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 <limits.h>
23 #include <stdio.h>
24 #include <stdlib.h>
25 #include <string.h>
26 #include <sha1.h>
27 #include <zlib.h>
28 #include <ctype.h>
29 
30 #include "got_error.h"
31 #include "got_object.h"
32 #include "got_cancel.h"
33 #include "got_commit_graph.h"
34 #include "got_path.h"
35 
36 #include "got_lib_delta.h"
37 #include "got_lib_inflate.h"
38 #include "got_lib_object.h"
39 #include "got_lib_object_idset.h"
40 
41 struct got_commit_graph_node {
42 	struct got_object_id id;
43 	time_t timestamp;
44 
45 	/* Used during graph iteration. */
46 	TAILQ_ENTRY(got_commit_graph_node) entry;
47 };
48 
49 TAILQ_HEAD(got_commit_graph_iter_list, got_commit_graph_node);
50 
51 struct got_commit_graph_branch_tip {
52 	struct got_object_id *commit_id;
53 	struct got_commit_object *commit;
54 	struct got_commit_graph_node *new_node;
55 };
56 
57 struct got_commit_graph {
58 	/* The set of all commits we have traversed. */
59 	struct got_object_idset *node_ids;
60 
61 	int flags;
62 #define GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL		0x01
63 
64 	/*
65 	 * A set of object IDs of known parent commits which we have not yet
66 	 * traversed. Each commit ID in this set represents a branch in commit
67 	 * history: Either the first-parent branch of the head node, or another
68 	 * branch corresponding to a traversed merge commit for which we have
69 	 * not traversed a branch point commit yet.
70 	 *
71 	 * Whenever we add a commit with a matching ID to the graph, we remove
72 	 * its corresponding element from this set, and add new elements for
73 	 * each of that commit's parent commits which were not traversed yet.
74 	 *
75 	 * When API users ask us to fetch more commits, we fetch commits from
76 	 * all currently open branches. This allows API users to process
77 	 * commits in linear order even though the history contains branches.
78 	 */
79 	struct got_object_idset *open_branches;
80 
81 	/* Array of branch tips for fetch_commits_from_open_branches(). */
82 	struct got_commit_graph_branch_tip *tips;
83 	int ntips;
84 
85 	/* Path of tree entry of interest to the API user. */
86 	char *path;
87 
88 	/* The next commit to return when the API user asks for one. */
89 	struct got_commit_graph_node *iter_node;
90 
91 	/* The graph iteration list contains all nodes in sorted order. */
92 	struct got_commit_graph_iter_list iter_list;
93 };
94 
95 static const struct got_error *
96 detect_changed_path(int *changed, struct got_commit_object *commit,
97     struct got_object_id *commit_id, const char *path,
98     struct got_repository *repo)
99 {
100 	const struct got_error *err = NULL;
101 	struct got_commit_object *pcommit = NULL;
102 	struct got_tree_object *tree = NULL, *ptree = NULL;
103 	struct got_object_qid *pid;
104 
105 	if (got_path_is_root_dir(path)) {
106 		*changed = 1;
107 		return NULL;
108 	}
109 
110 	*changed = 0;
111 
112 	pid = SIMPLEQ_FIRST(&commit->parent_ids);
113 	if (pid == NULL) {
114 		struct got_object_id *obj_id;
115 		err = got_object_id_by_path(&obj_id, repo, commit_id, path);
116 		if (err) {
117 			if (err->code == GOT_ERR_NO_TREE_ENTRY)
118 				err = NULL;
119 		} else
120 			*changed = 1; /* The path was created in this commit. */
121 		free(obj_id);
122 		return err;
123 	}
124 
125 	err = got_object_open_as_tree(&tree, repo, commit->tree_id);
126 	if (err)
127 		return err;
128 
129 	err = got_object_open_as_commit(&pcommit, repo, pid->id);
130 	if (err)
131 		goto done;
132 
133 	err = got_object_open_as_tree(&ptree, repo, pcommit->tree_id);
134 	if (err)
135 		goto done;
136 
