1 /*
2  * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/stat.h>
19 
20 #include <limits.h>
21 #include <stdio.h>
22 #include <stdlib.h>
23 #include <string.h>
24 #include <zlib.h>
25 #include <ctype.h>
26 
27 #include "got_compat.h"
28 
29 #include "got_error.h"
30 #include "got_object.h"
31 #include "got_cancel.h"
32 #include "got_commit_graph.h"
33 #include "got_path.h"
34 
35 #include "got_lib_delta.h"
36 #include "got_lib_inflate.h"
37 #include "got_lib_object.h"
38 #include "got_lib_object_idset.h"
39 
40 struct got_commit_graph_node {
41 	struct got_object_id id;
42 
43 	/* Used only during iteration. */
44 	time_t timestamp;
45 	TAILQ_ENTRY(got_commit_graph_node) entry;
46 };
47 
48 TAILQ_HEAD(got_commit_graph_iter_list, got_commit_graph_node);
49 
50 struct got_commit_graph_branch_tip {
51 	struct got_object_id *commit_id;
52 	struct got_commit_object *commit;
53 	struct got_commit_graph_node *new_node;
54 };
55 
56 struct got_commit_graph {
57 	/* The set of all commits we have traversed. */
58 	struct got_object_idset *node_ids;
59 
60 	int flags;
61 #define GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL		0x01
62 
63 	/*
64 	 * A set of object IDs of known parent commits which we have not yet
65 	 * traversed. Each commit ID in this set represents a branch in commit
66 	 * history: Either the first-parent branch of the head node, or another
67 	 * branch corresponding to a traversed merge commit for which we have
68 	 * not traversed a branch point commit yet.
69 	 *
70 	 * Whenever we add a commit with a matching ID to the graph, we remove
71 	 * its corresponding element from this set, and add new elements for
72 	 * each of that commit's parent commits which were not traversed yet.
73 	 *
74 	 * When API users ask us to fetch more commits, we fetch commits from
75 	 * all currently open branches. This allows API users to process
76 	 * commits in linear order even though the history contains branches.
77 	 */
78 	struct got_object_idset *open_branches;
79 
80 	/* Array of branch tips for fetch_commits_from_open_branches(). */
81 	struct got_commit_graph_branch_tip *tips;
82 	int ntips;
83 
84 	/* Path of tree entry of interest to the API user. */
85 	char *path;
86 
87 	/*
88 	 * Nodes which will be passed to the API user next, sorted by
89 	 * commit timestmap.
90 	 */
91 	struct got_commit_graph_iter_list iter_list;
92 };
93 
94 static const struct got_error *
95 detect_changed_path(int *changed, struct got_commit_object *commit,
96     struct got_object_id *commit_id, const char *path,
97     struct got_repository *repo)
98 {
99 	const struct got_error *err = NULL;
100 	struct got_commit_object *pcommit = NULL;
101 	struct got_tree_object *tree = NULL, *ptree = NULL;
102 	struct got_object_qid *pid;
103 
104 	if (got_path_is_root_dir(path)) {
105 		*changed = 1;
106 		return NULL;
107 	}
108 
109 	*changed = 0;
110 
111 	pid = STAILQ_FIRST(&commit->parent_ids);
112 	if (pid == NULL) {
113 		struct got_object_id *obj_id;
114 		err = got_object_id_by_path(&obj_id, repo, commit, path);
115 		if (err) {
116 			if (err->code == GOT_ERR_NO_TREE_ENTRY)
117 				err = NULL;
118 		} else
119 			*changed = 1; /* The path was created in this commit. */
120 		free(obj_id);
121 		return err;
122 	}
123 
124 	err = got_object_open_as_tree(&tree, repo, commit->tree_id);
125 	if (err)
126 		return err;
127 
128 	err = got_object_open_as_commit(&pcommit, repo, pid->id);
129 	if (err)
130 		goto done;
131 
132 	err = got_object_open_as_tree(&ptree, repo, pcommit->tree_id);
133 	if (err)
134 		goto done;
135 
136 	err = got_object_tree_path_changed(changed, tree, ptree, path, repo);
137 done:
138 	if (tree)
139 		got_object_tree_close(tree);
140 	if (ptree)
141 		got_object_tree_close(ptree);
142 	if (pcommit)
143 		got_object_commit_close(pcommit);
144 	return err;
145 }
146 
147 static void
