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_pathset.h"
43 #include "got_lib_sha1.h"
44 #include "got_lib_fileindex.h"
45 #include "got_lib_inflate.h"
46 #include "got_lib_delta.h"
47 #include "got_lib_object.h"
48 
49 #ifndef MIN
50 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
51 #endif
52 
53 static const struct got_error *
54 create_meta_file(const char *path_got, const char *name, const char *content)
55 {
56 	const struct got_error *err = NULL;
57 	char *path;
58 	int fd = -1;
59 
60 	if (asprintf(&path, "%s/%s", path_got, name) == -1) {
61 		err = got_error_from_errno();
62 		path = NULL;
63 		goto done;
64 	}
65 
66 	fd = open(path, O_RDWR | O_CREAT | O_EXCL | O_NOFOLLOW,
67 	    GOT_DEFAULT_FILE_MODE);
68 	if (fd == -1) {
69 		err = got_error_from_errno();
70 		goto done;
71 	}
72 
73 	if (content) {
74 		int len = dprintf(fd, "%s\n", content);
75 		if (len != strlen(content) + 1) {
76 			err = got_error_from_errno();
77 			goto done;
78 		}
79 	}
80 
81 done:
82 	if (fd != -1 && close(fd) == -1 && err == NULL)
83 		err = got_error_from_errno();
84 	free(path);
85 	return err;
86 }
87 
88 static const struct got_error *
89 update_meta_file(const char *path_got, const char *name, const char *content)
90 {
91 	const struct got_error *err = NULL;
92 	FILE *tmpfile = NULL;
93 	char *tmppath = NULL;
94 	char *path = NULL;
95 
96 	if (asprintf(&path, "%s/%s", path_got, name) == -1) {
97 		err = got_error_from_errno();
98 		path = NULL;
99 		goto done;
100 	}
101 
102 	err = got_opentemp_named(&tmppath, &tmpfile, path);
103 	if (err)
104 		goto done;
105 
106 	if (content) {
107 		int len = fprintf(tmpfile, "%s\n", content);
108 		if (len != strlen(content) + 1) {
109 			err = got_error_from_errno();
110 			goto done;
111 		}
112 	}
113 
114 	if (rename(tmppath, path) != 0) {
115 		err = got_error_from_errno();
116 		goto done;
117 	}
118 
119 done:
120 	free(tmppath);
121 	fclose(tmpfile);
122 	return err;
123 }
124 
125 static const struct got_error *
126 read_meta_file(char **content, const char *path_got, const char *name)
127 {
128 	const struct got_error *err = NULL;
129 	char *path;
130 	int fd = -1;
131 	ssize_t n;
132 	struct stat sb;
133 
134 	*content = NULL;
135 
136 	if (asprintf(&path, "%s/%s", path_got, name) == -1) {
137 		err = got_error_from_errno();
138 		path = NULL;
139 		goto done;
140 	}
141 
142 	fd = open(path, O_RDONLY | O_NOFOLLOW);
143 	if (fd == -1) {
144 		err = got_error_from_errno();
145 		goto done;
146 	}
147 	if (flock(fd, LOCK_SH | LOCK_NB) == -1) {
148 		err = (errno == EWOULDBLOCK ? got_error(GOT_ERR_WORKTREE_BUSY)
149 		    : got_error_from_errno());
150 		goto done;
151 	}
152 
153 	stat(path, &sb);
154 	*content = calloc(1, sb.st_size);
155 	if (*content == NULL) {
156 		err = got_error_from_errno();
157 		goto done;
158 	}
159 
160 	n = read(fd, *content, sb.st_size);
161 	if (n != sb.st_size) {
162 		err = (n == -1 ? got_error_from_errno() :
163 		    got_error(GOT_ERR_WORKTREE_META));
164 		goto done;
165 	}
166 	if ((*content)[sb.st_size - 1] != '\n') {
167 		err = got_error(GOT_ERR_WORKTREE_META);
168 		goto done;
169 	}
170 	(*content)[sb.st_size - 1] = '\0';
171 
172 done:
173 	if (fd != -1 && close(fd) == -1 && err == NULL)
174 		err = got_error_from_errno();
175 	free(path);
176 	if (err) {
177 		free(*content);
178 		*content = NULL;
179 	}
180 	return err;
181 }
182 
