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 <stdio.h>
23 #include <stdlib.h>
24 #include <string.h>
25 #include <sha1.h>
26 #include <zlib.h>
27 #include <ctype.h>
28 
29 #include "got_error.h"
30 #include "got_object.h"
31 #include "got_commit_graph.h"
32 
33 #include "got_lib_delta.h"
34 #include "got_lib_inflate.h"
35 #include "got_lib_object.h"
36 #include "got_lib_object_idset.h"
37 #include "got_lib_path.h"
38 
39 struct got_commit_graph_node {
40 	struct got_object_id id;
41 	time_t timestamp;
42 
43 	/* Used during graph iteration. */
44 	TAILQ_ENTRY(got_commit_graph_node) entry;
45 };
46 
47 TAILQ_HEAD(got_commit_graph_iter_list, got_commit_graph_node);
48 
49 struct got_commit_graph_branch_tip {
50 	struct got_object_id *commit_id;
51 	struct got_commit_object *commit;
52 	struct got_commit_graph_node *new_node;
53 	int changed;
54 	int branch_done;
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 	/* The commit at which traversal began (youngest commit in node_ids). */
62 	struct got_commit_graph_node *head_node;
63 
64 	int flags;
65 #define GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL		0x01
66 
67 	/*
68 	 * A set of object IDs of known parent commits which we have not yet
69 	 * traversed. Each commit ID in this set represents a branch in commit
70 	 * history: Either the first-parent branch of the head node, or another
71 	 * branch corresponding to a traversed merge commit for which we have
72 	 * not traversed a branch point commit yet.
73 	 *
74 	 * Whenever we add a commit with a matching ID to the graph, we remove
75 	 * its corresponding element from this set, and add new elements for
76 	 * each of that commit's parent commits which were not traversed yet.
77 	 *
78 	 * When API users ask us to fetch more commits, we fetch commits from
79 	 * all currently open branches. This allows API users to process
80 	 * commits in linear order even though the history contains branches.
81 	 */
82 	struct got_object_idset *open_branches;
83 
84 	/* Array of branch tips for fetch_commits_from_open_branches(). */
85 	struct got_commit_graph_branch_tip *tips;
86 	int ntips;
87 
88 	/* Path of tree entry of interest to the API user. */
89 	char *path;
90 
91 	/* The next commit to return when the API user asks for one. */
92 	struct got_commit_graph_node *iter_node;
93 
94 	/* The graph iteration list contains all nodes in sorted order. */
95 	struct got_commit_graph_iter_list iter_list;
96 };
97 
98 static struct got_commit_graph *
99 alloc_graph(const char *path)
100 {
101 	struct got_commit_graph *graph;
102 
103 	graph = calloc(1, sizeof(*graph));
104 	if (graph == NULL)
105 		return NULL;
106 
107 	graph->path = strdup(path);
108 	if (graph->path == NULL) {
109 		free(graph);
110 		return NULL;
111 	}
112 
113 	graph->node_ids = got_object_idset_alloc();
114 	if (graph->node_ids == NULL) {
115 		free(graph->path);
116 		free(graph);
117 		return NULL;
118 	}
119 
120 	graph->open_branches = got_object_idset_alloc();
121 	if (graph->open_branches == NULL) {
122 		got_object_idset_free(graph->node_ids);
123 		free(graph->path);
124 		free(graph);
125 		return NULL;
126 	}
127 
128 	TAILQ_INIT(&graph->iter_list);
129 	return graph;
130 }
131 
132 static const struct got_error *
133 detect_changed_path(int *changed, struct got_commit_object *commit,
