1 /*
2  * Copyright (c) 2018 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/stat.h>
18 #include <sys/limits.h>
19 #include <sys/queue.h>
20 
21 #include <stddef.h>
22 #include <string.h>
23 #include <stdio.h>
24 #include <stdlib.h>
25 #include <fcntl.h>
26 #include <errno.h>
27 #include <unistd.h>
28 #include <sha1.h>
29 #include <zlib.h>
30 #include <fnmatch.h>
31 #include <libgen.h>
32 
33 #include "got_error.h"
34 #include "got_repository.h"
35 #include "got_reference.h"
36 #include "got_object.h"
37 #include "got_worktree.h"
38 #include "got_opentemp.h"
39 
40 #include "got_lib_worktree.h"
41 #include "got_lib_path.h"
42 #include "got_lib_sha1.h"
43 #include "got_lib_fileindex.h"
44 #include "got_lib_inflate.h"
45 #include "got_lib_delta.h"
46 #include "got_lib_object.h"
47 
48 #ifndef MIN
49 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
50 #endif
51 
52 static const struct got_error *
53 create_meta_file(const char *path_got, const char *name, const char *content)
54 {
55 	const struct got_error *err = NULL;
56 	char *path;
57 	int fd = -1;
58 
59 	if (asprintf(&path, "%s/%s", path_got, name) == -1) {
60 		err = got_error_from_errno();
61 		path = NULL;
62 		goto done;
63 	}
64 
65 	fd = open(path, O_RDWR | O_CREAT | O_EXCL | O_NOFOLLOW,
66 	    GOT_DEFAULT_FILE_MODE);
67 	if (fd == -1) {
68 		err = got_error_from_errno();
69 		goto done;
70 	}
71 
72 	if (content) {
73 		int len = dprintf(fd, "%s\n", content);
74 		if (len != strlen(content) + 1) {
75 			err = got_error_from_errno();
76 			goto done;
77 		}
78 	}
79 
80 done:
81 	if (fd != -1 && close(fd) == -1 && err == NULL)
82 		err = got_error_from_errno();
83 	free(path);
84 	return err;
85 }
86 
87 static const struct got_error *
88 update_meta_file(const char *path_got, const char *name, const char *content)
89 {
90 	const struct got_error *err = NULL;
91 	FILE *tmpfile = NULL;
92 	char *tmppath = NULL;
93 	char *path = NULL;
94 
95 	if (asprintf(&path, "%s/%s", path_got, name) == -1) {
96 		err = got_error_from_errno();
97 		path = NULL;
98 		goto done;
99 	}
100 
101 	err = got_opentemp_named(&tmppath, &tmpfile, path);
102 	if (err)
103 		goto done;
104 
105 	if (content) {
106 		int len = fprintf(tmpfile, "%s\n", content);
107 		if (len != strlen(content) + 1) {
108 			err = got_error_from_errno();
109 			goto done;
110 		}
111 	}
112 
113 	if (rename(tmppath, path) != 0) {
114 		err = got_error_from_errno();
115 		goto done;
116 	}
117 
118 done:
119 	free(tmppath);
120 	fclose(tmpfile);
121 	return err;
122 }
123 
124 static const struct got_error *
125 read_meta_file(char **content, const char *path_got, const char *name)
126 {
127 	const struct got_error *err = NULL;
128 	char *path;
129 	int fd = -1;
130 	ssize_t n;
131 	struct stat sb;
132 
133 	*content = NULL;
134 
135 	if (asprintf(&path, "%s/%s", path_got, name) == -1) {
136 		err = got_error_from_errno();
137 		path = NULL;
138 		goto done;
139 	}
140 
141 	fd = open(path, O_RDONLY | O_NOFOLLOW);
142 	if (fd == -1) {
143 		err = got_error_from_errno();
144 		goto done;
145 	}
146 	if (flock(fd, LOCK_SH | LOCK_NB) == -1) {
147 		err = (errno == EWOULDBLOCK ? got_error(GOT_ERR_WORKTREE_BUSY)
148 		    : got_error_from_errno());
149 		goto done;
150 	}
151 
152 	stat(path, &sb);
153 	*content = calloc(1, sb.st_size);
154 	if (*content == NULL) {
155 		err = got_error_from_errno();
156 		goto done;
