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 	/*
89 	 * Nodes which will be passed to the API user next, sorted by
90 	 * commit timestmap.
91 	 */
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 		return;
157 	}
158 
159 	n = TAILQ_FIRST(&graph->iter_list);
160 	/* Ensure that an iteration in progress will see this new commit. */
161 	while (n) {
162 		next = TAILQ_NEXT(n, entry);
163 		if (next && node->timestamp >= next->timestamp) {
164 			TAILQ_INSERT_BEFORE(next, node, entry);
165 			return;
166 		}
167 		n = next;
168 	}
169 	TAILQ_INSERT_TAIL(&graph->iter_list, node, entry);
170 }
171 
172 static const struct got_error *
173 close_branch(struct got_commit_graph *graph, struct got_object_id *commit_id)
174 {
175 	const struct got_error *err;
176 
177 	err = got_object_idset_remove(NULL, graph->open_branches, commit_id);
178 	if (err && err->code != GOT_ERR_NO_OBJ)
179 		return err;
180 	return NULL;
181 }
182 
183 static const struct got_error *
184 advance_branch(struct got_commit_graph *graph, struct got_object_id *commit_id,
185     struct got_commit_object *commit, struct got_repository *repo)
186 {
187 	const struct got_error *err;
188 	struct got_object_qid *qid;
189 
190 	err = close_branch(graph, commit_id);
191 	if (err)
192 		return err;
193 
194 	if (graph->flags & GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL) {
195 		qid = SIMPLEQ_FIRST(&commit->parent_ids);
196 		if (qid == NULL ||
197 		    got_object_idset_contains(graph->open_branches, qid->id))
198 			return NULL;
199 		return got_object_idset_add(graph->open_branches,
200 		    qid->id, NULL);
201 	}
202 
203 	/*
204 	 * If we are graphing commits for a specific path, skip branches
205 	 * which do not contribute any content to this path.
206 	 */
207 	if (commit->nparents > 1 && !got_path_is_root_dir(graph->path)) {
208 		struct got_object_id *merged_id, *prev_id = NULL;
209 		int branches_differ = 0;
210 
211 		err = got_object_id_by_path(&merged_id, repo, commit_id,
212 		    graph->path);
213 		if (err)
214 			return err;
215 
216 		SIMPLEQ_FOREACH(qid, &commit->parent_ids, entry) {
217 			struct got_object_id *id;
218 
219 			if (got_object_idset_contains(graph->open_branches,
220 			    qid->id))
221 				continue;
222 
223 			err = got_object_id_by_path(&id, repo, qid->id,
224 			    graph->path);
225 			if (err) {
226 				if (err->code == GOT_ERR_NO_TREE_ENTRY) {
227 					branches_differ = 1;
228 					continue;
229 				}
230 				free(merged_id);
231 				free(prev_id);
232 				return err;
233 			}
234 
235 			if (prev_id) {
236 				if (!branches_differ &&
237 				    got_object_id_cmp(id, prev_id) != 0)
238 					branches_differ = 1;
239 				free(prev_id);
240 			}
241 			prev_id = id;
242 
243 			/*
244 			 * If a branch has created the merged content we can
245 			 * skip any other branches.
246 			 */
247 			if (got_object_id_cmp(merged_id, id) == 0) {
248 				err = got_object_idset_add(graph->open_branches,
249 				    qid->id, NULL);
250 				free(merged_id);
251 				free(id);
252 				return err;
253 			}
254 		}
255 
256 		free(prev_id);
257 		prev_id = NULL;
258 		free(merged_id);
259 		merged_id = NULL;
260 
261 		/*
262 		 * If the path's content is the same on all branches,
263 		 * follow the first parent only.
264 		 */
265 		if (!branches_differ) {
266 			qid = SIMPLEQ_FIRST(&commit->parent_ids);
267 			if (qid == NULL)
268 				return NULL;
269 			if (got_object_idset_contains(graph->open_branches,
270 			    qid->id))
271 				return NULL;
272 			if (got_object_idset_contains(graph->node_ids,
273 			    qid->id))
274 				return NULL; /* parent already traversed */
275 			return got_object_idset_add(graph->open_branches,
276 			    qid->id, NULL);
277 		}
278 	}
279 
280 	SIMPLEQ_FOREACH(qid, &commit->parent_ids, entry) {
281 		if (got_object_idset_contains(graph->open_branches, qid->id))
282 			continue;
283 		if (got_object_idset_contains(graph->node_ids, qid->id))
284 			continue; /* parent already traversed */
285 		err = got_object_idset_add(graph->open_branches, qid->id, NULL);
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, NULL);
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, commit_id, commit, repo);
455 		if (err)
456 			break;
457 	}
458 done:
459 	for (i = 0; i < arg.ntips; i++)
460 		got_object_commit_close(arg.tips[i].commit);
461 	return err;
462 }
463 
464 void
465 got_commit_graph_close(struct got_commit_graph *graph)
466 {
467 	if (graph->open_branches)
468 		got_object_idset_free(graph->open_branches);
469 	if (graph->node_ids)
470 		got_object_idset_free(graph->node_ids);
471 	free(graph->tips);
472 	free(graph->path);
473 	free(graph);
474 }
475 