134     struct got_object_id *commit_id, const char *path,
135     struct got_repository *repo)
136 {
137 	const struct got_error *err = NULL;
138 	struct got_commit_object *pcommit = NULL;
139 	struct got_tree_object *tree = NULL, *ptree = NULL;
140 	struct got_object_qid *pid;
141 
142 	if (got_path_is_root_dir(path)) {
143 		*changed = 1;
144 		return NULL;
145 	}
146 
147 	*changed = 0;
148 
149 	pid = SIMPLEQ_FIRST(&commit->parent_ids);
150 	if (pid == NULL) {
151 		struct got_object_id *obj_id;
152 		err = got_object_id_by_path(&obj_id, repo, commit_id, path);
153 		if (err) {
154 			if (err->code == GOT_ERR_NO_TREE_ENTRY)
155 				err = NULL;
156 		} else
157 			*changed = 1; /* The path was created in this commit. */
158 		free(obj_id);
159 		return err;
160 	}
161 
162 	err = got_object_open_as_tree(&tree, repo, commit->tree_id);
163 	if (err)
164 		return err;
165 
166 	err = got_object_open_as_commit(&pcommit, repo, pid->id);
167 	if (err)
168 		goto done;
169 
170 	err = got_object_open_as_tree(&ptree, repo, pcommit->tree_id);
171 	if (err)
172 		goto done;
173 
174 	err = got_object_tree_path_changed(changed, tree, ptree, path,
175 	    repo);
176 done:
177 	if (tree)
178 		got_object_tree_close(tree);
179 	if (ptree)
180 		got_object_tree_close(ptree);
181 	if (pcommit)
182 		got_object_commit_close(pcommit);
183 	return err;
184 }
185 
186 static void
187 add_node_to_iter_list(struct got_commit_graph *graph,
188     struct got_commit_graph_node *node,
189     struct got_commit_graph_node *child_node)
190 {
191 	struct got_commit_graph_node *n, *next;
192 
193 	if (TAILQ_EMPTY(&graph->iter_list)) {
194 		TAILQ_INSERT_HEAD(&graph->iter_list, node, entry);
195 		graph->iter_node = node;
196 		return;
197 	}
198 
199 	n = graph->iter_node;
200 	/* Ensure that an iteration in progress will see this new commit. */
201 	while (n) {
202 		next = TAILQ_NEXT(n, entry);
203 		if (next && node->timestamp >= next->timestamp) {
204 			TAILQ_INSERT_BEFORE(next, node, entry);
205 			return;
206 		}
207 		n = next;
208 	}
209 	TAILQ_INSERT_TAIL(&graph->iter_list, node, entry);
210 }
211 
212 static const struct got_error *
213 close_branch(struct got_commit_graph *graph, struct got_object_id *commit_id)
214 {
215 	const struct got_error *err;
216 
217 	err = got_object_idset_remove(NULL, graph->open_branches, commit_id);
218 	if (err && err->code != GOT_ERR_NO_OBJ)
219 		return err;
220 	return NULL;
221 }
222 
223 static const struct got_error *
224 advance_branch(struct got_commit_graph *graph,
225     struct got_commit_graph_node *node,
226     struct got_object_id *commit_id, struct got_commit_object *commit,
227     struct got_repository *repo)
228 {
229 	const struct got_error *err;
230 	struct got_object_qid *qid;
231 
232 	err = close_branch(graph, commit_id);
233 	if (err)
234 		return err;
235 
236 	if (graph->flags & GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL) {
237 		qid = SIMPLEQ_FIRST(&commit->parent_ids);
238 		if (qid == NULL ||
239 		    got_object_idset_get(graph->open_branches, qid->id))
240 			return NULL;
241 		return got_object_idset_add(graph->open_branches,
242 		    qid->id, node);
243 	}
244 
245 	/*
246 	 * If we are graphing commits for a specific path, skip branches
247 	 * which do not contribute any content to this path.
