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/types.h>
18 #include <sys/stat.h>
19 #include <sys/queue.h>
20 #include <sys/stdint.h>
21 
22 #include <stdio.h>
23 #include <stdlib.h>
24 #include <string.h>
25 #include <sha1.h>
26 #include <zlib.h>
27 #include <ctype.h>
28 
29 #include "got_error.h"
30 #include "got_object.h"
31 #include "got_commit_graph.h"
32 
33 #include "got_lib_delta.h"
34 #include "got_lib_inflate.h"
35 #include "got_lib_object.h"
36 #include "got_lib_object_idset.h"
37 #include "got_lib_path.h"
38 
39 struct got_commit_graph_node {
40 	struct got_object_id id;
41 
42 	/*
43 	 * Each graph node corresponds to a commit object.
44 	 * Graph vertices are modelled with an adjacency list.
45 	 * Adjacencies of a graph node are parent (older) commits.
46 	 */
47 	int nparents;
48 	struct got_object_id_queue parent_ids;
49 
50 	time_t commit_timestamp;
51 
52 	/* Used during graph iteration. */
53 	TAILQ_ENTRY(got_commit_graph_node) entry;
54 };
55 
56 TAILQ_HEAD(got_commit_graph_iter_list, got_commit_graph_node);
57 
58 struct got_commit_graph_branch_tip {
59 	struct got_object_id id;
60 	struct got_commit_graph_node *node;
61 };
62 
63 struct got_commit_graph {
64 	/* The set of all commits we have traversed. */
65 	struct got_object_idset *node_ids;
66 
67 	/* The commit at which traversal began (youngest commit in node_ids). */
68 	struct got_commit_graph_node *head_node;
69 
70 	int flags;
71 #define GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL		0x01
72 
73 	/*
74 	 * A set of object IDs of known parent commits which we have not yet
75 	 * traversed. Each commit ID in this set represents a branch in commit
76 	 * history: Either the first-parent branch of the head node, or another
77 	 * branch corresponding to a traversed merge commit for which we have
78 	 * not traversed a branch point commit yet.
79 	 *
80 	 * Whenever we add a commit with a matching ID to the graph, we remove
81 	 * its corresponding element from this set, and add new elements for
82 	 * each of that commit's parent commits which were not traversed yet.
83 	 *
84 	 * When API users ask us to fetch more commits, we fetch commits from
85 	 * all currently open branches. This allows API users to process
86 	 * commits in linear order even though the history contains branches.
87 	 */
88 	struct got_object_idset *open_branches;
89 
90 	/* Copy of known branch tips for fetch_commits_from_open_branches(). */
91 	struct got_commit_graph_branch_tip *tips;
92 	size_t ntips;
93 #define GOT_COMMIT_GRAPH_MIN_TIPS 100	/* minimum amount of tips to allocate */
94 
95 	/* Path of tree entry of interest to the API user. */
96 	char *path;
97 
98 	/* The next commit to return when the API user asks for one. */
99 	struct got_commit_graph_node *iter_node;
100 
101 	/* The graph iteration list contains all nodes in sorted order. */
102 	struct got_commit_graph_iter_list iter_list;
103 };
104 
105 static struct got_commit_graph *
106 alloc_graph(const char *path)
107 {
108 	struct got_commit_graph *graph;
109 
110 	graph = calloc(1, sizeof(*graph));
111 	if (graph == NULL)
112 		return NULL;
113 
114 	graph->path = strdup(path);
