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, struct got_object_id *commit_id,
186     struct got_commit_object *commit, struct got_repository *repo)
187 {
188 	const struct got_error *err;
189 	struct got_object_qid *qid;
190 
191 	err = close_branch(graph, commit_id);
192 	if (err)
193 		return err;
194 
195 	if (graph->flags & GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL) {
196 		qid = SIMPLEQ_FIRST(&commit->parent_ids);
197 		if (qid == NULL ||
198 		    got_object_idset_contains(graph->open_branches, qid->id))
199 			return NULL;
200 		return got_object_idset_add(graph->open_branches,
201 		    qid->id, NULL);
202 	}
203 
204 	/*
205 	 * If we are graphing commits for a specific path, skip branches
206 	 * which do not contribute any content to this path.
207 	 */
208 	if (commit->nparents > 1 && !got_path_is_root_dir(graph->path)) {
209 		struct got_object_id *merged_id, *prev_id = NULL;
210 		int branches_differ = 0;
211 
212 		err = got_object_id_by_path(&merged_id, repo, commit_id,
213 		    graph->path);
214 		if (err)
215 			return err;
216 
217 		SIMPLEQ_FOREACH(qid, &commit->parent_ids, entry) {
218 			struct got_object_id *id;
219 
220 			if (got_object_idset_contains(graph->open_branches,
221 			    qid->id))
222 				continue;
223 
224 			err = got_object_id_by_path(&id, repo, qid->id,
225 			    graph->path);
226 			if (err) {
227 				if (err->code == GOT_ERR_NO_TREE_ENTRY) {
228 					branches_differ = 1;
229 					continue;
230 				}
231 				free(merged_id);
232 				free(prev_id);
233 				return err;
234 			}
235 
236 			if (prev_id) {
237 				if (!branches_differ &&
238 				    got_object_id_cmp(id, prev_id) != 0)
239 					branches_differ = 1;
240 				free(prev_id);
241 			}
242 			prev_id = id;
243 
244 			/*
245 			 * If a branch has created the merged content we can
246 			 * skip any other branches.
247 			 */
248 			if (got_object_id_cmp(merged_id, id) == 0) {
249 				err = got_object_idset_add(graph->open_branches,
250 				    qid->id, NULL);
251 				free(merged_id);
252 				free(id);
253 				return err;
254 			}
255 		}
256 
257 		free(prev_id);
258 		prev_id = NULL;
259 		free(merged_id);
260 		merged_id = NULL;
261 
262 		/*
263 		 * If the path's content is the same on all branches,
264 		 * follow the first parent only.
265 		 */
266 		if (!branches_differ) {
267 			qid = SIMPLEQ_FIRST(&commit->parent_ids);
268 			if (qid == NULL)
269 				return NULL;
270 			if (got_object_idset_contains(graph->open_branches,
271 			    qid->id))
272 				return NULL;
273 			if (got_object_idset_contains(graph->node_ids,
274 			    qid->id))
275 				return NULL; /* parent already traversed */
276 			return got_object_idset_add(graph->open_branches,
277 			    qid->id, NULL);
278 		}
279 	}
280 
281 	SIMPLEQ_FOREACH(qid, &commit->parent_ids, entry) {
282 		if (got_object_idset_contains(graph->open_branches, qid->id))
283 			continue;
284 		if (got_object_idset_contains(graph->node_ids, qid->id))
285 			continue; /* parent already traversed */
286 		err = got_object_idset_add(graph->open_branches, qid->id, NULL);
287 		if (err)
288 			return err;
289 	}
290 
291 	return NULL;
292 }
293 
294 static const struct got_error *
295 add_node(struct got_commit_graph_node **new_node,
296     struct got_commit_graph *graph,
297     struct got_object_id *commit_id,
298     struct got_commit_object *commit,
299     struct got_repository *repo)
300 {
301 	const struct got_error *err = NULL;
302 	struct got_commit_graph_node *node;
303 
304 	*new_node = NULL;
305 
306 	node = calloc(1, sizeof(*node));
307 	if (node == NULL)
308 		return got_error_from_errno("calloc");
309 
310 	memcpy(&node->id, commit_id, sizeof(node->id));
311 	node->timestamp = commit->committer_time;
312 
313 	err = got_object_idset_add(graph->node_ids, &node->id, NULL);
