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 #include <sys/queue.h>
20 #include <sys/stdint.h>
21 
22 #include <limits.h>
23 #include <stdio.h>
24 #include <stdlib.h>
25 #include <string.h>
26 #include <sha1.h>
27 #include <zlib.h>
28 #include <ctype.h>
29 
30 #include "got_error.h"
31 #include "got_object.h"
32 #include "got_cancel.h"
33 #include "got_commit_graph.h"
34 #include "got_path.h"
35 
36 #include "got_lib_delta.h"
37 #include "got_lib_inflate.h"
38 #include "got_lib_object.h"
39 #include "got_lib_object_idset.h"
40 
41 struct got_commit_graph_node {
42 	struct got_object_id id;
43 
44 	/* Used only during iteration. */
45 	time_t timestamp;
46 	TAILQ_ENTRY(got_commit_graph_node) entry;
47 };
48 
49 TAILQ_HEAD(got_commit_graph_iter_list, got_commit_graph_node);
50 
51 struct got_commit_graph_branch_tip {
52 	struct got_object_id *commit_id;
53 	struct got_commit_object *commit;
54 	struct got_commit_graph_node *new_node;
55 };
56 
57 struct got_commit_graph {
58 	/* The set of all commits we have traversed. */
59 	struct got_object_idset *node_ids;
60 
61 	int flags;
62 #define GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL		0x01
63 
64 	/*
65 	 * A set of object IDs of known parent commits which we have not yet
66 	 * traversed. Each commit ID in this set represents a branch in commit
67 	 * history: Either the first-parent branch of the head node, or another
68 	 * branch corresponding to a traversed merge commit for which we have
69 	 * not traversed a branch point commit yet.
70 	 *
71 	 * Whenever we add a commit with a matching ID to the graph, we remove
72 	 * its corresponding element from this set, and add new elements for
73 	 * each of that commit's parent commits which were not traversed yet.
74 	 *
75 	 * When API users ask us to fetch more commits, we fetch commits from
76 	 * all currently open branches. This allows API users to process
77 	 * commits in linear order even though the history contains branches.
78 	 */
79 	struct got_object_idset *open_branches;
80 
81 	/* Array of branch tips for fetch_commits_from_open_branches(). */
82 	struct got_commit_graph_branch_tip *tips;
83 	int ntips;
84 
85 	/* Path of tree entry of interest to the API user. */
86 	char *path;
87 
88 	/*
89 	 * Nodes which will be passed to the API user next, sorted by
90 	 * commit timestmap.
91 	 */
92 	struct got_commit_graph_iter_list iter_list;
93 };
94 
95 static const struct got_error *
96 detect_changed_path(int *changed, struct got_commit_object *commit,
97     struct got_object_id *commit_id, const char *path,
98     struct got_repository *repo)
99 {
100 	const struct got_error *err = NULL;
101 	struct got_commit_object *pcommit = NULL;
102 	struct got_tree_object *tree = NULL, *ptree = NULL;
103 	struct got_object_qid *pid;
104 
105 	if (got_path_is_root_dir(path)) {
106 		*changed = 1;
107 		return NULL;
108 	}
109 
110 	*changed = 0;
111 
112 	pid = STAILQ_FIRST(&commit->parent_ids);
113 	if (pid == NULL) {
114 		struct got_object_id *obj_id;
115 		err = got_object_id_by_path(&obj_id, repo, commit, path);
116 		if (err) {
117 			if (err->code == GOT_ERR_NO_TREE_ENTRY)
118 				err = NULL;
119 		} else
120 			*changed = 1; /* The path was created in this commit. */
121 		free(obj_id);
122 		return err;
123 	}
124 
125 	err = got_object_open_as_tree(&tree, repo, commit->tree_id);
126 	if (err)
127 		return err;
128 
129 	err = got_object_open_as_commit(&pcommit, repo, &pid->id);
130 	if (err)
131 		goto done;
132 
133 	err = got_object_open_as_tree(&ptree, repo, pcommit->tree_id);
134 	if (err)
135 		goto done;
136 
137 	err = got_object_tree_path_changed(changed, tree, ptree, path, repo);
138 done:
139 	if (tree)
140 		got_object_tree_close(tree);
141 	if (ptree)
142 		got_object_tree_close(ptree);
143 	if (pcommit)
144 		got_object_commit_close(pcommit);
145 	return err;
146 }
147 
148 static void