115 	if (graph->path == NULL) {
116 		free(graph);
117 		return NULL;
118 	}
119 
120 	graph->node_ids = got_object_idset_alloc();
121 	if (graph->node_ids == NULL) {
122 		free(graph->path);
123 		free(graph);
124 		return NULL;
125 	}
126 
127 	graph->open_branches = got_object_idset_alloc();
128 	if (graph->open_branches == NULL) {
129 		got_object_idset_free(graph->node_ids);
130 		free(graph->path);
131 		free(graph);
132 		return NULL;
133 	}
134 
135 	TAILQ_INIT(&graph->iter_list);
136 	return graph;
137 }
138 
139 #if 0
140 static int
141 is_head_node(struct got_commit_graph_node *node)
142 {
143 	return node->nchildren == 0;
144 }
145 
146 int
147 is_branch_point(struct got_commit_graph_node *node)
148 {
149 	return node->nchildren > 1;
150 }
151 
152 static int
153 is_root_node(struct got_commit_graph_node *node)
154 {
155 	return node->nparents == 0;
156 }
157 #endif
158 
159 static int
160 is_merge_point(struct got_commit_graph_node *node)
161 {
162 	return node->nparents > 1;
163 }
164 
165 static const struct got_error *
166 detect_changed_path(int *changed, struct got_commit_object *commit,
167     struct got_object_id *commit_id, const char *path,
168     struct got_repository *repo)
169 {
170 	const struct got_error *err = NULL;
171 	struct got_commit_object *pcommit = NULL;
172 	struct got_tree_object *tree = NULL, *ptree = NULL;
173 	struct got_object_qid *pid;
174 
175 	if (got_path_is_root_dir(path)) {
176 		*changed = 1;
177 		return NULL;
178 	}
179 
180 	*changed = 0;
181 
182 	err = got_object_open_as_tree(&tree, repo, commit->tree_id);
183 	if (err)
184 		return err;
185 
186 	pid = SIMPLEQ_FIRST(&commit->parent_ids);
187 	if (pid == NULL) {
188 		struct got_object_id *obj_id;
189 		err = got_object_id_by_path(&obj_id, repo, commit_id, path);
190 		if (err)
191 			err = NULL;
192 		else
193 			*changed = 1;
194 		free(obj_id);
195 		goto done;
196 	}
197 
198 	err = got_object_open_as_commit(&pcommit, repo, pid->id);
199 	if (err)
200 		goto done;
201 
202 	err = got_object_open_as_tree(&ptree, repo, pcommit->tree_id);
203 	if (err)
204 		goto done;
205 
206 	err = got_object_tree_path_changed(changed, tree, ptree, path, repo);
207 done:
208 	if (tree)
209 		got_object_tree_close(tree);
210 	if (ptree)
211 		got_object_tree_close(ptree);
212 	if (pcommit)
213 		got_object_commit_close(pcommit);
214 	return err;
215 }
216 
217 static void
218 add_node_to_iter_list(struct got_commit_graph *graph,
219     struct got_commit_graph_node *node,
220     struct got_commit_graph_node *child_node)
221 {
222 	struct got_commit_graph_node *n, *next;
223 
224 	if (TAILQ_EMPTY(&graph->iter_list)) {
225 		TAILQ_INSERT_HEAD(&graph->iter_list, node, entry);
226 		graph->iter_node = node;
227 		return;
228 	}
229 
230 	n = graph->iter_node;
231 	/* Ensure that an iteration in progress will see this new commit. */
232 	while (n) {
233 		next = TAILQ_NEXT(n, entry);
234 		if (next && node->commit_timestamp >= next->commit_timestamp) {
235 			TAILQ_INSERT_BEFORE(next, node, entry);
236 			return;
237 		}
238 		n = next;
239 	}
240 	TAILQ_INSERT_AFTER(&graph->iter_list, graph->iter_node, node, entry);