314 	if (err)
315 		free(node);
316 	else
317 		*new_node = node;
318 	return err;
319 }
320 
321 const struct got_error *
322 got_commit_graph_open(struct got_commit_graph **graph,
323     const char *path, int first_parent_traversal)
324 {
325 	const struct got_error *err = NULL;
326 
327 	*graph = calloc(1, sizeof(**graph));
328 	if (*graph == NULL)
329 		return got_error_from_errno("calloc");
330 
331 	TAILQ_INIT(&(*graph)->iter_list);
332 
333 	(*graph)->path = strdup(path);
334 	if ((*graph)->path == NULL) {
335 		err = got_error_from_errno("strdup");
336 		goto done;
337 	}
338 
339 	(*graph)->node_ids = got_object_idset_alloc();
340 	if ((*graph)->node_ids == NULL) {
341 		err = got_error_from_errno("got_object_idset_alloc");
342 		goto done;
343 	}
344 
345 	(*graph)->open_branches = got_object_idset_alloc();
346 	if ((*graph)->open_branches == NULL) {
347 		err = got_error_from_errno("got_object_idset_alloc");
348 		goto done;
349 	}
350 
351 	if (first_parent_traversal)
352 		(*graph)->flags |= GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL;
353 done:
354 	if (err) {
355 		got_commit_graph_close(*graph);
356 		*graph = NULL;
357 	}
358 	return err;
359 }
360 
361 struct add_branch_tip_arg {
362 	struct got_commit_graph_branch_tip *tips;
363 	int ntips;
364 	struct got_repository *repo;
365 	struct got_commit_graph *graph;
366 };
367 
368 static const struct got_error *
369 add_branch_tip(struct got_object_id *commit_id, void *data, void *arg)
370 {
371 	const struct got_error *err;
372 	struct add_branch_tip_arg *a = arg;
373 	struct got_commit_graph_node *new_node;
374 	struct got_commit_object *commit;
375 
376 	err = got_object_open_as_commit(&commit, a->repo, commit_id);
377 	if (err)
378 		return err;
379 
380 	err = add_node(&new_node, a->graph, commit_id, commit, a->repo);
381 	if (err)
382 		return err;
383 
384 	a->tips[a->ntips].commit_id = new_node ? &new_node->id : NULL;
385 	a->tips[a->ntips].commit = commit;
386 	a->tips[a->ntips].new_node = new_node;
387 	a->ntips++;
388 
389 	return NULL;
390 }
391 
392 static const struct got_error *
393 fetch_commits_from_open_branches(struct got_commit_graph *graph,
394     struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
395 {
396 	const struct got_error *err;
397 	struct add_branch_tip_arg arg;
398 	int i, ntips;
399 
400 	ntips = got_object_idset_num_elements(graph->open_branches);
401 	if (ntips == 0)
402 		return NULL;
403 
404 	/* (Re-)allocate branch tips array if necessary. */
405 	if (graph->ntips < ntips) {
406 		struct got_commit_graph_branch_tip *tips;
407 		tips = recallocarray(graph->tips, graph->ntips, ntips,
408 		    sizeof(*tips));
409 		if (tips == NULL)
410 			return got_error_from_errno("recallocarray");
411 		graph->tips = tips;
412 		graph->ntips = ntips;
413 	}
414 	arg.tips = graph->tips;
415 	arg.ntips = 0; /* add_branch_tip() will increment */
416 	arg.repo = repo;
417 	arg.graph = graph;
418 	err = got_object_idset_for_each(graph->open_branches, add_branch_tip,
419 	    &arg);
420 	if (err)
421 		goto done;
422 
423 	for (i = 0; i < arg.ntips; i++) {
424 		struct got_object_id *commit_id;
425 		struct got_commit_object *commit;
426 		struct got_commit_graph_node *new_node;
427 		int changed;
428 
429 		if (cancel_cb) {
430 			err = (*cancel_cb)(cancel_arg);
431 			if (err)
432 				break;
433 		}
434 
435 		commit_id = arg.tips[i].commit_id;
436 		commit = arg.tips[i].commit;
437 		new_node = arg.tips[i].new_node;
438 
439 		err = detect_changed_path(&changed, commit, commit_id,
440 		    graph->path, repo);
441 		if (err) {
442 			if (err->code != GOT_ERR_NO_OBJ)
443 				break;
444 			/*
445 			 * History of the path stops here on the current
446 			 * branch. Keep going on other branches.