137 	err = got_object_tree_path_changed(changed, tree, ptree, path, repo);
138 done:
139 	if (tree)
140 		got_object_tree_close(tree);
141 	if (ptree)
142 		got_object_tree_close(ptree);
143 	if (pcommit)
144 		got_object_commit_close(pcommit);
145 	return err;
146 }
147 
148 static void
149 add_node_to_iter_list(struct got_commit_graph *graph,
150     struct got_commit_graph_node *node)
151 {
152 	struct got_commit_graph_node *n, *next;
153 
154 	if (TAILQ_EMPTY(&graph->iter_list)) {
155 		TAILQ_INSERT_HEAD(&graph->iter_list, node, entry);
156 		graph->iter_node = node;
157 		return;
158 	}
159 
160 	n = graph->iter_node;
161 	/* Ensure that an iteration in progress will see this new commit. */
162 	while (n) {
163 		next = TAILQ_NEXT(n, entry);
164 		if (next && node->timestamp >= next->timestamp) {
165 			TAILQ_INSERT_BEFORE(next, node, entry);
166 			return;
167 		}
168 		n = next;
169 	}
170 	TAILQ_INSERT_TAIL(&graph->iter_list, node, entry);
171 }
172 
173 static const struct got_error *
174 close_branch(struct got_commit_graph *graph, struct got_object_id *commit_id)
175 {
176 	const struct got_error *err;
177 
178 	err = got_object_idset_remove(NULL, graph->open_branches, commit_id);
179 	if (err && err->code != GOT_ERR_NO_OBJ)
180 		return err;
181 	return NULL;
182 }
183 
184 static const struct got_error *
185 advance_branch(struct got_commit_graph *graph,
186     struct got_commit_graph_node *node,
187     struct got_object_id *commit_id, struct got_commit_object *commit,
188     struct got_repository *repo)
189 {
190 	const struct got_error *err;
191 	struct got_object_qid *qid;
192 
193 	err = close_branch(graph, commit_id);
194 	if (err)
195 		return err;
196 
197 	if (graph->flags & GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL) {
198 		qid = SIMPLEQ_FIRST(&commit->parent_ids);
199 		if (qid == NULL ||
200 		    got_object_idset_get(graph->open_branches, qid->id))
201 			return NULL;
202 		return got_object_idset_add(graph->open_branches,
203 		    qid->id, node);
204 	}
205 
206 	/*
207 	 * If we are graphing commits for a specific path, skip branches
208 	 * which do not contribute any content to this path.
209 	 */
210 	if (commit->nparents > 1 && !got_path_is_root_dir(graph->path)) {
211 		struct got_object_id *merged_id, *prev_id = NULL;
212 		int branches_differ = 0;
213 
214 		err = got_object_id_by_path(&merged_id, repo, commit_id,
215 		    graph->path);
216 		if (err)
217 			return err;
218 
219 		SIMPLEQ_FOREACH(qid, &commit->parent_ids, entry) {
220 			struct got_object_id *id;
221 
222 			if (got_object_idset_get(graph->open_branches, qid->id))
223 				continue;
224 
225 			err = got_object_id_by_path(&id, repo, qid->id,
226 			    graph->path);
227 			if (err) {
228 				if (err->code == GOT_ERR_NO_TREE_ENTRY) {
229 					branches_differ = 1;
230 					continue;
231 				}
232 				free(merged_id);
233 				free(prev_id);
234 				return err;
235 			}
236 
237 			if (prev_id) {
238 				if (!branches_differ &&
239 				    got_object_id_cmp(id, prev_id) != 0)
240 					branches_differ = 1;
241 				free(prev_id);
242 			}
243 			prev_id = id;
244 
245 			/*
246 			 * If a branch has created the merged content we can
247 			 * skip any other branches.
248 			 */
249 			if (got_object_id_cmp(merged_id, id) == 0) {
250 				err = got_object_idset_add(graph->open_branches,
251 				    qid->id, node);
252 				free(merged_id);
253 				free(id);
254 				return err;
255 			}
256 		}
257 
258 		free(prev_id);
259 		prev_id = NULL;
260 		free(merged_id);
261 		merged_id = NULL;
262 
263 		/*
264 		 * If the path's content is the same on all branches,
265 		 * follow the first parent only.