183 const struct got_error *
184 got_worktree_init(const char *path, struct got_reference *head_ref,
185     const char *prefix, struct got_repository *repo)
186 {
187 	const struct got_error *err = NULL;
188 	struct got_object_id *commit_id = NULL;
189 	int obj_type;
190 	char *path_got = NULL;
191 	char *refstr = NULL;
192 	char *formatstr = NULL;
193 	char *absprefix = NULL;
194 	char *basestr = NULL;
195 
196 	err = got_ref_resolve(&commit_id, repo, head_ref);
197 	if (err)
198 		return err;
199 	err = got_object_get_type(&obj_type, repo, commit_id);
200 	if (err)
201 		return err;
202 	if (obj_type != GOT_OBJ_TYPE_COMMIT)
203 		return got_error(GOT_ERR_OBJ_TYPE);
204 
205 	if (!got_path_is_absolute(prefix)) {
206 		if (asprintf(&absprefix, "/%s", prefix) == -1)
207 			return got_error_from_errno();
208 	}
209 
210 	/* Create top-level directory (may already exist). */
211 	if (mkdir(path, GOT_DEFAULT_DIR_MODE) == -1 && errno != EEXIST) {
212 		err = got_error_from_errno();
213 		goto done;
214 	}
215 
216 	/* Create .got directory (may already exist). */
217 	if (asprintf(&path_got, "%s/%s", path, GOT_WORKTREE_GOT_DIR) == -1) {
218 		err = got_error_from_errno();
219 		goto done;
220 	}
221 	if (mkdir(path_got, GOT_DEFAULT_DIR_MODE) == -1 && errno != EEXIST) {
222 		err = got_error_from_errno();
223 		goto done;
224 	}
225 
226 	/* Create an empty lock file. */
227 	err = create_meta_file(path_got, GOT_WORKTREE_LOCK, NULL);
228 	if (err)
229 		goto done;
230 
231 	/* Create an empty file index. */
232 	err = create_meta_file(path_got, GOT_WORKTREE_FILE_INDEX, NULL);
233 	if (err)
234 		goto done;
235 
236 	/* Write the HEAD reference. */
237 	refstr = got_ref_to_str(head_ref);
238 	if (refstr == NULL) {
239 		err = got_error_from_errno();
240 		goto done;
241 	}
242 	err = create_meta_file(path_got, GOT_WORKTREE_HEAD_REF, refstr);
243 	if (err)
244 		goto done;
245 
246 	/* Record our base commit. */
247 	err = got_object_id_str(&basestr, commit_id);
248 	if (err)
249 		goto done;
250 	err = create_meta_file(path_got, GOT_WORKTREE_BASE_COMMIT, basestr);
251 	if (err)
252 		goto done;
253 
254 	/* Store path to repository. */
255 	err = create_meta_file(path_got, GOT_WORKTREE_REPOSITORY,
256 	    got_repo_get_path(repo));
257 	if (err)
258 		goto done;
259 
260 	/* Store in-repository path prefix. */
261 	err = create_meta_file(path_got, GOT_WORKTREE_PATH_PREFIX,
262 	    absprefix ? absprefix : prefix);
263 	if (err)
264 		goto done;
265 
266 	/* Stamp work tree with format file. */
267 	if (asprintf(&formatstr, "%d", GOT_WORKTREE_FORMAT_VERSION) == -1) {
268 		err = got_error_from_errno();
269 		goto done;
270 	}
271 	err = create_meta_file(path_got, GOT_WORKTREE_FORMAT, formatstr);
272 	if (err)
273 		goto done;
274 
275 done:
276 	free(commit_id);
277 	free(path_got);
278 	free(formatstr);
279 	free(refstr);
280 	free(absprefix);
281 	free(basestr);
282 	return err;
283 }
284 
285 const struct got_error *
286 got_worktree_open(struct got_worktree **worktree, const char *path)
287 {
288 	const struct got_error *err = NULL;
289 	char *path_got;
290 	char *formatstr = NULL;
291 	char *path_lock = NULL;
292 	char *base_commit_id_str = NULL;
293 	char *head_ref_str = NULL;
294 	int version, fd = -1;
295 	const char *errstr;
296 	struct got_repository *repo = NULL;
297 
298 	*worktree = NULL;
299 
300 	if (asprintf(&path_got, "%s/%s", path, GOT_WORKTREE_GOT_DIR) == -1) {
