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 	if (remotes)
379 		*remotes = NULL;
380 	if (nremotes)
381 		*nremotes = 0;
382 	if (gitconfig_owner)
383 		*gitconfig_owner = NULL;
384 
385 	fd = open(gitconfig_path, O_RDONLY);
386 	if (fd == -1) {
387 		if (errno == ENOENT)
388 			return NULL;
389 		return got_error_from_errno2("open", gitconfig_path);
390 	}
391 
392 	ibuf = calloc(1, sizeof(*ibuf));
393 	if (ibuf == NULL) {
394 		err = got_error_from_errno("calloc");
395 		goto done;
396 	}
397 
398 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
399 		err = got_error_from_errno("socketpair");
400 		goto done;
401 	}
402 
403 	pid = fork();
404 	if (pid == -1) {
405 		err = got_error_from_errno("fork");
406 		goto done;
407 	} else if (pid == 0) {
408 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
409 		    gitconfig_path);
410 		/* not reached */
411 	}
412 
413 	if (close(imsg_fds[1]) == -1) {
414 		err = got_error_from_errno("close");
415 		goto done;
416 	}
417 	imsg_fds[1] = -1;
418 	imsg_init(ibuf, imsg_fds[0]);
419 
420 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
421 	if (err)
422 		goto done;
423 	fd = -1;
424 
425 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
426 	if (err)
427 		goto done;
428 
429 	err = got_privsep_recv_gitconfig_int(
430 	    gitconfig_repository_format_version, ibuf);
431 	if (err)
432 		goto done;
433 
434 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
435 	if (err)
436 		goto done;
437 
438 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
439 	if (err)
440 		goto done;
441 
442 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
443 	if (err)
444 		goto done;
445 
446 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
447 	if (err)
448 		goto done;
449 
450 	if (remotes && nremotes) {
451 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
452 		if (err)
453 			goto done;
454 
455 		err = got_privsep_recv_gitconfig_remotes(remotes,
456 		    nremotes, ibuf);
457 		if (err)
458 			goto done;
459 	}
460 
461 	if (gitconfig_owner) {
462 		err = got_privsep_send_gitconfig_owner_req(ibuf);
463 		if (err)
464 			goto done;
465 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
466 		if (err)
467 			goto done;
468 	}
469 
470 	imsg_clear(ibuf);
471 	err = got_privsep_send_stop(imsg_fds[0]);
472 	child_err = got_privsep_wait_for_child(pid);
473 	if (child_err && err == NULL)
474 		err = child_err;
475 done:
476 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
477 		err = got_error_from_errno("close");
478 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
479 		err = got_error_from_errno("close");
480 	if (fd != -1 && close(fd) == -1 && err == NULL)
481 		err = got_error_from_errno2("close", gitconfig_path);
482 	free(ibuf);
483 	return err;
484 }
485 
486 static const struct got_error *
487 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
488 {
489 	const struct got_error *err = NULL;
490 	char *repo_gitconfig_path = NULL;
491 
492 	if (global_gitconfig_path) {
493 		/* Read settings from ~/.gitconfig. */
494 		int dummy_repo_version;
495 		err = parse_gitconfig_file(&dummy_repo_version,
496 		    &repo->global_gitconfig_author_name,
497 		    &repo->global_gitconfig_author_email,
498 		    NULL, NULL, NULL, global_gitconfig_path);
499 		if (err)
500 			return err;
501 	}
502 
503 	/* Read repository's .git/config file. */
504 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
505 	if (repo_gitconfig_path == NULL)
506 		return got_error_from_errno("got_repo_get_path_gitconfig");
507 
508 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
509 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
510 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
511 	    &repo->gitconfig_owner, repo_gitconfig_path);
512 	if (err)
513 		goto done;
514 done:
515 	free(repo_gitconfig_path);
516 	return err;
517 }
518 
519 const struct got_error *
520 got_repo_open(struct got_repository **repop, const char *path,
521     const char *global_gitconfig_path)
522 {
523 	struct got_repository *repo = NULL;
524 	const struct got_error *err = NULL;
525 	char *abspath;
526 	int i, tried_root = 0;
527 
528 	*repop = NULL;
529 
530 	if (got_path_is_absolute(path))
531 		abspath = strdup(path);
532 	else
533 		abspath = got_path_get_absolute(path);
534 	if (abspath == NULL)
535 		return got_error_path(path, GOT_ERR_BAD_PATH);
536 
537 	repo = calloc(1, sizeof(*repo));
538 	if (repo == NULL) {
