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 <string.h>
22 #include <stdio.h>
23 #include <stdlib.h>
24 #include <fcntl.h>
25 #include <errno.h>
26 #include <unistd.h>
27 #include <sha1.h>
28 #include <zlib.h>
29 #include <fnmatch.h>
30 
31 #include "got_error.h"
32 #include "got_repository.h"
33 #include "got_reference.h"
34 #include "got_object.h"
35 #include "got_worktree.h"
36 #include "got_opentemp.h"
37 
38 #include "got_lib_worktree.h"
39 #include "got_lib_path.h"
40 #include "got_lib_sha1.h"
41 #include "got_lib_fileindex.h"
42 #include "got_lib_inflate.h"
43 #include "got_lib_delta.h"
44 #include "got_lib_object.h"
45 
46 #ifndef MIN
47 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
48 #endif
49 
50 static const struct got_error *
51 create_meta_file(const char *path_got, const char *name, const char *content)
52 {
53 	const struct got_error *err = NULL;
54 	char *path;
55 	int fd = -1;
56 	char buf[4];
57 	ssize_t n;
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 	/* The file should be empty. */
73 	n = read(fd, buf, sizeof(buf));
74 	if (n != 0) {
75 		err = (n == -1 ? got_error_from_errno() :
76 		    got_error(GOT_ERR_WORKTREE_EXISTS));
77 		goto done;
78 	}
79 
80 	if (content) {
81 		int len = dprintf(fd, "%s\n", content);
82 		if (len != strlen(content) + 1) {
83 			err = got_error_from_errno();
84 			goto done;
85 		}
86 	}
87 
88 done:
89 	if (fd != -1 && close(fd) == -1 && err == NULL)
90 		err = got_error_from_errno();
91 	free(path);
92 	return err;
93 }
94 
95 static const struct got_error *
96 read_meta_file(char **content, const char *path_got, const char *name)
97 {
98 	const struct got_error *err = NULL;
99 	char *path;
100 	int fd = -1;
101 	ssize_t n;
102 	struct stat sb;
103 
104 	*content = NULL;
105 
106 	if (asprintf(&path, "%s/%s", path_got, name) == -1) {
107 		err = got_error_from_errno();
108 		path = NULL;
109 		goto done;
110 	}
111 
112 	fd = open(path, O_RDONLY | O_NOFOLLOW);
113 	if (fd == -1) {
114 		err = got_error_from_errno();
115 		goto done;
116 	}
117 	if (flock(fd, LOCK_SH | LOCK_NB) == -1) {
118 		err = (errno == EWOULDBLOCK ? got_error(GOT_ERR_WORKTREE_BUSY)
119 		    : got_error_from_errno());
120 		goto done;
121 	}
122 
123 	stat(path, &sb);
124 	*content = calloc(1, sb.st_size);
125 	if (*content == NULL) {
126 		err = got_error_from_errno();
127 		goto done;
128 	}
129 
130 	n = read(fd, *content, sb.st_size);
131 	if (n != sb.st_size) {
132 		err = (n == -1 ? got_error_from_errno() :
133 		    got_error(GOT_ERR_WORKTREE_META));
134 		goto done;
135 	}
136 	if ((*content)[sb.st_size - 1] != '\n') {
137 		err = got_error(GOT_ERR_WORKTREE_META);
138 		goto done;
139 	}
140 	(*content)[sb.st_size - 1] = '\0';
141 
142 done:
143 	if (fd != -1 && close(fd) == -1 && err == NULL)
144 		err = got_error_from_errno();
145 	free(path);
146 	if (err) {
147 		free(*content);
148 		*content = NULL;
149 	}
150 	return err;
151 }
152 
153 const struct got_error *
154 got_worktree_init(const char *path, struct got_reference *head_ref,
155     const char *prefix, struct got_repository *repo)
156 {
157 	const struct got_error *err = NULL;
