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