539 		err = got_error_from_errno("calloc");
540 		goto done;
541 	}
542 
543 	for (i = 0; i < nitems(repo->privsep_children); i++) {
544 		memset(&repo->privsep_children[i], 0,
545 		    sizeof(repo->privsep_children[0]));
546 		repo->privsep_children[i].imsg_fd = -1;
547 	}
548 
549 	err = got_object_cache_init(&repo->objcache,
550 	    GOT_OBJECT_CACHE_TYPE_OBJ);
551 	if (err)
552 		goto done;
553 	err = got_object_cache_init(&repo->treecache,
554 	    GOT_OBJECT_CACHE_TYPE_TREE);
555 	if (err)
556 		goto done;
557 	err = got_object_cache_init(&repo->commitcache,
558 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
559 	if (err)
560 		goto done;
561 	err = got_object_cache_init(&repo->tagcache,
562 	    GOT_OBJECT_CACHE_TYPE_TAG);
563 	if (err)
564 		goto done;
565 
566 	path = realpath(abspath, NULL);
567 	if (path == NULL) {
568 		err = got_error_from_errno2("realpath", abspath);
569 		goto done;
570 	}
571 
572 	do {
573 		err = open_repo(repo, path);
574 		if (err == NULL)
575 			break;
576 		if (err->code != GOT_ERR_NOT_GIT_REPO)
577 			break;
578 		if (path[0] == '/' && path[1] == '\0') {
579 			if (tried_root) {
580 				err = got_error(GOT_ERR_NOT_GIT_REPO);
581 				goto done;
582 			}
583 			tried_root = 1;
584 		}
585 		path = dirname(path);
586 		if (path == NULL) {
587 			err = got_error_from_errno2("dirname", path);
588 			goto done;
589 		}
590 	} while (path);
591 
592 	err = read_gitconfig(repo, global_gitconfig_path);
593 	if (err)
594 		goto done;
595 	if (repo->gitconfig_repository_format_version != 0)
596 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
597 done:
598 	if (err)
599 		got_repo_close(repo);
600 	else
601 		*repop = repo;
602 	free(abspath);
603 	return err;
604 }
605 
606 const struct got_error *
607 got_repo_close(struct got_repository *repo)
608 {
609 	const struct got_error *err = NULL, *child_err;
610 	int i;
611 
612 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
613 		if (repo->packidx_cache[i] == NULL)
614 			break;
615 		got_packidx_close(repo->packidx_cache[i]);
616 	}
617 
618 	for (i = 0; i < nitems(repo->packs); i++) {
619 		if (repo->packs[i].path_packfile == NULL)
620 			break;
621 		got_pack_close(&repo->packs[i]);
622 	}
623 
624 	free(repo->path);
625 	free(repo->path_git_dir);
626 
627 	got_object_cache_close(&repo->objcache);
628 	got_object_cache_close(&repo->treecache);
629 	got_object_cache_close(&repo->commitcache);
630 	got_object_cache_close(&repo->tagcache);
631 
632 	for (i = 0; i < nitems(repo->privsep_children); i++) {
633 		if (repo->privsep_children[i].imsg_fd == -1)
634 			continue;
635 		imsg_clear(repo->privsep_children[i].ibuf);
636 		free(repo->privsep_children[i].ibuf);
637 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
638 		child_err = got_privsep_wait_for_child(
639 		    repo->privsep_children[i].pid);
640 		if (child_err && err == NULL)
641 			err = child_err;
642 		if (close(repo->privsep_children[i].imsg_fd) != 0 &&
643 		    err == NULL)
644 			err = got_error_from_errno("close");
645 	}
646 
647 	free(repo->gitconfig_author_name);
648 	free(repo->gitconfig_author_email);
649 	for (i = 0; i < repo->ngitconfig_remotes; i++) {
650 		free(repo->gitconfig_remotes[i].name);
651 		free(repo->gitconfig_remotes[i].url);
652 	}
653 	free(repo->gitconfig_remotes);
654 	free(repo);
655 
656 	return err;
657 }
658 
659 const struct got_error *
660 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
661     const char *input_path, int check_disk)
662 {
663 	const struct got_error *err = NULL;
664 	const char *repo_abspath = NULL;
665 	size_t repolen, len;
666 	char *canonpath, *path = NULL;
667 
668 	*in_repo_path = NULL;
669 
670 	canonpath = strdup(input_path);
671 	if (canonpath == NULL) {
672 		err = got_error_from_errno("strdup");
673 		goto done;
674 	}
675 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
676 	if (err)
677 		goto done;
678 
679 	repo_abspath = got_repo_get_path(repo);
680 
681 	if (!check_disk || canonpath[0] == '\0') {
682 		path = strdup(canonpath);
683 		if (path == NULL) {
684 			err = got_error_from_errno("strdup");
685 			goto done;
686 		}
687 	} else {
688 		path = realpath(canonpath, NULL);
689 		if (path == NULL) {
690 			if (errno != ENOENT) {
691 				err = got_error_from_errno2("realpath",
692 				    canonpath);
693 				goto done;
694 			}
695 			/*
696 			 * Path is not on disk.
697 			 * Assume it is already relative to repository root.