148 add_node_to_iter_list(struct got_commit_graph *graph,
149     struct got_commit_graph_node *node, time_t committer_time)
150 {
151 	struct got_commit_graph_node *n, *next;
152 
153 	node->timestamp = committer_time;
154 
155 	n = TAILQ_FIRST(&graph->iter_list);
156 	while (n) {
157 		next = TAILQ_NEXT(n, entry);
158 		if (next && node->timestamp >= next->timestamp) {
159 			TAILQ_INSERT_BEFORE(next, node, entry);
160 			return;
161 		}
162 		n = next;
163 	}
164 	TAILQ_INSERT_TAIL(&graph->iter_list, node, entry);
165 }
166 
167 static const struct got_error *
168 add_node(struct got_commit_graph_node **new_node,
169     struct got_commit_graph *graph, struct got_object_id *commit_id,
170     struct got_repository *repo)
171 {
172 	const struct got_error *err = NULL;
173 	struct got_commit_graph_node *node;
174 
175 	*new_node = NULL;
176 
177 	node = calloc(1, sizeof(*node));
178 	if (node == NULL)
179 		return got_error_from_errno("calloc");
180 
181 	memcpy(&node->id, commit_id, sizeof(node->id));
182 	err = got_object_idset_add(graph->node_ids, &node->id, NULL);
183 	if (err)
184 		free(node);
185 	else
186 		*new_node = node;
187 	return err;
188 }
189 
190 /*
191  * Ask got-read-pack to traverse first-parent history until a commit is
192  * encountered which modified graph->path, or until the pack file runs
193  * out of relevant commits. This is faster than sending an individual
194  * request for each commit stored in the pack file.
195  */
196 static const struct got_error *
197 packed_first_parent_traversal(int *ncommits_traversed,
198     struct got_commit_graph *graph, struct got_object_id *commit_id,
199     struct got_repository *repo)
200 {
201 	const struct got_error *err = NULL;
202 	struct got_object_id_queue traversed_commits;
203 	struct got_object_qid *qid;
204 
205 	STAILQ_INIT(&traversed_commits);
206 	*ncommits_traversed = 0;
207 
208 	err = got_traverse_packed_commits(&traversed_commits,
209 	    commit_id, graph->path, repo);
210 	if (err)
211 		return err;
212 
213 	/* Add all traversed commits to the graph... */
214 	STAILQ_FOREACH(qid, &traversed_commits, entry) {
215 		struct got_commit_graph_node *node;
216 
217 		if (got_object_idset_contains(graph->open_branches, qid->id))
218 			continue;
219 		if (got_object_idset_contains(graph->node_ids, qid->id))
220 			continue;
221 
222 		(*ncommits_traversed)++;
223 
224 		/* ... except the last commit is the new branch tip. */
225 		if (STAILQ_NEXT(qid, entry) == NULL) {
226 			err = got_object_idset_add(graph->open_branches,
227 			    qid->id, NULL);
228 			break;
229 		}
230 
231 		err = add_node(&node, graph, qid->id, repo);
232 		if (err)
233 			break;
234 	}
235 
236 	got_object_id_queue_free(&traversed_commits);
237 	return err;
238 }
239 
240 static const struct got_error *
241 close_branch(struct got_commit_graph *graph, struct got_object_id *commit_id)
242 {
243 	const struct got_error *err;
244 
245 	err = got_object_idset_remove(NULL, graph->open_branches, commit_id);
246 	if (err && err->code != GOT_ERR_NO_OBJ)
247 		return err;
248 	return NULL;
249 }
250 
251 static const struct got_error *
252 advance_branch(struct got_commit_graph *graph, struct got_object_id *commit_id,
253     struct got_commit_object *commit, struct got_repository *repo)
254 {
255 	const struct got_error *err;
256 	struct got_object_qid *qid;
257 
258 	err = close_branch(graph, commit_id);
259 	if (err)
260 		return err;
261 
262 	if (graph->flags & GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL) {
263 		qid = STAILQ_FIRST(&commit->parent_ids);
264 		if (qid == NULL ||
265 		    got_object_idset_contains(graph->open_branches, qid->id))
266 			return NULL;
267 		/*
268 		 * The root directory always changes by definition, and when
269 		 * logging the root we want to traverse consecutive commits
270 		 * even if they point at the same tree.
