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