]));
789 	repo->packidx_cache[0] = packidx;
790 
791 	return NULL;
792 }
793 
794 static int
795 is_packidx_filename(const char *name, size_t len)
796 {
797 	if (len != GOT_PACKIDX_NAMELEN)
798 		return 0;
799 
800 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
801 		return 0;
802 
803 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
804 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
805 		return 0;
806 
807 	return 1;
808 }
809 
810 const struct got_error *
811 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
812     struct got_repository *repo, struct got_object_id *id)
813 {
814 	const struct got_error *err;
815 	char *path_packdir;
816 	DIR *packdir;
817 	struct dirent *dent;
818 	char *path_packidx;
819 	int i;
820 
821 	/* Search pack index cache. */
822 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
823 		if (repo->packidx_cache[i] == NULL)
824 			break;
825 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
826 		if (*idx != -1) {
827 			*packidx = repo->packidx_cache[i];
828 			/*
829 			 * Move this cache entry to the front. Repeatedly
830 			 * searching a wrong pack index can be expensive.
831 			 */
832 			if (i > 0) {
833 				struct got_packidx *p;
834 				p = repo->packidx_cache[0];
835 				repo->packidx_cache[0] = *packidx;
836 				repo->packidx_cache[i] = p;
837 			}
838 			return NULL;
839 		}
840 	}
841 	/* No luck. Search the filesystem. */
842 
843 	path_packdir = got_repo_get_path_objects_pack(repo);
844 	if (path_packdir == NULL)
845 		return got_error_from_errno("got_repo_get_path_objects_pack");
846 
847 	packdir = opendir(path_packdir);
848 	if (packdir == NULL) {
849 		if (errno == ENOENT)
850 			err = got_error_no_obj(id);
851 		else
852 			err = got_error_from_errno2("opendir", path_packdir);
853 		goto done;
854 	}
855 
856 	while ((dent = readdir(packdir)) != NULL) {
857 		int is_cached = 0;
858 
859 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
860 			continue;
861 
862 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
863 		    dent->d_name) == -1) {
864 			err = got_error_from_errno("asprintf");
865 			goto done;
866 		}
867 
868 		for (i = 0; i < nitems(repo->packidx_cache); i++) {
869 			if (repo->packidx_cache[i] == NULL)
870 				break;
871 			if (strcmp(repo->packidx_cache[i]->path_packidx,
872 			    path_packidx) == 0) {
873 				is_cached = 1;
874 				break;
875 			}
876 		}
877 		if (is_cached) {
878 			free(path_packidx);
879 			continue; /* already searched */
880 		}
881 
882 		err = got_packidx_open(packidx, path_packidx, 0);
883 		if (err) {
884 			free(path_packidx);
885 			goto done;
886 		}
887 
888 		err = cache_packidx(repo, *packidx, path_packidx);
889 		free(path_packidx);
890 		if (err)
891 			goto done;
892 
893 		*idx = got_packidx_get_object_idx(*packidx, id);
894 		if (*idx != -1) {
895 			err = NULL; /* found the object */
896 			goto done;
897 		}
898 	}
899 
900 	err = got_error_no_obj(id);
901 done:
902 	free(path_packdir);
903 	if (packdir && closedir(packdir) != 0 && err == NULL)
904 		err = got_error_from_errno("closedir");
905 	return err;
906 }
907 
908 static const struct got_error *
909 read_packfile_hdr(int fd, struct got_packidx *packidx)
910 {
911 	const struct got_error *err = NULL;
912 	uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
913 	struct got_packfile_hdr hdr;
914 	ssize_t n;
915 
916 	n = read(fd, &hdr, sizeof(hdr));
917 	if (n < 0)
918 		return got_error_from_errno("read");
919 	if (n != sizeof(hdr))