271 		 * But if we are looking for a specific path then we can avoid
272 		 * fetching packed commits which did not modify the path and
273 		 * only fetch their IDs. This speeds up 'got blame'.
274 		 */
275 		if (!got_path_is_root_dir(graph->path) &&
276 		    (commit->flags & GOT_COMMIT_FLAG_PACKED)) {
277 			int ncommits = 0;
278 			err = packed_first_parent_traversal(&ncommits,
279 			    graph, qid->id, repo);
280 			if (err || ncommits > 0)
281 				return err;
282 		}
283 		return got_object_idset_add(graph->open_branches,
284 		    qid->id, NULL);
285 	}
286 
287 	/*
288 	 * If we are graphing commits for a specific path, skip branches
289 	 * which do not contribute any content to this path.
290 	 */
291 	if (commit->nparents > 1 && !got_path_is_root_dir(graph->path)) {
292 		struct got_object_id *merged_id, *prev_id = NULL;
293 		int branches_differ = 0;
294 
295 		err = got_object_id_by_path(&merged_id, repo, commit,
296 		    graph->path);
297 		if (err)
298 			return err;
299 
300 		STAILQ_FOREACH(qid, &commit->parent_ids, entry) {
301 			struct got_object_id *id = NULL;
302 			struct got_commit_object *pcommit = NULL;
303 
304 			if (got_object_idset_contains(graph->open_branches,
305 			    qid->id))
306 				continue;
307 
308 			err = got_object_open_as_commit(&pcommit, repo,
309 			    qid->id);
310 			if (err) {
311 				free(merged_id);
312 				free(prev_id);
313 				return err;
314 			}
315 			err = got_object_id_by_path(&id, repo, pcommit,
316 			    graph->path);
317 			got_object_commit_close(pcommit);
318 			pcommit = NULL;
319 			if (err) {
320 				if (err->code == GOT_ERR_NO_TREE_ENTRY) {
321 					branches_differ = 1;
322 					continue;
323 				}
324 				free(merged_id);
325 				free(prev_id);
326 				return err;
327 			}
328 
329 			if (prev_id) {
330 				if (!branches_differ &&
331 				    got_object_id_cmp(id, prev_id) != 0)
332 					branches_differ = 1;
333 				free(prev_id);
334 			}
335 			prev_id = id;
336 
337 			/*
338 			 * If a branch has created the merged content we can
339 			 * skip any other branches.
340 			 */
341 			if (got_object_id_cmp(merged_id, id) == 0) {
342 				err = got_object_idset_add(graph->open_branches,
343 				    qid->id, NULL);
344 				free(merged_id);
345 				free(id);
346 				return err;
347 			}
348 		}
349 
350 		free(prev_id);
351 		prev_id = NULL;
352 		free(merged_id);
353 		merged_id = NULL;
354 
355 		/*
356 		 * If the path's content is the same on all branches,
357 		 * follow the first parent only.