266 		 */
267 		if (!branches_differ) {
268 			qid = SIMPLEQ_FIRST(&commit->parent_ids);
269 			if (qid == NULL)
270 				return NULL;
271 			if (got_object_idset_get(graph->open_branches, qid->id))
272 				return NULL;
273 			if (got_object_idset_get(graph->node_ids, qid->id))
274 				return NULL; /* parent already traversed */
275 			return got_object_idset_add(graph->open_branches,
276 			    qid->id, node);
277 		}
278 	}
279 
280 	SIMPLEQ_FOREACH(qid, &commit->parent_ids, entry) {
281 		if (got_object_idset_get(graph->open_branches, qid->id))
282 			continue;
283 		if (got_object_idset_get(graph->node_ids, qid->id))
284 			continue; /* parent already traversed */
285 		err = got_object_idset_add(graph->open_branches, qid->id, node);
286 		if (err)
287 			return err;
288 	}
289 
290 	return NULL;
291 }
292 
293 static const struct got_error *
294 add_node(struct got_commit_graph_node **new_node,
295     struct got_commit_graph *graph,
296     struct got_object_id *commit_id,
297     struct got_commit_object *commit,
298     struct got_repository *repo)
299 {
300 	const struct got_error *err = NULL;
301 	struct got_commit_graph_node *node;
302 
303 	*new_node = NULL;
304 
305 	node = calloc(1, sizeof(*node));
306 	if (node == NULL)
307 		return got_error_from_errno("calloc");
308 
309 	memcpy(&node->id, commit_id, sizeof(node->id));
310 	node->timestamp = commit->committer_time;
311 
312 	err = got_object_idset_add(graph->node_ids, &node->id, node);
313 	if (err)
314 		free(node);
315 	else
316 		*new_node = node;
317 	return err;
318 }
319 
320 const struct got_error *
321 got_commit_graph_open(struct got_commit_graph **graph,
322     const char *path, int first_parent_traversal)
323 {
324 	const struct got_error *err = NULL;
325 
326 	*graph = calloc(1, sizeof(**graph));
327 	if (*graph == NULL)
328 		return got_error_from_errno("calloc");
329 
330 	TAILQ_INIT(&(*graph)->iter_list);
331 
332 	(*graph)->path = strdup(path);
333 	if ((*graph)->path == NULL) {
334 		err = got_error_from_errno("strdup");
335 		goto done;
336 	}
337 
338 	(*graph)->node_ids = got_object_idset_alloc();
339 	if ((*graph)->node_ids == NULL) {
340 		err = got_error_from_errno("got_object_idset_alloc");
341 		goto done;
342 	}
343 
344 	(*graph)->open_branches = got_object_idset_alloc();
345 	if ((*graph)->open_branches == NULL) {
346 		err = got_error_from_errno("got_object_idset_alloc");
347 		goto done;
348 	}
349 
350 	if (first_parent_traversal)
351 		(*graph)->flags |= GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL;
352 done:
353 	if (err) {
354 		got_commit_graph_close(*graph);
355 		*graph = NULL;
356 	}
357 	return err;
358 }
359 
360 struct add_branch_tip_arg {
361 	struct got_commit_graph_branch_tip *tips;
362 	int ntips;
363 	struct got_repository *repo;
364 	struct got_commit_graph *graph;
365 };
366 
367 static const struct got_error *
368 add_branch_tip(struct got_object_id *commit_id, void *data, void *arg)
369 {
370 	const struct got_error *err;
371 	struct add_branch_tip_arg *a = arg;
372 	struct got_commit_graph_node *new_node;
373 	struct got_commit_object *commit;
374 
375 	err = got_object_open_as_commit(&commit, a->repo, commit_id);
376 	if (err)
377 		return err;
378 
379 	err = add_node(&new_node, a->graph, commit_id, commit, a->repo);
380 	if (err)
381 		return err;
382 
383 	a->tips[a->ntips].commit_id = new_node ? &new_node->id : NULL;
384 	a->tips[a->ntips].commit = commit;
385 	a->tips[a->ntips].new_node = new_node;
386 	a->ntips++;
387 
388 	return NULL;
389 }
390 
391 static const struct got_error *
392 fetch_commits_from_open_branches(struct got_commit_graph *graph,
393     struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
394 {
395 	const struct got_error *err;
396 	struct add_branch_tip_arg arg;
397 	int i, ntips;
398 
399 	ntips = got_object_idset_num_elements(graph->open_branches);
400 	if (ntips == 0)
401 		return NULL;
402 
403 	/* (Re-)allocate branch tips array if necessary. */
404 	if (graph->ntips < ntips) {
405 		struct got_commit_graph_branch_tip *tips;
406 		tips = recallocarray(graph->tips, graph->ntips, ntips,
407 		    sizeof(*tips));
408 		if (tips == NULL)
409 			return got_error_from_errno("recallocarray");
410 		graph->tips = tips;
411 		graph->ntips = ntips;
412 	}
413 	arg.tips = graph->tips;
414 	arg.ntips = 0; /* add_branch_tip() will increment */
415 	arg.repo = repo;
416 	arg.graph = graph;
417 	err = got_object_idset_for_each(graph->open_branches, add_branch_tip,
418 	    &arg);
419 	if (err)
420 		goto done;
421 
422 	for (i = 0; i < arg.ntips; i++) {
423 		struct got_object_id *commit_id;
424 		struct got_commit_object *commit;
425 		struct got_commit_graph_node *new_node;
426 		int changed;
427 
428 		if (cancel_cb) {
429 			err = (*cancel_cb)(cancel_arg);
430 			if (err)
431 				break;
432 		}
433 
434 		commit_id = arg.tips[i].commit_id;
435 		commit = arg.tips[i].commit;
436 		new_node = arg.tips[i].new_node;
437 
438 		err = detect_changed_path(&changed, commit, commit_id,
439 		    graph->path, repo);
440 		if (err) {
441 			if (err->code != GOT_ERR_NO_OBJ)
442 				break;
443 			/*
444 			 * History of the path stops here on the current
445 			 * branch. Keep going on other branches.
