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 	n = TAILQ_FIRST(&graph->iter_list);
155 	/* Ensure that an iteration in progress will see this new commit. */
156 	while (n) {
157 		next = TAILQ_NEXT(n, entry);
158 		if (next && node->timestamp >= next->timestamp) {
159 			TAILQ_INSERT_BEFORE(next, node, entry);
160 			return;
161 		}
162 		n = next;
163 	}
164 	TAILQ_INSERT_TAIL(&graph->iter_list, node, entry);
165 }
166 
167 static const struct got_error *
168 close_branch(struct got_commit_graph *graph, struct got_object_id *commit_id)
169 {
170 	const struct got_error *err;
171 
172 	err = got_object_idset_remove(NULL, graph->open_branches, commit_id);
173 	if (err && err->code != GOT_ERR_NO_OBJ)
174 		return err;
175 	return NULL;
176 }
177 
178 static const struct got_error *
179 advance_branch(struct got_commit_graph *graph, struct got_object_id *commit_id,
180     struct got_commit_object *commit, struct got_repository *repo)
181 {
182 	const struct got_error *err;
183 	struct got_object_qid *qid;
184 
185 	err = close_branch(graph, commit_id);
186 	if (err)
187 		return err;
188 
189 	if (graph->flags & GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL) {
190 		qid = SIMPLEQ_FIRST(&commit->parent_ids);
191 		if (qid == NULL ||
192 		    got_object_idset_contains(graph->open_branches, qid->id))
193 			return NULL;
194 		return got_object_idset_add(graph->open_branches,
195 		    qid->id, NULL);
196 	}
197 
198 	/*
199 	 * If we are graphing commits for a specific path, skip branches
200 	 * which do not contribute any content to this path.
201 	 */
202 	if (commit->nparents > 1 && !got_path_is_root_dir(graph->path)) {
203 		struct got_object_id *merged_id, *prev_id = NULL;
204 		int branches_differ = 0;
205 
206 		err = got_object_id_by_path(&merged_id, repo, commit_id,
207 		    graph->path);
208 		if (err)
209 			return err;
210 
211 		SIMPLEQ_FOREACH(qid, &commit->parent_ids, entry) {
212 			struct got_object_id *id;
213 
214 			if (got_object_idset_contains(graph->open_branches,
215 			    qid->id))
216 				continue;
217 
218 			err = got_object_id_by_path(&id, repo, qid->id,
219 			    graph->path);
220 			if (err) {
221 				if (err->code == GOT_ERR_NO_TREE_ENTRY) {
222 					branches_differ = 1;
223 					continue;
224 				}
225 				free(merged_id);
226 				free(prev_id);
227 				return err;
228 			}
229 
230 			if (prev_id) {
231 				if (!branches_differ &&
232 				    got_object_id_cmp(id, prev_id) != 0)
233 					branches_differ = 1;
234 				free(prev_id);
235 			}
236 			prev_id = id;
237 
238 			/*
239 			 * If a branch has created the merged content we can
240 			 * skip any other branches.
241 			 */
242 			if (got_object_id_cmp(merged_id, id) == 0) {
243 				err = got_object_idset_add(graph->open_branches,
244 				    qid->id, NULL);
245 				free(merged_id);
246 				free(id);
247 				return err;
248 			}
249 		}
250 
251 		free(prev_id);
252 		prev_id = NULL;
253 		free(merged_id);
254 		merged_id = NULL;
255 
256 		/*
257 		 * If the path's content is the same on all branches,
258 		 * follow the first parent only.