698 			 */
699 			path = strdup(canonpath);
700 			if (path == NULL) {
701 				err = got_error_from_errno("strdup");
702 				goto done;
703 			}
704 		}
705 
706 		repolen = strlen(repo_abspath);
707 		len = strlen(path);
708 
709 
710 		if (strcmp(path, repo_abspath) == 0) {
711 			free(path);
712 			path = strdup("");
713 			if (path == NULL) {
714 				err = got_error_from_errno("strdup");
715 				goto done;
716 			}
717 		} else if (len > repolen &&
718 		    got_path_is_child(path, repo_abspath, repolen)) {
719 			/* Matched an on-disk path inside repository. */
720 			if (got_repo_is_bare(repo)) {
721 				/*
722 				 * Matched an on-disk path inside repository
723 				 * database. Treat as repository-relative.
724 				 */
725 			} else {
726 				char *child;
727 				/* Strip common prefix with repository path. */
728 				err = got_path_skip_common_ancestor(&child,
729 				    repo_abspath, path);
730 				if (err)
731 					goto done;
732 				free(path);
733 				path = child;
734 			}
735 		} else {
736 			/*
737 			 * Matched unrelated on-disk path.
738 			 * Treat it as repository-relative.
739 			 */
740 		}
741 	}
742 
743 	/* Make in-repository path absolute */
744 	if (path[0] != '/') {
745 		char *abspath;
746 		if (asprintf(&abspath, "/%s", path) == -1) {
747 			err = got_error_from_errno("asprintf");
748 			goto done;
749 		}
750 		free(path);
751 		path = abspath;
752 	}
753 
754 done:
755 	free(canonpath);
756 	if (err)
757 		free(path);
758 	else
759 		*in_repo_path = path;
760 	return err;
761 }
762 
763 static const struct got_error *
764 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
765     const char *path_packidx)
766 {
767 	const struct got_error *err = NULL;
768 	int i;
769 
770 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
771 		if (repo->packidx_cache[i] == NULL)
772 			break;
773 		if (strcmp(repo->packidx_cache[i]->path_packidx,
774 		    path_packidx) == 0) {
775 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
776 		}
777 	}
778 	if (i == nitems(repo->packidx_cache)) {
779 		err = got_packidx_close(repo->packidx_cache[i - 1]);
780 		if (err)
781 			return err;
782 	}
783 
784 	/*
785 	 * Insert the new pack index at the front so it will
786 	 * be searched first in the future.
787 	 */
788 	memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
789 	    sizeof(repo->packidx_cache) -
790 	    sizeof(repo->packidx_cache[0]));
791 	repo->packidx_cache[0] = packidx;
792 
793 	return NULL;
794 }
795 
796 static int
797 is_packidx_filename(const char *name, size_t len)
798 {
799 	if (len != GOT_PACKIDX_NAMELEN)
800 		return 0;
801 
802 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
803 		return 0;
804 
805 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
806 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
807 		return 0;
808 
809 	return 1;
810 }
811 
812 const struct got_error *
813 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
814     struct got_repository *repo, struct got_object_id *id)
815 {
816 	const struct got_error *err;
817 	char *path_packdir;
818 	DIR *packdir;
819 	struct dirent *dent;
820 	char *path_packidx;
821 	int i;
822 
823 	/* Search pack index cache. */
824 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
825 		if (repo->packidx_cache[i] == NULL)
826 			break;
827 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
828 		if (*idx != -1) {
829 			*packidx = repo->packidx_cache[i];
830 			/*
831 			 * Move this cache entry to the front. Repeatedly
832 			 * searching a wrong pack index can be expensive.
833 			 */
834 			if (i > 0) {
835 				struct got_packidx *p;
836 				p = repo->packidx_cache[0];
837 				repo->packidx_cache[0] = *packidx;
838 				repo->packidx_cache[i] = p;
839 			}
840 			return NULL;
841 		}
842 	}
843 	/* No luck. Search the filesystem. */
844 
845 	path_packdir = got_repo_get_path_objects_pack(repo);
846 	if (path_packdir == NULL)
847 		return got_error_from_errno("got_repo_get_path_objects_pack");
848 
849 	packdir = opendir(path_packdir);
850 	if (packdir == NULL) {
851 		if (errno == ENOENT)
852 			err = got_error_no_obj(id);
853 		else
854 			err = got_error_from_errno2("opendir", path_packdir);
855 		goto done;
856 	}
857 
858 	while ((dent = readdir(packdir)) != NULL) {
859 		int is_cached = 0;
860 
861 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
862 			continue;
863 
864 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
865 		    dent->d_name) == -1) {
866 			err = got_error_from_errno("asprintf");
867 			goto done;
868 		}
869 
870 		for (i = 0; i < nitems(repo->packidx_cache); i++) {
871 			if (repo->packidx_cache[i] == NULL)
872 				break;
873 			if (strcmp(repo->packidx_cache[i]->path_packidx,
874 			    path_packidx) == 0) {
875 				is_cached = 1;
876 				break;
877 			}