241 }
242 
243 static const struct got_error *
244 add_vertex(struct got_object_id_queue *ids, struct got_object_id *id)
245 {
246 	struct got_object_qid *qid;
247 
248 	qid = calloc(1, sizeof(*qid));
249 	if (qid == NULL)
250 		return got_error_from_errno();
251 
252 	qid->id = got_object_id_dup(id);
253 	if (qid->id == NULL) {
254 		const struct got_error *err = got_error_from_errno();
255 		got_object_qid_free(qid);
256 		return err;
257 	}
258 
259 	SIMPLEQ_INSERT_TAIL(ids, qid, entry);
260 	return NULL;
261 }
262 
263 static const struct got_error *
264 close_branch(struct got_commit_graph *graph, struct got_object_id *commit_id)
265 {
266 	const struct got_error *err;
267 
268 	err = got_object_idset_remove(NULL, graph->open_branches, commit_id);
269 	if (err && err->code != GOT_ERR_NO_OBJ)
270 		return err;
271 	return NULL;
272 }
273 
274 static const struct got_error *
275 advance_branch(struct got_commit_graph *graph,
276     struct got_commit_graph_node *node,
277     struct got_object_id *commit_id, struct got_commit_object *commit,
278     struct got_repository *repo)
279 {
280 	const struct got_error *err;
281 	struct got_object_qid *qid;
282 
283 	err = close_branch(graph, commit_id);
284 	if (err)
285 		return err;
286 
287 	if (graph->flags & GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL) {
288 		qid = SIMPLEQ_FIRST(&commit->parent_ids);
289 		if (qid == NULL)
290 			return NULL;
291 		err = got_object_idset_add(NULL, graph->open_branches, qid->id,
292 		    node);
293 		if (err && err->code != GOT_ERR_OBJ_EXISTS)
294 			return err;
295 		return NULL;
296 	}
297 
298 	/*
299 	 * If we are graphing commits for a specific path, skip branches
300 	 * which do not contribute any content to this path.
301 	 */
302 	if (is_merge_point(node) && !got_path_is_root_dir(graph->path)) {
303 		struct got_object_id *id, *merged_id, *prev_id = NULL;
304 		int branches_differ = 0;
305 		#if 0
306 		int n = 1;
307 		#endif
308 
309 		err = got_object_id_by_path(&merged_id, repo, commit_id,
310 		    graph->path);
311 		if (err)
312 			return err;
313 
314 		SIMPLEQ_FOREACH(qid, &commit->parent_ids, entry) {
315 			if (got_object_idset_get(graph->node_ids, qid->id))
316 				continue; /* parent already traversed */
317 
318 			err = got_object_id_by_path(&id, repo, qid->id,
319 			    graph->path);
320 			if (err) {
321 				if (err->code == GOT_ERR_NO_OBJ) {
322 					branches_differ = 1;
323 					continue;
324 				}
325 				return err;
326 			}
327 
328 			if (prev_id) {
329 				if (!branches_differ &&
330 				    got_object_id_cmp(merged_id, prev_id) != 0)
331 					branches_differ = 1;
332 			} else
333 				prev_id = id;
334 
335 			/*
336 			 * If a branch has created the merged content we can
337 			 * skip any other branches.
338 			 */
339 			if (got_object_id_cmp(merged_id, id) == 0) {
340 				err = got_object_idset_add(NULL,
341 				    graph->open_branches, qid->id, node);
342 				if (err && err->code != GOT_ERR_OBJ_EXISTS)
343 					return err;
344 				return NULL;
345 			}
346 		}
347 
348 		/*
349 		 * If the path's content is the same on all branches,
350 		 * follow the first parent only.
