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/queue.h>
18 #include <sys/stat.h>
19 
20 #include <limits.h>
21 #include <stdlib.h>
22 #include <stdio.h>
23 #include <sha1.h>
24 #include <string.h>
25 #include <zlib.h>
26 #include <errno.h>
27 #include <libgen.h>
28 
29 #include "got_error.h"
30 #include "got_reference.h"
31 #include "got_repository.h"
32 #include "got_worktree.h"
33 #include "got_object.h"
34 
35 #include "got_lib_path.h"
36 #include "got_lib_delta.h"
37 #include "got_lib_inflate.h"
38 #include "got_lib_object.h"
39 #include "got_lib_pack.h"
40 #include "got_lib_repository.h"
41 #include "got_lib_worktree.h"
42 #include "got_lib_object_idcache.h"
43 
44 #ifndef nitems
45 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
46 #endif
47 
48 #define GOT_GIT_DIR	".git"
49 
50 /* Mandatory files and directories inside the git directory. */
51 #define GOT_OBJECTS_DIR		"objects"
52 #define GOT_REFS_DIR		"refs"
53 #define GOT_HEAD_FILE		"HEAD"
54 
55 /* Other files and directories inside the git directory. */
56 #define GOT_FETCH_HEAD_FILE	"FETCH_HEAD"
57 #define GOT_ORIG_HEAD_FILE	"ORIG_HEAD"
58 #define GOT_OBJECTS_PACK_DIR	"objects/pack"
59 
60 char *
61 got_repo_get_path(struct got_repository *repo)
62 {
63 	return strdup(repo->path);
64 }
65 
66 char *
67 got_repo_get_path_git_dir(struct got_repository *repo)
68 {
69 	return strdup(repo->path_git_dir);
70 }
71 
72 int
73 got_repo_is_bare(struct got_repository *repo)
74 {
75 	return (strcmp(repo->path, repo->path_git_dir) == 0);
76 }
77 
78 static char *
79 get_path_git_child(struct got_repository *repo, const char *basename)
80 {
81 	char *path_child;
82 	
83 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
84 	    basename) == -1)
85 		return NULL;
86 
87 	return path_child;
88 }
89 
90 char *
91 got_repo_get_path_objects(struct got_repository *repo)
92 {
93 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
94 }
95 
96 char *
97 got_repo_get_path_objects_pack(struct got_repository *repo)
98 {
99 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
100 }
101 
102 char *
103 got_repo_get_path_refs(struct got_repository *repo)
104 {
105 	return get_path_git_child(repo, GOT_REFS_DIR);
106 }
107 
108 static char *
109 get_path_head(struct got_repository *repo)
110 {
111 	return get_path_git_child(repo, GOT_HEAD_FILE);
112 }
113 
114 static int
115 is_git_repo(struct got_repository *repo)
116 {
117 	char *path_git = got_repo_get_path_git_dir(repo);
118 	char *path_objects = got_repo_get_path_objects(repo);
119 	char *path_refs = got_repo_get_path_refs(repo);
120 	char *path_head = get_path_head(repo);
121 	int ret = 0;
122 	struct stat sb;
123 	struct got_reference *head_ref;
124 
125 	if (lstat(path_git, &sb) == -1)
126 		goto done;
127 	if (!S_ISDIR(sb.st_mode))
128 		goto done;
129 
130 	if (lstat(path_objects, &sb) == -1)
131 		goto done;
132 	if (!S_ISDIR(sb.st_mode))
133 		goto done;
134 
135 	if (lstat(path_refs, &sb) == -1)
136 		goto done;
137 	if (!S_ISDIR(sb.st_mode))
138 		goto done;
139 
140 	if (lstat(path_head, &sb) == -1)
141 		goto done;
142 	if (!S_ISREG(sb.st_mode))
143 		goto done;
144 
145 	/* Check if the HEAD reference can be opened. */
146 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD) != NULL)
147 		goto done;
148 	got_ref_close(head_ref);
149 
150 	ret = 1;
151 done:
152 	free(path_git);
153 	free(path_objects);
154 	free(path_refs);
155 	free(path_head);
156 	return ret;
157 
158 }
159 
160 #ifndef GOT_NO_OBJ_CACHE
161 static const struct got_error *