149 add_node_to_iter_list(struct got_commit_graph *graph,
150     struct got_commit_graph_node *node, time_t committer_time)
151 {
152 	struct got_commit_graph_node *n, *next;
153 
154 	node->timestamp = committer_time;
155 
156 	n = TAILQ_FIRST(&graph->iter_list);
157 	while (n) {
158 		next = TAILQ_NEXT(n, entry);
159 		if (next && node->timestamp >= next->timestamp) {
160 			TAILQ_INSERT_BEFORE(next, node, entry);
161 			return;
162 		}
163 		n = next;
164 	}
165 	TAILQ_INSERT_TAIL(&graph->iter_list, node, entry);
166 }
167 
168 static const struct got_error *
169 add_node(struct got_commit_graph_node **new_node,
170     struct got_commit_graph *graph, struct got_object_id *commit_id,
171     struct got_repository *repo)
172 {
173 	const struct got_error *err = NULL;
174 	struct got_commit_graph_node *node;
175 
176 	*new_node = NULL;
177 
178 	node = calloc(1, sizeof(*node));
179 	if (node == NULL)
180 		return got_error_from_errno("calloc");
181 
182 	memcpy(&node->id, commit_id, sizeof(node->id));
183 	err = got_object_idset_add(graph->node_ids, &node->id, NULL);
184 	if (err)
185 		free(node);
186 	else
187 		*new_node = node;
188 	return err;
189 }
190 
191 /*
192  * Ask got-read-pack to traverse first-parent history until a commit is
193  * encountered which modified graph->path, or until the pack file runs
194  * out of relevant commits. This is faster than sending an individual
195  * request for each commit stored in the pack file.
196  */
197 static const struct got_error *
198 packed_first_parent_traversal(int *ncommits_traversed,
199     struct got_commit_graph *graph, struct got_object_id *commit_id,
200     struct got_repository *repo)
201 {
202 	const struct got_error *err = NULL;
203 	struct got_object_id_queue traversed_commits;
204 	struct got_object_qid *qid;
205 
206 	STAILQ_INIT(&traversed_commits);
207 	*ncommits_traversed = 0;
208 
209 	err = got_traverse_packed_commits(&traversed_commits,
210 	    commit_id, graph->path, repo);
211 	if (err)
212 		return err;
213 
214 	/* Add all traversed commits to the graph... */
215 	STAILQ_FOREACH(qid, &traversed_commits, entry) {
216 		struct got_commit_graph_node *node;
217 
218 		if (got_object_idset_contains(graph->open_branches, &qid->id))
219 			continue;
220 		if (got_object_idset_contains(graph->node_ids, &qid->id))
221 			continue;
222 
223 		(*ncommits_traversed)++;
224 
225 		/* ... except the last commit is the new branch tip. */
226 		if (STAILQ_NEXT(qid, entry) == NULL) {
227 			err = got_object_idset_add(graph->open_branches,
228 			    &qid->id, NULL);
229 			break;
230 		}
231 
232 		err = add_node(&node, graph, &qid->id, repo);
233 		if (err)
234 			break;
235 	}
236 
237 	got_object_id_queue_free(&traversed_commits);
238 	return err;
239 }
240 
241 static const struct got_error *
242 close_branch(struct got_commit_graph *graph, struct got_object_id *commit_id)
243 {
244 	const struct got_error *err;
245 
246 	err = got_object_idset_remove(NULL, graph->open_branches, commit_id);
247 	if (err && err->code != GOT_ERR_NO_OBJ)
248 		return err;
249 	return NULL;
250 }
251 
252 static const struct got_error *
253 advance_branch(struct got_commit_graph *graph, struct got_object_id *commit_id,
254     struct got_commit_object *commit, struct got_repository *repo)
255 {
256 	const struct got_error *err;
257 	struct got_object_qid *qid;
258 
259 	err = close_branch(graph, commit_id);
260 	if (err)
261 		return err;
262 
263 	if (graph->flags & GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL) {
264 		qid = STAILQ_FIRST(&commit->parent_ids);
265 		if (qid == NULL ||
266 		    got_object_idset_contains(graph->open_branches, &qid->id))
267 			return NULL;
268 		/*
269 		 * The root directory always changes by definition, and when
270 		 * logging the root we want to traverse consecutive commits
271 		 * even if they point at the same tree.