157 	}
158 
159 	n = read(fd, *content, sb.st_size);
160 	if (n != sb.st_size) {
161 		err = (n == -1 ? got_error_from_errno() :
162 		    got_error(GOT_ERR_WORKTREE_META));
163 		goto done;
164 	}
165 	if ((*content)[sb.st_size - 1] != '\n') {
166 		err = got_error(GOT_ERR_WORKTREE_META);
167 		goto done;
168 	}
169 	(*content)[sb.st_size - 1] = '\0';
170 
171 done:
172 	if (fd != -1 && close(fd) == -1 && err == NULL)
173 		err = got_error_from_errno();
174 	free(path);
175 	if (err) {
176 		free(*content);
177 		*content = NULL;
178 	}
179 	return err;
180 }
181 
182 const struct got_error *
183 got_worktree_init(const char *path, struct got_reference *head_ref,
184     const char *prefix, struct got_repository *repo)
185 {
186 	const struct got_error *err = NULL;
187 	struct got_object_id *commit_id = NULL;
188 	int obj_type;
189 	char *path_got = NULL;
190 	char *refstr = NULL;
191 	char *formatstr = NULL;
192 	char *absprefix = NULL;
193 	char *basestr = NULL;
194 
195 	err = got_ref_resolve(&commit_id, repo, head_ref);
196 	if (err)
197 		return err;
198 	err = got_object_get_type(&obj_type, repo, commit_id);
199 	if (err)
200 		return err;
201 	if (obj_type != GOT_OBJ_TYPE_COMMIT)
202 		return got_error(GOT_ERR_OBJ_TYPE);
203 
204 	if (!got_path_is_absolute(prefix)) {
205 		if (asprintf(&absprefix, "/%s", prefix) == -1)
206 			return got_error_from_errno();
207 	}
208 
209 	/* Create top-level directory (may already exist). */
210 	if (mkdir(path, GOT_DEFAULT_DIR_MODE) == -1 && errno != EEXIST) {
211 		err = got_error_from_errno();
212 		goto done;
213 	}
214 
215 	/* Create .got directory (may already exist). */
216 	if (asprintf(&path_got, "%s/%s", path, GOT_WORKTREE_GOT_DIR) == -1) {
217 		err = got_error_from_errno();
218 		goto done;
219 	}
220 	if (mkdir(path_got, GOT_DEFAULT_DIR_MODE) == -1 && errno != EEXIST) {
221 		err = got_error_from_errno();
222 		goto done;
223 	}
224 
225 	/* Create an empty lock file. */
226 	err = create_meta_file(path_got, GOT_WORKTREE_LOCK, NULL);
227 	if (err)
228 		goto done;
229 
230 	/* Create an empty file index. */
231 	err = create_meta_file(path_got, GOT_WORKTREE_FILE_INDEX, NULL);
232 	if (err)
233 		goto done;
234 
235 	/* Write the HEAD reference. */
236 	refstr = got_ref_to_str(head_ref);
237 	if (refstr == NULL) {
238 		err = got_error_from_errno();
239 		goto done;
240 	}
241 	err = create_meta_file(path_got, GOT_WORKTREE_HEAD_REF, refstr);
242 	if (err)
243 		goto done;
244 
245 	/* Record our base commit. */
246 	err = got_object_id_str(&basestr, commit_id);
247 	if (err)
248 		goto done;
249 	err = create_meta_file(path_got, GOT_WORKTREE_BASE_COMMIT, basestr);
250 	if (err)
251 		goto done;
252 
253 	/* Store path to repository. */
254 	err = create_meta_file(path_got, GOT_WORKTREE_REPOSITORY,
255 	    got_repo_get_path(repo));
256 	if (err)
257 		goto done;
258 
259 	/* Store in-repository path prefix. */
260 	err = create_meta_file(path_got, GOT_WORKTREE_PATH_PREFIX,
261 	    absprefix ? absprefix : prefix);
262 	if (err)
263 		goto done;
264 
265 	/* Stamp work tree with format file. */
266 	if (asprintf(&formatstr, "%d", GOT_WORKTREE_FORMAT_VERSION) == -1) {
267 		err = got_error_from_errno();
268 		goto done;
269 	}
270 	err = create_meta_file(path_got, GOT_WORKTREE_FORMAT, formatstr);
271 	if (err)
272 		goto done;
273 
274 done:
275 	free(commit_id);
276 	free(path_got);