920 		return got_error(GOT_ERR_BAD_PACKFILE);
921 
922 	if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
923 	    betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
924 	    betoh32(hdr.nobjects) != totobj)
925 		err = got_error(GOT_ERR_BAD_PACKFILE);
926 
927 	return err;
928 }
929 
930 static const struct got_error *
931 open_packfile(int *fd, const char *path_packfile, struct got_packidx *packidx)
932 {
933 	const struct got_error *err = NULL;
934 
935 	*fd = open(path_packfile, O_RDONLY | O_NOFOLLOW);
936 	if (*fd == -1)
937 		return got_error_from_errno2("open", path_packfile);
938 
939 	if (packidx) {
940 		err = read_packfile_hdr(*fd, packidx);
941 		if (err) {
942 			close(*fd);
943 			*fd = -1;
944 		}
945 	}
946 
947 	return err;
948 }
949 
950 const struct got_error *
951 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
952     const char *path_packfile, struct got_packidx *packidx)
953 {
954 	const struct got_error *err = NULL;
955 	struct got_pack *pack = NULL;
956 	struct stat sb;
957 	int i;
958 
959 	if (packp)
960 		*packp = NULL;
961 
962 	for (i = 0; i < nitems(repo->packs); i++) {
963 		pack = &repo->packs[i];
964 		if (pack->path_packfile == NULL)
965 			break;
966 		if (strcmp(pack->path_packfile, path_packfile) == 0)
967 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
968 	}
969 
970 	if (i == nitems(repo->packs) - 1) {
971 		err = got_pack_close(&repo->packs[i - 1]);
972 		if (err)
973 			return err;
974 		memmove(&repo->packs[1], &repo->packs[0],
975 		    sizeof(repo->packs) - sizeof(repo->packs[0]));
976 		i = 0;
977 	}
978 
979 	pack = &repo->packs[i];
980 
981 	pack->path_packfile = strdup(path_packfile);
982 	if (pack->path_packfile == NULL) {
983 		err = got_error_from_errno("strdup");
984 		goto done;
985 	}
986 
987 	err = open_packfile(&pack->fd, path_packfile, packidx);
988 	if (err)
989 		goto done;
990 
991 	if (fstat(pack->fd, &sb) != 0) {
992 		err = got_error_from_errno("fstat");
993 		goto done;
994 	}
995 	pack->filesize = sb.st_size;
996 
997 	pack->privsep_child = NULL;
998 
999 #ifndef GOT_PACK_NO_MMAP
1000 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1001 	    pack->fd, 0);
1002 	if (pack->map == MAP_FAILED) {
1003 		if (errno != ENOMEM) {
1004 			err = got_error_from_errno("mmap");
1005 			goto done;
1006 		}
1007 		pack->map = NULL; /* fall back to read(2) */
1008 	}
1009 #endif
1010 done:
1011 	if (err) {
1012 		if (pack) {
1013 			free(pack->path_packfile);
1014 			memset(pack, 0, sizeof(*pack));
1015 		}
1016 	} else if (packp)
1017 		*packp = pack;
1018 	return err;
1019 }
1020 
1021 struct got_pack *
1022 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1023 {
1024 	struct got_pack *pack = NULL;
1025 	int i;
1026 
1027 	for (i = 0; i < nitems(repo->packs); i++) {
1028 		pack = &repo->packs[i];
1029 		if (pack->path_packfile == NULL)
1030 			break;
1031 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1032 			return pack;
1033 	}
1034 
1035 	return NULL;
1036 }
1037 
1038 const struct got_error *
1039 got_repo_init(const char *repo_path)
1040 {
1041 	const struct got_error *err = NULL;
1042 	const char *dirnames[] = {
1043 		GOT_OBJECTS_DIR,
1044 		GOT_OBJECTS_PACK_DIR,
1045 		GOT_REFS_DIR,
1046 	};
1047 	const char *description_str = "Unnamed repository; "
1048 	    "edit this file 'description' to name the repository.";
1049 	const char *headref_str = "ref: refs/heads/main";
1050 	const char *gitconfig_str = "[core]\n"