272 		 * But if we are looking for a specific path then we can avoid
273 		 * fetching packed commits which did not modify the path and
274 		 * only fetch their IDs. This speeds up 'got blame'.
275 		 */
276 		if (!got_path_is_root_dir(graph->path) &&
277 		    (commit->flags & GOT_COMMIT_FLAG_PACKED)) {
278 			int ncommits = 0;
279 			err = packed_first_parent_traversal(&ncommits,
280 			    graph, &qid->id, repo);
281 			if (err || ncommits > 0)
282 				return err;
283 		}
284 		return got_object_idset_add(graph->open_branches,
285 		    &qid->id, NULL);
286 	}
287 
288 	/*
289 	 * If we are graphing commits for a specific path, skip branches
290 	 * which do not contribute any content to this path.
291 	 */
292 	if (commit->nparents > 1 && !got_path_is_root_dir(graph->path)) {
293 		struct got_object_id *merged_id, *prev_id = NULL;
294 		int branches_differ = 0;
295 
296 		err = got_object_id_by_path(&merged_id, repo, commit,
297 		    graph->path);
298 		if (err)
299 			return err;
300 
301 		STAILQ_FOREACH(qid, &commit->parent_ids, entry) {
302 			struct got_object_id *id = NULL;
303 			struct got_commit_object *pcommit = NULL;
304 
305 			if (got_object_idset_contains(graph->open_branches,
306 			    &qid->id))
307 				continue;
308 
309 			err = got_object_open_as_commit(&pcommit, repo,
310 			    &qid->id);
311 			if (err) {
312 				free(merged_id);
313 				free(prev_id);
314 				return err;
315 			}
316 			err = got_object_id_by_path(&id, repo, pcommit,
317 			    graph->path);
318 			got_object_commit_close(pcommit);
319 			pcommit = NULL;
320 			if (err) {
321 				if (err->code == GOT_ERR_NO_TREE_ENTRY) {
322 					branches_differ = 1;
323 					continue;
324 				}
325 				free(merged_id);
326 				free(prev_id);
327 				return err;
328 			}
329 
330 			if (prev_id) {
331 				if (!branches_differ &&
332 				    got_object_id_cmp(id, prev_id) != 0)
333 					branches_differ = 1;
334 				free(prev_id);
335 			}
336 			prev_id = id;
337 
338 			/*
339 			 * If a branch has created the merged content we can
340 			 * skip any other branches.
341 			 */
342 			if (got_object_id_cmp(merged_id, id) == 0) {
343 				err = got_object_idset_add(graph->open_branches,
344 				    &qid->id, NULL);
345 				free(merged_id);
346 				free(id);
347 				return err;
348 			}
349 		}
350 
351 		free(prev_id);
352 		prev_id = NULL;
353 		free(merged_id);
354 		merged_id = NULL;
355 
356 		/*
357 		 * If the path's content is the same on all branches,
358 		 * follow the first parent only.
