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_id, 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_id,
296 		    graph->path);
297 		if (err)
298 			return err;
299 
300 		STAILQ_FOREACH(qid, &commit->parent_ids, entry) {
301 			struct got_object_id *id;
302 
303 			if (got_object_idset_contains(graph->open_branches,
304 			    qid->id))
305 				continue;
306 
307 			err = got_object_id_by_path(&id, repo, qid->id,
308 			    graph->path);
309 			if (err) {
310 				if (err->code == GOT_ERR_NO_TREE_ENTRY) {
311 					branches_differ = 1;
312 					continue;
313 				}
314 				free(merged_id);
315 				free(prev_id);
316 				return err;
317 			}
318 
319 			if (prev_id) {
320 				if (!branches_differ &&
321 				    got_object_id_cmp(id, prev_id) != 0)
322 					branches_differ = 1;
323 				free(prev_id);
324 			}
325 			prev_id = id;
326 
327 			/*
328 			 * If a branch has created the merged content we can
329 			 * skip any other branches.
330 			 */
331 			if (got_object_id_cmp(merged_id, id) == 0) {
332 				err = got_object_idset_add(graph->open_branches,
333 				    qid->id, NULL);
334 				free(merged_id);
335 				free(id);
336 				return err;
337 			}
338 		}
339 
340 		free(prev_id);
341 		prev_id = NULL;
342 		free(merged_id);
343 		merged_id = NULL;
344 
345 		/*
346 		 * If the path's content is the same on all branches,
347 		 * follow the first parent only.
348 		 */
349 		if (!branches_differ) {
350 			qid = STAILQ_FIRST(&commit->parent_ids);
351 			if (qid == NULL)
352 				return NULL;
353 			if (got_object_idset_contains(graph->open_branches,
354 			    qid->id))
355 				return NULL;
356 			if (got_object_idset_contains(graph->node_ids,
357 			    qid->id))
358 				return NULL; /* parent already traversed */
359 			return got_object_idset_add(graph->open_branches,
360 			    qid->id, NULL);
361 		}
362 	}
363 
364 	STAILQ_FOREACH(qid, &commit->parent_ids, entry) {
365 		if (got_object_idset_contains(graph->open_branches, qid->id))
366 			continue;
367 		if (got_object_idset_contains(graph->node_ids, qid->id))
368 			continue; /* parent already traversed */
369 		err = got_object_idset_add(graph->open_branches, qid->id, NULL);
370 		if (err)
371 			return err;
372 	}
373 
374 	return NULL;
375 }
376 
377 const struct got_error *
378 got_commit_graph_open(struct got_commit_graph **graph,
379     const char *path, int first_parent_traversal)
380 {
381 	const struct got_error *err = NULL;
382 
383 	*graph = calloc(1, sizeof(**graph));
384 	if (*graph == NULL)
385 		return got_error_from_errno("calloc");
386 
387 	TAILQ_INIT(&(*graph)->iter_list);
388 
389 	(*graph)->path = strdup(path);
390 	if ((*graph)->path == NULL) {
391 		err = got_error_from_errno("strdup");
392 		goto done;
393 	}
394 
395 	(*graph)->node_ids = got_object_idset_alloc();
396 	if ((*graph)->node_ids == NULL) {
397 		err = got_error_from_errno("got_object_idset_alloc");
398 		goto done;
399 	}
400 
401 	(*graph)->open_branches = got_object_idset_alloc();
402 	if ((*graph)->open_branches == NULL) {
403 		err = got_error_from_errno("got_object_idset_alloc");
404 		goto done;
405 	}
406 
407 	if (first_parent_traversal)
408 		(*graph)->flags |= GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL;
409 done:
410 	if (err) {
411 		got_commit_graph_close(*graph);
412 		*graph = NULL;
413 	}
414 	return err;
415 }
416 
417 struct add_branch_tip_arg {
418 	struct got_commit_graph_branch_tip *tips;
419 	int ntips;
420 	struct got_repository *repo;
421 	struct got_commit_graph *graph;
422 };
423 
424 static const struct got_error *
425 add_branch_tip(struct got_object_id *commit_id, void *data, void *arg)
426 {
427 	const struct got_error *err;
428 	struct add_branch_tip_arg *a = arg;
429 	struct got_commit_graph_node *new_node;
430 	struct got_commit_object *commit;
