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 static int
147 is_merge_point(struct got_commit_graph_node *node)
148 {
149 	return node->nparents > 1;
150 }
151 
152 int
153 is_branch_point(struct got_commit_graph_node *node)
154 {
155 	return node->nchildren > 1;
156 }
157 
158 static int
159 is_root_node(struct got_commit_graph_node *node)
160 {
161 	return node->nparents == 0;
162 }
163 #endif
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_TAIL(&graph->iter_list, node, entry);
226 		return;
227 	}
228 
229 	/* Ensure that an iteration in progress will see this new commit. */
230 	if (graph->iter_node) {
231 		n = graph->iter_node;
232 		while (n) {
233 			next = TAILQ_NEXT(n, entry);
234 			if (next &&
235 			    node->commit_timestamp >= next->commit_timestamp) {
236 				TAILQ_INSERT_BEFORE(next, node, entry);
237 				return;
238 			}
239 			n = next;
240 		}
241 		TAILQ_INSERT_AFTER(&graph->iter_list, graph->iter_node,
242 		    node, entry);
243 		return;
244 	}
245 
246 	/*
247 	 * If a child node is known, begin looping over the list there
248 	 * instead of beginning from the list head.
249 	 */
250 	n = child_node ? child_node : TAILQ_FIRST(&graph->iter_list);
251 
252 	/* Insert into list based on committer timestamp. */
253 	do {
254 		if (node->commit_timestamp == n->commit_timestamp) {
255 			TAILQ_INSERT_AFTER(&graph->iter_list, n, node, entry);
256 			break;
257 		} else if (node->commit_timestamp < n->commit_timestamp) {
258 			next = TAILQ_NEXT(n, entry);
259 			if (next == NULL) {
260 				TAILQ_INSERT_AFTER(&graph->iter_list, n,
261 				    node, entry);
262 				break;
263 			}
264 			if (node->commit_timestamp >= next->commit_timestamp) {
265 				TAILQ_INSERT_BEFORE(next, node, entry);
266 				break;
267 			}
268 		} else {
269 			TAILQ_INSERT_BEFORE(n, node, entry);
270 			break;
271 		}
272 		n = TAILQ_NEXT(n, entry);
273 	} while (n);
274 }
275 
276 static const struct got_error *
277 add_vertex(struct got_object_id_queue *ids, struct got_object_id *id)
278 {
279 	struct got_object_qid *qid;
280 
281 	qid = calloc(1, sizeof(*qid));
282 	if (qid == NULL)
283 		return got_error_from_errno();
284 
285 	qid->id = got_object_id_dup(id);
286 	if (qid->id == NULL) {
287 		const struct got_error *err = got_error_from_errno();
288 		got_object_qid_free(qid);
289 		return err;
290 	}
291 
292 	SIMPLEQ_INSERT_TAIL(ids, qid, entry);
293 	return NULL;
294 }
295 
296 static const struct got_error *
297 close_branch(struct got_commit_graph *graph, struct got_object_id *commit_id)
298 {
299 	const struct got_error *err;
300 
301 	err = got_object_idset_remove(NULL, graph->open_branches, commit_id);
302 	if (err && err->code != GOT_ERR_NO_OBJ)
303 		return err;
304 	return NULL;
305 }
306 
307 static const struct got_error *
308 advance_branch(struct got_commit_graph *graph,
309     struct got_commit_graph_node *node,
310     struct got_object_id *commit_id, struct got_commit_object *commit)
311 {
312 	const struct got_error *err;
313 	struct got_object_qid *qid;
314 
315 	err = close_branch(graph, commit_id);
316 	if (err)
317 		return err;
318 
319 	if (graph->flags & GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL) {
320 		qid = SIMPLEQ_FIRST(&commit->parent_ids);
321 		if (qid == NULL)
322 			return NULL;
323 		err = got_object_idset_add(NULL, graph->open_branches, qid->id,
324 		    node);
325 		if (err && err->code != GOT_ERR_OBJ_EXISTS)
326 			return err;
327 		return NULL;
328 	}
329 