359 		 */
360 		if (!branches_differ) {
361 			qid = STAILQ_FIRST(&commit->parent_ids);
362 			if (qid == NULL)
363 				return NULL;
364 			if (got_object_idset_contains(graph->open_branches,
365 			    &qid->id))
366 				return NULL;
367 			if (got_object_idset_contains(graph->node_ids,
368 			    &qid->id))
369 				return NULL; /* parent already traversed */
370 			return got_object_idset_add(graph->open_branches,
371 			    &qid->id, NULL);
372 		}
373 	}
374 
375 	STAILQ_FOREACH(qid, &commit->parent_ids, entry) {
376 		if (got_object_idset_contains(graph->open_branches, &qid->id))
377 			continue;
378 		if (got_object_idset_contains(graph->node_ids, &qid->id))
379 			continue; /* parent already traversed */
380 		err = got_object_idset_add(graph->open_branches, &qid->id,
381 		    NULL);
382 		if (err)
383 			return err;
384 	}
385 
386 	return NULL;
387 }
388 
389 const struct got_error *
390 got_commit_graph_open(struct got_commit_graph **graph,
391     const char *path, int first_parent_traversal)
392 {
393 	const struct got_error *err = NULL;
394 
395 	*graph = calloc(1, sizeof(**graph));
396 	if (*graph == NULL)
397 		return got_error_from_errno("calloc");
398 
399 	TAILQ_INIT(&(*graph)->iter_list);
400 
401 	(*graph)->path = strdup(path);
402 	if ((*graph)->path == NULL) {
403 		err = got_error_from_errno("strdup");
404 		goto done;
405 	}
406 
407 	(*graph)->node_ids = got_object_idset_alloc();
408 	if ((*graph)->node_ids == NULL) {
409 		err = got_error_from_errno("got_object_idset_alloc");
410 		goto done;
411 	}
412 
413 	(*graph)->open_branches = got_object_idset_alloc();
414 	if ((*graph)->open_branches == NULL) {
415 		err = got_error_from_errno("got_object_idset_alloc");
416 		goto done;
417 	}
418 
419 	if (first_parent_traversal)
420 		(*graph)->flags |= GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL;
421 done:
422 	if (err) {
423 		got_commit_graph_close(*graph);
424 		*graph = NULL;
425 	}
426 	return err;
427 }
428 
429 struct add_branch_tip_arg {
430 	struct got_commit_graph_branch_tip *tips;
431 	int ntips;
432 	struct got_repository *repo;
433 	struct got_commit_graph *graph;
434 };
435 
436 static const struct got_error *
437 add_branch_tip(struct got_object_id *commit_id, void *data, void *arg)
438 {
439 	const struct got_error *err;
440 	struct add_branch_tip_arg *a = arg;
441 	struct got_commit_graph_node *new_node;
442 	struct got_commit_object *commit;
443 
444 	err = got_object_open_as_commit(&commit, a->repo, commit_id);
445 	if (err)
446 		return err;
447 
448 	err = add_node(&new_node, a->graph, commit_id, a->repo);
449 	if (err)
450 		return err;
451 
452 	a->tips[a->ntips].commit_id = new_node ? &new_node->id : NULL;
453 	a->tips[a->ntips].commit = commit;
454 	a->tips[a->ntips].new_node = new_node;
455 	a->ntips++;
456 
457 	return NULL;
458 }
459 
460 static const struct got_error *
461 fetch_commits_from_open_branches(struct got_commit_graph *graph,
462     struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
463 {
464 	const struct got_error *err;
465 	struct add_branch_tip_arg arg;
466 	int i, ntips;
467 
468 	ntips = got_object_idset_num_elements(graph->open_branches);
469 	if (ntips == 0)
470 		return NULL;
471 
472 	/* (Re-)allocate branch tips array if necessary. */
473 	if (graph->ntips < ntips) {
474 		struct got_commit_graph_branch_tip *tips;