1051 	    "\trepositoryformatversion = 0\n"
1052 	    "\tfilemode = true\n"
1053 	    "\tbare = true\n";
1054 	char *path;
1055 	int i;
1056 
1057 	if (!got_path_dir_is_empty(repo_path))
1058 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1059 
1060 	for (i = 0; i < nitems(dirnames); i++) {
1061 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1062 			return got_error_from_errno("asprintf");
1063 		}
1064 		err = got_path_mkdir(path);
1065 		free(path);
1066 		if (err)
1067 			return err;
1068 	}
1069 
1070 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1071 		return got_error_from_errno("asprintf");
1072 	err = got_path_create_file(path, description_str);
1073 	free(path);
1074 	if (err)
1075 		return err;
1076 
1077 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1078 		return got_error_from_errno("asprintf");
1079 	err = got_path_create_file(path, headref_str);
1080 	free(path);
1081 	if (err)
1082 		return err;
1083 
1084 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1085 		return got_error_from_errno("asprintf");
1086 	err = got_path_create_file(path, gitconfig_str);
1087 	free(path);
1088 	if (err)
1089 		return err;
1090 
1091 	return NULL;
1092 }
1093 
1094 static const struct got_error *
1095 match_packed_object(struct got_object_id **unique_id,
1096     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1097 {
1098 	const struct got_error *err = NULL;
1099 	char *path_packdir;
1100 	DIR *packdir;
1101 	struct dirent *dent;
1102 	char *path_packidx;
1103 	struct got_object_id_queue matched_ids;
1104 
1105 	SIMPLEQ_INIT(&matched_ids);
1106 
1107 	path_packdir = got_repo_get_path_objects_pack(repo);
1108 	if (path_packdir == NULL)
1109 		return got_error_from_errno("got_repo_get_path_objects_pack");
1110 
1111 	packdir = opendir(path_packdir);
1112 	if (packdir == NULL) {
1113 		if (errno != ENOENT)
1114 			err = got_error_from_errno2("opendir", path_packdir);
1115 		goto done;
1116 	}
1117 
1118 	while ((dent = readdir(packdir)) != NULL) {
1119 		struct got_packidx *packidx;
1120 		struct got_object_qid *qid;
1121 
1122 
1123 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
1124 			continue;
1125 
1126 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
1127 		    dent->d_name) == -1) {
1128 			err = got_error_from_errno("asprintf");
1129 			break;
1130 		}
1131 
1132 		err = got_packidx_open(&packidx, path_packidx, 0);
1133 		free(path_packidx);
1134 		if (err)
1135 			break;
1136 
1137 		err = got_packidx_match_id_str_prefix(&matched_ids,
1138 		    packidx, id_str_prefix);
1139 		if (err) {
1140 			got_packidx_close(packidx);
1141 			break;
1142 		}
1143 		err = got_packidx_close(packidx);
1144 		if (err)
1145 			break;
1146 
1147 		SIMPLEQ_FOREACH(qid, &matched_ids, entry) {
1148 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1149 				int matched_type;
1150 				err = got_object_get_type(&matched_type, repo,
1151 				    qid->id);
1152 				if (err)
1153 					goto done;
1154 				if (matched_type != obj_type)
1155 					continue;
1156 			}
1157 			if (*unique_id == NULL) {
1158 				*unique_id = got_object_id_dup(qid->id);
1159 				if (*unique_id == NULL) {
1160 					err = got_error_from_errno("malloc");
1161 					goto done;
1162 				}
1163 			} else {
1164 				if (got_object_id_cmp(*unique_id, qid->id) == 0)
1165 					continue; /* packed multiple times */
1166 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1167 				goto done;
1168 			}
1169 		}
1170 	}
1171 done:
1172 	got_object_id_queue_free(&matched_ids);