301 		err = got_error_from_errno();
302 		path_got = NULL;
303 		goto done;
304 	}
305 
306 	if (asprintf(&path_lock, "%s/%s", path_got, GOT_WORKTREE_LOCK) == -1) {
307 		err = got_error_from_errno();
308 		path_lock = NULL;
309 		goto done;
310 	}
311 
312 	fd = open(path_lock, O_RDWR | O_EXLOCK | O_NONBLOCK);
313 	if (fd == -1) {
314 		err = (errno == EWOULDBLOCK ? got_error(GOT_ERR_WORKTREE_BUSY)
315 		    : got_error_from_errno());
316 		goto done;
317 	}
318 
319 	err = read_meta_file(&formatstr, path_got, GOT_WORKTREE_FORMAT);
320 	if (err)
321 		goto done;
322 
323 	version = strtonum(formatstr, 1, INT_MAX, &errstr);
324 	if (errstr) {
325 		err = got_error(GOT_ERR_WORKTREE_META);
326 		goto done;
327 	}
328 	if (version != GOT_WORKTREE_FORMAT_VERSION) {
329 		err = got_error(GOT_ERR_WORKTREE_VERS);
330 		goto done;
331 	}
332 
333 	*worktree = calloc(1, sizeof(**worktree));
334 	if (*worktree == NULL) {
335 		err = got_error_from_errno();
336 		goto done;
337 	}
338 	(*worktree)->lockfd = -1;
339 
340 	(*worktree)->root_path = strdup(path);
341 	if ((*worktree)->root_path == NULL) {
342 		err = got_error_from_errno();
343 		goto done;
344 	}
345 	err = read_meta_file(&(*worktree)->repo_path, path_got,
346 	    GOT_WORKTREE_REPOSITORY);
347 	if (err)
348 		goto done;
349 
350 	err = read_meta_file(&(*worktree)->path_prefix, path_got,
351 	    GOT_WORKTREE_PATH_PREFIX);
352 	if (err)
353 		goto done;
354 
355 	err = read_meta_file(&base_commit_id_str, path_got,
356 	    GOT_WORKTREE_BASE_COMMIT);
357 	if (err)
358 		goto done;
359 
360 	err = got_repo_open(&repo, (*worktree)->repo_path);
361 	if (err)
362 		goto done;
363 
364 	err = got_object_resolve_id_str(&(*worktree)->base_commit_id, repo,
365 	    base_commit_id_str);
366 	if (err)
367 		goto done;
368 
369 	err = read_meta_file(&head_ref_str, path_got, GOT_WORKTREE_HEAD_REF);
370 	if (err)
371 		goto done;
372 
373 	err = got_ref_open(&(*worktree)->head_ref, repo, head_ref_str);
374 done:
375 	if (repo)
376 		got_repo_close(repo);
377 	free(path_got);
378 	free(path_lock);
379 	free(head_ref_str);
380 	free(base_commit_id_str);
381 	if (err) {
382 		if (fd != -1)
383 			close(fd);
384 		if (*worktree != NULL)
385 			got_worktree_close(*worktree);
386 		*worktree = NULL;
387 	} else
388 		(*worktree)->lockfd = fd;
389 
390 	return err;
391 }
392 
393 void
394 got_worktree_close(struct got_worktree *worktree)
395 {
396 	free(worktree->root_path);
397 	free(worktree->repo_path);
398 	free(worktree->path_prefix);
399 	free(worktree->base_commit_id);
400 	if (worktree->head_ref)
401 		got_ref_close(worktree->head_ref);
402 	if (worktree->lockfd != -1)
403 		close(worktree->lockfd);
404 	free(worktree);
405 }
406 
407 const char *
408 got_worktree_get_repo_path(struct got_worktree *worktree)
409 {
410 	return worktree->repo_path;
411 }
412 
413 const char *
414 got_worktree_get_path_prefix(struct got_worktree *worktree)
415 {
416 	return worktree->path_prefix;
417 }
418 
419 const struct got_error *
420 got_worktree_match_path_prefix(int *match, struct got_worktree *worktree,
421     const char *path_prefix)
422 {
423 	char *absprefix = NULL;
424 
425 	if (!got_path_is_absolute(path_prefix)) {
426 		if (asprintf(&absprefix, "/%s", path_prefix) == -1)
427 			return got_error_from_errno();
428 	}
429 	*match = (strcmp(absprefix ? absprefix : path_prefix,