330 	SIMPLEQ_FOREACH(qid, &commit->parent_ids, entry) {
331 		if (got_object_idset_get(graph->node_ids, qid->id))
332 			continue; /* parent already traversed */
333 		err = got_object_idset_add(NULL, graph->open_branches,
334 		    qid->id, node);
335 		if (err && err->code != GOT_ERR_OBJ_EXISTS)
336 			return err;
337 	}
338 
339 	return NULL;
340 }
341 
342 static void
343 free_node(struct got_commit_graph_node *node)
344 {
345 	while (!SIMPLEQ_EMPTY(&node->parent_ids)) {
346 		struct got_object_qid *pid = SIMPLEQ_FIRST(&node->parent_ids);
347 		SIMPLEQ_REMOVE_HEAD(&node->parent_ids, entry);
348 		got_object_qid_free(pid);
349 	}
350 	free(node);
351 }
352 
353 static const struct got_error *
354 add_node(struct got_commit_graph_node **new_node,
355     struct got_commit_graph *graph, struct got_object_id *commit_id,
356     struct got_commit_object *commit, struct got_commit_graph_node *child_node,
357     struct got_repository *repo)
358 {
359 	const struct got_error *err = NULL;
360 	struct got_commit_graph_node *node;
361 	struct got_object_qid *pid;
362 	int changed = 0, branch_done = 0;
363 
364 	*new_node = NULL;
365 
366 	node = calloc(1, sizeof(*node));
367 	if (node == NULL)
368 		return got_error_from_errno();
369 
370 	memcpy(&node->id, commit_id, sizeof(node->id));
371 	SIMPLEQ_INIT(&node->parent_ids);
372 	SIMPLEQ_FOREACH(pid, &commit->parent_ids, entry) {
373 		err = add_vertex(&node->parent_ids, pid->id);
374 		if (err) {
375 			free_node(node);
376 			return err;
377 		}
378 		node->nparents++;
379 	}
380 	node->commit_timestamp = mktime(&commit->tm_committer); 
381 	if (node->commit_timestamp == -1) {
382 		free_node(node);
383 		return got_error_from_errno();
384 	}
385 
386 	err = got_object_idset_add(NULL, graph->node_ids, &node->id, node);
387 	if (err) {
388 		if (err->code == GOT_ERR_OBJ_EXISTS)
389 			err = NULL;
390 		free_node(node);
391 		return err;
392 	}
393 
394 	err = detect_changed_path(&changed, commit, commit_id, graph->path,
395 	    repo);
396 	if (err) {
397 		if (err->code == GOT_ERR_NO_OBJ) {
398 			/*
399 			 * History of the path stops here on the current
400 			 * branch. Keep going on other branches.
401 			 */
402 			err = NULL;
403 			branch_done = 1;
404 		} else {
405 			free_node(node);
406 			return err;
407 		}
408 	}
409 
410 	if (changed)
411 		add_node_to_iter_list(graph, node, child_node);
412 
413 	if (branch_done)
414 		err = close_branch(graph, commit_id);
415 	else
416 		err = advance_branch(graph, node, commit_id, commit);
417 	if (err)
418 		free_node(node);
419 	else
420 		*new_node = node;
421 
422 	return err;
423 }
424 
425 const struct got_error *
426 got_commit_graph_open(struct got_commit_graph **graph,
427     struct got_object_id *commit_id, const char *path,
428     int first_parent_traversal, struct got_repository *repo)
429 {
430 	const struct got_error *err = NULL;
431 	struct got_commit_object *commit;
432 
433 	*graph = NULL;
434 
435 	err = got_object_open_as_commit(&commit, repo, commit_id);
436 	if (err)
437 		return err;
438 
439 	/* The path must exist in our initial commit. */
440 	if (!got_path_is_root_dir(path)) {
441 		struct got_object_id *obj_id;
442 		err = got_object_id_by_path(&obj_id, repo, commit_id, path);
443 		if (err)
444 			return err;
445 		free(obj_id);
446 	}
447 
448 	*graph = alloc_graph(path);
449 	if (*graph == NULL) {
450 		got_object_commit_close(commit);
451 		return got_error_from_errno();
452 	}
453 
454 	if (first_parent_traversal)