162 cache_add(struct got_object_cache *cache, struct got_object_id *id, void *item)
163 {
164 	const struct got_error *err = NULL;
165 	struct got_object_cache_entry *ce;
166 	int nelem;
167 
168 	nelem = got_object_idcache_num_elements(cache->idcache);
169 	if (nelem >= cache->size) {
170 		err = got_object_idcache_remove_least_used((void **)&ce,
171 		    cache->idcache);
172 		if (err)
173 			return err;
174 		switch (cache->type) {
175 		case GOT_OBJECT_CACHE_TYPE_OBJ:
176 			got_object_close(ce->data.obj);
177 			break;
178 		case GOT_OBJECT_CACHE_TYPE_TREE:
179 			got_object_tree_close(ce->data.tree);
180 			break;
181 		case GOT_OBJECT_CACHE_TYPE_COMMIT:
182 			got_object_commit_close(ce->data.commit);
183 			break;
184 		}
185 		free(ce);
186 	}
187 
188 	ce = calloc(1, sizeof(*ce));
189 	if (ce == NULL)
190 		return got_error_from_errno();
191 	memcpy(&ce->id, id, sizeof(ce->id));
192 	switch (cache->type) {
193 	case GOT_OBJECT_CACHE_TYPE_OBJ:
194 		ce->data.obj = (struct got_object *)item;
195 		break;
196 	case GOT_OBJECT_CACHE_TYPE_TREE:
197 		ce->data.tree = (struct got_tree_object *)item;
198 		break;
199 	case GOT_OBJECT_CACHE_TYPE_COMMIT:
200 		ce->data.commit = (struct got_commit_object *)item;
201 		break;
202 	}
203 
204 	err = got_object_idcache_add(cache->idcache, id, ce);
205 	if (err) {
206 		if (err->code == GOT_ERR_OBJ_EXISTS) {
207 			free(ce);
208 			err = NULL;
209 		}
210 	}
211 	return err;
212 }
213 #endif
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 = cache_add(&repo->objcache, id, obj);
222 	if (err)
223 		return err;
224 	obj->refcnt++;
225 #endif
226 	return NULL;
227 }
228 
229 struct got_object *
230 got_repo_get_cached_object(struct got_repository *repo,
231     struct got_object_id *id)
232 {
233 	struct got_object_cache_entry *ce;
234 
235 	ce = got_object_idcache_get(repo->objcache.idcache, id);
236 	if (ce) {
237 		repo->objcache.cache_hit++;
238 		return ce->data.obj;
239 	}
240 
241 	repo->objcache.cache_miss++;
242 	return NULL;
243 }
244 
245 const struct got_error *
246 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
247     struct got_tree_object *tree)
248 {
249 #ifndef GOT_NO_OBJ_CACHE
250 	const struct got_error *err = NULL;
251 	err = cache_add(&repo->treecache, id, tree);
252 	if (err)
253 		return err;
254 	tree->refcnt++;
255 #endif
256 	return NULL;
257 }
258 
259 struct got_tree_object *
260 got_repo_get_cached_tree(struct got_repository *repo,
261     struct got_object_id *id)
262 {
263 	struct got_object_cache_entry *ce;
264 
265 	ce = got_object_idcache_get(repo->treecache.idcache, id);
266 	if (ce) {
267 		repo->treecache.cache_hit++;
268 		return ce->data.tree;
269 	}
270 
271 	repo->treecache.cache_miss++;
272 	return NULL;
273 }
274 
275 const struct got_error *
276 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
277     struct got_commit_object *commit)
278 {
279 #ifndef GOT_NO_OBJ_CACHE
280 	const struct got_error *err = NULL;
281 	err = cache_add(&repo->commitcache, id, commit);
282 	if (err)
283 		return err;
284 
285 	commit->refcnt++;
286 #endif
287 	return NULL;
288 }
289 
290 struct got_commit_object *
291 got_repo_get_cached_commit(struct got_repository *repo,
292     struct got_object_id *id)
293 {
294 	struct got_object_cache_entry *ce;
295 
296 	ce = got_object_idcache_get(repo->commitcache.idcache, id);
297 	if (ce) {
298 		repo->commitcache.cache_hit++;
299 		return ce->data.commit;
300 	}
301 
302 	repo->commitcache.cache_miss++;
303 	return NULL;
304 }
305 
306 const struct got_error *
