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 			if (err->code == GOT_ERR_NO_OBJ)
192 				err = NULL;
193 		} else
194 			*changed = 1; /* The path was created in this commit. */
195 		free(obj_id);
196 	} else {
197 		err = got_object_open_as_commit(&pcommit, repo, pid->id);
198 		if (err)
199 			goto done;
200 
201 		err = got_object_open_as_tree(&ptree, repo, pcommit->tree_id);
202 		if (err)
203 			goto done;
204 
205 		err = got_object_tree_path_changed(changed, tree, ptree, path,
206 		    repo);
207 	}
208 done:
209 	if (tree)
210 		got_object_tree_close(tree);
211 	if (ptree)
212 		got_object_tree_close(ptree);
213 	if (pcommit)
214 		got_object_commit_close(pcommit);
215 	return err;
216 }
217 
218 static void
219 add_node_to_iter_list(struct got_commit_graph *graph,
220     struct got_commit_graph_node *node,
221     struct got_commit_graph_node *child_node)
222 {
223 	struct got_commit_graph_node *n, *next;
224 
225 	if (TAILQ_EMPTY(&graph->iter_list)) {
226 		TAILQ_INSERT_HEAD(&graph->iter_list, node, entry);
227 		graph->iter_node = node;
228 		return;
229 	}
230 
231 	n = graph->iter_node;
232 	/* Ensure that an iteration in progress will see this new commit. */
233 	while (n) {
234 		next = TAILQ_NEXT(n, entry);
235 		if (next && node->commit_timestamp >= next->commit_timestamp) {
236 			TAILQ_INSERT_BEFORE(next, node, entry);
237 			return;
238 		}
239 		n = next;
240 	}
241 	TAILQ_INSERT_TAIL(&graph->iter_list, node, entry);
242 }
243 
244 static const struct got_error *
245 add_vertex(struct got_object_id_queue *ids, struct got_object_id *id)
246 {
247 	struct got_object_qid *qid;
248 
249 	qid = calloc(1, sizeof(*qid));
250 	if (qid == NULL)
251 		return got_error_from_errno();
252 
253 	qid->id = got_object_id_dup(id);
254 	if (qid->id == NULL) {
255 		const struct got_error *err = got_error_from_errno();
256 		got_object_qid_free(qid);
257 		return err;
258 	}
259 
260 	SIMPLEQ_INSERT_TAIL(ids, qid, entry);
261 	return NULL;
262 }
263 
264 static const struct got_error *
265 close_branch(struct got_commit_graph *graph, struct got_object_id *commit_id)
266 {
267 	const struct got_error *err;
268 
269 	err = got_object_idset_remove(NULL, graph->open_branches, commit_id);
270 	if (err && err->code != GOT_ERR_NO_OBJ)
271 		return err;
272 	return NULL;
273 }
274 
275 static const struct got_error *
276 advance_branch(struct got_commit_graph *graph,
277     struct got_commit_graph_node *node,
278     struct got_object_id *commit_id, struct got_commit_object *commit,
279     struct got_repository *repo)
280 {
281 	const struct got_error *err;
282 	struct got_object_qid *qid;
283 
284 	err = close_branch(graph, commit_id);
285 	if (err)
286 		return err;
287 
288 	if (graph->flags & GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL) {
289 		qid = SIMPLEQ_FIRST(&commit->parent_ids);
290 		if (qid == NULL)
291 			return NULL;
292 		err = got_object_idset_add(graph->open_branches, qid->id, node);
293 		return err;
294 	}
295 
296 	/*
297 	 * If we are graphing commits for a specific path, skip branches
298 	 * which do not contribute any content to this path.
299 	 */
300 	if (is_merge_point(node) && !got_path_is_root_dir(graph->path)) {
301 		struct got_object_id *id, *merged_id, *prev_id = NULL;
302 		int branches_differ = 0;
303 
304 		err = got_object_id_by_path(&merged_id, repo, commit_id,
305 		    graph->path);
306 		if (err)
307 			return err;
308 
309 		SIMPLEQ_FOREACH(qid, &commit->parent_ids, entry) {
310 			if (got_object_idset_get(graph->node_ids, qid->id))
311 				continue; /* parent already traversed */
312 
313 			err = got_object_id_by_path(&id, repo, qid->id,
314 			    graph->path);
315 			if (err) {
316 				if (err->code == GOT_ERR_NO_OBJ) {
317 					branches_differ = 1;
318 					continue;
319 				}
320 				return err;
321 			}
322 
323 			if (prev_id) {
324 				if (!branches_differ &&
325 				    got_object_id_cmp(merged_id, prev_id) != 0)
326 					branches_differ = 1;
327 			} else
328 				prev_id = id;
329 
330 			/*
331 			 * If a branch has created the merged content we can
332 			 * skip any other branches.