446 			 */
447 			err = close_branch(graph, commit_id);
448 			if (err)
449 				break;
450 			continue;
451 		}
452 		if (changed)
453 			add_node_to_iter_list(graph, new_node);
454 		err = advance_branch(graph, new_node, commit_id,
455 		    commit, repo);
456 		if (err)
457 			break;
458 	}
459 done:
460 	for (i = 0; i < arg.ntips; i++)
461 		got_object_commit_close(arg.tips[i].commit);
462 	return err;
463 }
464 
465 static const struct got_error *
466 free_node_iter(struct got_object_id *id, void *data, void *arg)
467 {
468 	struct got_commit_graph_node *node = data;
469 	free(node);
470 	return NULL;
471 }
472 
473 void
474 got_commit_graph_close(struct got_commit_graph *graph)
475 {
476 	if (graph->open_branches)
477 		got_object_idset_free(graph->open_branches);
478 	got_object_idset_for_each(graph->node_ids, free_node_iter, NULL);
479 	if (graph->node_ids)
480 		got_object_idset_free(graph->node_ids);
481 	free(graph->tips);
482 	free(graph->path);
483 	free(graph);
484 }
485 
486 const struct got_error *
487 got_commit_graph_iter_start(struct got_commit_graph *graph,
488     struct got_object_id *id, struct got_repository *repo,
489     got_cancel_cb cancel_cb, void *cancel_arg)
490 {
491 	const struct got_error *err = NULL;
492 	struct got_commit_graph_node *start_node;
493 	struct got_commit_object *commit;
494 	int changed;
495 
496 	if (!TAILQ_EMPTY(&graph->iter_list))
497 		return got_error(GOT_ERR_ITER_BUSY);
498 
499 	err = got_object_open_as_commit(&commit, repo, id);
500 	if (err)
501 		return err;
502 
503 	err = add_node(&start_node, graph, id, commit, repo);
504 	if (err)
505 		goto done;
506 
507 	err = detect_changed_path(&changed, commit, id, graph->path, repo);
508 	if (err)
509 		goto done;
510 	if (changed)
511 		add_node_to_iter_list(graph, start_node);
512 
513 	err = advance_branch(graph, start_node, id, commit, repo);
514 	if (err)
515 		goto done;
516 
517 	if (!changed) {
518 		/* Locate first commit which changed graph->path. */
519 		while (graph->iter_node == NULL &&
520 		    got_object_idset_num_elements(graph->open_branches) > 0) {
521 			err = fetch_commits_from_open_branches(graph, repo,
522 			    cancel_cb, cancel_arg);
523 			if (err)
524 				break;
525 		}
526 	}
527 done:
528 	got_object_commit_close(commit);
529 	return err;
530 }
531 
532 const struct got_error *
533 got_commit_graph_iter_next(struct got_object_id **id,
534     struct got_commit_graph *graph, struct got_repository *repo,
535     got_cancel_cb cancel_cb, void *cancel_arg)
536 {
537 	const struct got_error *err = NULL;
538 
539 	*id = NULL;
540 
541 	if (graph->iter_node == NULL) {
542 		/* We are done iterating, or iteration was not started. */
543 		return got_error(GOT_ERR_ITER_COMPLETED);
544 	}
545 
546 	if (graph->iter_node ==
547 	    TAILQ_LAST(&graph->iter_list, got_commit_graph_iter_list) &&
548 	    got_object_idset_num_elements(graph->open_branches) == 0) {
549 		/* We are done iterating. */
550 		*id = &graph->iter_node->id;
551 		graph->iter_node = NULL;
552 		return NULL;
553 	}
554 