277 	free(formatstr);
278 	free(refstr);
279 	free(absprefix);
280 	free(basestr);
281 	return err;
282 }
283 
284 const struct got_error *
285 got_worktree_open(struct got_worktree **worktree, const char *path)
286 {
287 	const struct got_error *err = NULL;
288 	char *path_got;
289 	char *formatstr = NULL;
290 	char *path_lock = NULL;
291 	char *base_commit_id_str = NULL;
292 	char *head_ref_str = NULL;
293 	int version, fd = -1;
294 	const char *errstr;
295 	struct got_repository *repo = NULL;
296 
297 	*worktree = NULL;
298 
299 	if (asprintf(&path_got, "%s/%s", path, GOT_WORKTREE_GOT_DIR) == -1) {
300 		err = got_error_from_errno();
301 		path_got = NULL;
302 		goto done;
303 	}
304 
305 	if (asprintf(&path_lock, "%s/%s", path_got, GOT_WORKTREE_LOCK) == -1) {
306 		err = got_error_from_errno();
307 		path_lock = NULL;
308 		goto done;
309 	}
310 
311 	fd = open(path_lock, O_RDWR | O_EXLOCK | O_NONBLOCK);
312 	if (fd == -1) {
313 		err = (errno == EWOULDBLOCK ? got_error(GOT_ERR_WORKTREE_BUSY)
314 		    : got_error_from_errno());
315 		goto done;
316 	}
317 
318 	err = read_meta_file(&formatstr, path_got, GOT_WORKTREE_FORMAT);
319 	if (err)
320 		goto done;
321 
322 	version = strtonum(formatstr, 1, INT_MAX, &errstr);
323 	if (errstr) {
324 		err = got_error(GOT_ERR_WORKTREE_META);
325 		goto done;
326 	}
327 	if (version != GOT_WORKTREE_FORMAT_VERSION) {
328 		err = got_error(GOT_ERR_WORKTREE_VERS);
329 		goto done;
330 	}
331 
332 	*worktree = calloc(1, sizeof(**worktree));
333 	if (*worktree == NULL) {
334 		err = got_error_from_errno();
335 		goto done;
336 	}
337 	(*worktree)->lockfd = -1;
338 
339 	(*worktree)->root_path = strdup(path);
340 	if ((*worktree)->root_path == NULL) {
341 		err = got_error_from_errno();
342 		goto done;
343 	}
344 	err = read_meta_file(&(*worktree)->repo_path, path_got,
345 	    GOT_WORKTREE_REPOSITORY);
346 	if (err)
347 		goto done;
348 
349 	err = read_meta_file(&(*worktree)->path_prefix, path_got,
350 	    GOT_WORKTREE_PATH_PREFIX);
351 	if (err)
352 		goto done;
353 
354 	err = read_meta_file(&base_commit_id_str, path_got,
355 	    GOT_WORKTREE_BASE_COMMIT);
356 	if (err)
357 		goto done;
358 
359 	err = got_repo_open(&repo, (*worktree)->repo_path);
360 	if (err)
361 		goto done;
362 
363 	err = got_object_resolve_id_str(&(*worktree)->base_commit_id, repo,
364 	    base_commit_id_str);
365 	if (err)
366 		goto done;
367 
368 	err = read_meta_file(&head_ref_str, path_got, GOT_WORKTREE_HEAD_REF);
369 	if (err)
370 		goto done;
371 
372 	err = got_ref_open(&(*worktree)->head_ref, repo, head_ref_str);
373 done:
374 	if (repo)
375 		got_repo_close(repo);
376 	free(path_got);
377 	free(path_lock);
378 	free(head_ref_str);
379 	free(base_commit_id_str);
380 	if (err) {
381 		if (fd != -1)
382 			close(fd);
383 		if (*worktree != NULL)
384 			got_worktree_close(*worktree);
385 		*worktree = NULL;
386 	} else
387 		(*worktree)->lockfd = fd;
388 
389 	return err;
390 }
391 
392 void
393 got_worktree_close(struct got_worktree *worktree)
394 {
395 	free(worktree->root_path);
396 	free(worktree->repo_path);
397 	free(worktree->path_prefix);
398 	free(worktree->base_commit_id);
399 	if (worktree->head_ref)
400 		got_ref_close(worktree->head_ref);
401 	if (worktree->lockfd != -1)
402 		close(worktree->lockfd);
403 	free(worktree);
404 }
405 
406 const char *
407 got_worktree_get_repo_path(struct got_worktree *worktree)