430 	    worktree->path_prefix) == 0);
431 	free(absprefix);
432 	return NULL;
433 }
434 
435 char *
436 got_worktree_get_head_ref_name(struct got_worktree *worktree)
437 {
438 	return got_ref_to_str(worktree->head_ref);
439 }
440 
441 struct got_reference *
442 got_worktree_get_head_ref(struct got_worktree *worktree)
443 {
444 	return got_ref_dup(worktree->head_ref);
445 }
446 
447 const struct got_object_id *
448 got_worktree_get_base_commit_id(struct got_worktree *worktree)
449 {
450 	return worktree->base_commit_id;
451 }
452 
453 const struct got_error *
454 got_worktree_set_base_commit_id(struct got_worktree *worktree,
455     struct got_repository *repo, struct got_object_id *commit_id)
456 {
457 	const struct got_error *err;
458 	struct got_object *obj = NULL;
459 	char *id_str = NULL;
460 	char *path_got = NULL;
461 
462 	if (asprintf(&path_got, "%s/%s", worktree->root_path,
463 	    GOT_WORKTREE_GOT_DIR) == -1) {
464 		err = got_error_from_errno();
465 		path_got = NULL;
466 		goto done;
467 	}
468 
469 	err = got_object_open(&obj, repo, commit_id);
470 	if (err)
471 		return err;
472 
473 	if (obj->type != GOT_OBJ_TYPE_COMMIT) {
474 		err = got_error(GOT_ERR_OBJ_TYPE);
475 		goto done;
476 	}
477 
478 	/* Record our base commit. */
479 	err = got_object_id_str(&id_str, commit_id);
480 	if (err)
481 		goto done;
482 	err = update_meta_file(path_got, GOT_WORKTREE_BASE_COMMIT, id_str);
483 	if (err)
484 		goto done;
485 
486 	free(worktree->base_commit_id);
487 	worktree->base_commit_id = got_object_id_dup(commit_id);
488 	if (worktree->base_commit_id == NULL) {
489 		err = got_error_from_errno();
490 		goto done;
491 	}
492 done:
493 	if (obj)
494 		got_object_close(obj);
495 	free(id_str);
496 	free(path_got);
497 	return err;
498 }
499 
500 static const struct got_error *
501 lock_worktree(struct got_worktree *worktree, int operation)
502 {
503 	if (flock(worktree->lockfd, operation | LOCK_NB) == -1)
504 		return (errno == EWOULDBLOCK ? got_error(GOT_ERR_WORKTREE_BUSY)
505 		    : got_error_from_errno());
506 	return NULL;
507 }
508 
509 static const char *
510 apply_path_prefix(struct got_worktree *worktree, const char *path)
511 {
512 	const char *p = path;
513 	p += strlen(worktree->path_prefix);
514 	if (*p == '/')
515 		p++;
516 	return p;
517 }
518 
519 static const struct got_error *
520 blob_checkout(struct got_worktree *worktree, struct got_fileindex *fileindex,
521    struct got_fileindex_entry *entry, const char *path,
522    struct got_blob_object *blob, struct got_repository *repo,
523    got_worktree_checkout_cb progress_cb, void *progress_arg,
524    const char *progress_path)
525 {
526 	const struct got_error *err = NULL;
527 	char *ondisk_path;
528 	int fd = -1;
529 	size_t len, hdrlen;
530 	int update = 0;
531 	char *tmppath = NULL;
532 
533 	if (asprintf(&ondisk_path, "%s/%s", worktree->root_path,
534 	    apply_path_prefix(worktree, path)) == -1)
535 		return got_error_from_errno();
536 
537 	fd = open(ondisk_path, O_RDWR | O_CREAT | O_EXCL | O_NOFOLLOW,
538 	    GOT_DEFAULT_FILE_MODE);
539 	if (fd == -1) {
540 		err = got_error_from_errno();
541 		if (errno == EEXIST) {
542 			struct stat sb;
543 			if (lstat(ondisk_path, &sb) == -1) {
544 				err = got_error_from_errno();
545 				goto done;
546 			} else if (!S_ISREG(sb.st_mode)) {
547 				/* TODO file is obstructed; do something */