455 		(*graph)->flags |= GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL;
456 
457 	err = add_node(&(*graph)->head_node, *graph, commit_id, commit, NULL,
458 	    repo);
459 	got_object_commit_close(commit);
460 	if (err) {
461 		got_commit_graph_close(*graph);
462 		*graph = NULL;
463 		return err;
464 	}
465 	
466 	return NULL;
467 }
468 
469 struct gather_branch_tips_arg {
470 	struct got_commit_graph_branch_tip *tips;
471 	int ntips;
472 };
473 
474 static void
475 gather_branch_tips(struct got_object_id *id, void *data, void *arg)
476 {
477 	struct gather_branch_tips_arg *a = arg;
478 	memcpy(&a->tips[a->ntips].id, id, sizeof(*id));
479 	a->tips[a->ntips].node = data;
480 	a->ntips++;
481 }
482 
483 static const struct got_error *
484 fetch_commits_from_open_branches(int *ncommits, int *wanted_id_added,
485     struct got_object_id **changed_id, struct got_commit_graph *graph,
486     struct got_repository *repo, struct got_object_id *wanted_id)
487 {
488 	const struct got_error *err;
489 	struct gather_branch_tips_arg arg;
490 	int i;
491 
492 	*ncommits = 0;
493 	if (wanted_id_added)
494 		*wanted_id_added = 0;
495 	if (changed_id)
496 		*changed_id = NULL;
497 
498 	arg.ntips = got_object_idset_num_elements(graph->open_branches);
499 	if (arg.ntips == 0)
500 		return NULL;
501 
502 	/*
503 	 * Adding nodes to the graph might change the graph's open
504 	 * branches state. Create a local copy of the current state.
505 	 */
506 	if (graph->ntips < arg.ntips) {
507 		struct got_commit_graph_branch_tip *tips;
508 		if (arg.ntips < GOT_COMMIT_GRAPH_MIN_TIPS)
509 			arg.ntips = GOT_COMMIT_GRAPH_MIN_TIPS;
510 		tips = reallocarray(graph->tips, arg.ntips, sizeof(*tips));
511 		if (tips == NULL)
512 			return got_error_from_errno();
513 		graph->tips = tips;
514 		graph->ntips = arg.ntips;
515 	}
516 	arg.ntips = 0; /* reset; gather_branch_tips() will increment */
517 	arg.tips = graph->tips;
518 	got_object_idset_for_each(graph->open_branches,
519 	    gather_branch_tips, &arg);
520 
521 	for (i = 0; i < arg.ntips; i++) {
522 		struct got_object_id *commit_id;
523 		struct got_commit_graph_node *child_node, *new_node;
524 		struct got_commit_object *commit;
525 		int changed;
526 
527 		commit_id = &graph->tips[i].id;
528 		child_node = graph->tips[i].node;
529 
530 		err = got_object_open_as_commit(&commit, repo, commit_id);
531 		if (err)
532 			break;
533 
534 		err = detect_changed_path(&changed, commit, commit_id,
535 		    graph->path, repo);
536 		if (err) {
537 			got_object_commit_close(commit);
538 			if (err->code != GOT_ERR_NO_OBJ)
539 				break;
540 			err = close_branch(graph, commit_id);
541 			if (err)
542 				break;
543 			continue;
544 		}
545 		if (changed && changed_id && *changed_id == NULL)
546 			*changed_id = commit_id;
547 		err = add_node(&new_node, graph, commit_id, commit, child_node,
548 		    repo);
549 		got_object_commit_close(commit);
550 		if (err)
551 			break;
552 		if (new_node)
553 			(*ncommits)++;
554 		if (wanted_id && got_object_id_cmp(commit_id, wanted_id) == 0)
555 			*wanted_id_added = 1;
556 	}
557 
558 	return err;
559 }
560 
561 const struct got_error *
562 got_commit_graph_fetch_commits(struct got_commit_graph *graph, int limit,
563     struct got_repository *repo)
564 {
565 	const struct got_error *err;
566 	int nfetched = 0, ncommits;
567 	struct got_object_id *changed_id = NULL;
568 
569 	while (nfetched < limit) {
570 		err = fetch_commits_from_open_branches(&ncommits, NULL,
571 		    &changed_id, graph, repo, NULL);