333 			 */
334 			if (got_object_id_cmp(merged_id, id) == 0) {
335 				err = got_object_idset_add(graph->open_branches,
336 				    qid->id, node);
337 				return err;
338 			}
339 		}
340 
341 		/*
342 		 * If the path's content is the same on all branches,
343 		 * follow the first parent only.
344 		 */
345 		if (!branches_differ) {
346 			qid = SIMPLEQ_FIRST(&commit->parent_ids);
347 			if (qid == NULL)
348 				return NULL;
349 			if (got_object_idset_get(graph->node_ids, qid->id))
350 				return NULL; /* parent already traversed */
351 			if (got_object_idset_get(graph->open_branches, qid->id))
352 				return NULL;
353 			return got_object_idset_add(graph->open_branches,
354 			    qid->id, node);
355 		}
356 	}
357 
358 	SIMPLEQ_FOREACH(qid, &commit->parent_ids, entry) {
359 		if (got_object_idset_get(graph->node_ids, qid->id))
360 			continue; /* parent already traversed */
361 		if (got_object_idset_get(graph->open_branches, qid->id))
362 			continue;
363 		err = got_object_idset_add(graph->open_branches, qid->id, node);
364 		if (err)
365 			return err;
366 	}
367 
368 	return NULL;
369 }
370 
371 static void
372 free_node(struct got_commit_graph_node *node)
373 {
374 	while (!SIMPLEQ_EMPTY(&node->parent_ids)) {
375 		struct got_object_qid *pid = SIMPLEQ_FIRST(&node->parent_ids);
376 		SIMPLEQ_REMOVE_HEAD(&node->parent_ids, entry);
377 		got_object_qid_free(pid);
378 	}
379 	free(node);
380 }
381 
382 static const struct got_error *
383 add_node(struct got_commit_graph_node **new_node,
384     struct got_commit_graph *graph, struct got_object_id *commit_id,
385     struct got_commit_object *commit, struct got_commit_graph_node *child_node,
386     struct got_repository *repo)
387 {
388 	const struct got_error *err = NULL;
389 	struct got_commit_graph_node *node;
390 	struct got_object_qid *pid;
391 	int changed = 0, branch_done = 0;
392 
393 	*new_node = NULL;
394 
395 	node = calloc(1, sizeof(*node));
396 	if (node == NULL)
397 		return got_error_from_errno();
398 
399 	memcpy(&node->id, commit_id, sizeof(node->id));
400 	SIMPLEQ_INIT(&node->parent_ids);
401 	SIMPLEQ_FOREACH(pid, &commit->parent_ids, entry) {
402 		err = add_vertex(&node->parent_ids, pid->id);
403 		if (err) {
404 			free_node(node);
405 			return err;
406 		}
407 		node->nparents++;
408 	}
409 
410 	err = got_object_idset_add(graph->node_ids, &node->id, node);
411 	if (err) {
412 		free_node(node);
413 		return err;
414 	}
415 
416 	err = detect_changed_path(&changed, commit, commit_id, graph->path,
417 	    repo);
418 	if (err) {
419 		if (err->code == GOT_ERR_NO_OBJ) {
420 			/*
421 			 * History of the path stops here on the current
422 			 * branch. Keep going on other branches.