878 		}
879 		if (is_cached) {
880 			free(path_packidx);
881 			continue; /* already searched */
882 		}
883 
884 		err = got_packidx_open(packidx, path_packidx, 0);
885 		if (err) {
886 			free(path_packidx);
887 			goto done;
888 		}
889 
890 		err = cache_packidx(repo, *packidx, path_packidx);
891 		free(path_packidx);
892 		if (err)
893 			goto done;
894 
895 		*idx = got_packidx_get_object_idx(*packidx, id);
896 		if (*idx != -1) {
897 			err = NULL; /* found the object */
898 			goto done;
899 		}
900 	}
901 
902 	err = got_error_no_obj(id);
903 done:
904 	free(path_packdir);
905 	if (packdir && closedir(packdir) != 0 && err == NULL)
906 		err = got_error_from_errno("closedir");
907 	return err;
908 }
909 
910 static const struct got_error *
911 read_packfile_hdr(int fd, struct got_packidx *packidx)
912 {
913 	const struct got_error *err = NULL;
914 	uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
915 	struct got_packfile_hdr hdr;
916 	ssize_t n;
917 
918 	n = read(fd, &hdr, sizeof(hdr));
919 	if (n < 0)
920 		return got_error_from_errno("read");
921 	if (n != sizeof(hdr))
922 		return got_error(GOT_ERR_BAD_PACKFILE);
923 
924 	if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
925 	    betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
926 	    betoh32(hdr.nobjects) != totobj)
927 		err = got_error(GOT_ERR_BAD_PACKFILE);
928 
929 	return err;
930 }
931 
932 static const struct got_error *
933 open_packfile(int *fd, const char *path_packfile, struct got_packidx *packidx)
934 {
935 	const struct got_error *err = NULL;
936 
937 	*fd = open(path_packfile, O_RDONLY | O_NOFOLLOW);
938 	if (*fd == -1)
939 		return got_error_from_errno2("open", path_packfile);
940 
941 	if (packidx) {
942 		err = read_packfile_hdr(*fd, packidx);
943 		if (err) {
944 			close(*fd);
945 			*fd = -1;
946 		}
947 	}
948 
949 	return err;
950 }
951 
952 const struct got_error *
953 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
954     const char *path_packfile, struct got_packidx *packidx)
955 {
956 	const struct got_error *err = NULL;
957 	struct got_pack *pack = NULL;
958 	struct stat sb;
959 	int i;
960 
961 	if (packp)
962 		*packp = NULL;
963 
964 	for (i = 0; i < nitems(repo->packs); i++) {
965 		pack = &repo->packs[i];
966 		if (pack->path_packfile == NULL)
967 			break;
968 		if (strcmp(pack->path_packfile, path_packfile) == 0)
969 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
970 	}
971 
972 	if (i == nitems(repo->packs) - 1) {
973 		err = got_pack_close(&repo->packs[i - 1]);
974 		if (err)
975 			return err;
976 		memmove(&repo->packs[1], &repo->packs[0],
977 		    sizeof(repo->packs) - sizeof(repo->packs[0]));
978 		i = 0;
979 	}
980 
981 	pack = &repo->packs[i];
982 
983 	pack->path_packfile = strdup(path_packfile);
984 	if (pack->path_packfile == NULL) {
985 		err = got_error_from_errno("strdup");
986 		goto done;
987 	}
988 
989 	err = open_packfile(&pack->fd, path_packfile, packidx);
990 	if (err)
991 		goto done;
992 
993 	if (fstat(pack->fd, &sb) != 0) {
994 		err = got_error_from_errno("fstat");
995 		goto done;
996 	}
997 	pack->filesize = sb.st_size;
998 
999 	pack->privsep_child = NULL;
1000 
1001 #ifndef GOT_PACK_NO_MMAP
1002 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1003 	    pack->fd, 0);
1004 	if (pack->map == MAP_FAILED) {
1005 		if (errno != ENOMEM) {
1006 			err = got_error_from_errno("mmap");
1007 			goto done;
1008 		}
1009 		pack->map = NULL; /* fall back to read(2) */
1010 	}
1011 #endif
1012 done:
1013 	if (err) {
1014 		if (pack) {
1015 			free(pack->path_packfile);
1016 			memset(pack, 0, sizeof(*pack));
1017 		}
1018 	} else if (packp)
1019 		*packp = pack;
1020 	return err;
1021 }
1022 
1023 struct got_pack *
1024 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1025 {
1026 	struct got_pack *pack = NULL;
1027 	int i;
1028 
1029 	for (i = 0; i < nitems(repo->packs); i++) {
1030 		pack = &repo->packs[i];
1031 		if (pack->path_packfile == NULL)
1032 			break;
1033 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1034 			return pack;
1035 	}
1036 
1037 	return NULL;
1038 }
1039 
1040 const struct got_error *
1041 got_repo_init(const char *repo_path)
1042 {
1043 	const struct got_error *err = NULL;
1044 	const char *dirnames[] = {
1045 		GOT_OBJECTS_DIR,
1046 		GOT_OBJECTS_PACK_DIR,
1047 		GOT_REFS_DIR,
1048 	};
1049 	const char *description_str = "Unnamed repository; "
1050 	    "edit this file 'description' to name the repository.";
1051 	const char *headref_str = "ref: refs/heads/main";
1052 	const char *gitconfig_str = "[core]\n"
1053 	    "\trepositoryformatversion = 0\n"
1054 	    "\tfilemode = true\n"
1055 	    "\tbare = true\n";
1056 	char *path;
1057 	int i;
1058 
1059 	if (!got_path_dir_is_empty(repo_path))