408 {
409 	return worktree->repo_path;
410 }
411 
412 const char *
413 got_worktree_get_path_prefix(struct got_worktree *worktree)
414 {
415 	return worktree->path_prefix;
416 }
417 
418 const struct got_error *
419 got_worktree_match_path_prefix(int *match, struct got_worktree *worktree,
420     const char *path_prefix)
421 {
422 	char *absprefix = NULL;
423 
424 	if (!got_path_is_absolute(path_prefix)) {
425 		if (asprintf(&absprefix, "/%s", path_prefix) == -1)
426 			return got_error_from_errno();
427 	}
428 	*match = (strcmp(absprefix ? absprefix : path_prefix,
429 	    worktree->path_prefix) == 0);
430 	free(absprefix);
431 	return NULL;
432 }
433 
434 char *
435 got_worktree_get_head_ref_name(struct got_worktree *worktree)
436 {
437 	return got_ref_to_str(worktree->head_ref);
438 }
439 
440 struct got_reference *
441 got_worktree_get_head_ref(struct got_worktree *worktree)
442 {
443 	return got_ref_dup(worktree->head_ref);
444 }
445 
446 const struct got_object_id *
447 got_worktree_get_base_commit_id(struct got_worktree *worktree)
448 {
449 	return worktree->base_commit_id;
450 }
451 
452 const struct got_error *
453 got_worktree_set_base_commit_id(struct got_worktree *worktree,
454     struct got_repository *repo, struct got_object_id *commit_id)
455 {
456 	const struct got_error *err;
457 	struct got_object *obj = NULL;
458 	char *id_str = NULL;
459 	char *path_got = NULL;
460 
461 	if (asprintf(&path_got, "%s/%s", worktree->root_path,
462 	    GOT_WORKTREE_GOT_DIR) == -1) {
463 		err = got_error_from_errno();
464 		path_got = NULL;
465 		goto done;
466 	}
467 
468 	err = got_object_open(&obj, repo, commit_id);
469 	if (err)
470 		return err;
471 
472 	if (obj->type != GOT_OBJ_TYPE_COMMIT) {
473 		err = got_error(GOT_ERR_OBJ_TYPE);
474 		goto done;
475 	}
476 
477 	/* Record our base commit. */
478 	err = got_object_id_str(&id_str, commit_id);
479 	if (err)
480 		goto done;
481 	err = update_meta_file(path_got, GOT_WORKTREE_BASE_COMMIT, id_str);
482 	if (err)
483 		goto done;
484 
485 	free(worktree->base_commit_id);
486 	worktree->base_commit_id = got_object_id_dup(commit_id);
487 	if (worktree->base_commit_id == NULL) {
488 		err = got_error_from_errno();
489 		goto done;
490 	}
491 done:
492 	if (obj)
493 		got_object_close(obj);
494 	free(id_str);
495 	free(path_got);
496 	return err;
497 }
498 
499 static const struct got_error *
500 lock_worktree(struct got_worktree *worktree, int operation)
501 {
502 	if (flock(worktree->lockfd, operation | LOCK_NB) == -1)
503 		return (errno == EWOULDBLOCK ? got_error(GOT_ERR_WORKTREE_BUSY)
504 		    : got_error_from_errno());
505 	return NULL;
506 }
507 
508 static const char *
509 apply_path_prefix(struct got_worktree *worktree, const char *path)
510 {
511 	const char *p = path;
512 	p += strlen(worktree->path_prefix);
513 	if (*p == '/')
514 		p++;
515 	return p;
516 }
517 
518 static const struct got_error *
519 blob_checkout(struct got_worktree *worktree, struct got_fileindex *fileindex,
520    struct got_fileindex_entry *entry, const char *path,
521    struct got_blob_object *blob, struct got_repository *repo,
522    got_worktree_checkout_cb progress_cb, void *progress_arg,
523    const char *progress_path)
524 {
525 	const struct got_error *err = NULL;
526 	char *ondisk_path;
527 	int fd = -1;
528 	size_t len, hdrlen;
529 	int update = 0;
530 	char *tmppath = NULL;
