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