1060 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1061 
1062 	for (i = 0; i < nitems(dirnames); i++) {
1063 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1064 			return got_error_from_errno("asprintf");
1065 		}
1066 		err = got_path_mkdir(path);
1067 		free(path);
1068 		if (err)
1069 			return err;
1070 	}
1071 
1072 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1073 		return got_error_from_errno("asprintf");
1074 	err = got_path_create_file(path, description_str);
1075 	free(path);
1076 	if (err)
1077 		return err;
1078 
1079 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1080 		return got_error_from_errno("asprintf");
1081 	err = got_path_create_file(path, headref_str);
1082 	free(path);
1083 	if (err)
1084 		return err;
1085 
1086 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1087 		return got_error_from_errno("asprintf");
1088 	err = got_path_create_file(path, gitconfig_str);
1089 	free(path);
1090 	if (err)
1091 		return err;
1092 
1093 	return NULL;
1094 }
1095 
1096 static const struct got_error *
1097 match_packed_object(struct got_object_id **unique_id,
1098     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1099 {
1100 	const struct got_error *err = NULL;
1101 	char *path_packdir;
1102 	DIR *packdir;
1103 	struct dirent *dent;
1104 	char *path_packidx;
1105 	struct got_object_id_queue matched_ids;
1106 
1107 	SIMPLEQ_INIT(&matched_ids);
1108 
1109 	path_packdir = got_repo_get_path_objects_pack(repo);
1110 	if (path_packdir == NULL)
1111 		return got_error_from_errno("got_repo_get_path_objects_pack");
1112 
1113 	packdir = opendir(path_packdir);
1114 	if (packdir == NULL) {
1115 		if (errno != ENOENT)
1116 			err = got_error_from_errno2("opendir", path_packdir);
1117 		goto done;
1118 	}
1119 
1120 	while ((dent = readdir(packdir)) != NULL) {
1121 		struct got_packidx *packidx;
1122 		struct got_object_qid *qid;
1123 
1124 
1125 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
1126 			continue;
1127 
1128 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
1129 		    dent->d_name) == -1) {
1130 			err = got_error_from_errno("asprintf");
1131 			break;
1132 		}
1133 
1134 		err = got_packidx_open(&packidx, path_packidx, 0);
1135 		free(path_packidx);
1136 		if (err)
1137 			break;
1138 
1139 		err = got_packidx_match_id_str_prefix(&matched_ids,
1140 		    packidx, id_str_prefix);
1141 		if (err) {
1142 			got_packidx_close(packidx);
1143 			break;
1144 		}
1145 		err = got_packidx_close(packidx);
1146 		if (err)
1147 			break;
1148 
1149 		SIMPLEQ_FOREACH(qid, &matched_ids, entry) {
1150 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1151 				int matched_type;
1152 				err = got_object_get_type(&matched_type, repo,
1153 				    qid->id);
1154 				if (err)
1155 					goto done;
1156 				if (matched_type != obj_type)
1157 					continue;
1158 			}
1159 			if (*unique_id == NULL) {
1160 				*unique_id = got_object_id_dup(qid->id);
1161 				if (*unique_id == NULL) {
1162 					err = got_error_from_errno("malloc");
1163 					goto done;
1164 				}
1165 			} else {
1166 				if (got_object_id_cmp(*unique_id, qid->id) == 0)
1167 					continue; /* packed multiple times */
1168 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1169 				goto done;
1170 			}
1171 		}
1172 	}
1173 done:
1174 	got_object_id_queue_free(&matched_ids);
1175 	free(path_packdir);
1176 	if (packdir && closedir(packdir) != 0 && err == NULL)
1177 		err = got_error_from_errno("closedir");
1178 	if (err) {
1179 		free(*unique_id);
1180 		*unique_id = NULL;
1181 	}
1182 	return err;
1183 }
1184 
1185 static const struct got_error *
1186 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1187     const char *object_dir, const char *id_str_prefix, int obj_type,
1188     struct got_repository *repo)
1189 {
1190 	const struct got_error *err = NULL;
1191 	char *path;
1192 	DIR *dir = NULL;
1193 	struct dirent *dent;
1194 	struct got_object_id id;
1195 
1196 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1197 		err = got_error_from_errno("asprintf");
1198 		goto done;
1199 	}
1200 
1201 	dir = opendir(path);
1202 	if (dir == NULL) {
1203 		if (errno == ENOENT) {
1204 			err = NULL;
1205 			goto done;
1206 		}
1207 		err = got_error_from_errno2("opendir", path);
1208 		goto done;
1209 	}
1210 	while ((dent = readdir(dir)) != NULL) {
1211 		char *id_str;
1212 		int cmp;
1213 
1214 		if (strcmp(dent->d_name, ".") == 0 ||
1215 		    strcmp(dent->d_name, "..") == 0)
1216 			continue;
1217 
1218 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1219 			err = got_error_from_errno("asprintf");
1220 			goto done;
1221 		}
1222 
1223 		if (!got_parse_sha1_digest(id.sha1, id_str))
1224 			continue;
1225 
1226 		/*
1227 		 * Directory entries do not necessarily appear in
1228 		 * sorted order, so we must iterate over all of them.