358 		 */
359 		if (!branches_differ) {
360 			qid = STAILQ_FIRST(&commit->parent_ids);
361 			if (qid == NULL)
362 				return NULL;
363 			if (got_object_idset_contains(graph->open_branches,
364 			    qid->id))
365 				return NULL;
366 			if (got_object_idset_contains(graph->node_ids,
367 			    qid->id))
368 				return NULL; /* parent already traversed */
369 			return got_object_idset_add(graph->open_branches,
370 			    qid->id, NULL);
371 		}
372 	}
373 
374 	STAILQ_FOREACH(qid, &commit->parent_ids, entry) {
375 		if (got_object_idset_contains(graph->open_branches, qid->id))
376 			continue;
377 		if (got_object_idset_contains(graph->node_ids, qid->id))
378 			continue; /* parent already traversed */
379 		err = got_object_idset_add(graph->open_branches, qid->id, NULL);
380 		if (err)
381 			return err;
382 	}
383 
384 	return NULL;
385 }
386 
387 const struct got_error *
388 got_commit_graph_open(struct got_commit_graph **graph,
389     const char *path, int first_parent_traversal)
390 {
391 	const struct got_error *err = NULL;
392 
393 	*graph = calloc(1, sizeof(**graph));
394 	if (*graph == NULL)
395 		return got_error_from_errno("calloc");
396 
397 	TAILQ_INIT(&(*graph)->iter_list);
398 
399 	(*graph)->path = strdup(path);
400 	if ((*graph)->path == NULL) {
401 		err = got_error_from_errno("strdup");
402 		goto done;
403 	}
404 
405 	(*graph)->node_ids = got_object_idset_alloc();
406 	if ((*graph)->node_ids == NULL) {
407 		err = got_error_from_errno("got_object_idset_alloc");
408 		goto done;
409 	}
410 
411 	(*graph)->open_branches = got_object_idset_alloc();
412 	if ((*graph)->open_branches == NULL) {
413 		err = got_error_from_errno("got_object_idset_alloc");
414 		goto done;
415 	}
416 
417 	if (first_parent_traversal)
418 		(*graph)->flags |= GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL;
419 done:
420 	if (err) {
421 		got_commit_graph_close(*graph);
422 		*graph = NULL;
423 	}
424 	return err;
425 }
426 
427 struct add_branch_tip_arg {
428 	struct got_commit_graph_branch_tip *tips;
429 	int ntips;
430 	struct got_repository *repo;
431 	struct got_commit_graph *graph;
432 };
433 
434 static const struct got_error *
435 add_branch_tip(struct got_object_id *commit_id, void *data, void *arg)
436 {
437 	const struct got_error *err;
438 	struct add_branch_tip_arg *a = arg;
439 	struct got_commit_graph_node *new_node;
440 	struct got_commit_object *commit;
441 
442 	err = got_object_open_as_commit(&commit, a->repo, commit_id);
443 	if (err)
444 		return err;
445 
446 	err = add_node(&new_node, a->graph, commit_id, a->repo);
447 	if (err)
448 		return err;
449 
450 	a->tips[a->ntips].commit_id = new_node ? &new_node->id : NULL;
451 	a->tips[a->ntips].commit = commit;
452 	a->tips[a->ntips].new_node = new_node;
453 	a->ntips++;
454 
455 	return NULL;
456 }
457 
458 static const struct got_error *
459 fetch_commits_from_open_branches(struct got_commit_graph *graph,
460     struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
461 {
462 	const struct got_error *err;
463 	struct add_branch_tip_arg arg;
464 	int i, ntips;
465 
466 	ntips = got_object_idset_num_elements(graph->open_branches);
467 	if (ntips == 0)
468 		return NULL;
469 
470 	/* (Re-)allocate branch tips array if necessary. */
471 	if (graph->ntips < ntips) {
472 		struct got_commit_graph_branch_tip *tips;