548 				err = got_error(GOT_ERR_FILE_OBSTRUCTED);
549 				goto done;
550 			} else {
551 				err = got_opentemp_named_fd(&tmppath, &fd,
552 				    ondisk_path);
553 				if (err)
554 					goto done;
555 				update = 1;
556 			}
557 		} else
558 			return err;
559 	}
560 
561 	(*progress_cb)(progress_arg,
562 	    update ? GOT_STATUS_UPDATE : GOT_STATUS_ADD, progress_path);
563 
564 	hdrlen = got_object_blob_get_hdrlen(blob);
565 	do {
566 		const uint8_t *buf = got_object_blob_get_read_buf(blob);
567 		err = got_object_blob_read_block(&len, blob);
568 		if (err)
569 			break;
570 		if (len > 0) {
571 			/* Skip blob object header first time around. */
572 			ssize_t outlen = write(fd, buf + hdrlen, len - hdrlen);
573 			if (outlen == -1) {
574 				err = got_error_from_errno();
575 				goto done;
576 			} else if (outlen != len - hdrlen) {
577 				err = got_error(GOT_ERR_IO);
578 				goto done;
579 			}
580 			hdrlen = 0;
581 		}
582 	} while (len != 0);
583 
584 	fsync(fd);
585 
586 	if (update) {
587 		if (rename(tmppath, ondisk_path) != 0) {
588 			err = got_error_from_errno();
589 			goto done;
590 		}
591 	}
592 
593 	if (entry)
594 		err = got_fileindex_entry_update(entry, ondisk_path,
595 		    blob->id.sha1, worktree->base_commit_id->sha1);
596 	else {
597 		err = got_fileindex_entry_alloc(&entry, ondisk_path,
598 		    apply_path_prefix(worktree, path), blob->id.sha1,
599 		    worktree->base_commit_id->sha1);
600 		if (err)
601 			goto done;
602 		err = got_fileindex_entry_add(fileindex, entry);
603 	}
604 	if (err)
605 		goto done;
606 done:
607 	if (fd != -1)
608 		close(fd);
609 	free(ondisk_path);
610 	free(tmppath);
611 	return err;
612 }
613 
614 static const struct got_error *
615 add_dir_on_disk(struct got_worktree *worktree, const char *path)
616 {
617 	const struct got_error *err = NULL;
618 	char *abspath;
619 
620 	if (asprintf(&abspath, "%s/%s", worktree->root_path,
621 	    apply_path_prefix(worktree, path)) == -1)
622 		return got_error_from_errno();
623 
624 	/* XXX queue work rather than editing disk directly? */
625 	if (mkdir(abspath, GOT_DEFAULT_DIR_MODE) == -1) {
626 		struct stat sb;
627 
628 		if (errno != EEXIST) {
629 			err = got_error_from_errno();
630 			goto done;
631 		}
632 
633 		if (lstat(abspath, &sb) == -1) {
634 			err = got_error_from_errno();
635 			goto done;
636 		}
637 
638 		if (!S_ISDIR(sb.st_mode)) {
639 			/* TODO directory is obstructed; do something */
640 			return got_error(GOT_ERR_FILE_OBSTRUCTED);
641 		}
642 	}
643 
644 done:
645 	free(abspath);
646 	return err;
647 }
648 
649 static const struct got_error *
650 tree_checkout(struct got_worktree *, struct got_fileindex *,
651     struct got_tree_object *, const char *, struct got_repository *,
652     got_worktree_checkout_cb progress_cb, void *progress_arg,
653     got_worktree_cancel_cb cancel_cb, void *cancel_arg);
654 
655 static const struct got_error *
656 tree_checkout_entry(struct got_worktree *worktree,
657     struct got_fileindex *fileindex, struct got_tree_entry *te,
658     const char *parent, struct got_repository *repo,
659     got_worktree_checkout_cb progress_cb, void *progress_arg,
660     got_worktree_cancel_cb cancel_cb, void *cancel_arg)
661 {
662 	const struct got_error *err = NULL;
663 	struct got_object *obj = NULL;
664 	struct got_blob_object *blob = NULL;
665 	struct got_fileindex_entry *entry = NULL;
666 	struct got_tree_object *tree = NULL;