158 	struct got_object_id *commit_id = NULL;
159 	int obj_type;
160 	char *path_got = NULL;
161 	char *refstr = NULL;
162 	char *repo_path = NULL;
163 	char *formatstr = NULL;
164 	char *absprefix = NULL;
165 	char *basestr = NULL;
166 
167 	err = got_ref_resolve(&commit_id, repo, head_ref);
168 	if (err)
169 		return err;
170 	err = got_object_get_type(&obj_type, repo, commit_id);
171 	if (err)
172 		return err;
173 	if (obj_type != GOT_OBJ_TYPE_COMMIT)
174 		return got_error(GOT_ERR_OBJ_TYPE);
175 
176 	if (!got_path_is_absolute(prefix)) {
177 		if (asprintf(&absprefix, "/%s", prefix) == -1)
178 			return got_error_from_errno();
179 	}
180 
181 	/* Create top-level directory (may already exist). */
182 	if (mkdir(path, GOT_DEFAULT_DIR_MODE) == -1 && errno != EEXIST) {
183 		err = got_error_from_errno();
184 		goto done;
185 	}
186 
187 	/* Create .got directory (may already exist). */
188 	if (asprintf(&path_got, "%s/%s", path, GOT_WORKTREE_GOT_DIR) == -1) {
189 		err = got_error_from_errno();
190 		goto done;
191 	}
192 	if (mkdir(path_got, GOT_DEFAULT_DIR_MODE) == -1 && errno != EEXIST) {
193 		err = got_error_from_errno();
194 		goto done;
195 	}
196 
197 	/* Create an empty lock file. */
198 	err = create_meta_file(path_got, GOT_WORKTREE_LOCK, NULL);
199 	if (err)
200 		goto done;
201 
202 	/* Create an empty file index. */
203 	err = create_meta_file(path_got, GOT_WORKTREE_FILE_INDEX, NULL);
204 	if (err)
205 		goto done;
206 
207 	/* Write the HEAD reference. */
208 	refstr = got_ref_to_str(head_ref);
209 	if (refstr == NULL) {
210 		err = got_error_from_errno();
211 		goto done;
212 	}
213 	err = create_meta_file(path_got, GOT_WORKTREE_HEAD_REF, refstr);
214 	if (err)
215 		goto done;
216 
217 	/* Record our base commit. */
218 	err = got_object_id_str(&basestr, commit_id);
219 	if (err)
220 		goto done;
221 	err = create_meta_file(path_got, GOT_WORKTREE_BASE_COMMIT, basestr);
222 	if (err)
223 		goto done;
224 
225 	/* Store path to repository. */
226 	repo_path = got_repo_get_path(repo);
227 	if (repo_path == NULL) {
228 		err = got_error_from_errno();
229 		goto done;
230 	}
231 	err = create_meta_file(path_got, GOT_WORKTREE_REPOSITORY, repo_path);
232 	if (err)
233 		goto done;
234 
235 	/* Store in-repository path prefix. */
236 	err = create_meta_file(path_got, GOT_WORKTREE_PATH_PREFIX,
237 	    absprefix ? absprefix : prefix);
238 	if (err)
239 		goto done;
240 
241 	/* Stamp work tree with format file. */
242 	if (asprintf(&formatstr, "%d", GOT_WORKTREE_FORMAT_VERSION) == -1) {
243 		err = got_error_from_errno();
244 		goto done;
245 	}
246 	err = create_meta_file(path_got, GOT_WORKTREE_FORMAT, formatstr);
247 	if (err)
248 		goto done;
249 
250 done:
251 	free(commit_id);
252 	free(path_got);
253 	free(formatstr);
254 	free(refstr);
255 	free(repo_path);
256 	free(absprefix);
257 	free(basestr);
258 	return err;
259 }
260 
261 const struct got_error *
262 got_worktree_open(struct got_worktree **worktree, const char *path)
263 {
264 	const struct got_error *err = NULL;
265 	char *path_got;
266 	char *formatstr = NULL;
267 	char *path_lock = NULL;