351 		 */
352 		if (!branches_differ) {
353 			qid = SIMPLEQ_FIRST(&commit->parent_ids);
354 			if (qid == NULL)
355 				return NULL;
356 			if (got_object_idset_get(graph->node_ids, qid->id))
357 				return NULL; /* parent already traversed */
358 			err = got_object_idset_add(NULL,
359 			    graph->open_branches, qid->id, node);
360 			if (err && err->code != GOT_ERR_OBJ_EXISTS)
361 				return err;
362 			return NULL;
363 		}
364 	}
365 
366 	SIMPLEQ_FOREACH(qid, &commit->parent_ids, entry) {
367 		if (got_object_idset_get(graph->node_ids, qid->id))
368 			continue; /* parent already traversed */
369 		err = got_object_idset_add(NULL, graph->open_branches,
370 		    qid->id, node);
371 		if (err && err->code != GOT_ERR_OBJ_EXISTS)
372 			return err;
373 	}
374 
375 	return NULL;
376 }
377 
378 static void
379 free_node(struct got_commit_graph_node *node)
380 {
381 	while (!SIMPLEQ_EMPTY(&node->parent_ids)) {
382 		struct got_object_qid *pid = SIMPLEQ_FIRST(&node->parent_ids);
383 		SIMPLEQ_REMOVE_HEAD(&node->parent_ids, entry);
384 		got_object_qid_free(pid);
385 	}
386 	free(node);
387 }
388 
389 static const struct got_error *
390 add_node(struct got_commit_graph_node **new_node,
391     struct got_commit_graph *graph, struct got_object_id *commit_id,
392     struct got_commit_object *commit, struct got_commit_graph_node *child_node,
393     struct got_repository *repo)
394 {
395 	const struct got_error *err = NULL;
396 	struct got_commit_graph_node *node;
397 	struct got_object_qid *pid;
398 	int changed = 0, branch_done = 0;
399 
400 	*new_node = NULL;
401 
402 	node = calloc(1, sizeof(*node));
403 	if (node == NULL)
404 		return got_error_from_errno();
405 
406 	memcpy(&node->id, commit_id, sizeof(node->id));
407 	SIMPLEQ_INIT(&node->parent_ids);
408 	SIMPLEQ_FOREACH(pid, &commit->parent_ids, entry) {
409 		err = add_vertex(&node->parent_ids, pid->id);
410 		if (err) {
411 			free_node(node);
412 			return err;
413 		}
414 		node->nparents++;
415 	}
416 
417 	err = got_object_idset_add(NULL, graph->node_ids, &node->id, node);
418 	if (err) {
419 		if (err->code == GOT_ERR_OBJ_EXISTS)
420 			err = NULL;
421 		free_node(node);
422 		return err;
423 	}
424 
425 	err = detect_changed_path(&changed, commit, commit_id, graph->path,
426 	    repo);
427 	if (err) {
428 		if (err->code == GOT_ERR_NO_OBJ) {
429 			/*
430 			 * History of the path stops here on the current
431 			 * branch. Keep going on other branches.
432 			 */
433 			err = NULL;
434 			branch_done = 1;
435 		} else {
436 			free_node(node);
437 			return err;
438 		}
439 	}
440 
441 	node->commit_timestamp = mktime(&commit->tm_committer);
442 	if (node->commit_timestamp == -1) {
443 		free_node(node);
444 		return got_error_from_errno();
445 	}
446 
447 	if (changed && !is_merge_point(node))
448 		add_node_to_iter_list(graph, node, child_node);
449 
450 	if (branch_done)
451 		err = close_branch(graph, commit_id);
452 	else
453 		err = advance_branch(graph, node, commit_id, commit, repo);
454 	if (err)
455 		free_node(node);
456 	else
457 		*new_node = node;
458 
459 	return err;
460 }
461 
462 const struct got_error *
463 got_commit_graph_open(struct got_commit_graph **graph,
464     struct got_object_id *commit_id, const char *path,
465     int first_parent_traversal, struct got_repository *repo)
466 {
467 	const struct got_error *err = NULL;
468 	struct got_commit_object *commit;
469 
470 	*graph = NULL;
471 
472 	err = got_object_open_as_commit(&commit, repo, commit_id);
473 	if (err)
474 		return err;
475 
476 	/* The path must exist in our initial commit. */
477 	if (!got_path_is_root_dir(path)) {
478 		struct got_object_id *obj_id;