259 		 */
260 		if (!branches_differ) {
261 			qid = SIMPLEQ_FIRST(&commit->parent_ids);
262 			if (qid == NULL)
263 				return NULL;
264 			if (got_object_idset_contains(graph->open_branches,
265 			    qid->id))
266 				return NULL;
267 			if (got_object_idset_contains(graph->node_ids,
268 			    qid->id))
269 				return NULL; /* parent already traversed */
270 			return got_object_idset_add(graph->open_branches,
271 			    qid->id, NULL);
272 		}
273 	}
274 
275 	SIMPLEQ_FOREACH(qid, &commit->parent_ids, entry) {
276 		if (got_object_idset_contains(graph->open_branches, qid->id))
277 			continue;
278 		if (got_object_idset_contains(graph->node_ids, qid->id))
279 			continue; /* parent already traversed */
280 		err = got_object_idset_add(graph->open_branches, qid->id, NULL);
281 		if (err)
282 			return err;
283 	}
284 
285 	return NULL;
286 }
287 
288 static const struct got_error *
289 add_node(struct got_commit_graph_node **new_node,
290     struct got_commit_graph *graph,
291     struct got_object_id *commit_id,
292     struct got_commit_object *commit,
293     struct got_repository *repo)
294 {
295 	const struct got_error *err = NULL;
296 	struct got_commit_graph_node *node;
297 
298 	*new_node = NULL;
299 
300 	node = calloc(1, sizeof(*node));
301 	if (node == NULL)
302 		return got_error_from_errno("calloc");
303 
304 	memcpy(&node->id, commit_id, sizeof(node->id));
305 	node->timestamp = commit->committer_time;
306 
307 	err = got_object_idset_add(graph->node_ids, &node->id, NULL);
308 	if (err)
309 		free(node);
310 	else
311 		*new_node = node;
312 	return err;
313 }
314 
315 const struct got_error *
316 got_commit_graph_open(struct got_commit_graph **graph,
317     const char *path, int first_parent_traversal)
318 {
319 	const struct got_error *err = NULL;
320 
321 	*graph = calloc(1, sizeof(**graph));
322 	if (*graph == NULL)
323 		return got_error_from_errno("calloc");
324 
325 	TAILQ_INIT(&(*graph)->iter_list);
326 
327 	(*graph)->path = strdup(path);
328 	if ((*graph)->path == NULL) {
329 		err = got_error_from_errno("strdup");
330 		goto done;
331 	}
332 
333 	(*graph)->node_ids = got_object_idset_alloc();
334 	if ((*graph)->node_ids == NULL) {
335 		err = got_error_from_errno("got_object_idset_alloc");
336 		goto done;
337 	}
338 
339 	(*graph)->open_branches = got_object_idset_alloc();
340 	if ((*graph)->open_branches == NULL) {
341 		err = got_error_from_errno("got_object_idset_alloc");
342 		goto done;
343 	}
344 
345 	if (first_parent_traversal)
346 		(*graph)->flags |= GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL;
347 done:
348 	if (err) {
349 		got_commit_graph_close(*graph);
350 		*graph = NULL;
351 	}
352 	return err;
353 }
354 
355 struct add_branch_tip_arg {
356 	struct got_commit_graph_branch_tip *tips;
357 	int ntips;
358 	struct got_repository *repo;
359 	struct got_commit_graph *graph;
360 };
361 
362 static const struct got_error *
363 add_branch_tip(struct got_object_id *commit_id, void *data, void *arg)
364 {
365 	const struct got_error *err;
366 	struct add_branch_tip_arg *a = arg;
367 	struct got_commit_graph_node *new_node;
368 	struct got_commit_object *commit;
369 
370 	err = got_object_open_as_commit(&commit, a->repo, commit_id);
371 	if (err)
372 		return err;
373 
374 	err = add_node(&new_node, a->graph, commit_id, commit, a->repo);
375 	if (err)
376 		return err;
377 
378 	a->tips[a->ntips].commit_id = new_node ? &new_node->id : NULL;
379 	a->tips[a->ntips].commit = commit;
380 	a->tips[a->ntips].new_node = new_node;
381 	a->ntips++;
382 