531 
532 	if (asprintf(&ondisk_path, "%s/%s", worktree->root_path,
533 	    apply_path_prefix(worktree, path)) == -1)
534 		return got_error_from_errno();
535 
536 	fd = open(ondisk_path, O_RDWR | O_CREAT | O_EXCL | O_NOFOLLOW,
537 	    GOT_DEFAULT_FILE_MODE);
538 	if (fd == -1) {
539 		err = got_error_from_errno();
540 		if (errno == EEXIST) {
541 			struct stat sb;
542 			if (lstat(ondisk_path, &sb) == -1) {
543 				err = got_error_from_errno();
544 				goto done;
545 			} else if (!S_ISREG(sb.st_mode)) {
546 				/* TODO file is obstructed; do something */
547 				err = got_error(GOT_ERR_FILE_OBSTRUCTED);
548 				goto done;
549 			} else {
550 				err = got_opentemp_named_fd(&tmppath, &fd,
551 				    ondisk_path);
552 				if (err)
553 					goto done;
554 				update = 1;
555 			}
556 		} else
557 			return err;
558 	}
559 
560 	(*progress_cb)(progress_arg,
561 	    update ? GOT_STATUS_UPDATE : GOT_STATUS_ADD, progress_path);
562 
563 	hdrlen = got_object_blob_get_hdrlen(blob);
564 	do {
565 		const uint8_t *buf = got_object_blob_get_read_buf(blob);
566 		err = got_object_blob_read_block(&len, blob);
567 		if (err)
568 			break;
569 		if (len > 0) {
570 			/* Skip blob object header first time around. */
571 			ssize_t outlen = write(fd, buf + hdrlen, len - hdrlen);
572 			if (outlen == -1) {
573 				err = got_error_from_errno();
574 				goto done;
575 			} else if (outlen != len - hdrlen) {
576 				err = got_error(GOT_ERR_IO);
577 				goto done;
578 			}
579 			hdrlen = 0;
580 		}
581 	} while (len != 0);
582 
583 	fsync(fd);
584 
585 	if (update) {
586 		if (rename(tmppath, ondisk_path) != 0) {
587 			err = got_error_from_errno();
588 			goto done;
589 		}
590 	}
591 
592 	if (entry)
593 		err = got_fileindex_entry_update(entry, ondisk_path,
594 		    blob->id.sha1, worktree->base_commit_id->sha1);
595 	else {
596 		err = got_fileindex_entry_alloc(&entry, ondisk_path,
597 		    apply_path_prefix(worktree, path), blob->id.sha1,
598 		    worktree->base_commit_id->sha1);
599 		if (err)
600 			goto done;
601 		err = got_fileindex_entry_add(fileindex, entry);
602 	}
603 	if (err)
604 		goto done;
605 done:
606 	if (fd != -1)
607 		close(fd);
608 	free(ondisk_path);
609 	free(tmppath);
610 	return err;
611 }
612 
613 static const struct got_error *
614 add_dir_on_disk(struct got_worktree *worktree, const char *path)
615 {
616 	const struct got_error *err = NULL;
617 	char *abspath;
618 
619 	if (asprintf(&abspath, "%s/%s", worktree->root_path,
620 	    apply_path_prefix(worktree, path)) == -1)
621 		return got_error_from_errno();
622 
623 	/* XXX queue work rather than editing disk directly? */
624 	if (mkdir(abspath, GOT_DEFAULT_DIR_MODE) == -1) {
625 		struct stat sb;
626 
627 		if (errno != EEXIST) {
628 			err = got_error_from_errno();
629 			goto done;
630 		}
631 
632 		if (lstat(abspath, &sb) == -1) {
633 			err = got_error_from_errno();
634 			goto done;
635 		}
636 
637 		if (!S_ISDIR(sb.st_mode)) {
638 			/* TODO directory is obstructed; do something */
639 			return got_error(GOT_ERR_FILE_OBSTRUCTED);
640 		}
641 	}
642 
643 done:
644 	free(abspath);
645 	return err;
646 }
647 
648 static const struct got_error *
649 tree_checkout(struct got_worktree *, struct got_fileindex *,
650     struct got_tree_object *, const char *, struct got_repository *,
651     got_worktree_checkout_cb progress_cb, void *progress_arg,