268 	char *base_commit_id_str = NULL;
269 	char *head_ref_str = NULL;
270 	int version, fd = -1;
271 	const char *errstr;
272 	struct got_repository *repo = NULL;
273 
274 	*worktree = NULL;
275 
276 	if (asprintf(&path_got, "%s/%s", path, GOT_WORKTREE_GOT_DIR) == -1) {
277 		err = got_error_from_errno();
278 		path_got = NULL;
279 		goto done;
280 	}
281 
282 	if (asprintf(&path_lock, "%s/%s", path_got, GOT_WORKTREE_LOCK) == -1) {
283 		err = got_error_from_errno();
284 		path_lock = NULL;
285 		goto done;
286 	}
287 
288 	fd = open(path_lock, O_RDWR | O_EXLOCK | O_NONBLOCK);
289 	if (fd == -1) {
290 		err = (errno == EWOULDBLOCK ? got_error(GOT_ERR_WORKTREE_BUSY)
291 		    : got_error_from_errno());
292 		goto done;
293 	}
294 
295 	err = read_meta_file(&formatstr, path_got, GOT_WORKTREE_FORMAT);
296 	if (err)
297 		goto done;
298 
299 	version = strtonum(formatstr, 1, INT_MAX, &errstr);
300 	if (errstr) {
301 		err = got_error(GOT_ERR_WORKTREE_META);
302 		goto done;
303 	}
304 	if (version != GOT_WORKTREE_FORMAT_VERSION) {
305 		err = got_error(GOT_ERR_WORKTREE_VERS);
306 		goto done;
307 	}
308 
309 	*worktree = calloc(1, sizeof(**worktree));
310 	if (*worktree == NULL) {
311 		err = got_error_from_errno();
312 		goto done;
313 	}
314 	(*worktree)->lockfd = -1;
315 
316 	(*worktree)->root_path = strdup(path);
317 	if ((*worktree)->root_path == NULL) {
318 		err = got_error_from_errno();
319 		goto done;
320 	}
321 	err = read_meta_file(&(*worktree)->repo_path, path_got,
322 	    GOT_WORKTREE_REPOSITORY);
323 	if (err)
324 		goto done;
325 
326 	err = read_meta_file(&(*worktree)->path_prefix, path_got,
327 	    GOT_WORKTREE_PATH_PREFIX);
328 	if (err)
329 		goto done;
330 
331 	err = read_meta_file(&base_commit_id_str, path_got,
332 	    GOT_WORKTREE_BASE_COMMIT);
333 	if (err)
334 		goto done;
335 
336 	err = got_repo_open(&repo, (*worktree)->repo_path);
337 	if (err)
338 		goto done;
339 
340 	err = got_object_resolve_id_str(&(*worktree)->base_commit_id, repo,
341 	    base_commit_id_str);
342 	if (err)
343 		goto done;
344 
345 	err = read_meta_file(&head_ref_str, path_got, GOT_WORKTREE_HEAD_REF);
346 	if (err)
347 		goto done;
348 
349 	err = got_ref_open(&(*worktree)->head_ref, repo, head_ref_str);
350 done:
351 	if (repo)
352 		got_repo_close(repo);
353 	free(path_got);
354 	free(path_lock);
355 	free(head_ref_str);
356 	free(base_commit_id_str);
357 	if (err) {
358 		if (fd != -1)
359 			close(fd);
360 		if (*worktree != NULL)
361 			got_worktree_close(*worktree);
362 		*worktree = NULL;
363 	} else
364 		(*worktree)->lockfd = fd;
365 
366 	return err;
367 }
368 
369 void
370 got_worktree_close(struct got_worktree *worktree)
371 {
372 	free(worktree->root_path);
373 	free(worktree->repo_path);
374 	free(worktree->path_prefix);
375 	free(worktree->base_commit_id);
376 	if (worktree->head_ref)
377 		got_ref_close(worktree->head_ref);
378 	if (worktree->lockfd != -1)
379 		close(worktree->lockfd);
380 	free(worktree);
381 }
382 
383 char *
384 got_worktree_get_repo_path(struct got_worktree *worktree)
385 {
386 	return strdup(worktree->repo_path);
387 }
388 
389 char *