307 open_repo(struct got_repository *repo, const char *path)
308 {
309 	const struct got_error *err = NULL;
310 	struct got_worktree *worktree = NULL;
311 
312 	/* bare git repository? */
313 	repo->path_git_dir = strdup(path);
314 	if (repo->path_git_dir == NULL) {
315 		err = got_error_from_errno();
316 		goto done;
317 	}
318 	if (is_git_repo(repo)) {
319 		repo->path = strdup(repo->path_git_dir);
320 		if (repo->path == NULL) {
321 			err = got_error_from_errno();
322 			goto done;
323 		}
324 		return NULL;
325 	}
326 
327 	/* git repository with working tree? */
328 	free(repo->path_git_dir);
329 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
330 		err = got_error_from_errno();
331 		goto done;
332 	}
333 	if (is_git_repo(repo)) {
334 		repo->path = strdup(path);
335 		if (repo->path == NULL) {
336 			err = got_error_from_errno();
337 			goto done;
338 		}
339 		return NULL;
340 	}
341 
342 	/* got work tree checked out from bare git repository? */
343 	free(repo->path_git_dir);
344 	repo->path_git_dir = NULL;
345 	err = got_worktree_open(&worktree, path);
346 	if (err) {
347 		if (err->code == GOT_ERR_ERRNO && errno == ENOENT)
348 			err = got_error(GOT_ERR_NOT_GIT_REPO);
349 		goto done;
350 	}
351 	repo->path_git_dir = strdup(worktree->repo_path);
352 	if (repo->path_git_dir == NULL) {
353 		err = got_error_from_errno();
354 		goto done;
355 	}
356 
357 	/* got work tree checked out from git repository with working tree? */
358 	if (!is_git_repo(repo)) {
359 		free(repo->path_git_dir);
360 		if (asprintf(&repo->path_git_dir, "%s/%s", worktree->repo_path,
361 		    GOT_GIT_DIR) == -1) {
362 			err = got_error_from_errno();
363 			repo->path_git_dir = NULL;
364 			goto done;
365 		}
366 		if (!is_git_repo(repo)) {
367 			err = got_error(GOT_ERR_NOT_GIT_REPO);
368 			goto done;
369 		}
370 		repo->path = strdup(worktree->repo_path);
371 		if (repo->path == NULL) {
372 			err = got_error_from_errno();
373 			goto done;
374 		}
375 	} else {
376 		repo->path = strdup(repo->path_git_dir);
377 		if (repo->path == NULL) {
378 			err = got_error_from_errno();
379 			goto done;
380 		}
381 	}
382 done:
383 	if (worktree)
384 		got_worktree_close(worktree);
385 	return err;
386 }
387 
388 const struct got_error *
389 got_repo_open(struct got_repository **repop, const char *path)
390 {
391 	struct got_repository *repo = NULL;
392 	const struct got_error *err = NULL;
393 	char *abspath, *normpath = NULL;
394 	int tried_root = 0;
395 
396 	*repop = NULL;
397 
398 	if (got_path_is_absolute(path))
399 		abspath = strdup(path);
400 	else
401 		abspath = got_path_get_absolute(path);
402 	if (abspath == NULL)
403 		return got_error(GOT_ERR_BAD_PATH);
404 
405 	repo = calloc(1, sizeof(*repo));
406 	if (repo == NULL) {
407 		err = got_error_from_errno();
408 		goto done;
409 	}
410 
411 	repo->objcache.type = GOT_OBJECT_CACHE_TYPE_OBJ;
412 	repo->objcache.size = GOT_OBJECT_CACHE_SIZE_OBJ;
413 	repo->objcache.idcache = got_object_idcache_alloc(repo->objcache.size);
414 	if (repo->objcache.idcache == NULL) {
415 		err = got_error_from_errno();
416 		goto done;
417 	}
418 
419 	repo->treecache.type = GOT_OBJECT_CACHE_TYPE_TREE;
420 	repo->treecache.size = GOT_OBJECT_CACHE_SIZE_TREE;
421 	repo->treecache.idcache =
422 	    got_object_idcache_alloc(repo->treecache.size);
423 	if (repo->treecache.idcache == NULL) {
424 		err = got_error_from_errno();
425 		goto done;
426 	}
427 
428 	repo->commitcache.type = GOT_OBJECT_CACHE_TYPE_COMMIT;
429 	repo->commitcache.size = GOT_OBJECT_CACHE_SIZE_COMMIT;
430 	repo->commitcache.idcache =