572 		if (err)
573 			return err;
574 		if (ncommits == 0)
575 			break;
576 		if (changed_id)
577 			nfetched += ncommits;
578 	}
579 
580 	return NULL;
581 }
582 
583 const struct got_error *
584 got_commit_graph_fetch_commits_up_to(int *nfetched,
585     struct got_commit_graph *graph, struct got_object_id *wanted_id,
586     struct got_repository *repo)
587 {
588 	const struct got_error *err;
589 	int ncommits, wanted_id_added = 0;
590 
591 	*nfetched = 0;
592 
593 	if (got_object_idset_get(graph->node_ids, wanted_id) != NULL)
594 		return NULL;
595 
596 	while (!wanted_id_added) {
597 		err = fetch_commits_from_open_branches(&ncommits,
598 		    &wanted_id_added, NULL, graph, repo, wanted_id);
599 		if (err)
600 			return err;
601 		if (ncommits == 0)
602 			return NULL;
603 		*nfetched += ncommits;
604 	}
605 
606 	return NULL;
607 }
608 
609 static void
610 free_node_iter(struct got_object_id *id, void *data, void *arg)
611 {
612 	struct got_commit_graph_node *node = data;
613 	free_node(node);
614 }
615 
616 void
617 got_commit_graph_close(struct got_commit_graph *graph)
618 {
619 	got_object_idset_free(graph->open_branches);
620 	got_object_idset_for_each(graph->node_ids, free_node_iter, NULL);
621 	got_object_idset_free(graph->node_ids);
622 	free(graph->tips);
623 	free(graph->path);
624 	free(graph);
625 }
626 
627 const struct got_error *
628 got_commit_graph_iter_start(struct got_commit_graph *graph,
629     struct got_object_id *id, struct got_repository *repo)
630 {
631 	const struct got_error *err = NULL;
632 	struct got_commit_graph_node *start_node;
633 	struct got_commit_object *commit;
634 	int changed;
635 
636 	start_node = got_object_idset_get(graph->node_ids, id);
637 	if (start_node == NULL)
638 		return got_error(GOT_ERR_NO_OBJ);
639 
640 
641 	err = got_object_open_as_commit(&commit, repo, &start_node->id);
642 	if (err)
643 		return err;
644 
645 	err = detect_changed_path(&changed, commit, &start_node->id,
646 	    graph->path, repo);
647 	if (err) {
648 		got_object_commit_close(commit);
649 		return err;
650 	}
651 
652 	if (!changed) {
653 		/* Locate first commit which changed graph->path. */
654 		struct got_object_id *changed_id = NULL;
655 		while (changed_id == NULL) {
656 			int ncommits;
657 			err = fetch_commits_from_open_branches(&ncommits, NULL,
658 			    &changed_id, graph, repo, NULL);
659 			if (err) {
660 				got_object_commit_close(commit);
661 				return err;
662 			}
663 		}
664 		start_node = got_object_idset_get(graph->node_ids, changed_id);
665 	}
666 	got_object_commit_close(commit);
667 
668 	graph->iter_node = start_node;
669 	return NULL;
670 }
671 
672 const struct got_error *
673 got_commit_graph_iter_next(struct got_object_id **id,
674     struct got_commit_graph *graph)
675 {
676 	*id = NULL;
677 
678 	if (graph->iter_node == NULL) {
679 		/* We are done interating, or iteration was not started. */
680 		return got_error(GOT_ERR_ITER_COMPLETED);
681 	}
682 
683 	if (graph->iter_node ==
684 	    TAILQ_LAST(&graph->iter_list, got_commit_graph_iter_list) &&
685 	    got_object_idset_num_elements(graph->open_branches) == 0) {
686 		*id = &graph->iter_node->id;
687 		/* We are done interating. */
688 		graph->iter_node = NULL;
689 		return NULL;
690 	}
691 
692 	if (TAILQ_NEXT(graph->iter_node, entry) == NULL)
693 		return got_error(GOT_ERR_ITER_NEED_MORE);
694 
695 	*id = &graph->iter_node->id;
696 	graph->iter_node = TAILQ_NEXT(graph->iter_node, entry);
697 	return NULL;
698 }