555 	while (TAILQ_NEXT(graph->iter_node, entry) == NULL &&
556 	    got_object_idset_num_elements(graph->open_branches) > 0) {
557 		err = fetch_commits_from_open_branches(graph, repo,
558 		    cancel_cb, cancel_arg);
559 		if (err)
560 			return err;
561 	}
562 
563 	*id = &graph->iter_node->id;
564 	graph->iter_node = TAILQ_NEXT(graph->iter_node, entry);
565 	return NULL;
566 }
567 
568 const struct got_error *
569 got_commit_graph_find_youngest_common_ancestor(struct got_object_id **yca_id,
570     struct got_object_id *commit_id, struct got_object_id *commit_id2,
571     struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
572 {
573 	const struct got_error *err = NULL;
574 	struct got_commit_graph *graph = NULL, *graph2 = NULL;
575 	int completed = 0, completed2 = 0;
576 	struct got_object_idset *commit_ids;
577 
578 	*yca_id = NULL;
579 
580 	commit_ids = got_object_idset_alloc();
581 	if (commit_ids == NULL)
582 		return got_error_from_errno("got_object_idset_alloc");
583 
584 	err = got_commit_graph_open(&graph, "/", 1);
585 	if (err)
586 		goto done;
587 
588 	err = got_commit_graph_open(&graph2, "/", 1);
589 	if (err)
590 		goto done;
591 
592 	err = got_commit_graph_iter_start(graph, commit_id, repo,
593 	    cancel_cb, cancel_arg);
594 	if (err)
595 		goto done;
596 
597 	err = got_commit_graph_iter_start(graph2, commit_id2, repo,
598 	    cancel_cb, cancel_arg);
599 	if (err)
600 		goto done;
601 
602 	for (;;) {
603 		struct got_object_id *id = NULL, *id2 = NULL;
604 
605 		if (cancel_cb) {
606 			err = (*cancel_cb)(cancel_arg);
607 			if (err)
608 				break;
609 		}
610 
611 		if (!completed) {
612 			err = got_commit_graph_iter_next(&id, graph, repo,
613 			    cancel_cb, cancel_arg);
614 			if (err) {
615 				if (err->code != GOT_ERR_ITER_COMPLETED)
616 					break;
617 				err = NULL;
618 				completed = 1;
619 			}
620 		}
621 
622 		if (!completed2) {
623 			err = got_commit_graph_iter_next(&id2, graph2, repo,
624 			    cancel_cb, cancel_arg);
625 			if (err) {
626 				if (err->code != GOT_ERR_ITER_COMPLETED)
627 					break;
628 				err = NULL;
629 				completed2 = 1;
630 			}
631 		}
632 
633 		if (id) {
634 			if (got_object_idset_get(commit_ids, id)) {
635 				*yca_id = got_object_id_dup(id);
636 				if (*yca_id)
637 					break;
638 				err = got_error_from_errno("got_object_id_dup");
639 				break;
640 
641 			}
642 			err = got_object_idset_add(commit_ids, id, id);
643 			if (err)
644 				break;
645 		}
646 		if (id2) {
647 			if (got_object_idset_get(commit_ids, id2)) {
648 				*yca_id = got_object_id_dup(id2);
649 				if (*yca_id)
650 					break;
651 				err = got_error_from_errno("got_object_id_dup");
652 				break;
653 
654 			}
655 			err = got_object_idset_add(commit_ids, id2, id2);
656 			if (err)
657 				break;
658 		}
659 
660 		if (completed && completed2) {
661 			err = got_error(GOT_ERR_ANCESTRY);
662 			break;
663 		}
664 
665 	}
666 done:
667 	got_object_idset_free(commit_ids);
668 	if (graph)
669 		got_commit_graph_close(graph);
670 	if (graph2)
671 		got_commit_graph_close(graph2);
672 	return err;
673 }