475 		tips = recallocarray(graph->tips, graph->ntips, ntips,
476 		    sizeof(*tips));
477 		if (tips == NULL)
478 			return got_error_from_errno("recallocarray");
479 		graph->tips = tips;
480 		graph->ntips = ntips;
481 	}
482 	arg.tips = graph->tips;
483 	arg.ntips = 0; /* add_branch_tip() will increment */
484 	arg.repo = repo;
485 	arg.graph = graph;
486 	err = got_object_idset_for_each(graph->open_branches, add_branch_tip,
487 	    &arg);
488 	if (err)
489 		goto done;
490 
491 	for (i = 0; i < arg.ntips; i++) {
492 		struct got_object_id *commit_id;
493 		struct got_commit_object *commit;
494 		struct got_commit_graph_node *new_node;
495 		int changed;
496 
497 		if (cancel_cb) {
498 			err = (*cancel_cb)(cancel_arg);
499 			if (err)
500 				break;
501 		}
502 
503 		commit_id = arg.tips[i].commit_id;
504 		commit = arg.tips[i].commit;
505 		new_node = arg.tips[i].new_node;
506 
507 		err = detect_changed_path(&changed, commit, commit_id,
508 		    graph->path, repo);
509 		if (err) {
510 			if (err->code != GOT_ERR_NO_OBJ)
511 				break;
512 			/*
513 			 * History of the path stops here on the current
514 			 * branch. Keep going on other branches.
515 			 */
516 			err = close_branch(graph, commit_id);
517 			if (err)
518 				break;
519 			continue;
520 		}
521 		if (changed)
522 			add_node_to_iter_list(graph, new_node,
523 			    got_object_commit_get_committer_time(commit));
524 		err = advance_branch(graph, commit_id, commit, repo);
525 		if (err)
526 			break;
527 	}
528 done:
529 	for (i = 0; i < arg.ntips; i++)
530 		got_object_commit_close(arg.tips[i].commit);
531 	return err;
532 }
533 
534 void
535 got_commit_graph_close(struct got_commit_graph *graph)
536 {
537 	if (graph->open_branches)
538 		got_object_idset_free(graph->open_branches);
539 	if (graph->node_ids)
540 		got_object_idset_free(graph->node_ids);
541 	free(graph->tips);
542 	free(graph->path);
543 	free(graph);
544 }
545 
546 const struct got_error *
547 got_commit_graph_iter_start(struct got_commit_graph *graph,
548     struct got_object_id *id, struct got_repository *repo,
549     got_cancel_cb cancel_cb, void *cancel_arg)
550 {
551 	const struct got_error *err = NULL;
552 
553 	if (!TAILQ_EMPTY(&graph->iter_list))
554 		return got_error(GOT_ERR_ITER_BUSY);
555 
556 	err = got_object_idset_add(graph->open_branches, id, NULL);
557 	if (err)
558 		return err;
559 
560 	/* Locate first commit which changed graph->path. */
561 	while (TAILQ_EMPTY(&graph->iter_list) &&
562 	    got_object_idset_num_elements(graph->open_branches) > 0) {
563 		err = fetch_commits_from_open_branches(graph, repo,
564 		    cancel_cb, cancel_arg);
565 		if (err)
566 			return err;
567 	}
568 
569 	if (TAILQ_EMPTY(&graph->iter_list)) {
570 		const char *path;
571 		if (got_path_is_root_dir(graph->path))
572 			return got_error_no_obj(id);
573 		path = graph->path;
574 		while (path[0] == '/')
575 			path++;
576 		return got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
577 	}
578 
579 	return NULL;
580 }
581 
582 const struct got_error *
583 got_commit_graph_iter_next(struct got_object_id **id,
584     struct got_commit_graph *graph, struct got_repository *repo,
585     got_cancel_cb cancel_cb, void *cancel_arg)
586 {
587 	const struct got_error *err = NULL;
588 	struct got_commit_graph_node *node;
589 
590 	*id = NULL;
591 