248 	 */
249 	if (commit->nparents > 1 && !got_path_is_root_dir(graph->path)) {
250 		struct got_object_id *merged_id, *prev_id = NULL;
251 		int branches_differ = 0;
252 
253 		err = got_object_id_by_path(&merged_id, repo, commit_id,
254 		    graph->path);
255 		if (err)
256 			return err;
257 
258 		SIMPLEQ_FOREACH(qid, &commit->parent_ids, entry) {
259 			struct got_object_id *id;
260 
261 			if (got_object_idset_get(graph->node_ids, qid->id))
262 				continue; /* parent already traversed */
263 			if (got_object_idset_get(graph->open_branches, qid->id))
264 				continue;
265 
266 			err = got_object_id_by_path(&id, repo, qid->id,
267 			    graph->path);
268 			if (err) {
269 				if (err->code == GOT_ERR_NO_TREE_ENTRY) {
270 					branches_differ = 1;
271 					continue;
272 				}
273 				free(merged_id);
274 				free(prev_id);
275 				return err;
276 			}
277 
278 			if (prev_id) {
279 				if (!branches_differ &&
280 				    got_object_id_cmp(id, prev_id) != 0)
281 					branches_differ = 1;
282 				free(prev_id);
283 			}
284 			prev_id = id;
285 
286 			/*
287 			 * If a branch has created the merged content we can
288 			 * skip any other branches.
289 			 */
290 			if (got_object_id_cmp(merged_id, id) == 0) {
291 				err = got_object_idset_add(graph->open_branches,
292 				    qid->id, node);
293 				free(merged_id);
294 				free(id);
295 				return err;
296 			}
297 		}
298 
299 		free(prev_id);
300 		prev_id = NULL;
301 		free(merged_id);
302 		merged_id = NULL;
303 
304 		/*
305 		 * If the path's content is the same on all branches,
306 		 * follow the first parent only.
307 		 */
308 		if (!branches_differ) {
309 			qid = SIMPLEQ_FIRST(&commit->parent_ids);
310 			if (qid == NULL)
311 				return NULL;
312 			if (got_object_idset_get(graph->node_ids, qid->id))
313 				return NULL; /* parent already traversed */
314 			if (got_object_idset_get(graph->open_branches, qid->id))
315 				return NULL;
316 			return got_object_idset_add(graph->open_branches,
317 			    qid->id, node);
318 		}
319 	}
320 
321 	SIMPLEQ_FOREACH(qid, &commit->parent_ids, entry) {
322 		if (got_object_idset_get(graph->node_ids, qid->id))
323 			continue; /* parent already traversed */
324 		if (got_object_idset_get(graph->open_branches, qid->id))
325 			continue;
326 		err = got_object_idset_add(graph->open_branches, qid->id, node);
327 		if (err)
328 			return err;
329 	}
330 
331 	return NULL;
332 }
333 
334 static const struct got_error *
335 add_node(struct got_commit_graph_node **new_node, int *changed,
336     int *branch_done, struct got_commit_graph *graph,
337     struct got_object_id *commit_id, struct got_commit_object *commit,
338     struct got_commit_graph_node *child_node, struct got_repository *repo)
339 {
340 	const struct got_error *err = NULL;
341 	struct got_commit_graph_node *node;
342 
343 	*new_node = NULL;
344 	*changed = 0;
345 	*branch_done = 0;
346 
347 	node = calloc(1, sizeof(*node));
348 	if (node == NULL)
349 		return got_error_from_errno();
350 
351 	memcpy(&node->id, commit_id, sizeof(node->id));
352 	node->timestamp = commit->committer_time;
353 
354 	err = got_object_idset_add(graph->node_ids, &node->id, node);
355 	if (err) {
356 		free(node);
357 		return err;
358 	}
359 
360 	err = detect_changed_path(changed, commit, commit_id, graph->path,
361 	    repo);
362 	if (err) {
363 		if (err->code == GOT_ERR_NO_OBJ) {
364 			/*
365 			 * History of the path stops here on the current
366 			 * branch. Keep going on other branches.