390 got_worktree_get_head_ref_name(struct got_worktree *worktree)
391 {
392 	return got_ref_to_str(worktree->head_ref);
393 }
394 
395 static const struct got_error *
396 lock_worktree(struct got_worktree *worktree, int operation)
397 {
398 	if (flock(worktree->lockfd, operation | LOCK_NB) == -1)
399 		return (errno == EWOULDBLOCK ? got_error(GOT_ERR_WORKTREE_BUSY)
400 		    : got_error_from_errno());
401 	return NULL;
402 }
403 
404 static const char *
405 apply_path_prefix(struct got_worktree *worktree, const char *path)
406 {
407 	const char *p = path;
408 	p += strlen(worktree->path_prefix);
409 	if (*p == '/')
410 		p++;
411 	return p;
412 }
413 
414 static const struct got_error *
415 add_file_on_disk(struct got_worktree *worktree, struct got_fileindex *fileindex,
416    const char *path, struct got_blob_object *blob, struct got_repository *repo)
417 {
418 	const struct got_error *err = NULL;
419 	char *ondisk_path;
420 	int fd;
421 	size_t len, hdrlen;
422 	struct got_fileindex_entry *entry;
423 
424 	if (asprintf(&ondisk_path, "%s/%s", worktree->root_path,
425 	    apply_path_prefix(worktree, path)) == -1)
426 		return got_error_from_errno();
427 
428 	fd = open(ondisk_path, O_RDWR | O_CREAT | O_EXCL | O_NOFOLLOW,
429 	    GOT_DEFAULT_FILE_MODE);
430 	if (fd == -1) {
431 		err = got_error_from_errno();
432 		if (errno == EEXIST) {
433 			struct stat sb;
434 			if (lstat(ondisk_path, &sb) == -1) {
435 				err = got_error_from_errno();
436 			} else if (!S_ISREG(sb.st_mode)) {
437 				/* TODO file is obstructed; do something */
438 				err = got_error(GOT_ERR_FILE_OBSTRUCTED);
439 			}
440 		}
441 		return err;
442 	}
443 
444 	hdrlen = got_object_blob_get_hdrlen(blob);
445 	do {
446 		const uint8_t *buf = got_object_blob_get_read_buf(blob);
447 		err = got_object_blob_read_block(&len, blob);
448 		if (err)
449 			break;
450 		if (len > 0) {
451 			/* Skip blob object header first time around. */
452 			ssize_t outlen = write(fd, buf + hdrlen, len - hdrlen);
453 			if (outlen == -1) {
454 				err = got_error_from_errno();
455 				goto done;
456 			} else if (outlen != len - hdrlen) {
457 				err = got_error(GOT_ERR_IO);
458 				goto done;
459 			}
460 			hdrlen = 0;
461 		}
462 	} while (len != 0);
463 
464 	fsync(fd);
465 
466 	err = got_fileindex_entry_alloc(&entry, ondisk_path,
467 	    apply_path_prefix(worktree, path), blob->id.sha1,
468 	    worktree->base_commit_id->sha1);
469 	if (err)
470 		goto done;
471 
472 	err = got_fileindex_entry_add(fileindex, entry);
473 	if (err)
474 		goto done;
475 done:
476 	close(fd);
477 	free(ondisk_path);
478 	return err;
479 }
480 
481 static const struct got_error *
482 add_dir_on_disk(struct got_worktree *worktree, const char *path)
483 {
484 	const struct got_error *err = NULL;
485 	char *abspath;
486 
487 	if (asprintf(&abspath, "%s/%s", worktree->root_path,
488 	    apply_path_prefix(worktree, path)) == -1)
489 		return got_error_from_errno();
490 
491 	/* XXX queue work rather than editing disk directly? */
492 	if (mkdir(abspath, GOT_DEFAULT_DIR_MODE) == -1) {
493 		struct stat sb;
494 
495 		if (errno != EEXIST) {
496 			err = got_error_from_errno();
497 			goto done;