383 	return NULL;
384 }
385 
386 static const struct got_error *
387 fetch_commits_from_open_branches(struct got_commit_graph *graph,
388     struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
389 {
390 	const struct got_error *err;
391 	struct add_branch_tip_arg arg;
392 	int i, ntips;
393 
394 	ntips = got_object_idset_num_elements(graph->open_branches);
395 	if (ntips == 0)
396 		return NULL;
397 
398 	/* (Re-)allocate branch tips array if necessary. */
399 	if (graph->ntips < ntips) {
400 		struct got_commit_graph_branch_tip *tips;
401 		tips = recallocarray(graph->tips, graph->ntips, ntips,
402 		    sizeof(*tips));
403 		if (tips == NULL)
404 			return got_error_from_errno("recallocarray");
405 		graph->tips = tips;
406 		graph->ntips = ntips;
407 	}
408 	arg.tips = graph->tips;
409 	arg.ntips = 0; /* add_branch_tip() will increment */
410 	arg.repo = repo;
411 	arg.graph = graph;
412 	err = got_object_idset_for_each(graph->open_branches, add_branch_tip,
413 	    &arg);
414 	if (err)
415 		goto done;
416 
417 	for (i = 0; i < arg.ntips; i++) {
418 		struct got_object_id *commit_id;
419 		struct got_commit_object *commit;
420 		struct got_commit_graph_node *new_node;
421 		int changed;
422 
423 		if (cancel_cb) {
424 			err = (*cancel_cb)(cancel_arg);
425 			if (err)
426 				break;
427 		}
428 
429 		commit_id = arg.tips[i].commit_id;
430 		commit = arg.tips[i].commit;
431 		new_node = arg.tips[i].new_node;
432 
433 		err = detect_changed_path(&changed, commit, commit_id,
434 		    graph->path, repo);
435 		if (err) {
436 			if (err->code != GOT_ERR_NO_OBJ)
437 				break;
438 			/*
439 			 * History of the path stops here on the current
440 			 * branch. Keep going on other branches.
441 			 */
442 			err = close_branch(graph, commit_id);
443 			if (err)
444 				break;
445 			continue;
446 		}
447 		if (changed)
448 			add_node_to_iter_list(graph, new_node);
449 		err = advance_branch(graph, commit_id, commit, repo);
450 		if (err)
451 			break;
452 	}
453 done:
454 	for (i = 0; i < arg.ntips; i++)
455 		got_object_commit_close(arg.tips[i].commit);
456 	return err;
457 }
458 
459 void
460 got_commit_graph_close(struct got_commit_graph *graph)
461 {
462 	if (graph->open_branches)
463 		got_object_idset_free(graph->open_branches);
464 	if (graph->node_ids)
465 		got_object_idset_free(graph->node_ids);
466 	free(graph->tips);
467 	free(graph->path);
468 	free(graph);
469 }
470 
471 const struct got_error *
472 got_commit_graph_iter_start(struct got_commit_graph *graph,
473     struct got_object_id *id, struct got_repository *repo,
474     got_cancel_cb cancel_cb, void *cancel_arg)
475 {
476 	const struct got_error *err = NULL;
477 
478 	if (!TAILQ_EMPTY(&graph->iter_list))
479 		return got_error(GOT_ERR_ITER_BUSY);
480 
481 	err = got_object_idset_add(graph->open_branches, id, NULL);
482 	if (err)
483 		return err;
484 
485 	/* Locate first commit which changed graph->path. */
486 	while (TAILQ_EMPTY(&graph->iter_list) &&
487 	    got_object_idset_num_elements(graph->open_branches) > 0) {
488 		err = fetch_commits_from_open_branches(graph, repo,
489 		    cancel_cb, cancel_arg);
490 		if (err)
491 			return err;
492 	}
493 
494 	if (TAILQ_EMPTY(&graph->iter_list)) {
495 		const char *path;
496 		if (got_path_is_root_dir(graph->path))
497 			return got_error_no_obj(id);
498 		path = graph->path;
499 		while (path[0] == '/')
500 			path++;
501 		return got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
502 	}
503 
504 	return NULL;
505 }
506 
507 const struct got_error *