667 	char *path = NULL;
668 	char *progress_path = NULL;
669 	size_t len;
670 
671 	if (parent[0] == '/' && parent[1] == '\0')
672 		parent = "";
673 	if (asprintf(&path, "%s/%s", parent, te->name) == -1)
674 		return got_error_from_errno();
675 
676 	/* Skip this entry if it is outside of our path prefix. */
677 	len = MIN(strlen(worktree->path_prefix), strlen(path));
678 	if (strncmp(path, worktree->path_prefix, len) != 0) {
679 		free(path);
680 		return NULL;
681 	}
682 
683 	err = got_object_open(&obj, repo, te->id);
684 	if (err)
685 		goto done;
686 
687 	progress_path = path;
688 	if (strncmp(progress_path, worktree->path_prefix, len) == 0)
689 		progress_path += len;
690 
691 	switch (obj->type) {
692 	case GOT_OBJ_TYPE_BLOB:
693 		if (strlen(worktree->path_prefix) >= strlen(path))
694 			break;
695 		entry = got_fileindex_entry_get(fileindex,
696 		    apply_path_prefix(worktree, path));
697 		if (entry &&
698 		    memcmp(entry->commit_sha1, worktree->base_commit_id->sha1,
699 		    SHA1_DIGEST_LENGTH) == 0) {
700 			(*progress_cb)(progress_arg, GOT_STATUS_EXISTS,
701 			    progress_path);
702 			break;
703 		}
704 		if (entry && memcmp(entry->blob_sha1, obj->id.sha1,
705 		    SHA1_DIGEST_LENGTH) == 0)
706 			break;
707 		err = got_object_blob_open(&blob, repo, obj, 8192);
708 		if (err)
709 			goto done;
710 		err = blob_checkout(worktree, fileindex, entry, path, blob,
711 		    repo, progress_cb, progress_arg, progress_path);
712 		break;
713 	case GOT_OBJ_TYPE_TREE:
714 		if (strlen(worktree->path_prefix) < strlen(path)) {
715 			err = add_dir_on_disk(worktree, path);
716 			if (err)
717 				break;
718 		}
719 		err = got_object_tree_open(&tree, repo, obj);
720 		if (err)
721 			goto done;
722 		/* XXX infinite recursion possible */
723 		err = tree_checkout(worktree, fileindex, tree, path, repo,
724 		    progress_cb, progress_arg, cancel_cb, cancel_arg);
725 		break;
726 	default:
727 		break;
728 	}
729 
730 done:
731 	if (blob)
732 		got_object_blob_close(blob);
733 	if (tree)
734 		got_object_tree_close(tree);
735 	if (obj)
736 		got_object_close(obj);
737 	free(path);
738 	return err;
739 }
740 
741 struct collect_missing_entry_args {
742 	struct got_fileindex *fileindex;
743 	const struct got_tree_entries *entries;
744 	struct got_pathset *missing_entries;
745 	const char *current_subdir;
746 };
747 
748 static const struct got_error *
749 collect_missing_file(void *args, struct got_fileindex_entry *entry)
750 {
751 	struct collect_missing_entry_args *a = args;
752 	char *start, *end;
753 	ptrdiff_t len;
754 	struct got_tree_entry *te;
755 	int found = 0;
756 
757 	if (a->current_subdir[0] != '\0' &&
758 	    strncmp(a->current_subdir, entry->path,
759 	    strlen(a->current_subdir)) != 0)
760 		return NULL;
761 
762 	start = entry->path + strlen(a->current_subdir);
763 	while (start[0] == '/')
764 		start++;
765 	end = strchr(start, '/');
766 	if (end == NULL) {
767 		end = strchr(start, '\0');
768 		if (end == NULL)
769 			return got_error(GOT_ERR_BAD_PATH);
770 	}
771 	len = end - start;
772 
773 	SIMPLEQ_FOREACH(te, &a->entries->head, entry) {
774 		if (strncmp(start, te->name, len) == 0 &&
775 		    te->name[len] == '\0') {
776 			found = 1;
777 			break;
778 		}
779 	}
780 
781 	if (found)
782 		return NULL;
783 
784 	return got_pathset_add(a->missing_entries, entry->path, entry);
785 }
786 