1229 		 */
1230 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1231 		if (cmp != 0) {
1232 			free(id_str);
1233 			continue;
1234 		}
1235 
1236 		if (*unique_id == NULL) {
1237 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1238 				int matched_type;
1239 				err = got_object_get_type(&matched_type, repo,
1240 				    &id);
1241 				if (err)
1242 					goto done;
1243 				if (matched_type != obj_type)
1244 					continue;
1245 			}
1246 			*unique_id = got_object_id_dup(&id);
1247 			if (*unique_id == NULL) {
1248 				err = got_error_from_errno("got_object_id_dup");
1249 				free(id_str);
1250 				goto done;
1251 			}
1252 		} else {
1253 			if (got_object_id_cmp(*unique_id, &id) == 0)
1254 				continue; /* both packed and loose */
1255 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1256 			free(id_str);
1257 			goto done;
1258 		}
1259 	}
1260 done:
1261 	if (dir && closedir(dir) != 0 && err == NULL)
1262 		err = got_error_from_errno("closedir");
1263 	if (err) {
1264 		free(*unique_id);
1265 		*unique_id = NULL;
1266 	}
1267 	free(path);
1268 	return err;
1269 }
1270 
1271 const struct got_error *
1272 got_repo_match_object_id_prefix(struct got_object_id **id,
1273     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1274 {
1275 	const struct got_error *err = NULL;
1276 	char *path_objects = got_repo_get_path_objects(repo);
1277 	char *object_dir = NULL;
1278 	size_t len;
1279 	int i;
1280 
1281 	*id = NULL;
1282 
1283 	for (i = 0; i < strlen(id_str_prefix); i++) {
1284 		if (isxdigit((unsigned char)id_str_prefix[i]))
1285 			continue;
1286 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1287 	}
1288 
1289 	len = strlen(id_str_prefix);
1290 	if (len >= 2) {
1291 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1292 		if (err)
1293 			goto done;
1294 		object_dir = strndup(id_str_prefix, 2);
1295 		if (object_dir == NULL) {
1296 			err = got_error_from_errno("strdup");
1297 			goto done;
1298 		}
1299 		err = match_loose_object(id, path_objects, object_dir,
1300 		    id_str_prefix, obj_type, repo);
1301 	} else if (len == 1) {
1302 		int i;
1303 		for (i = 0; i < 0xf; i++) {
1304 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1305 			    == -1) {
1306 				err = got_error_from_errno("asprintf");
1307 				goto done;
1308 			}
1309 			err = match_packed_object(id, repo, object_dir,
1310 			    obj_type);
1311 			if (err)
1312 				goto done;
1313 			err = match_loose_object(id, path_objects, object_dir,
1314 			    id_str_prefix, obj_type, repo);
1315 			if (err)
1316 				goto done;
1317 		}
1318 	} else {
1319 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1320 		goto done;
1321 	}
1322 done:
1323 	free(object_dir);
1324 	if (err) {
1325 		free(*id);
1326 		*id = NULL;
1327 	} else if (*id == NULL)
1328 		err = got_error(GOT_ERR_NO_OBJ);
1329 
1330 	return err;
1331 }
1332 
1333 const struct got_error *
1334 got_repo_match_object_id(struct got_object_id **id, char **label,
1335     const char *id_str, int obj_type, int resolve_tags,
1336     struct got_repository *repo)
1337 {
1338 	const struct got_error *err;
1339 	struct got_tag_object *tag;
1340 	struct got_reference *ref = NULL;
1341 
1342 	*id = NULL;
1343 	if (label)
1344 		*label = NULL;
1345 
1346 	if (resolve_tags) {
1347 		err = got_repo_object_match_tag(&tag, id_str, GOT_OBJ_TYPE_ANY,
1348 		    repo);
1349 		if (err == NULL) {
1350 			*id = got_object_id_dup(
1351 			    got_object_tag_get_object_id(tag));
1352 			if (*id == NULL)
1353 				err = got_error_from_errno("got_object_id_dup");
1354 			else if (label && asprintf(label, "refs/tags/%s",
1355 			    got_object_tag_get_name(tag)) == -1) {
1356 				err = got_error_from_errno("asprintf");
1357 				free(*id);
1358 				*id = NULL;
1359 			}
1360 			got_object_tag_close(tag);
1361 			return err;
1362 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1363 		    err->code != GOT_ERR_NO_OBJ)
1364 			return err;
1365 	}
1366 
1367 	err = got_repo_match_object_id_prefix(id, id_str, obj_type, repo);
1368 	if (err) {
1369 		if (err->code != GOT_ERR_BAD_OBJ_ID_STR)
1370 			return err;
1371 		err = got_ref_open(&ref, repo, id_str, 0);