508 got_commit_graph_iter_next(struct got_object_id **id,
509     struct got_commit_graph *graph, struct got_repository *repo,
510     got_cancel_cb cancel_cb, void *cancel_arg)
511 {
512 	const struct got_error *err = NULL;
513 	struct got_commit_graph_node *node;
514 
515 	*id = NULL;
516 
517 	if (TAILQ_EMPTY(&graph->iter_list)) {
518 		/* We are done iterating, or iteration was not started. */
519 		return got_error(GOT_ERR_ITER_COMPLETED);
520 	}
521 
522 	node = TAILQ_FIRST(&graph->iter_list);
523 	while (TAILQ_NEXT(node, entry) == NULL &&
524 	    got_object_idset_num_elements(graph->open_branches) > 0) {
525 		err = fetch_commits_from_open_branches(graph, repo,
526 		    cancel_cb, cancel_arg);
527 		if (err)
528 			return err;
529 	}
530 
531 	*id = &node->id;
532 	TAILQ_REMOVE(&graph->iter_list, node, entry);
533 	return NULL;
534 }
535 
536 const struct got_error *
537 got_commit_graph_find_youngest_common_ancestor(struct got_object_id **yca_id,
538     struct got_object_id *commit_id, struct got_object_id *commit_id2,
539     struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
540 {
541 	const struct got_error *err = NULL;
542 	struct got_commit_graph *graph = NULL, *graph2 = NULL;
543 	int completed = 0, completed2 = 0;
544 	struct got_object_idset *commit_ids;
545 
546 	*yca_id = NULL;
547 
548 	commit_ids = got_object_idset_alloc();
549 	if (commit_ids == NULL)
550 		return got_error_from_errno("got_object_idset_alloc");
551 
552 	err = got_commit_graph_open(&graph, "/", 1);
553 	if (err)
554 		goto done;
555 
556 	err = got_commit_graph_open(&graph2, "/", 1);
557 	if (err)
558 		goto done;
559 
560 	err = got_commit_graph_iter_start(graph, commit_id, repo,
561 	    cancel_cb, cancel_arg);
562 	if (err)
563 		goto done;
564 
565 	err = got_commit_graph_iter_start(graph2, commit_id2, repo,
566 	    cancel_cb, cancel_arg);
567 	if (err)
568 		goto done;
569 
570 	for (;;) {
571 		struct got_object_id *id = NULL, *id2 = NULL;
572 
573 		if (cancel_cb) {
574 			err = (*cancel_cb)(cancel_arg);
575 			if (err)
576 				break;
577 		}
578 
579 		if (!completed) {
580 			err = got_commit_graph_iter_next(&id, graph, repo,
581 			    cancel_cb, cancel_arg);
582 			if (err) {
583 				if (err->code != GOT_ERR_ITER_COMPLETED)
584 					break;
585 				err = NULL;
586 				completed = 1;
587 			}
588 		}
589 
590 		if (!completed2) {
591 			err = got_commit_graph_iter_next(&id2, graph2, repo,
592 			    cancel_cb, cancel_arg);
593 			if (err) {
594 				if (err->code != GOT_ERR_ITER_COMPLETED)
595 					break;
596 				err = NULL;
597 				completed2 = 1;
598 			}
599 		}
600 
601 		if (id) {
602 			if (got_object_idset_contains(commit_ids, id)) {
603 				*yca_id = got_object_id_dup(id);
604 				if (*yca_id)
605 					break;
606 				err = got_error_from_errno("got_object_id_dup");
607 				break;
608 
609 			}
610 			err = got_object_idset_add(commit_ids, id, NULL);
611 			if (err)
612 				break;
613 		}
614 		if (id2) {
615 			if (got_object_idset_contains(commit_ids, id2)) {
616 				*yca_id = got_object_id_dup(id2);
617 				if (*yca_id)
618 					break;
619 				err = got_error_from_errno("got_object_id_dup");
620 				break;
621 
622 			}
623 			err = got_object_idset_add(commit_ids, id2, NULL);
624 			if (err)
625 				break;
626 		}
627 
628 		if (completed && completed2) {
629 			err = got_error(GOT_ERR_ANCESTRY);
630 			break;
631 		}
632 
633 	}
634 done:
635 	got_object_idset_free(commit_ids);
636 	if (graph)
637 		got_commit_graph_close(graph);
638 	if (graph2)
639 		got_commit_graph_close(graph2);
640 	return err;
641 }