431 
432 	err = got_object_open_as_commit(&commit, a->repo, commit_id);
433 	if (err)
434 		return err;
435 
436 	err = add_node(&new_node, a->graph, commit_id, a->repo);
437 	if (err)
438 		return err;
439 
440 	a->tips[a->ntips].commit_id = new_node ? &new_node->id : NULL;
441 	a->tips[a->ntips].commit = commit;
442 	a->tips[a->ntips].new_node = new_node;
443 	a->ntips++;
444 
445 	return NULL;
446 }
447 
448 static const struct got_error *
449 fetch_commits_from_open_branches(struct got_commit_graph *graph,
450     struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
451 {
452 	const struct got_error *err;
453 	struct add_branch_tip_arg arg;
454 	int i, ntips;
455 
456 	ntips = got_object_idset_num_elements(graph->open_branches);
457 	if (ntips == 0)
458 		return NULL;
459 
460 	/* (Re-)allocate branch tips array if necessary. */
461 	if (graph->ntips < ntips) {
462 		struct got_commit_graph_branch_tip *tips;
463 		tips = recallocarray(graph->tips, graph->ntips, ntips,
464 		    sizeof(*tips));
465 		if (tips == NULL)
466 			return got_error_from_errno("recallocarray");
467 		graph->tips = tips;
468 		graph->ntips = ntips;
469 	}
470 	arg.tips = graph->tips;
471 	arg.ntips = 0; /* add_branch_tip() will increment */
472 	arg.repo = repo;
473 	arg.graph = graph;
474 	err = got_object_idset_for_each(graph->open_branches, add_branch_tip,
475 	    &arg);
476 	if (err)
477 		goto done;
478 
479 	for (i = 0; i < arg.ntips; i++) {
480 		struct got_object_id *commit_id;
481 		struct got_commit_object *commit;
482 		struct got_commit_graph_node *new_node;
483 		int changed;
484 
485 		if (cancel_cb) {
486 			err = (*cancel_cb)(cancel_arg);
487 			if (err)
488 				break;
489 		}
490 
491 		commit_id = arg.tips[i].commit_id;
492 		commit = arg.tips[i].commit;
493 		new_node = arg.tips[i].new_node;
494 
495 		err = detect_changed_path(&changed, commit, commit_id,
496 		    graph->path, repo);
497 		if (err) {
498 			if (err->code != GOT_ERR_NO_OBJ)
499 				break;
500 			/*
501 			 * History of the path stops here on the current
502 			 * branch. Keep going on other branches.
503 			 */
504 			err = close_branch(graph, commit_id);
505 			if (err)
506 				break;
507 			continue;
508 		}
509 		if (changed)
510 			add_node_to_iter_list(graph, new_node,
511 			    got_object_commit_get_committer_time(commit));
512 		err = advance_branch(graph, commit_id, commit, repo);
513 		if (err)
514 			break;
515 	}
516 done:
517 	for (i = 0; i < arg.ntips; i++)
518 		got_object_commit_close(arg.tips[i].commit);
519 	return err;
520 }
521 
522 void
523 got_commit_graph_close(struct got_commit_graph *graph)
524 {
525 	if (graph->open_branches)
526 		got_object_idset_free(graph->open_branches);
527 	if (graph->node_ids)
528 		got_object_idset_free(graph->node_ids);
529 	free(graph->tips);
530 	free(graph->path);
531 	free(graph);
532 }
533 
534 const struct got_error *
535 got_commit_graph_iter_start(struct got_commit_graph *graph,
536     struct got_object_id *id, struct got_repository *repo,
537     got_cancel_cb cancel_cb, void *cancel_arg)
538 {
539 	const struct got_error *err = NULL;
540 
541 	if (!TAILQ_EMPTY(&graph->iter_list))
542 		return got_error(GOT_ERR_ITER_BUSY);
543 
544 	err = got_object_idset_add(graph->open_branches, id, NULL);
545 	if (err)
546 		return err;
547 
548 	/* Locate first commit which changed graph->path. */
549 	while (TAILQ_EMPTY(&graph->iter_list) &&
550 	    got_object_idset_num_elements(graph->open_branches) > 0) {
551 		err = fetch_commits_from_open_branches(graph, repo,
552 		    cancel_cb, cancel_arg);
553 		if (err)
554 			return err;
555 	}
556 
557 	if (TAILQ_EMPTY(&graph->iter_list)) {
558 		const char *path;
559 		if (got_path_is_root_dir(graph->path))
560 			return got_error_no_obj(id);
561 		path = graph->path;
562 		while (path[0] == '/')