1372 		if (err != NULL)
1373 			goto done;
1374 		if (label) {
1375 			*label = strdup(got_ref_get_name(ref));
1376 			if (*label == NULL) {
1377 				err = got_error_from_errno("strdup");
1378 				goto done;
1379 			}
1380 		}
1381 		err = got_ref_resolve(id, repo, ref);
1382 	} else if (label) {
1383 		err = got_object_id_str(label, *id);
1384 		if (*label == NULL) {
1385 			err = got_error_from_errno("strdup");
1386 			goto done;
1387 		}
1388 	}
1389 done:
1390 	if (ref)
1391 		got_ref_close(ref);
1392 	return err;
1393 }
1394 
1395 const struct got_error *
1396 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1397     int obj_type, struct got_repository *repo)
1398 {
1399 	const struct got_error *err;
1400 	struct got_reflist_head refs;
1401 	struct got_reflist_entry *re;
1402 	struct got_object_id *tag_id;
1403 
1404 	SIMPLEQ_INIT(&refs);
1405 	*tag = NULL;
1406 
1407 	err = got_ref_list(&refs, repo, "refs/tags", got_ref_cmp_by_name, NULL);
1408 	if (err)
1409 		return err;
1410 
1411 	SIMPLEQ_FOREACH(re, &refs, entry) {
1412 		const char *refname;
1413 		refname = got_ref_get_name(re->ref);
1414 		if (got_ref_is_symbolic(re->ref))
1415 			continue;
1416 		refname += strlen("refs/tags/");
1417 		if (strcmp(refname, name) != 0)
1418 			continue;
1419 		err = got_ref_resolve(&tag_id, repo, re->ref);
1420 		if (err)
1421 			break;
1422 		err = got_object_open_as_tag(tag, repo, tag_id);
1423 		free(tag_id);
1424 		if (err)
1425 			break;
1426 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1427 		    got_object_tag_get_object_type(*tag) == obj_type)
1428 			break;
1429 		got_object_tag_close(*tag);
1430 		*tag = NULL;
1431 	}
1432 
1433 	got_ref_list_free(&refs);
1434 	if (err == NULL && *tag == NULL)
1435 		err = got_error(GOT_ERR_NO_OBJ);
1436 	return err;
1437 }
1438 
1439 static const struct got_error *
1440 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1441     const char *name, mode_t mode, struct got_object_id *blob_id)
1442 {
1443 	const struct got_error *err = NULL;
1444 
1445 	 *new_te = NULL;
1446 
1447 	*new_te = calloc(1, sizeof(**new_te));
1448 	if (*new_te == NULL)
1449 		return got_error_from_errno("calloc");
1450 
1451 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1452 	    sizeof((*new_te)->name)) {
1453 		err = got_error(GOT_ERR_NO_SPACE);
1454 		goto done;
1455 	}
1456 
1457 	(*new_te)->mode = S_IFREG | (mode & ((S_IRWXU | S_IRWXG | S_IRWXO)));
1458 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1459 done:
1460 	if (err && *new_te) {
1461 		free(*new_te);
1462 		*new_te = NULL;
1463 	}
1464 	return err;
1465 }
1466 
1467 static const struct got_error *
1468 import_file(struct got_tree_entry **new_te, struct dirent *de,
1469     const char *path, struct got_repository *repo)
1470 {
1471 	const struct got_error *err;
1472 	struct got_object_id *blob_id = NULL;
1473 	char *filepath;
1474 	struct stat sb;
1475 
1476 	if (asprintf(&filepath, "%s%s%s", path,
1477 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1478 		return got_error_from_errno("asprintf");
1479 
1480 	if (lstat(filepath, &sb) != 0) {
1481 		err = got_error_from_errno2("lstat", path);
1482 		goto done;
1483 	}
1484 
1485 	err = got_object_blob_create(&blob_id, filepath, repo);
1486 	if (err)
1487 		goto done;
1488 
1489 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1490 	    blob_id);
1491 done:
1492 	free(filepath);
1493 	if (err)
1494 		free(blob_id);
1495 	return err;
1496 }
1497 
1498 static const struct got_error *
1499 insert_tree_entry(struct got_tree_entry *new_te,
1500     struct got_pathlist_head *paths)
1501 {
1502 	const struct got_error *err = NULL;
1503 	struct got_pathlist_entry *new_pe;
1504 
1505 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1506 	if (err)
1507 		return err;
1508 	if (new_pe == NULL)
1509 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1510 	return NULL;
1511 }
1512 
1513 static const struct got_error *write_tree(struct got_object_id **,
1514     const char *, struct got_pathlist_head *, struct got_repository *,
1515     got_repo_import_cb progress_cb, void *progress_arg);
1516 
1517 static const struct got_error *