1173 	free(path_packdir);
1174 	if (packdir && closedir(packdir) != 0 && err == NULL)
1175 		err = got_error_from_errno("closedir");
1176 	if (err) {
1177 		free(*unique_id);
1178 		*unique_id = NULL;
1179 	}
1180 	return err;
1181 }
1182 
1183 static const struct got_error *
1184 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1185     const char *object_dir, const char *id_str_prefix, int obj_type,
1186     struct got_repository *repo)
1187 {
1188 	const struct got_error *err = NULL;
1189 	char *path;
1190 	DIR *dir = NULL;
1191 	struct dirent *dent;
1192 	struct got_object_id id;
1193 
1194 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1195 		err = got_error_from_errno("asprintf");
1196 		goto done;
1197 	}
1198 
1199 	dir = opendir(path);
1200 	if (dir == NULL) {
1201 		if (errno == ENOENT) {
1202 			err = NULL;
1203 			goto done;
1204 		}
1205 		err = got_error_from_errno2("opendir", path);
1206 		goto done;
1207 	}
1208 	while ((dent = readdir(dir)) != NULL) {
1209 		char *id_str;
1210 		int cmp;
1211 
1212 		if (strcmp(dent->d_name, ".") == 0 ||
1213 		    strcmp(dent->d_name, "..") == 0)
1214 			continue;
1215 
1216 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1217 			err = got_error_from_errno("asprintf");
1218 			goto done;
1219 		}
1220 
1221 		if (!got_parse_sha1_digest(id.sha1, id_str))
1222 			continue;
1223 
1224 		/*
1225 		 * Directory entries do not necessarily appear in
1226 		 * sorted order, so we must iterate over all of them.
1227 		 */
1228 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1229 		if (cmp != 0) {
1230 			free(id_str);
1231 			continue;
1232 		}
1233 
1234 		if (*unique_id == NULL) {
1235 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1236 				int matched_type;
1237 				err = got_object_get_type(&matched_type, repo,
1238 				    &id);
1239 				if (err)
1240 					goto done;
1241 				if (matched_type != obj_type)
1242 					continue;
1243 			}
1244 			*unique_id = got_object_id_dup(&id);
1245 			if (*unique_id == NULL) {
1246 				err = got_error_from_errno("got_object_id_dup");
1247 				free(id_str);
1248 				goto done;
1249 			}
1250 		} else {
1251 			if (got_object_id_cmp(*unique_id, &id) == 0)
1252 				continue; /* both packed and loose */
1253 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1254 			free(id_str);
1255 			goto done;
1256 		}
1257 	}
1258 done:
1259 	if (dir && closedir(dir) != 0 && err == NULL)
1260 		err = got_error_from_errno("closedir");
1261 	if (err) {
1262 		free(*unique_id);
1263 		*unique_id = NULL;
1264 	}
1265 	free(path);
1266 	return err;
1267 }
1268 
1269 const struct got_error *
1270 got_repo_match_object_id_prefix(struct got_object_id **id,
1271     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1272 {
1273 	const struct got_error *err = NULL;
1274 	char *path_objects = got_repo_get_path_objects(repo);
1275 	char *object_dir = NULL;
1276 	size_t len;
1277 	int i;
1278 
1279 	*id = NULL;
1280 
1281 	for (i = 0; i < strlen(id_str_prefix); i++) {
1282 		if (isxdigit((unsigned char)id_str_prefix[i]))
1283 			continue;
1284 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1285 	}
1286 
1287 	len = strlen(id_str_prefix);
1288 	if (len >= 2) {
1289 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1290 		if (err)
1291 			goto done;
1292 		object_dir = strndup(id_str_prefix, 2);
1293 		if (object_dir == NULL) {
1294 			err = got_error_from_errno("strdup");
1295 			goto done;
1296 		}