592 	node = TAILQ_FIRST(&graph->iter_list);
593 	if (node == NULL) {
594 		/* We are done iterating, or iteration was not started. */
595 		return got_error(GOT_ERR_ITER_COMPLETED);
596 	}
597 
598 	while (TAILQ_NEXT(node, entry) == NULL &&
599 	    got_object_idset_num_elements(graph->open_branches) > 0) {
600 		err = fetch_commits_from_open_branches(graph, repo,
601 		    cancel_cb, cancel_arg);
602 		if (err)
603 			return err;
604 	}
605 
606 	*id = &node->id;
607 	TAILQ_REMOVE(&graph->iter_list, node, entry);
608 	return NULL;
609 }
610 
611 const struct got_error *
612 got_commit_graph_find_youngest_common_ancestor(struct got_object_id **yca_id,
613     struct got_object_id *commit_id, struct got_object_id *commit_id2,
614     int first_parent_traversal,
615     struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
616 {
617 	const struct got_error *err = NULL;
618 	struct got_commit_graph *graph = NULL, *graph2 = NULL;
619 	int completed = 0, completed2 = 0;
620 	struct got_object_idset *commit_ids;
621 
622 	*yca_id = NULL;
623 
624 	commit_ids = got_object_idset_alloc();
625 	if (commit_ids == NULL)
626 		return got_error_from_errno("got_object_idset_alloc");
627 
628 	err = got_commit_graph_open(&graph, "/", first_parent_traversal);
629 	if (err)
630 		goto done;
631 
632 	err = got_commit_graph_open(&graph2, "/", first_parent_traversal);
633 	if (err)
634 		goto done;
635 
636 	err = got_commit_graph_iter_start(graph, commit_id, repo,
637 	    cancel_cb, cancel_arg);
638 	if (err)
639 		goto done;
640 
641 	err = got_commit_graph_iter_start(graph2, commit_id2, repo,
642 	    cancel_cb, cancel_arg);
643 	if (err)
644 		goto done;
645 
646 	for (;;) {
647 		struct got_object_id *id = NULL, *id2 = NULL;
648 
649 		if (cancel_cb) {
650 			err = (*cancel_cb)(cancel_arg);
651 			if (err)
652 				break;
653 		}
654 
655 		if (!completed) {
656 			err = got_commit_graph_iter_next(&id, graph, repo,
657 			    cancel_cb, cancel_arg);
658 			if (err) {
659 				if (err->code != GOT_ERR_ITER_COMPLETED)
660 					break;
661 				err = NULL;
662 				completed = 1;
663 			}
664 		}
665 
666 		if (!completed2) {
667 			err = got_commit_graph_iter_next(&id2, graph2, repo,
668 			    cancel_cb, cancel_arg);
669 			if (err) {
670 				if (err->code != GOT_ERR_ITER_COMPLETED)
671 					break;
672 				err = NULL;
673 				completed2 = 1;
674 			}
675 		}
676 
677 		if (id) {
678 			if (got_object_idset_contains(commit_ids, id)) {
679 				*yca_id = got_object_id_dup(id);
680 				if (*yca_id)
681 					break;
682 				err = got_error_from_errno("got_object_id_dup");
683 				break;
684 
685 			}
686 			err = got_object_idset_add(commit_ids, id, NULL);
687 			if (err)
688 				break;
689 		}
690 		if (id2) {
691 			if (got_object_idset_contains(commit_ids, id2)) {
692 				*yca_id = got_object_id_dup(id2);
693 				if (*yca_id)
694 					break;
695 				err = got_error_from_errno("got_object_id_dup");
696 				break;
697 
698 			}
699 			err = got_object_idset_add(commit_ids, id2, NULL);
700 			if (err)
701 				break;
702 		}
703 
704 		if (completed && completed2) {
705 			err = got_error(GOT_ERR_ANCESTRY);
706 			break;
707 		}
708 
709 	}
710 done:
711 	got_object_idset_free(commit_ids);
712 	if (graph)
713 		got_commit_graph_close(graph);
714 	if (graph2)
715 		got_commit_graph_close(graph2);
716 	return err;
717 }