431 	    got_object_idcache_alloc(repo->commitcache.size);
432 	if (repo->commitcache.idcache == NULL) {
433 		err = got_error_from_errno();
434 		goto done;
435 	}
436 
437 	normpath = got_path_normalize(abspath);
438 	if (normpath == NULL) {
439 		err = got_error(GOT_ERR_BAD_PATH);
440 		goto done;
441 	}
442 
443 	path = normpath;
444 	do {
445 		err = open_repo(repo, path);
446 		if (err == NULL)
447 			break;
448 		if (err->code != GOT_ERR_NOT_GIT_REPO)
449 			break;
450 		if (path[0] == '/' && path[1] == '\0') {
451 			if (tried_root) {
452 				err = got_error(GOT_ERR_NOT_GIT_REPO);
453 				break;
454 			}
455 			tried_root = 1;
456 		}
457 		path = dirname(path);
458 		if (path == NULL)
459 			err = got_error_from_errno();
460 	} while (path);
461 done:
462 	if (err)
463 		got_repo_close(repo);
464 	else
465 		*repop = repo;
466 	free(abspath);
467 	free(normpath);
468 	return err;
469 }
470 
471 #if 0
472 static void
473 print_cache_stats(struct got_object_cache *cache, const char *name)
474 {
475 	fprintf(stderr, "%s cache: %d elements, %d hits, %d missed\n",
476 	    name, got_object_idcache_num_elements(cache->idcache),
477 	    cache->cache_hit, cache->cache_miss);
478 }
479 
480 void check_refcount(struct got_object_id *id, void *data, void *arg)
481 {
482 	struct got_object_cache *cache = arg;
483 	struct got_object_cache_entry *ce = data;
484 	struct got_object *obj;
485 	struct got_tree_object *tree;
486 	struct got_commit_object *commit;
487 	char *id_str;
488 
489 	if (got_object_id_str(&id_str, id) != NULL)
490 		return;
491 
492 	switch (cache->type) {
493 	case GOT_OBJECT_CACHE_TYPE_OBJ:
494 		obj = ce->data.obj;
495 		if (obj->refcnt == 1)
496 			break;
497 		fprintf(stderr, "object %s has %d unclaimed references\n",
498 		    id_str, obj->refcnt - 1);
499 		break;
500 	case GOT_OBJECT_CACHE_TYPE_TREE:
501 		tree = ce->data.tree;
502 		if (tree->refcnt == 1)
503 			break;
504 		fprintf(stderr, "tree %s has %d unclaimed references\n",
505 		    id_str, tree->refcnt - 1);
506 		break;
507 	case GOT_OBJECT_CACHE_TYPE_COMMIT:
508 		commit = ce->data.commit;
509 		if (commit->refcnt == 1)
510 			break;
511 		fprintf(stderr, "commit %s has %d unclaimed references\n",
512 		    id_str, commit->refcnt);
513 		break;
514 	}
515 	free(id_str);
516 }
517 #endif
518 
519 void
520 got_repo_close(struct got_repository *repo)
521 {
522 	int i;
523 
524 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
525 		if (repo->packidx_cache[i] == NULL)
526 			break;
527 		got_packidx_close(repo->packidx_cache[i]);
528 	}
529 
530 	for (i = 0; i < nitems(repo->packs); i++) {
531 		if (repo->packs[i].path_packfile == NULL)
532 			break;
533 		got_pack_close(&repo->packs[i]);
534 	}
535 
536 	free(repo->path);
537 	free(repo->path_git_dir);
538 
539 #if 0
540 	print_cache_stats(&repo->objcache, "object");
541 	print_cache_stats(&repo->treecache, "tree");
542 	print_cache_stats(&repo->commitcache, "commit");
543 	got_object_idcache_for_each(repo->objcache.idcache, check_refcount,
544 	    &repo->objcache);
545 	got_object_idcache_for_each(repo->treecache.idcache, check_refcount,
546 	    &repo->treecache);
547 	got_object_idcache_for_each(repo->commitcache.idcache, check_refcount,
548 	    &repo->commitcache);
549 #endif
550 
551 	if (repo->objcache.idcache)
552 		got_object_idcache_free(repo->objcache.idcache);
553 	if (repo->treecache.idcache)
554 		got_object_idcache_free(repo->treecache.idcache);
555 	if (repo->commitcache.idcache)
556 		got_object_idcache_free(repo->commitcache.idcache);