652     got_worktree_cancel_cb cancel_cb, void *cancel_arg);
653 
654 static const struct got_error *
655 tree_checkout_entry(struct got_worktree *worktree,
656     struct got_fileindex *fileindex, struct got_tree_entry *te,
657     const char *parent, struct got_repository *repo,
658     got_worktree_checkout_cb progress_cb, void *progress_arg,
659     got_worktree_cancel_cb cancel_cb, void *cancel_arg)
660 {
661 	const struct got_error *err = NULL;
662 	struct got_object *obj = NULL;
663 	struct got_blob_object *blob = NULL;
664 	struct got_fileindex_entry *entry = NULL;
665 	struct got_tree_object *tree = NULL;
666 	char *path = NULL;
667 	char *progress_path = NULL;
668 	size_t len;
669 
670 	if (parent[0] == '/' && parent[1] == '\0')
671 		parent = "";
672 	if (asprintf(&path, "%s/%s", parent, te->name) == -1)
673 		return got_error_from_errno();
674 
675 	/* Skip this entry if it is outside of our path prefix. */
676 	len = MIN(strlen(worktree->path_prefix), strlen(path));
677 	if (strncmp(path, worktree->path_prefix, len) != 0) {
678 		free(path);
679 		return NULL;
680 	}
681 
682 	err = got_object_open(&obj, repo, te->id);
683 	if (err)
684 		goto done;
685 
686 	progress_path = path;
687 	if (strncmp(progress_path, worktree->path_prefix, len) == 0)
688 		progress_path += len;
689 
690 	switch (obj->type) {
691 	case GOT_OBJ_TYPE_BLOB:
692 		if (strlen(worktree->path_prefix) >= strlen(path))
693 			break;
694 		entry = got_fileindex_entry_get(fileindex,
695 		    apply_path_prefix(worktree, path));
696 		if (entry &&
697 		    memcmp(entry->commit_sha1, worktree->base_commit_id->sha1,
698 		    SHA1_DIGEST_LENGTH) == 0) {
699 			(*progress_cb)(progress_arg, GOT_STATUS_EXISTS,
700 			    progress_path);
701 			break;
702 		}
703 		if (entry && memcmp(entry->blob_sha1, obj->id.sha1,
704 		    SHA1_DIGEST_LENGTH) == 0)
705 			break;
706 		err = got_object_blob_open(&blob, repo, obj, 8192);
707 		if (err)
708 			goto done;
709 		err = blob_checkout(worktree, fileindex, entry, path, blob,
710 		    repo, progress_cb, progress_arg, progress_path);
711 		break;
712 	case GOT_OBJ_TYPE_TREE:
713 		if (strlen(worktree->path_prefix) < strlen(path)) {
714 			err = add_dir_on_disk(worktree, path);
715 			if (err)
716 				break;
717 		}
718 		err = got_object_tree_open(&tree, repo, obj);
719 		if (err)
720 			goto done;
721 		/* XXX infinite recursion possible */
722 		err = tree_checkout(worktree, fileindex, tree, path, repo,
723 		    progress_cb, progress_arg, cancel_cb, cancel_arg);
724 		break;
725 	default:
726 		break;
727 	}
728 
729 done:
730 	if (blob)
731 		got_object_blob_close(blob);
732 	if (tree)
733 		got_object_tree_close(tree);
734 	if (obj)
735 		got_object_close(obj);
736 	free(path);
737 	return err;
738 }
739 
740 struct collect_missing_entry_args {
741 	struct got_fileindex *fileindex;
742 	const struct got_tree_entries *entries;
743 	struct got_fileindex missing_entries;
744 	const char *current_subdir;
745 };
746 
747 static const struct got_error *
748 collect_missing_file(void *args, struct got_fileindex_entry *entry)
749 {
750 	struct collect_missing_entry_args *a = args;
751 	char *start, *end;
752 	ptrdiff_t len;
753 	struct got_tree_entry *te;
754 	int found = 0;
755 
756 	if (a->current_subdir[0] != '\0' &&
757 	    strncmp(a->current_subdir, entry->path,
758 	    strlen(a->current_subdir)) != 0)
759 		return NULL;
760 