1518 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1519     const char *path, struct got_pathlist_head *ignores,
1520     struct got_repository *repo,
1521     got_repo_import_cb progress_cb, void *progress_arg)
1522 {
1523 	const struct got_error *err;
1524 	struct got_object_id *id = NULL;
1525 	char *subdirpath;
1526 
1527 	if (asprintf(&subdirpath, "%s%s%s", path,
1528 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1529 		return got_error_from_errno("asprintf");
1530 
1531 	(*new_te) = calloc(1, sizeof(**new_te));
1532 	if (*new_te == NULL)
1533 		return got_error_from_errno("calloc");
1534 	(*new_te)->mode = S_IFDIR;
1535 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
1536 	    sizeof((*new_te)->name)) {
1537 		err = got_error(GOT_ERR_NO_SPACE);
1538 		goto done;
1539 	}
1540 	err = write_tree(&id, subdirpath, ignores,  repo,
1541 	    progress_cb, progress_arg);
1542 	if (err)
1543 		goto done;
1544 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
1545 	
1546 done:
1547 	free(id);
1548 	free(subdirpath);
1549 	if (err) {
1550 		free(*new_te);
1551 		*new_te = NULL;
1552 	}
1553 	return err;
1554 }
1555 
1556 static const struct got_error *
1557 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1558     struct got_pathlist_head *ignores, struct got_repository *repo,
1559     got_repo_import_cb progress_cb, void *progress_arg)
1560 {
1561 	const struct got_error *err = NULL;
1562 	DIR *dir;
1563 	struct dirent *de;
1564 	int nentries;
1565 	struct got_tree_entry *new_te = NULL;
1566 	struct got_pathlist_head paths;
1567 	struct got_pathlist_entry *pe;
1568 
1569 	*new_tree_id = NULL;
1570 
1571 	TAILQ_INIT(&paths);
1572 
1573 	dir = opendir(path_dir);
1574 	if (dir == NULL) {
1575 		err = got_error_from_errno2("opendir", path_dir);
1576 		goto done;
1577 	}
1578 
1579 	nentries = 0;
1580 	while ((de = readdir(dir)) != NULL) {
1581 		int ignore = 0;
1582 
1583 		if (strcmp(de->d_name, ".") == 0 ||
1584 		    strcmp(de->d_name, "..") == 0)
1585 			continue;
1586 
1587 		TAILQ_FOREACH(pe, ignores, entry) {
1588 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
1589 				ignore = 1;
1590 				break;
1591 			}
1592 		}
1593 		if (ignore)
1594 			continue;
1595 		if (de->d_type == DT_DIR) {
1596 			err = import_subdir(&new_te, de, path_dir,
1597 			    ignores, repo, progress_cb, progress_arg);
1598 			if (err) {
1599 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
1600 					goto done;
1601 				err = NULL;
1602 				continue;
1603 			}
1604 		} else if (de->d_type == DT_REG) {
1605 			err = import_file(&new_te, de, path_dir, repo);
1606 			if (err)
1607 				goto done;
1608 		} else
1609 			continue;
1610 
1611 		err = insert_tree_entry(new_te, &paths);
1612 		if (err)
1613 			goto done;
1614 		nentries++;
1615 	}
1616 
1617 	if (TAILQ_EMPTY(&paths)) {
1618 		err = got_error(GOT_ERR_NO_TREE_ENTRY);
1619 		goto done;
1620 	}
1621 
1622 	TAILQ_FOREACH(pe, &paths, entry) {
1623 		struct got_tree_entry *te = pe->data;
1624 		char *path;
1625 		if (!S_ISREG(te->mode))
1626 			continue;
1627 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
1628 			err = got_error_from_errno("asprintf");
1629 			goto done;
1630 		}
1631 		err = (*progress_cb)(progress_arg, path);
1632 		free(path);
1633 		if (err)
1634 			goto done;
1635 	}
1636 
1637 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
1638 done:
1639 	if (dir)
1640 		closedir(dir);
1641 	got_pathlist_free(&paths);
1642 	return err;
1643 }
1644 
1645 const struct got_error *
1646 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
1647     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
1648     struct got_repository *repo, got_repo_import_cb progress_cb,
1649     void *progress_arg)
1650 {
1651 	const struct got_error *err;
1652 	struct got_object_id *new_tree_id;
1653 
1654 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
1655 	    progress_cb, progress_arg);
1656 	if (err)
1657 		return err;
1658 
1659 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
1660 	    author, time(NULL), author, time(NULL), logmsg, repo);
1661 	free(new_tree_id);
1662 	return err;
1663 }