447 			 */
448 			err = close_branch(graph, commit_id);
449 			if (err)
450 				break;
451 			continue;
452 		}
453 		if (changed)
454 			add_node_to_iter_list(graph, new_node);
455 		err = advance_branch(graph, commit_id, 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 void
466 got_commit_graph_close(struct got_commit_graph *graph)
467 {
468 	if (graph->open_branches)
469 		got_object_idset_free(graph->open_branches);
470 	if (graph->node_ids)
471 		got_object_idset_free(graph->node_ids);
472 	free(graph->tips);
473 	free(graph->path);
474 	free(graph);
475 }
476 
477 const struct got_error *
478 got_commit_graph_iter_start(struct got_commit_graph *graph,
479     struct got_object_id *id, struct got_repository *repo,
480     got_cancel_cb cancel_cb, void *cancel_arg)
481 {
482 	const struct got_error *err = NULL;
483 
484 	if (!TAILQ_EMPTY(&graph->iter_list))
485 		return got_error(GOT_ERR_ITER_BUSY);
486 
487 	err = got_object_idset_add(graph->open_branches, id, NULL);
488 	if (err)
489 		return err;
490 
491 	/* Locate first commit which changed graph->path. */
492 	while (graph->iter_node == NULL &&
493 	    got_object_idset_num_elements(graph->open_branches) > 0) {
494 		err = fetch_commits_from_open_branches(graph, repo,
495 		    cancel_cb, cancel_arg);
496 		if (err)
497 			return err;
498 	}
499 
500 	return NULL;
501 }
502 
503 const struct got_error *
504 got_commit_graph_iter_next(struct got_object_id **id,
505     struct got_commit_graph *graph, struct got_repository *repo,
506     got_cancel_cb cancel_cb, void *cancel_arg)
507 {
508 	const struct got_error *err = NULL;
509 
510 	*id = NULL;
511 
512 	if (graph->iter_node == NULL) {
513 		/* We are done iterating, or iteration was not started. */
514 		return got_error(GOT_ERR_ITER_COMPLETED);
515 	}
516 
517 	if (graph->iter_node ==
518 	    TAILQ_LAST(&graph->iter_list, got_commit_graph_iter_list) &&
519 	    got_object_idset_num_elements(graph->open_branches) == 0) {
520 		/* We are done iterating. */
521 		*id = &graph->iter_node->id;
522 		graph->iter_node = NULL;
523 		return NULL;
524 	}
525 
526 	while (TAILQ_NEXT(graph->iter_node, entry) == NULL &&
527 	    got_object_idset_num_elements(graph->open_branches) > 0) {
528 		err = fetch_commits_from_open_branches(graph, repo,
529 		    cancel_cb, cancel_arg);
530 		if (err)
531 			return err;
532 	}
533 
534 	*id = &graph->iter_node->id;
535 	graph->iter_node = TAILQ_NEXT(graph->iter_node, entry);
536 	return NULL;
537 }
538 
539 const struct got_error *
540 got_commit_graph_find_youngest_common_ancestor(struct got_object_id **yca_id,
541     struct got_object_id *commit_id, struct got_object_id *commit_id2,
542     struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
543 {
544 	const struct got_error *err = NULL;
545 	struct got_commit_graph *graph = NULL, *graph2 = NULL;
546 	int completed = 0, completed2 = 0;
547 	struct got_object_idset *commit_ids;
548 
549 	*yca_id = NULL;
550 
551 	commit_ids = got_object_idset_alloc();
552 	if (commit_ids == NULL)
553 		return got_error_from_errno("got_object_idset_alloc");
554 
555 	err = got_commit_graph_open(&graph, "/", 1);
556 	if (err)
557 		goto done;
558 
559 	err = got_commit_graph_open(&graph2, "/", 1);
560 	if (err)
561 		goto done;
562 
563 	err = got_commit_graph_iter_start(graph, commit_id, repo,
564 	    cancel_cb, cancel_arg);
565 	if (err)
566 		goto done;
567 
568 	err = got_commit_graph_iter_start(graph2, commit_id2, repo,
569 	    cancel_cb, cancel_arg);
570 	if (err)
571 		goto done;
572 
573 	for (;;) {
574 		struct got_object_id *id = NULL, *id2 = NULL;
575 
576 		if (cancel_cb) {
577 			err = (*cancel_cb)(cancel_arg);
578 			if (err)
579 				break;
580 		}
581 
582 		if (!completed) {
583 			err = got_commit_graph_iter_next(&id, graph, repo,
584 			    cancel_cb, cancel_arg);
585 			if (err) {
586 				if (err->code != GOT_ERR_ITER_COMPLETED)
587 					break;
588 				err = NULL;
589 				completed = 1;
590 			}
591 		}
592 
593 		if (!completed2) {
594 			err = got_commit_graph_iter_next(&id2, graph2, repo,
595 			    cancel_cb, cancel_arg);
596 			if (err) {
597 				if (err->code != GOT_ERR_ITER_COMPLETED)
598 					break;
599 				err = NULL;
600 				completed2 = 1;
601 			}
602 		}
603 
604 		if (id) {
605 			if (got_object_idset_contains(commit_ids, id)) {
606 				*yca_id = got_object_id_dup(id);
607 				if (*yca_id)
608 					break;
609 				err = got_error_from_errno("got_object_id_dup");
610 				break;
611 
612 			}
613 			err = got_object_idset_add(commit_ids, id, NULL);
614 			if (err)
615 				break;
616 		}
617 		if (id2) {
618 			if (got_object_idset_contains(commit_ids, id2)) {
619 				*yca_id = got_object_id_dup(id2);
620 				if (*yca_id)
621 					break;
622 				err = got_error_from_errno("got_object_id_dup");
623 				break;
624 
625 			}
626 			err = got_object_idset_add(commit_ids, id2, NULL);
627 			if (err)
628 				break;
629 		}
630 
631 		if (completed && completed2) {
632 			err = got_error(GOT_ERR_ANCESTRY);
633 			break;
634 		}
635 
636 	}
637 done:
638 	got_object_idset_free(commit_ids);
639 	if (graph)
640 		got_commit_graph_close(graph);
641 	if (graph2)
642 		got_commit_graph_close(graph2);
643 	return err;
644 }