787 struct remove_missing_file_args {
788     const char *root_path;
789     struct got_fileindex *fileindex;
790     got_worktree_checkout_cb progress_cb;
791     void *progress_arg;
792     got_worktree_cancel_cb cancel_cb;
793     void *cancel_arg;
794 };
795 
796 static const struct got_error *
797 remove_missing_file(const char *path, void *data, void *arg)
798 {
799 	const struct got_error *err = NULL;
800 	char *ondisk_path = NULL;
801 	struct remove_missing_file_args *a = arg;
802 	struct got_fileindex_entry *entry = data;
803 
804 	if (a->cancel_cb) {
805 		err = (*a->cancel_cb)(a->cancel_arg);
806 		if (err)
807 			return err;
808 	}
809 
810 	(*a->progress_cb)(a->progress_arg, GOT_STATUS_DELETE, path);
811 
812 	if (asprintf(&ondisk_path, "%s/%s", a->root_path, path) == -1)
813 		return got_error_from_errno();
814 
815 	if (unlink(ondisk_path) == -1)
816 		err = got_error_from_errno();
817 	else {
818 		char *parent = dirname(ondisk_path);
819 		while (parent && strcmp(parent, a->root_path) != 0) {
820 			if (rmdir(parent) == -1) {
821 				if (errno != ENOTEMPTY)
822 					err = got_error_from_errno();
823 				break;
824 			}
825 			parent = dirname(parent);
826 		}
827 	}
828 	free(ondisk_path);
829 
830 	if (err == NULL) {
831 		got_fileindex_entry_remove(a->fileindex, entry);
832 		got_fileindex_entry_free(entry);
833 	}
834 	return err;
835 }
836 
837 /* Remove files which exist in the file index but not in the tree. */
838 static const struct got_error *
839 remove_missing_files(struct got_worktree *worktree, const char *path,
840     struct got_fileindex *fileindex, const struct got_tree_entries *entries,
841     got_worktree_checkout_cb progress_cb, void *progress_arg,
842     got_worktree_cancel_cb cancel_cb, void *cancel_arg)
843 {
844 	const struct got_error *err = NULL;
845 	struct collect_missing_entry_args a;
846 	struct remove_missing_file_args a2;
847 
848 	a.fileindex = fileindex;
849 	a.entries = entries;
850 	a.missing_entries = got_pathset_alloc();
851 	if (a.missing_entries == NULL)
852 		return got_error_from_errno();
853 	a.current_subdir = apply_path_prefix(worktree, path);
854 	err = got_fileindex_for_each_entry_safe(fileindex,
855 	    collect_missing_file, &a);
856 	if (err)
857 		return err;
858 
859 	a2.root_path = worktree->root_path;
860 	a2.fileindex = fileindex;
861 	a2.cancel_cb = cancel_cb;
862 	a2.cancel_arg = cancel_arg;
863 	a2.progress_cb = progress_cb;
864 	a2.progress_arg = progress_arg;
865 	err = got_pathset_for_each_safe(a.missing_entries, remove_missing_file,
866 	    &a2);
867 	got_pathset_free(a.missing_entries);
868 	return err;
869 }
870 
871 static const struct got_error *
872 tree_checkout(struct got_worktree *worktree,
873     struct got_fileindex *fileindex, struct got_tree_object *tree,
874     const char *path, struct got_repository *repo,
875     got_worktree_checkout_cb progress_cb, void *progress_arg,
876     got_worktree_cancel_cb cancel_cb, void *cancel_arg)
877 {
878 	const struct got_error *err = NULL;
879 	const struct got_tree_entries *entries;
880 	struct got_tree_entry *te;
881 	size_t len;
882 
883 	/* Skip this tree if it shares no path components with the prefix. */
884 	len = MIN(strlen(worktree->path_prefix), strlen(path));
885 	if (strncmp(path, worktree->path_prefix, len) != 0)
886 		return NULL;
887 
888 	entries = got_object_tree_get_entries(tree);