498 		}
499 
500 		if (lstat(abspath, &sb) == -1) {
501 			err = got_error_from_errno();
502 			goto done;
503 		}
504 
505 		if (!S_ISDIR(sb.st_mode)) {
506 			/* TODO directory is obstructed; do something */
507 			return got_error(GOT_ERR_FILE_OBSTRUCTED);
508 		}
509 	}
510 
511 done:
512 	free(abspath);
513 	return err;
514 }
515 
516 static const struct got_error *
517 tree_checkout(struct got_worktree *, struct got_fileindex *,
518     struct got_tree_object *, const char *, struct got_repository *,
519     got_worktree_checkout_cb progress_cb, void *progress_arg,
520     got_worktree_cancel_cb cancel_cb, void *cancel_arg);
521 
522 static const struct got_error *
523 tree_checkout_entry(struct got_worktree *worktree,
524     struct got_fileindex *fileindex, struct got_tree_entry *te,
525     const char *parent, struct got_repository *repo,
526     got_worktree_checkout_cb progress_cb, void *progress_arg,
527     got_worktree_cancel_cb cancel_cb, void *cancel_arg)
528 {
529 	const struct got_error *err = NULL;
530 	struct got_object *obj = NULL;
531 	struct got_blob_object *blob = NULL;
532 	struct got_tree_object *tree = NULL;
533 	char *path = NULL;
534 	char *progress_path = NULL;
535 	size_t len;
536 
537 	if (parent[0] == '/' && parent[1] == '\0')
538 		parent = "";
539 	if (asprintf(&path, "%s/%s", parent, te->name) == -1)
540 		return got_error_from_errno();
541 
542 	/* Skip this entry if it is outside of our path prefix. */
543 	len = MIN(strlen(worktree->path_prefix), strlen(path));
544 	if (strncmp(path, worktree->path_prefix, len) != 0) {
545 		free(path);
546 		return NULL;
547 	}
548 
549 	err = got_object_open(&obj, repo, te->id);
550 	if (err)
551 		goto done;
552 
553 	progress_path = path;
554 	if (strncmp(progress_path, worktree->path_prefix, len) == 0)
555 		progress_path += len;
556 	(*progress_cb)(progress_arg, progress_path);
557 
558 	switch (obj->type) {
559 	case GOT_OBJ_TYPE_BLOB:
560 		if (strlen(worktree->path_prefix) >= strlen(path))
561 			break;
562 		err = got_object_blob_open(&blob, repo, obj, 8192);
563 		if (err)
564 			goto done;
565 		err = add_file_on_disk(worktree, fileindex, path, blob, repo);
566 		break;
567 	case GOT_OBJ_TYPE_TREE:
568 		err = got_object_tree_open(&tree, repo, obj);
569 		if (err)
570 			goto done;
571 		if (strlen(worktree->path_prefix) < strlen(path)) {
572 			err = add_dir_on_disk(worktree, path);
573 			if (err)
574 				break;
575 		}
576 		/* XXX infinite recursion possible */
577 		err = tree_checkout(worktree, fileindex, tree, path, repo,
578 		    progress_cb, progress_arg, cancel_cb, cancel_arg);
579 		break;
580 	default:
581 		break;
582 	}
583 
584 done:
585 	if (blob)
586 		got_object_blob_close(blob);
587 	if (tree)
588 		got_object_tree_close(tree);
589 	if (obj)
590 		got_object_close(obj);
591 	free(path);
592 	return err;
593 }
594 
595 static const struct got_error *
596 tree_checkout(struct got_worktree *worktree,
597     struct got_fileindex *fileindex, struct got_tree_object *tree,
598     const char *path, struct got_repository *repo,
599     got_worktree_checkout_cb progress_cb, void *progress_arg,