479 		err = got_object_id_by_path(&obj_id, repo, commit_id, path);
480 		if (err)
481 			return err;
482 		free(obj_id);
483 	}
484 
485 	*graph = alloc_graph(path);
486 	if (*graph == NULL) {
487 		got_object_commit_close(commit);
488 		return got_error_from_errno();
489 	}
490 
491 	if (first_parent_traversal)
492 		(*graph)->flags |= GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL;
493 
494 	err = add_node(&(*graph)->head_node, *graph, commit_id, commit, NULL,
495 	    repo);
496 	got_object_commit_close(commit);
497 	if (err) {
498 		got_commit_graph_close(*graph);
499 		*graph = NULL;
500 		return err;
501 	}
502 	
503 	return NULL;
504 }
505 
506 struct gather_branch_tips_arg {
507 	struct got_commit_graph_branch_tip *tips;
508 	int ntips;
509 };
510 
511 static void
512 gather_branch_tips(struct got_object_id *id, void *data, void *arg)
513 {
514 	struct gather_branch_tips_arg *a = arg;
515 	memcpy(&a->tips[a->ntips].id, id, sizeof(*id));
516 	a->tips[a->ntips].node = data;
517 	a->ntips++;
518 }
519 
520 static const struct got_error *
521 fetch_commits_from_open_branches(int *ncommits, int *wanted_id_added,
522     struct got_object_id **changed_id, struct got_commit_graph *graph,
523     struct got_repository *repo, struct got_object_id *wanted_id)
524 {
525 	const struct got_error *err;
526 	struct gather_branch_tips_arg arg;
527 	int i;
528 
529 	*ncommits = 0;
530 	if (wanted_id_added)
531 		*wanted_id_added = 0;
532 	if (changed_id)
533 		*changed_id = NULL;
534 
535 	arg.ntips = got_object_idset_num_elements(graph->open_branches);
536 	if (arg.ntips == 0)
537 		return NULL;
538 
539 	/*
540 	 * Adding nodes to the graph might change the graph's open
541 	 * branches state. Create a local copy of the current state.
542 	 */
543 	if (graph->ntips < arg.ntips) {
544 		struct got_commit_graph_branch_tip *tips;
545 		if (arg.ntips < GOT_COMMIT_GRAPH_MIN_TIPS)
546 			arg.ntips = GOT_COMMIT_GRAPH_MIN_TIPS;
547 		tips = reallocarray(graph->tips, arg.ntips, sizeof(*tips));
548 		if (tips == NULL)
549 			return got_error_from_errno();
550 		graph->tips = tips;
551 		graph->ntips = arg.ntips;
552 	}
553 	arg.ntips = 0; /* reset; gather_branch_tips() will increment */
554 	arg.tips = graph->tips;
555 	got_object_idset_for_each(graph->open_branches,
556 	    gather_branch_tips, &arg);
557 
558 	for (i = 0; i < arg.ntips; i++) {
559 		struct got_object_id *commit_id;
560 		struct got_commit_graph_node *child_node, *new_node;
561 		struct got_commit_object *commit;
562 		int changed;
563 
564 		commit_id = &graph->tips[i].id;
565 		child_node = graph->tips[i].node;
566 
567 		err = got_object_open_as_commit(&commit, repo, commit_id);
568 		if (err)
569 			break;
570 
571 		err = detect_changed_path(&changed, commit, commit_id,
572 		    graph->path, repo);
573 		if (err) {
574 			got_object_commit_close(commit);
575 			if (err->code != GOT_ERR_NO_OBJ)
576 				break;
577 			err = close_branch(graph, commit_id);
578 			if (err)
579 				break;
580 			continue;
581 		}
582 		if (changed && changed_id && *changed_id == NULL)
583 			*changed_id = commit_id;
584 		err = add_node(&new_node, graph, commit_id, commit, child_node,
585 		    repo);
586 		got_object_commit_close(commit);
587 		if (err)
588 			break;
589 		if (new_node)
590 			(*ncommits)++;
591 		if (wanted_id && got_object_id_cmp(commit_id, wanted_id) == 0)
592 			*wanted_id_added = 1;
593 	}
594 
595 	return err;
596 }
597 
598 const struct got_error *
599 got_commit_graph_fetch_commits(struct got_commit_graph *graph, int limit,
600     struct got_repository *repo)
601 {
602 	const struct got_error *err;
603 	int nfetched = 0, ncommits;