473 		tips = recallocarray(graph->tips, graph->ntips, ntips,
474 		    sizeof(*tips));
475 		if (tips == NULL)
476 			return got_error_from_errno("recallocarray");
477 		graph->tips = tips;
478 		graph->ntips = ntips;
479 	}
480 	arg.tips = graph->tips;
481 	arg.ntips = 0; /* add_branch_tip() will increment */
482 	arg.repo = repo;
483 	arg.graph = graph;
484 	err = got_object_idset_for_each(graph->open_branches, add_branch_tip,
485 	    &arg);
486 	if (err)
487 		goto done;
488 
489 	for (i = 0; i < arg.ntips; i++) {
490 		struct got_object_id *commit_id;
491 		struct got_commit_object *commit;
492 		struct got_commit_graph_node *new_node;
493 		int changed;
494 
495 		if (cancel_cb) {
496 			err = (*cancel_cb)(cancel_arg);
497 			if (err)
498 				break;
499 		}
500 
501 		commit_id = arg.tips[i].commit_id;
502 		commit = arg.tips[i].commit;
503 		new_node = arg.tips[i].new_node;
504 
505 		err = detect_changed_path(&changed, commit, commit_id,
506 		    graph->path, repo);
507 		if (err) {
508 			if (err->code != GOT_ERR_NO_OBJ)
509 				break;
510 			/*
511 			 * History of the path stops here on the current
512 			 * branch. Keep going on other branches.
513 			 */
514 			err = close_branch(graph, commit_id);
515 			if (err)
516 				break;
517 			continue;
518 		}
519 		if (changed)
520 			add_node_to_iter_list(graph, new_node,
521 			    got_object_commit_get_committer_time(commit));
522 		err = advance_branch(graph, commit_id, commit, repo);
523 		if (err)
524 			break;
525 	}
526 done:
527 	for (i = 0; i < arg.ntips; i++)
528 		got_object_commit_close(arg.tips[i].commit);
529 	return err;
530 }
531 
532 void
533 got_commit_graph_close(struct got_commit_graph *graph)
534 {
535 	if (graph->open_branches)
536 		got_object_idset_free(graph->open_branches);
537 	if (graph->node_ids)
538 		got_object_idset_free(graph->node_ids);
539 	free(graph->tips);
540 	free(graph->path);
541 	free(graph);
542 }
543 
544 const struct got_error *
545 got_commit_graph_iter_start(struct got_commit_graph *graph,
546     struct got_object_id *id, struct got_repository *repo,
547     got_cancel_cb cancel_cb, void *cancel_arg)
548 {
549 	const struct got_error *err = NULL;
550 
551 	if (!TAILQ_EMPTY(&graph->iter_list))
552 		return got_error(GOT_ERR_ITER_BUSY);
553 
554 	err = got_object_idset_add(graph->open_branches, id, NULL);
555 	if (err)
556 		return err;
557 
558 	/* Locate first commit which changed graph->path. */
559 	while (TAILQ_EMPTY(&graph->iter_list) &&
560 	    got_object_idset_num_elements(graph->open_branches) > 0) {
561 		err = fetch_commits_from_open_branches(graph, repo,
562 		    cancel_cb, cancel_arg);
563 		if (err)
564 			return err;
565 	}
566 
567 	if (TAILQ_EMPTY(&graph->iter_list)) {
568 		const char *path;
569 		if (got_path_is_root_dir(graph->path))
570 			return got_error_no_obj(id);
571 		path = graph->path;
572 		while (path[0] == '/')
573 			path++;
574 		return got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
575 	}
576 
577 	return NULL;
578 }
579 
580 const struct got_error *
581 got_commit_graph_iter_next(struct got_object_id **id,
582     struct got_commit_graph *graph, struct got_repository *repo,
583     got_cancel_cb cancel_cb, void *cancel_arg)
584 {
585 	const struct got_error *err = NULL;
586 	struct got_commit_graph_node *node;
587 
588 	*id = NULL;
589 