600     got_worktree_cancel_cb cancel_cb, void *cancel_arg)
601 {
602 	const struct got_error *err = NULL;
603 	const struct got_tree_entries *entries;
604 	struct got_tree_entry *te;
605 	size_t len;
606 
607 	/* Skip this tree if it is outside of our path prefix. */
608 	len = MIN(strlen(worktree->path_prefix), strlen(path));
609 	if (strncmp(path, worktree->path_prefix, len) != 0)
610 		return NULL;
611 
612 	entries = got_object_tree_get_entries(tree);
613 	SIMPLEQ_FOREACH(te, &entries->head, entry) {
614 		if (cancel_cb) {
615 			err = (*cancel_cb)(cancel_arg);
616 			if (err)
617 				break;
618 		}
619 		err = tree_checkout_entry(worktree, fileindex, te, path, repo,
620 		    progress_cb, progress_arg, cancel_cb, cancel_arg);
621 		if (err)
622 			break;
623 	}
624 
625 	return err;
626 }
627 
628 const struct got_error *
629 got_worktree_checkout_files(struct got_worktree *worktree,
630     struct got_repository *repo, got_worktree_checkout_cb progress_cb,
631     void *progress_arg, got_worktree_cancel_cb cancel_cb, void *cancel_arg)
632 {
633 	const struct got_error *err = NULL, *unlockerr, *checkout_err = NULL;
634 	struct got_commit_object *commit = NULL;
635 	struct got_tree_object *tree = NULL;
636 	char *fileindex_path = NULL, *new_fileindex_path = NULL;
637 	struct got_fileindex *fileindex = NULL;
638 	FILE *new_index = NULL;
639 
640 	err = lock_worktree(worktree, LOCK_EX);
641 	if (err)
642 		return err;
643 
644 	fileindex = got_fileindex_alloc();
645 	if (fileindex == NULL) {
646 		err = got_error_from_errno();
647 		goto done;
648 	}
649 
650 	if (asprintf(&fileindex_path, "%s/%s/%s", worktree->root_path,
651 	    GOT_WORKTREE_GOT_DIR, GOT_WORKTREE_FILE_INDEX) == -1) {
652 		err = got_error_from_errno();
653 		fileindex_path = NULL;
654 		goto done;
655 	}
656 
657 	err = got_opentemp_named(&new_fileindex_path, &new_index,
658 	    fileindex_path);
659 	if (err)
660 		goto done;
661 
662 	err = got_object_open_as_commit(&commit, repo,
663 	   worktree->base_commit_id);
664 	if (err)
665 		goto done;
666 
667 	err = got_object_open_as_tree(&tree, repo, commit->tree_id);
668 	if (err)
669 		goto done;
670 
671 	checkout_err = tree_checkout(worktree, fileindex, tree, "/", repo,
672 	    progress_cb, progress_arg, cancel_cb, cancel_arg);
673 
674 	/* Try to sync the fileindex back to disk in any case. */
675 	err = got_fileindex_write(fileindex, new_index);
676 	if (err)
677 		goto done;
678 
679 	if (rename(new_fileindex_path, fileindex_path) != 0) {
680 		err = got_error_from_errno();
681 		goto done;
682 	}
683 
684 	free(new_fileindex_path);
685 	new_fileindex_path = NULL;
686 
687 done:
688 	if (tree)
689 		got_object_tree_close(tree);
690 	if (commit)
691 		got_object_commit_close(commit);
692 	if (new_fileindex_path)
693 		unlink(new_fileindex_path);
694 	if (new_index)
695 		fclose(new_index);
696 	free(new_fileindex_path);
697 	free(fileindex_path);
698 	got_fileindex_free(fileindex);
699 	if (checkout_err)
700 		err = checkout_err;
701 	unlockerr = lock_worktree(worktree, LOCK_SH);
702 	if (unlockerr && err == NULL)
703 		err = unlockerr;
704 	return err;
705 }