1297 		err = match_loose_object(id, path_objects, object_dir,
1298 		    id_str_prefix, obj_type, repo);
1299 	} else if (len == 1) {
1300 		int i;
1301 		for (i = 0; i < 0xf; i++) {
1302 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1303 			    == -1) {
1304 				err = got_error_from_errno("asprintf");
1305 				goto done;
1306 			}
1307 			err = match_packed_object(id, repo, object_dir,
1308 			    obj_type);
1309 			if (err)
1310 				goto done;
1311 			err = match_loose_object(id, path_objects, object_dir,
1312 			    id_str_prefix, obj_type, repo);
1313 			if (err)
1314 				goto done;
1315 		}
1316 	} else {
1317 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1318 		goto done;
1319 	}
1320 done:
1321 	free(object_dir);
1322 	if (err) {
1323 		free(*id);
1324 		*id = NULL;
1325 	} else if (*id == NULL)
1326 		err = got_error(GOT_ERR_NO_OBJ);
1327 
1328 	return err;
1329 }
1330 
1331 const struct got_error *
1332 got_repo_match_object_id(struct got_object_id **id, char **label,
1333     const char *id_str, int obj_type, int resolve_tags,
1334     struct got_repository *repo)
1335 {
1336 	const struct got_error *err;
1337 	struct got_tag_object *tag;
1338 	struct got_reference *ref = NULL;
1339 
1340 	*id = NULL;
1341 	if (label)
1342 		*label = NULL;
1343 
1344 	if (resolve_tags) {
1345 		err = got_repo_object_match_tag(&tag, id_str, GOT_OBJ_TYPE_ANY,
1346 		    repo);
1347 		if (err == NULL) {
1348 			*id = got_object_id_dup(
1349 			    got_object_tag_get_object_id(tag));
1350 			if (*id == NULL)
1351 				err = got_error_from_errno("got_object_id_dup");
1352 			else if (label && asprintf(label, "refs/tags/%s",
1353 			    got_object_tag_get_name(tag)) == -1) {
1354 				err = got_error_from_errno("asprintf");
1355 				free(*id);
1356 				*id = NULL;
1357 			}
1358 			got_object_tag_close(tag);
1359 			return err;
1360 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1361 		    err->code != GOT_ERR_NO_OBJ)
1362 			return err;
1363 	}
1364 
1365 	err = got_repo_match_object_id_prefix(id, id_str, obj_type, repo);
1366 	if (err) {
1367 		if (err->code != GOT_ERR_BAD_OBJ_ID_STR)
1368 			return err;
1369 		err = got_ref_open(&ref, repo, id_str, 0);
1370 		if (err != NULL)
1371 			goto done;
1372 		if (label) {
1373 			*label = strdup(got_ref_get_name(ref));
1374 			if (*label == NULL) {
1375 				err = got_error_from_errno("strdup");
1376 				goto done;
1377 			}
1378 		}
1379 		err = got_ref_resolve(id, repo, ref);
1380 	} else if (label) {
1381 		err = got_object_id_str(label, *id);
1382 		if (*label == NULL) {
1383 			err = got_error_from_errno("strdup");
1384 			goto done;
1385 		}
1386 	}
1387 done:
1388 	if (ref)
1389 		got_ref_close(ref);
1390 	return err;
1391 }
1392 
1393 const struct got_error *
1394 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1395     int obj_type, struct got_repository *repo)
1396 {
1397 	const struct got_error *err;
1398 	struct got_reflist_head refs;
1399 	struct got_reflist_entry *re;
1400 	struct got_object_id *tag_id;
1401 
1402 	SIMPLEQ_INIT(&refs);
1403 	*tag = NULL;
1404 
1405 	err = got_ref_list(&refs, repo, "refs/tags", got_ref_cmp_by_name, NULL);
1406 	if (err)
1407 		return err;
1408 
1409 	SIMPLEQ_FOREACH(re, &refs, entry) {
1410 		const char *refname;
1411 		refname = got_ref_get_name(re->ref);
1412 		if (got_ref_is_symbolic(re->ref))
1413 			continue;
1414 		refname += strlen("refs/tags/");
1415 		if (strcmp(refname, name) != 0)
1416 			continue;
1417 		err = got_ref_resolve(&tag_id, repo, re->ref);