557 	free(repo);
558 }
559 
560 const struct got_error *
561 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
562     const char *input_path)
563 {
564 	const struct got_error *err = NULL;
565 	char *repo_abspath = NULL, *cwd = NULL;
566 	struct stat sb;
567 	size_t repolen, cwdlen, len;
568 	char *canonpath, *path;
569 
570 	*in_repo_path = NULL;
571 
572 	cwd = getcwd(NULL, 0);
573 	if (cwd == NULL)
574 		return got_error_from_errno();
575 
576 	canonpath = strdup(input_path);
577 	if (canonpath == NULL) {
578 		err = got_error_from_errno();
579 		goto done;
580 	}
581 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
582 	if (err)
583 		goto done;
584 
585 	repo_abspath = got_repo_get_path(repo);
586 	if (repo_abspath == NULL) {
587 		err = got_error_from_errno();
588 		goto done;
589 	}
590 
591 	/* TODO: Call "get in-repository path of work-tree node" API. */
592 
593 	if (lstat(canonpath, &sb) != 0) {
594 		if (errno != ENOENT) {
595 			err = got_error_from_errno();
596 			goto done;
597 		}
598 		/*
599 		 * Path is not on disk.
600 		 * Assume it is already relative to repository root.
601 		 */
602 		path = strdup(canonpath);
603 	} else {
604 		int is_repo_child = 0, is_cwd_child = 0;
605 
606 		path = realpath(canonpath, NULL);
607 		if (path == NULL) {
608 			err = got_error_from_errno();
609 			goto done;
610 		}
611 
612 		repolen = strlen(repo_abspath);
613 		cwdlen = strlen(cwd);
614 		len = strlen(path);
615 
616 		if (len > repolen && strncmp(path, repo_abspath, repolen) == 0)
617 			is_repo_child = 1;
618 		if (len > cwdlen && strncmp(path, cwd, cwdlen) == 0)
619 			is_cwd_child = 1;
620 
621 		if (strcmp(path, repo_abspath) == 0) {
622 			free(path);
623 			path = strdup("");
624 			if (path == NULL) {
625 				err = got_error_from_errno();
626 				goto done;
627 			}
628 		} else if (is_repo_child && is_cwd_child) {
629 			char *child;
630 			/* TODO: Is path inside a got worktree? */
631 			/* Strip common prefix with repository path. */
632 			err = got_path_skip_common_ancestor(&child,
633 			    repo_abspath, path);
634 			if (err)
635 				goto done;
636 			free(path);
637 			path = child;
638 		} else if (is_repo_child) {
639 			/* Matched an on-disk path inside repository. */
640 			if (got_repo_is_bare(repo)) {
641 				/*
642 				 * Matched an on-disk path inside repository
643 				 * database. Treat as repository-relative.
644 				 */
645 			} else {
646 				char *child;
647 				/* Strip common prefix with repository path. */
648 				err = got_path_skip_common_ancestor(&child,
649 				    repo_abspath, path);
650 				if (err)
651 					goto done;
652 				free(path);
653 				path = child;
654 			}
655 		} else if (is_cwd_child) {
656 			char *child;
657 			/* TODO: Is path inside a got worktree? */
658 			/* Strip common prefix with cwd. */
659 			err = got_path_skip_common_ancestor(&child, cwd,
660 			    path);
661 			if (err)
662 				goto done;
663 			free(path);
664 			path = child;
665 		} else {
666 			/*
667 			 * Matched unrelated on-disk path.
668 			 * Treat it as repository-relative.
669 			 */
670 		}
671 	}
672 
673 	/* Make in-repository path absolute */
674 	if (path[0] != '/') {
675 		char *abspath;
676 		if (asprintf(&abspath, "/%s", path) == -1) {
677 			err = got_error_from_errno();
678 			goto done;
679 		}
680 		free(path);
681 		path = abspath;
682 	}
683 
684 done:
685 	free(repo_abspath);
686 	free(cwd);
687 	free(canonpath);
688 	if (err)
689 		free(path);
690 	else
691 		*in_repo_path = path;
692 	return err;
693 }