563 			path++;
564 		return got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
565 	}
566 
567 	return NULL;
568 }
569 
570 const struct got_error *
571 got_commit_graph_iter_next(struct got_object_id **id,
572     struct got_commit_graph *graph, struct got_repository *repo,
573     got_cancel_cb cancel_cb, void *cancel_arg)
574 {
575 	const struct got_error *err = NULL;
576 	struct got_commit_graph_node *node;
577 
578 	*id = NULL;
579 
580 	node = TAILQ_FIRST(&graph->iter_list);
581 	if (node == NULL) {
582 		/* We are done iterating, or iteration was not started. */
583 		return got_error(GOT_ERR_ITER_COMPLETED);
584 	}
585 
586 	while (TAILQ_NEXT(node, entry) == NULL &&
587 	    got_object_idset_num_elements(graph->open_branches) > 0) {
588 		err = fetch_commits_from_open_branches(graph, repo,
589 		    cancel_cb, cancel_arg);
590 		if (err)
591 			return err;
592 	}
593 
594 	*id = &node->id;
595 	TAILQ_REMOVE(&graph->iter_list, node, entry);
596 	return NULL;
597 }
598 
599 const struct got_error *
600 got_commit_graph_find_youngest_common_ancestor(struct got_object_id **yca_id,
601     struct got_object_id *commit_id, struct got_object_id *commit_id2,
602     int first_parent_traversal,
603     struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
604 {
605 	const struct got_error *err = NULL;
606 	struct got_commit_graph *graph = NULL, *graph2 = NULL;
607 	int completed = 0, completed2 = 0;
608 	struct got_object_idset *commit_ids;
609 
610 	*yca_id = NULL;
611 
612 	commit_ids = got_object_idset_alloc();
613 	if (commit_ids == NULL)
614 		return got_error_from_errno("got_object_idset_alloc");
615 
616 	err = got_commit_graph_open(&graph, "/", first_parent_traversal);
617 	if (err)
618 		goto done;
619 
620 	err = got_commit_graph_open(&graph2, "/", first_parent_traversal);
621 	if (err)
622 		goto done;
623 
624 	err = got_commit_graph_iter_start(graph, commit_id, repo,
625 	    cancel_cb, cancel_arg);
626 	if (err)
627 		goto done;
628 
629 	err = got_commit_graph_iter_start(graph2, commit_id2, repo,
630 	    cancel_cb, cancel_arg);
631 	if (err)
632 		goto done;
633 
634 	for (;;) {
635 		struct got_object_id *id = NULL, *id2 = NULL;
636 
637 		if (cancel_cb) {
638 			err = (*cancel_cb)(cancel_arg);
639 			if (err)
640 				break;
641 		}
642 
643 		if (!completed) {
644 			err = got_commit_graph_iter_next(&id, graph, repo,
645 			    cancel_cb, cancel_arg);
646 			if (err) {
647 				if (err->code != GOT_ERR_ITER_COMPLETED)
648 					break;
649 				err = NULL;
650 				completed = 1;
651 			}
652 		}
653 
654 		if (!completed2) {
655 			err = got_commit_graph_iter_next(&id2, graph2, repo,
656 			    cancel_cb, cancel_arg);
657 			if (err) {
658 				if (err->code != GOT_ERR_ITER_COMPLETED)
659 					break;
660 				err = NULL;
661 				completed2 = 1;
662 			}
663 		}
664 
665 		if (id) {
666 			if (got_object_idset_contains(commit_ids, id)) {
667 				*yca_id = got_object_id_dup(id);
668 				if (*yca_id)
669 					break;
670 				err = got_error_from_errno("got_object_id_dup");
671 				break;
672 
673 			}
674 			err = got_object_idset_add(commit_ids, id, NULL);
675 			if (err)
676 				break;
677 		}
678 		if (id2) {
679 			if (got_object_idset_contains(commit_ids, id2)) {
680 				*yca_id = got_object_id_dup(id2);
681 				if (*yca_id)
682 					break;
683 				err = got_error_from_errno("got_object_id_dup");
684 				break;
685 
686 			}
687 			err = got_object_idset_add(commit_ids, id2, NULL);
688 			if (err)
689 				break;
690 		}
691 
692 		if (completed && completed2) {
693 			err = got_error(GOT_ERR_ANCESTRY);
694 			break;
695 		}
696 
697 	}
698 done:
699 	got_object_idset_free(commit_ids);
700 	if (graph)
701 		got_commit_graph_close(graph);
702 	if (graph2)
703 		got_commit_graph_close(graph2);
704 	return err;
705 }