423 			 */
424 			err = NULL;
425 			branch_done = 1;
426 		} else {
427 			free_node(node);
428 			return err;
429 		}
430 	}
431 
432 	node->commit_timestamp = mktime(&commit->tm_committer);
433 	if (node->commit_timestamp == -1) {
434 		free_node(node);
435 		return got_error_from_errno();
436 	}
437 
438 	if (changed)
439 		add_node_to_iter_list(graph, node, child_node);
440 
441 	if (branch_done)
442 		err = close_branch(graph, commit_id);
443 	else
444 		err = advance_branch(graph, node, commit_id, commit, repo);
445 	if (err)
446 		free_node(node);
447 	else
448 		*new_node = node;
449 
450 	return err;
451 }
452 
453 const struct got_error *
454 got_commit_graph_open(struct got_commit_graph **graph,
455     struct got_object_id *commit_id, const char *path,
456     int first_parent_traversal, struct got_repository *repo)
457 {
458 	const struct got_error *err = NULL;
459 	struct got_commit_object *commit;
460 
461 	*graph = NULL;
462 
463 	err = got_object_open_as_commit(&commit, repo, commit_id);
464 	if (err)
465 		return err;
466 
467 	/* The path must exist in our initial commit. */
468 	if (!got_path_is_root_dir(path)) {
469 		struct got_object_id *obj_id;
470 		err = got_object_id_by_path(&obj_id, repo, commit_id, path);
471 		if (err)
472 			return err;
473 		free(obj_id);
474 	}
475 
476 	*graph = alloc_graph(path);
477 	if (*graph == NULL) {
478 		got_object_commit_close(commit);
479 		return got_error_from_errno();
480 	}
481 
482 	if (first_parent_traversal)
483 		(*graph)->flags |= GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL;
484 
485 	err = add_node(&(*graph)->head_node, *graph, commit_id, commit, NULL,
486 	    repo);
487 	got_object_commit_close(commit);
488 	if (err) {
489 		got_commit_graph_close(*graph);
490 		*graph = NULL;
491 		return err;
492 	}
493 	
494 	return NULL;
495 }
496 
497 struct gather_branch_tips_arg {
498 	struct got_commit_graph_branch_tip *tips;
499 	int ntips;
500 };
501 
502 static void
503 gather_branch_tips(struct got_object_id *id, void *data, void *arg)
504 {
505 	struct gather_branch_tips_arg *a = arg;
506 	memcpy(&a->tips[a->ntips].id, id, sizeof(*id));
507 	a->tips[a->ntips].node = data;
508 	a->ntips++;
509 }
510 
511 static const struct got_error *
512 fetch_commits_from_open_branches(int *ncommits,
513     struct got_object_id **changed_id, struct got_commit_graph *graph,
514     struct got_repository *repo)
515 {
516 	const struct got_error *err;
517 	struct gather_branch_tips_arg arg;
518 	int i;
519 
520 	*ncommits = 0;
521 	if (changed_id)
522 		*changed_id = NULL;
523 
524 	arg.ntips = got_object_idset_num_elements(graph->open_branches);
525 	if (arg.ntips == 0)
526 		return NULL;
527 
528 	/*
529 	 * Adding nodes to the graph might change the graph's open
530 	 * branches state. Create a local copy of the current state.
531 	 */
532 	if (graph->ntips < arg.ntips) {
533 		struct got_commit_graph_branch_tip *tips;
534 		if (arg.ntips < GOT_COMMIT_GRAPH_MIN_TIPS)
535 			arg.ntips = GOT_COMMIT_GRAPH_MIN_TIPS;
536 		tips = reallocarray(graph->tips, arg.ntips, sizeof(*tips));
537 		if (tips == NULL)
538 			return got_error_from_errno();
539 		graph->tips = tips;
540 		graph->ntips = arg.ntips;
541 	}
542 	arg.ntips = 0; /* reset; gather_branch_tips() will increment */
543 	arg.tips = graph->tips;
544 	got_object_idset_for_each(graph->open_branches,
545 	    gather_branch_tips, &arg);
546 
547 	for (i = 0; i < arg.ntips; i++) {
548 		struct got_object_id *commit_id;
549 		struct got_commit_graph_node *child_node, *new_node;
550 		struct got_commit_object *commit;
551 		int changed;
552 
553 		commit_id = &graph->tips[i].id;
554 		child_node = graph->tips[i].node;
555 
556 		err = got_object_open_as_commit(&commit, repo, commit_id);
557 		if (err)
558 			break;
559 