761 	start = entry->path + strlen(a->current_subdir);
762 	while (start[0] == '/')
763 		start++;
764 	end = strchr(start, '/');
765 	if (end == NULL) {
766 		end = strchr(start, '\0');
767 		if (end == NULL)
768 			return got_error(GOT_ERR_BAD_PATH);
769 	}
770 	len = end - start;
771 
772 	SIMPLEQ_FOREACH(te, &a->entries->head, entry) {
773 		if (strncmp(start, te->name, len) == 0 &&
774 		    te->name[len] == '\0') {
775 			found = 1;
776 			break;
777 		}
778 	}
779 
780 	if (found)
781 		return NULL;
782 
783 	got_fileindex_entry_remove(a->fileindex, entry);
784 	return got_fileindex_entry_add(&a->missing_entries, entry);
785 }
786 
787 /* Remove files which exist in the file index but not in the tree. */
788 static const struct got_error *
789 remove_missing_files(struct got_worktree *worktree, const char *path,
790     struct got_fileindex *fileindex, const struct got_tree_entries *entries,
791     got_worktree_checkout_cb progress_cb, void *progress_arg,
792     got_worktree_cancel_cb cancel_cb, void *cancel_arg)
793 {
794 	const struct got_error *err = NULL;
795 	struct collect_missing_entry_args a;
796 	struct got_fileindex_entry *entry, *tmp;
797 
798 	a.fileindex = fileindex;
799 	a.entries = entries;
800 	a.missing_entries.nentries = 0;
801 	a.current_subdir = apply_path_prefix(worktree, path);
802 	TAILQ_INIT(&a.missing_entries.entries);
803 	err = got_fileindex_for_each_entry_safe(fileindex,
804 	    collect_missing_file, &a);
805 	if (err)
806 		return err;
807 
808 	TAILQ_FOREACH_SAFE(entry, &a.missing_entries.entries, entry, tmp) {
809 		char *ondisk_path = NULL;
810 
811 		if (cancel_cb) {
812 			err = (*cancel_cb)(cancel_arg);
813 			if (err)
814 				break;
815 		}
816 
817 		(*progress_cb)(progress_arg, GOT_STATUS_DELETE, entry->path);
818 
819 		if (asprintf(&ondisk_path, "%s/%s", worktree->root_path,
820 		    entry->path) == -1) {
821 			err = got_error_from_errno();
822 			break;
823 		}
824 
825 		if (unlink(ondisk_path) == -1)
826 			err = got_error_from_errno();
827 		else {
828 			char *parent = dirname(ondisk_path);
829 			if (rmdir(parent) == -1 && errno != ENOTEMPTY)
830 				err = got_error_from_errno();
831 		}
832 		free(ondisk_path);
833 		if (err)
834 			break;
835 
836 		TAILQ_REMOVE(&a.missing_entries.entries, entry, entry);
837 		got_fileindex_entry_free(entry);
838 	}
839 
840 	if (err) {
841 		while (!TAILQ_EMPTY(&a.missing_entries.entries)) {
842 			entry = TAILQ_FIRST(&a.missing_entries.entries);
843 			TAILQ_REMOVE(&a.missing_entries.entries, entry, entry);
844 			got_fileindex_entry_free(entry);
845 		}
846 	}
847 
848 	return err;
849 }
850 
851 static const struct got_error *
852 tree_checkout(struct got_worktree *worktree,
853     struct got_fileindex *fileindex, struct got_tree_object *tree,
854     const char *path, struct got_repository *repo,
855     got_worktree_checkout_cb progress_cb, void *progress_arg,
856     got_worktree_cancel_cb cancel_cb, void *cancel_arg)
857 {
858 	const struct got_error *err = NULL;
859 	const struct got_tree_entries *entries;
860 	struct got_tree_entry *te;
861 	size_t len;
862 
863 	/* Skip this tree if it shares no path components with the prefix. */
864 	len = MIN(strlen(worktree->path_prefix), strlen(path));
865 	if (strncmp(path, worktree->path_prefix, len) != 0)
866 		return NULL;
867 
868 	entries = got_object_tree_get_entries(tree);
