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