1418 		if (err)
1419 			break;
1420 		err = got_object_open_as_tag(tag, repo, tag_id);
1421 		free(tag_id);
1422 		if (err)
1423 			break;
1424 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1425 		    got_object_tag_get_object_type(*tag) == obj_type)
1426 			break;
1427 		got_object_tag_close(*tag);
1428 		*tag = NULL;
1429 	}
1430 
1431 	got_ref_list_free(&refs);
1432 	if (err == NULL && *tag == NULL)
1433 		err = got_error(GOT_ERR_NO_OBJ);
1434 	return err;
1435 }
1436 
1437 static const struct got_error *
1438 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1439     const char *name, mode_t mode, struct got_object_id *blob_id)
1440 {
1441 	const struct got_error *err = NULL;
1442 
1443 	 *new_te = NULL;
1444 
1445 	*new_te = calloc(1, sizeof(**new_te));
1446 	if (*new_te == NULL)
1447 		return got_error_from_errno("calloc");
1448 
1449 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1450 	    sizeof((*new_te)->name)) {
1451 		err = got_error(GOT_ERR_NO_SPACE);
1452 		goto done;
1453 	}
1454 
1455 	(*new_te)->mode = S_IFREG | (mode & ((S_IRWXU | S_IRWXG | S_IRWXO)));
1456 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1457 done:
1458 	if (err && *new_te) {
1459 		free(*new_te);
1460 		*new_te = NULL;
1461 	}
1462 	return err;
1463 }
1464 
1465 static const struct got_error *
1466 import_file(struct got_tree_entry **new_te, struct dirent *de,
1467     const char *path, struct got_repository *repo)
1468 {
1469 	const struct got_error *err;
1470 	struct got_object_id *blob_id = NULL;
1471 	char *filepath;
1472 	struct stat sb;
1473 
1474 	if (asprintf(&filepath, "%s%s%s", path,
1475 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1476 		return got_error_from_errno("asprintf");
1477 
1478 	if (lstat(filepath, &sb) != 0) {
1479 		err = got_error_from_errno2("lstat", path);
1480 		goto done;
1481 	}
1482 
1483 	err = got_object_blob_create(&blob_id, filepath, repo);
1484 	if (err)
1485 		goto done;
1486 
1487 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1488 	    blob_id);
1489 done:
1490 	free(filepath);
1491 	if (err)
1492 		free(blob_id);
1493 	return err;
1494 }
1495 
1496 static const struct got_error *
1497 insert_tree_entry(struct got_tree_entry *new_te,
1498     struct got_pathlist_head *paths)
1499 {
1500 	const struct got_error *err = NULL;
1501 	struct got_pathlist_entry *new_pe;
1502 
1503 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1504 	if (err)
1505 		return err;
1506 	if (new_pe == NULL)
1507 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1508 	return NULL;
1509 }
1510 
1511 static const struct got_error *write_tree(struct got_object_id **,
1512     const char *, struct got_pathlist_head *, struct got_repository *,
1513     got_repo_import_cb progress_cb, void *progress_arg);
1514 
1515 static const struct got_error *
1516 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1517     const char *path, struct got_pathlist_head *ignores,
1518     struct got_repository *repo,
1519     got_repo_import_cb progress_cb, void *progress_arg)
1520 {
1521 	const struct got_error *err;
1522 	struct got_object_id *id = NULL;
1523 	char *subdirpath;
1524 
1525 	if (asprintf(&subdirpath, "%s%s%s", path,
1526 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1527 		return got_error_from_errno("asprintf");
1528 
1529 	(*new_te) = calloc(1, sizeof(**new_te));
1530 	if (*new_te == NULL)
1531 		return got_error_from_errno("calloc");
1532 	(*new_te)->mode = S_IFDIR;