560 		err = detect_changed_path(&changed, commit, commit_id,
561 		    graph->path, repo);
562 		if (err) {
563 			got_object_commit_close(commit);
564 			if (err->code != GOT_ERR_NO_OBJ)
565 				break;
566 			err = close_branch(graph, commit_id);
567 			if (err)
568 				break;
569 			continue;
570 		}
571 		if (changed && changed_id && *changed_id == NULL)
572 			*changed_id = commit_id;
573 		err = add_node(&new_node, graph, commit_id, commit, child_node,
574 		    repo);
575 		got_object_commit_close(commit);
576 		if (err)
577 			break;
578 		if (new_node)
579 			(*ncommits)++;
580 	}
581 
582 	return err;
583 }
584 
585 const struct got_error *
586 got_commit_graph_fetch_commits(struct got_commit_graph *graph, int limit,
587     struct got_repository *repo)
588 {
589 	const struct got_error *err;
590 	int nfetched = 0, ncommits;
591 	struct got_object_id *changed_id = NULL;
592 
593 	while (nfetched < limit) {
594 		err = fetch_commits_from_open_branches(&ncommits,
595 		    &changed_id, graph, repo);
596 		if (err)
597 			return err;
598 		if (ncommits == 0)
599 			break;
600 		if (changed_id)
601 			nfetched += ncommits;
602 	}
603 
604 	return NULL;
605 }
606 
607 static void
608 free_node_iter(struct got_object_id *id, void *data, void *arg)
609 {
610 	struct got_commit_graph_node *node = data;
611 	free_node(node);
612 }
613 
614 void
615 got_commit_graph_close(struct got_commit_graph *graph)
616 {
617 	got_object_idset_free(graph->open_branches);
618 	got_object_idset_for_each(graph->node_ids, free_node_iter, NULL);
619 	got_object_idset_free(graph->node_ids);
620 	free(graph->tips);
621 	free(graph->path);
622 	free(graph);
623 }
624 
625 const struct got_error *
626 got_commit_graph_iter_start(struct got_commit_graph *graph,
627     struct got_object_id *id, struct got_repository *repo)
628 {
629 	const struct got_error *err = NULL;
630 	struct got_commit_graph_node *start_node;
631 	struct got_commit_object *commit;
632 	int changed;
633 
634 	start_node = got_object_idset_get(graph->node_ids, id);
635 	if (start_node == NULL)
636 		return got_error(GOT_ERR_NO_OBJ);
637 
638 	err = got_object_open_as_commit(&commit, repo, &start_node->id);
639 	if (err)
640 		return err;
641 
642 	err = detect_changed_path(&changed, commit, &start_node->id,
643 	    graph->path, repo);
644 	if (err) {
645 		got_object_commit_close(commit);
646 		return err;
647 	}
648 
649 	if (!changed) {
650 		/* Locate first commit which changed graph->path. */
651 		struct got_object_id *changed_id = NULL;
652 		while (changed_id == NULL) {
653 			int ncommits;
654 			err = fetch_commits_from_open_branches(&ncommits,
655 			    &changed_id, graph, repo);
656 			if (err) {
657 				got_object_commit_close(commit);
658 				return err;
659 			}
660 		}
661 		start_node = got_object_idset_get(graph->node_ids, changed_id);
662 	}
663 	got_object_commit_close(commit);
664 
665 	graph->iter_node = start_node;
666 	return NULL;
667 }
668 
669 const struct got_error *
670 got_commit_graph_iter_next(struct got_object_id **id,
671     struct got_commit_graph *graph)
672 {
673 	*id = NULL;
674 
675 	if (graph->iter_node == NULL) {
676 		/* We are done iterating, or iteration was not started. */
677 		return got_error(GOT_ERR_ITER_COMPLETED);
678 	}
679 
680 	if (graph->iter_node ==
681 	    TAILQ_LAST(&graph->iter_list, got_commit_graph_iter_list) &&
682 	    got_object_idset_num_elements(graph->open_branches) == 0) {
683 		/* We are done iterating. */
684 		*id = &graph->iter_node->id;
685 		graph->iter_node = NULL;
686 		return NULL;
687 	}
688 
689 	if (TAILQ_NEXT(graph->iter_node, entry) == NULL)
690 		return got_error(GOT_ERR_ITER_NEED_MORE);
691 
692 	*id = &graph->iter_node->id;
693 	graph->iter_node = TAILQ_NEXT(graph->iter_node, entry);
694 	return NULL;
695 }