869 	SIMPLEQ_FOREACH(te, &entries->head, entry) {
870 		if (cancel_cb) {
871 			err = (*cancel_cb)(cancel_arg);
872 			if (err)
873 				return err;
874 		}
875 		err = tree_checkout_entry(worktree, fileindex, te, path, repo,
876 		    progress_cb, progress_arg, cancel_cb, cancel_arg);
877 		if (err)
878 			return err;
879 	}
880 
881 	len = strlen(worktree->path_prefix);
882 	if (strncmp(worktree->path_prefix, path, len) == 0) {
883 		err = remove_missing_files(worktree, path, fileindex, entries,
884 		    progress_cb, progress_arg, cancel_cb, cancel_arg);
885 		if (err)
886 			return err;
887 	}
888 
889 	return err;
890 }
891 
892 const struct got_error *
893 got_worktree_checkout_files(struct got_worktree *worktree,
894     struct got_repository *repo, got_worktree_checkout_cb progress_cb,
895     void *progress_arg, got_worktree_cancel_cb cancel_cb, void *cancel_arg)
896 {
897 	const struct got_error *err = NULL, *unlockerr, *checkout_err = NULL;
898 	struct got_commit_object *commit = NULL;
899 	struct got_tree_object *tree = NULL;
900 	char *fileindex_path = NULL, *new_fileindex_path = NULL;
901 	struct got_fileindex *fileindex = NULL;
902 	FILE *index = NULL, *new_index = NULL;
903 
904 	err = lock_worktree(worktree, LOCK_EX);
905 	if (err)
906 		return err;
907 
908 	fileindex = got_fileindex_alloc();
909 	if (fileindex == NULL) {
910 		err = got_error_from_errno();
911 		goto done;
912 	}
913 
914 	if (asprintf(&fileindex_path, "%s/%s/%s", worktree->root_path,
915 	    GOT_WORKTREE_GOT_DIR, GOT_WORKTREE_FILE_INDEX) == -1) {
916 		err = got_error_from_errno();
917 		fileindex_path = NULL;
918 		goto done;
919 	}
920 
921 	/*
922 	 * Read the file index.
923 	 * Checking out files is supposed to be an idempotent operation.
924 	 * If the on-disk file index is incomplete we will try to complete it.
925 	 */
926 	index = fopen(fileindex_path, "rb");
927 	if (index == NULL) {
928 		err = got_error_from_errno();
929 		goto done;
930 	}
931 	err = got_fileindex_read(fileindex, index);
932 	fclose(index);
933 	if (err)
934 		goto done;
935 
936 	err = got_opentemp_named(&new_fileindex_path, &new_index,
937 	    fileindex_path);
938 	if (err)
939 		goto done;
940 
941 	err = got_object_open_as_commit(&commit, repo,
942 	   worktree->base_commit_id);
943 	if (err)
944 		goto done;
945 
946 	err = got_object_open_as_tree(&tree, repo, commit->tree_id);
947 	if (err)
948 		goto done;
949 
950 	checkout_err = tree_checkout(worktree, fileindex, tree, "/", repo,
951 	    progress_cb, progress_arg, cancel_cb, cancel_arg);
952 
953 	/* Try to sync the fileindex back to disk in any case. */
954 	err = got_fileindex_write(fileindex, new_index);
955 	if (err)
956 		goto done;
957 
958 	if (rename(new_fileindex_path, fileindex_path) != 0) {
959 		err = got_error_from_errno();
960 		goto done;
961 	}
962 
963 	free(new_fileindex_path);
964 	new_fileindex_path = NULL;
965 
966 done:
967 	if (tree)
968 		got_object_tree_close(tree);
969 	if (commit)
970 		got_object_commit_close(commit);
971 	if (new_fileindex_path)
972 		unlink(new_fileindex_path);
973 	if (new_index)
974 		fclose(new_index);
975 	free(new_fileindex_path);
976 	free(fileindex_path);
977 	got_fileindex_free(fileindex);
978 	if (checkout_err)
979 		err = checkout_err;
980 	unlockerr = lock_worktree(worktree, LOCK_SH);
981 	if (unlockerr && err == NULL)
982 		err = unlockerr;
983 	return err;
984 }