889 	SIMPLEQ_FOREACH(te, &entries->head, entry) {
890 		if (cancel_cb) {
891 			err = (*cancel_cb)(cancel_arg);
892 			if (err)
893 				return err;
894 		}
895 		err = tree_checkout_entry(worktree, fileindex, te, path, repo,
896 		    progress_cb, progress_arg, cancel_cb, cancel_arg);
897 		if (err)
898 			return err;
899 	}
900 
901 	len = strlen(worktree->path_prefix);
902 	if (strncmp(worktree->path_prefix, path, len) == 0) {
903 		err = remove_missing_files(worktree, path, fileindex, entries,
904 		    progress_cb, progress_arg, cancel_cb, cancel_arg);
905 		if (err)
906 			return err;
907 	}
908 
909 	return err;
910 }
911 
912 const struct got_error *
913 got_worktree_checkout_files(struct got_worktree *worktree,
914     struct got_repository *repo, got_worktree_checkout_cb progress_cb,
915     void *progress_arg, got_worktree_cancel_cb cancel_cb, void *cancel_arg)
916 {
917 	const struct got_error *err = NULL, *unlockerr, *checkout_err = NULL;
918 	struct got_commit_object *commit = NULL;
919 	struct got_tree_object *tree = NULL;
920 	char *fileindex_path = NULL, *new_fileindex_path = NULL;
921 	struct got_fileindex *fileindex = NULL;
922 	FILE *index = NULL, *new_index = NULL;
923 
924 	err = lock_worktree(worktree, LOCK_EX);
925 	if (err)
926 		return err;
927 
928 	fileindex = got_fileindex_alloc();
929 	if (fileindex == NULL) {
930 		err = got_error_from_errno();
931 		goto done;
932 	}
933 
934 	if (asprintf(&fileindex_path, "%s/%s/%s", worktree->root_path,
935 	    GOT_WORKTREE_GOT_DIR, GOT_WORKTREE_FILE_INDEX) == -1) {
936 		err = got_error_from_errno();
937 		fileindex_path = NULL;
938 		goto done;
939 	}
940 
941 	/*
942 	 * Read the file index.
943 	 * Checking out files is supposed to be an idempotent operation.
944 	 * If the on-disk file index is incomplete we will try to complete it.
945 	 */
946 	index = fopen(fileindex_path, "rb");
947 	if (index == NULL) {
948 		err = got_error_from_errno();
949 		goto done;
950 	}
951 	err = got_fileindex_read(fileindex, index);
952 	fclose(index);
953 	if (err)
954 		goto done;
955 
956 	err = got_opentemp_named(&new_fileindex_path, &new_index,
957 	    fileindex_path);
958 	if (err)
959 		goto done;
960 
961 	err = got_object_open_as_commit(&commit, repo,
962 	   worktree->base_commit_id);
963 	if (err)
964 		goto done;
965 
966 	err = got_object_open_as_tree(&tree, repo, commit->tree_id);
967 	if (err)
968 		goto done;
969 
970 	checkout_err = tree_checkout(worktree, fileindex, tree, "/", repo,
971 	    progress_cb, progress_arg, cancel_cb, cancel_arg);
972 
973 	/* Try to sync the fileindex back to disk in any case. */
974 	err = got_fileindex_write(fileindex, new_index);
975 	if (err)
976 		goto done;
977 
978 	if (rename(new_fileindex_path, fileindex_path) != 0) {
979 		err = got_error_from_errno();
980 		goto done;
981 	}
982 
983 	free(new_fileindex_path);
984 	new_fileindex_path = NULL;
985 
986 done:
987 	if (tree)
988 		got_object_tree_close(tree);
989 	if (commit)
990 		got_object_commit_close(commit);
991 	if (new_fileindex_path)
992 		unlink(new_fileindex_path);
993 	if (new_index)
994 		fclose(new_index);
995 	free(new_fileindex_path);
996 	free(fileindex_path);
997 	got_fileindex_free(fileindex);
998 	if (checkout_err)
999 		err = checkout_err;
1000 	unlockerr = lock_worktree(worktree, LOCK_SH);
1001 	if (unlockerr && err == NULL)
1002 		err = unlockerr;
1003 	return err;
1004 }