476 const struct got_error *
477 got_commit_graph_iter_start(struct got_commit_graph *graph,
478     struct got_object_id *id, struct got_repository *repo,
479     got_cancel_cb cancel_cb, void *cancel_arg)
480 {
481 	const struct got_error *err = NULL;
482 
483 	if (!TAILQ_EMPTY(&graph->iter_list))
484 		return got_error(GOT_ERR_ITER_BUSY);
485 
486 	err = got_object_idset_add(graph->open_branches, id, NULL);
487 	if (err)
488 		return err;
489 
490 	/* Locate first commit which changed graph->path. */
491 	while (TAILQ_EMPTY(&graph->iter_list) &&
492 	    got_object_idset_num_elements(graph->open_branches) > 0) {
493 		err = fetch_commits_from_open_branches(graph, repo,
494 		    cancel_cb, cancel_arg);
495 		if (err)
496 			return err;
497 	}
498 
499 	if (TAILQ_EMPTY(&graph->iter_list)) {
500 		const char *path;
501 		if (got_path_is_root_dir(graph->path))
502 			return got_error_no_obj(id);
503 		path = graph->path;
504 		while (path[0] == '/')
505 			path++;
506 		return got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
507 	}
508 
509 	return NULL;
510 }
511 
512 const struct got_error *
513 got_commit_graph_iter_next(struct got_object_id **id,
514     struct got_commit_graph *graph, struct got_repository *repo,
515     got_cancel_cb cancel_cb, void *cancel_arg)
516 {
517 	const struct got_error *err = NULL;
518 	struct got_commit_graph_node *node;
519 
520 	*id = NULL;
521 
522 	if (TAILQ_EMPTY(&graph->iter_list)) {
523 		/* We are done iterating, or iteration was not started. */
524 		return got_error(GOT_ERR_ITER_COMPLETED);
525 	}
526 
527 	node = TAILQ_FIRST(&graph->iter_list);
528 	while (TAILQ_NEXT(node, entry) == NULL &&
529 	    got_object_idset_num_elements(graph->open_branches) > 0) {
530 		err = fetch_commits_from_open_branches(graph, repo,
531 		    cancel_cb, cancel_arg);
532 		if (err)
533 			return err;
534 	}
535 
536 	*id = &node->id;
537 	TAILQ_REMOVE(&graph->iter_list, node, entry);
538 	return NULL;
539 }
540 
541 const struct got_error *
542 got_commit_graph_find_youngest_common_ancestor(struct got_object_id **yca_id,
543     struct got_object_id *commit_id, struct got_object_id *commit_id2,
544     struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
545 {
546 	const struct got_error *err = NULL;
547 	struct got_commit_graph *graph = NULL, *graph2 = NULL;
548 	int completed = 0, completed2 = 0;
549 	struct got_object_idset *commit_ids;
550 
551 	*yca_id = NULL;
552 
553 	commit_ids = got_object_idset_alloc();
554 	if (commit_ids == NULL)
555 		return got_error_from_errno("got_object_idset_alloc");
556 
557 	err = got_commit_graph_open(&graph, "/", 1);
558 	if (err)
559 		goto done;
560 
561 	err = got_commit_graph_open(&graph2, "/", 1);
562 	if (err)
563 		goto done;
564 
565 	err = got_commit_graph_iter_start(graph, commit_id, repo,
566 	    cancel_cb, cancel_arg);
567 	if (err)
568 		goto done;
569 
570 	err = got_commit_graph_iter_start(graph2, commit_id2, repo,
571 	    cancel_cb, cancel_arg);
572 	if (err)
573 		goto done;
574 
575 	for (;;) {
576 		struct got_object_id *id = NULL, *id2 = NULL;
577 
578 		if (cancel_cb) {
579 			err = (*cancel_cb)(cancel_arg);
580 			if (err)
581 				break;
582 		}
583 
584 		if (!completed) {
585 			err = got_commit_graph_iter_next(&id, graph, repo,
586 			    cancel_cb, cancel_arg);
587 			if (err) {
588 				if (err->code != GOT_ERR_ITER_COMPLETED)
589 					break;
590 				err = NULL;
591 				completed = 1;
592 			}
593 		}
594 
595 		if (!completed2) {
596 			err = got_commit_graph_iter_next(&id2, graph2, repo,
597 			    cancel_cb, cancel_arg);
598 			if (err) {
599 				if (err->code != GOT_ERR_ITER_COMPLETED)
600 					break;
601 				err = NULL;
602 				completed2 = 1;
603 			}
604 		}
605 
606 		if (id) {
607 			if (got_object_idset_contains(commit_ids, id)) {
608 				*yca_id = got_object_id_dup(id);
609 				if (*yca_id)
610 					break;
611 				err = got_error_from_errno("got_object_id_dup");
612 				break;
613 
614 			}
615 			err = got_object_idset_add(commit_ids, id, NULL);
616 			if (err)
617 				break;
618 		}
619 		if (id2) {
620 			if (got_object_idset_contains(commit_ids, id2)) {
621 				*yca_id = got_object_id_dup(id2);
622 				if (*yca_id)
623 					break;
624 				err = got_error_from_errno("got_object_id_dup");
625 				break;
626 
627 			}
628 			err = got_object_idset_add(commit_ids, id2, NULL);
629 			if (err)
630 				break;
631 		}
632 
633 		if (completed && completed2) {
634 			err = got_error(GOT_ERR_ANCESTRY);
635 			break;
636 		}
637 
638 	}
639 done:
640 	got_object_idset_free(commit_ids);
641 	if (graph)
642 		got_commit_graph_close(graph);
643 	if (graph2)
644 		got_commit_graph_close(graph2);
645 	return err;
646 }