1533 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
1534 	    sizeof((*new_te)->name)) {
1535 		err = got_error(GOT_ERR_NO_SPACE);
1536 		goto done;
1537 	}
1538 	err = write_tree(&id, subdirpath, ignores,  repo,
1539 	    progress_cb, progress_arg);
1540 	if (err)
1541 		goto done;
1542 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
1543 	
1544 done:
1545 	free(id);
1546 	free(subdirpath);
1547 	if (err) {
1548 		free(*new_te);
1549 		*new_te = NULL;
1550 	}
1551 	return err;
1552 }
1553 
1554 static const struct got_error *
1555 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1556     struct got_pathlist_head *ignores, struct got_repository *repo,
1557     got_repo_import_cb progress_cb, void *progress_arg)
1558 {
1559 	const struct got_error *err = NULL;
1560 	DIR *dir;
1561 	struct dirent *de;
1562 	int nentries;
1563 	struct got_tree_entry *new_te = NULL;
1564 	struct got_pathlist_head paths;
1565 	struct got_pathlist_entry *pe;
1566 
1567 	*new_tree_id = NULL;
1568 
1569 	TAILQ_INIT(&paths);
1570 
1571 	dir = opendir(path_dir);
1572 	if (dir == NULL) {
1573 		err = got_error_from_errno2("opendir", path_dir);
1574 		goto done;
1575 	}
1576 
1577 	nentries = 0;
1578 	while ((de = readdir(dir)) != NULL) {
1579 		int ignore = 0;
1580 
1581 		if (strcmp(de->d_name, ".") == 0 ||
1582 		    strcmp(de->d_name, "..") == 0)
1583 			continue;
1584 
1585 		TAILQ_FOREACH(pe, ignores, entry) {
1586 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
1587 				ignore = 1;
1588 				break;
1589 			}
1590 		}
1591 		if (ignore)
1592 			continue;
1593 		if (de->d_type == DT_DIR) {
1594 			err = import_subdir(&new_te, de, path_dir,
1595 			    ignores, repo, progress_cb, progress_arg);
1596 			if (err) {
1597 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
1598 					goto done;
1599 				err = NULL;
1600 				continue;
1601 			}
1602 		} else if (de->d_type == DT_REG) {
1603 			err = import_file(&new_te, de, path_dir, repo);
1604 			if (err)
1605 				goto done;
1606 		} else
1607 			continue;
1608 
1609 		err = insert_tree_entry(new_te, &paths);
1610 		if (err)
1611 			goto done;
1612 		nentries++;
1613 	}
1614 
1615 	if (TAILQ_EMPTY(&paths)) {
1616 		err = got_error(GOT_ERR_NO_TREE_ENTRY);
1617 		goto done;
1618 	}
1619 
1620 	TAILQ_FOREACH(pe, &paths, entry) {
1621 		struct got_tree_entry *te = pe->data;
1622 		char *path;
1623 		if (!S_ISREG(te->mode))
1624 			continue;
1625 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
1626 			err = got_error_from_errno("asprintf");
1627 			goto done;
1628 		}
1629 		err = (*progress_cb)(progress_arg, path);
1630 		free(path);
1631 		if (err)
1632 			goto done;
1633 	}
1634 
1635 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
1636 done:
1637 	if (dir)
1638 		closedir(dir);
1639 	got_pathlist_free(&paths);
1640 	return err;
1641 }
1642 
1643 const struct got_error *
1644 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
1645     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
1646     struct got_repository *repo, got_repo_import_cb progress_cb,
1647     void *progress_arg)
1648 {
1649 	const struct got_error *err;
1650 	struct got_object_id *new_tree_id;
1651 
1652 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
1653 	    progress_cb, progress_arg);
1654 	if (err)
1655 		return err;
1656 
1657 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
1658 	    author, time(NULL), author, time(NULL), logmsg, repo);
1659 	free(new_tree_id);
1660 	return err;
1661 }