590 	node = TAILQ_FIRST(&graph->iter_list);
591 	if (node == NULL) {
592 		/* We are done iterating, or iteration was not started. */
593 		return got_error(GOT_ERR_ITER_COMPLETED);
594 	}
595 
596 	while (TAILQ_NEXT(node, entry) == NULL &&
597 	    got_object_idset_num_elements(graph->open_branches) > 0) {
598 		err = fetch_commits_from_open_branches(graph, repo,
599 		    cancel_cb, cancel_arg);
600 		if (err)
601 			return err;
602 	}
603 
604 	*id = &node->id;
605 	TAILQ_REMOVE(&graph->iter_list, node, entry);
606 	return NULL;
607 }
608 
609 const struct got_error *
610 got_commit_graph_find_youngest_common_ancestor(struct got_object_id **yca_id,
611     struct got_object_id *commit_id, struct got_object_id *commit_id2,
612     int first_parent_traversal,
613     struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
614 {
615 	const struct got_error *err = NULL;
616 	struct got_commit_graph *graph = NULL, *graph2 = NULL;
617 	int completed = 0, completed2 = 0;
618 	struct got_object_idset *commit_ids;
619 
620 	*yca_id = NULL;
621 
622 	commit_ids = got_object_idset_alloc();
623 	if (commit_ids == NULL)
624 		return got_error_from_errno("got_object_idset_alloc");
625 
626 	err = got_commit_graph_open(&graph, "/", first_parent_traversal);
627 	if (err)
628 		goto done;
629 
630 	err = got_commit_graph_open(&graph2, "/", first_parent_traversal);
631 	if (err)
632 		goto done;
633 
634 	err = got_commit_graph_iter_start(graph, commit_id, repo,
635 	    cancel_cb, cancel_arg);
636 	if (err)
637 		goto done;
638 
639 	err = got_commit_graph_iter_start(graph2, commit_id2, repo,
640 	    cancel_cb, cancel_arg);
641 	if (err)
642 		goto done;
643 
644 	for (;;) {
645 		struct got_object_id *id = NULL, *id2 = NULL;
646 
647 		if (cancel_cb) {
648 			err = (*cancel_cb)(cancel_arg);
649 			if (err)
650 				break;
651 		}
652 
653 		if (!completed) {
654 			err = got_commit_graph_iter_next(&id, graph, repo,
655 			    cancel_cb, cancel_arg);
656 			if (err) {
657 				if (err->code != GOT_ERR_ITER_COMPLETED)
658 					break;
659 				err = NULL;
660 				completed = 1;
661 			}
662 		}
663 
664 		if (!completed2) {
665 			err = got_commit_graph_iter_next(&id2, graph2, repo,
666 			    cancel_cb, cancel_arg);
667 			if (err) {
668 				if (err->code != GOT_ERR_ITER_COMPLETED)
669 					break;
670 				err = NULL;
671 				completed2 = 1;
672 			}
673 		}
674 
675 		if (id) {
676 			if (got_object_idset_contains(commit_ids, id)) {
677 				*yca_id = got_object_id_dup(id);
678 				if (*yca_id)
679 					break;
680 				err = got_error_from_errno("got_object_id_dup");
681 				break;
682 
683 			}
684 			err = got_object_idset_add(commit_ids, id, NULL);
685 			if (err)
686 				break;
687 		}
688 		if (id2) {
689 			if (got_object_idset_contains(commit_ids, id2)) {
690 				*yca_id = got_object_id_dup(id2);
691 				if (*yca_id)
692 					break;
693 				err = got_error_from_errno("got_object_id_dup");
694 				break;
695 
696 			}
697 			err = got_object_idset_add(commit_ids, id2, NULL);
698 			if (err)
699 				break;
700 		}
701 
702 		if (completed && completed2) {
703 			err = got_error(GOT_ERR_ANCESTRY);
704 			break;
705 		}
706 
707 	}
708 done:
709 	got_object_idset_free(commit_ids);
710 	if (graph)
711 		got_commit_graph_close(graph);
712 	if (graph2)
713 		got_commit_graph_close(graph2);
714 	return err;
715 }