604 	struct got_object_id *changed_id = NULL;
605 
606 	while (nfetched < limit) {
607 		err = fetch_commits_from_open_branches(&ncommits, NULL,
608 		    &changed_id, graph, repo, NULL);
609 		if (err)
610 			return err;
611 		if (ncommits == 0)
612 			break;
613 		if (changed_id)
614 			nfetched += ncommits;
615 	}
616 
617 	return NULL;
618 }
619 
620 const struct got_error *
621 got_commit_graph_fetch_commits_up_to(int *nfetched,
622     struct got_commit_graph *graph, struct got_object_id *wanted_id,
623     struct got_repository *repo)
624 {
625 	const struct got_error *err;
626 	int ncommits, wanted_id_added = 0;
627 
628 	*nfetched = 0;
629 
630 	if (got_object_idset_get(graph->node_ids, wanted_id) != NULL)
631 		return NULL;
632 
633 	while (!wanted_id_added) {
634 		err = fetch_commits_from_open_branches(&ncommits,
635 		    &wanted_id_added, NULL, graph, repo, wanted_id);
636 		if (err)
637 			return err;
638 		if (ncommits == 0)
639 			return NULL;
640 		*nfetched += ncommits;
641 	}
642 
643 	return NULL;
644 }
645 
646 static void
647 free_node_iter(struct got_object_id *id, void *data, void *arg)
648 {
649 	struct got_commit_graph_node *node = data;
650 	free_node(node);
651 }
652 
653 void
654 got_commit_graph_close(struct got_commit_graph *graph)
655 {
656 	got_object_idset_free(graph->open_branches);
657 	got_object_idset_for_each(graph->node_ids, free_node_iter, NULL);
658 	got_object_idset_free(graph->node_ids);
659 	free(graph->tips);
660 	free(graph->path);
661 	free(graph);
662 }
663 
664 const struct got_error *
665 got_commit_graph_iter_start(struct got_commit_graph *graph,
666     struct got_object_id *id, struct got_repository *repo)
667 {
668 	const struct got_error *err = NULL;
669 	struct got_commit_graph_node *start_node;
670 	struct got_commit_object *commit;
671 	int changed;
672 
673 	start_node = got_object_idset_get(graph->node_ids, id);
674 	if (start_node == NULL)
675 		return got_error(GOT_ERR_NO_OBJ);
676 
677 
678 	err = got_object_open_as_commit(&commit, repo, &start_node->id);
679 	if (err)
680 		return err;
681 
682 	err = detect_changed_path(&changed, commit, &start_node->id,
683 	    graph->path, repo);
684 	if (err) {
685 		got_object_commit_close(commit);
686 		return err;
687 	}
688 
689 	if (!changed) {
690 		/* Locate first commit which changed graph->path. */
691 		struct got_object_id *changed_id = NULL;
692 		while (changed_id == NULL) {
693 			int ncommits;
694 			err = fetch_commits_from_open_branches(&ncommits, NULL,
695 			    &changed_id, graph, repo, NULL);
696 			if (err) {
697 				got_object_commit_close(commit);
698 				return err;
699 			}
700 		}
701 		start_node = got_object_idset_get(graph->node_ids, changed_id);
702 	}
703 	got_object_commit_close(commit);
704 
705 	graph->iter_node = start_node;
706 	return NULL;
707 }
708 
709 const struct got_error *
710 got_commit_graph_iter_next(struct got_object_id **id,
711     struct got_commit_graph *graph)
712 {
713 	*id = NULL;
714 
715 	if (graph->iter_node == NULL) {
716 		/* We are done iterating, or iteration was not started. */
717 		return got_error(GOT_ERR_ITER_COMPLETED);
718 	}
719 
720 	if (graph->iter_node ==
721 	    TAILQ_LAST(&graph->iter_list, got_commit_graph_iter_list) &&
722 	    got_object_idset_num_elements(graph->open_branches) == 0) {
723 		/* We are done iterating. */
724 		*id = &graph->iter_node->id;
725 		graph->iter_node = NULL;
726 		return NULL;
727 	}
728 
729 	if (TAILQ_NEXT(graph->iter_node, entry) == NULL)
730 		return got_error(GOT_ERR_ITER_NEED_MORE);
731 
732 	*id = &graph->iter_node->id;
733 	graph->iter_node = TAILQ_NEXT(graph->iter_node, entry);
734 	return NULL;
735 }