367 			 */
368 			err = NULL;
369 			*branch_done = 1;
370 		} else {
371 			got_object_idset_remove(NULL, graph->node_ids,
372 			    &node->id);
373 			free(node);
374 			return err;
375 		}
376 	}
377 
378 	if (*changed)
379 		add_node_to_iter_list(graph, node, child_node);
380 	*new_node = node;
381 	return NULL;
382 }
383 
384 const struct got_error *
385 got_commit_graph_open(struct got_commit_graph **graph,
386     struct got_object_id *commit_id, const char *path,
387     int first_parent_traversal, struct got_repository *repo)
388 {
389 	const struct got_error *err = NULL;
390 	struct got_commit_object *commit;
391 	int changed, branch_done;
392 
393 	*graph = NULL;
394 
395 	err = got_object_open_as_commit(&commit, repo, commit_id);
396 	if (err)
397 		return err;
398 
399 	/* The path must exist in our initial commit. */
400 	if (!got_path_is_root_dir(path)) {
401 		struct got_object_id *obj_id;
402 		err = got_object_id_by_path(&obj_id, repo, commit_id, path);
403 		if (err)
404 			return err;
405 		free(obj_id);
406 	}
407 
408 	*graph = alloc_graph(path);
409 	if (*graph == NULL) {
410 		got_object_commit_close(commit);
411 		return got_error_from_errno();
412 	}
413 
414 	if (first_parent_traversal)
415 		(*graph)->flags |= GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL;
416 
417 	err = add_node(&(*graph)->head_node, &changed, &branch_done, *graph,
418 	    commit_id, commit, NULL, repo);
419 	if (err == NULL) {
420 		err = advance_branch(*graph, (*graph)->head_node, commit_id,
421 		    commit, repo);
422 	}
423 	got_object_commit_close(commit);
424 	if (err) {
425 		got_commit_graph_close(*graph);
426 		*graph = NULL;
427 		return err;
428 	}
429 	
430 	return NULL;
431 }
432 
433 struct add_branch_tip_arg {
434 	struct got_commit_graph_branch_tip *tips;
435 	int ntips;
436 	struct got_repository *repo;
437 	struct got_commit_graph *graph;
438 };
439 
440 static const struct got_error *
441 add_branch_tip(struct got_object_id *commit_id, void *data, void *arg)
442 {
443 	const struct got_error *err;
444 	struct got_commit_graph_node *child_node = data;
445 	struct add_branch_tip_arg *a = arg;
446 	struct got_commit_graph_node *new_node;
447 	struct got_commit_object *commit;
448 	int changed, branch_done;
449 
450 	err = got_object_open_as_commit(&commit, a->repo, commit_id);
451 	if (err)
452 		return err;
453 
454 	err = add_node(&new_node, &changed, &branch_done, a->graph,
455 	    commit_id, commit, child_node, a->repo);
456 	if (err)
457 		return err;
458 
459 	a->tips[a->ntips].commit_id = new_node ? &new_node->id : NULL;
460 	a->tips[a->ntips].commit = commit;
461 	a->tips[a->ntips].new_node = new_node;
462 	a->tips[a->ntips].changed = changed;
463 	a->tips[a->ntips].branch_done = branch_done;
464 	a->ntips++;
465 
466 	return NULL;
467 }
468 
469 static const struct got_error *
470 fetch_commits_from_open_branches(int *nfetched,
471     struct got_object_id **changed_id, struct got_commit_graph *graph,
472     struct got_repository *repo)
473 {
474 	const struct got_error *err;
475 	struct add_branch_tip_arg arg;
476 	int i, ntips;
477 
478 	*nfetched = 0;
479 	*changed_id = NULL;
480 
481 	ntips = got_object_idset_num_elements(graph->open_branches);
482 	if (ntips == 0)
483 		return NULL;
484 
485 	/* (Re-)allocate branch tips array if necessary. */
486 	if (graph->ntips < ntips) {
487 		struct got_commit_graph_branch_tip *tips;
488 		tips = recallocarray(graph->tips, graph->ntips, ntips,
489 		    sizeof(*tips));
490 		if (tips == NULL)
491 			return got_error_from_errno();
492 		graph->tips = tips;
493 		graph->ntips = ntips;
494 	}
495 	arg.tips = graph->tips;
496 	arg.ntips = 0; /* add_branch_tip() will increment */
497 	arg.repo = repo;
498 	arg.graph = graph;
499 	err = got_object_idset_for_each(graph->open_branches, add_branch_tip,
500 	    &arg);
501 	if (err)
502 		goto done;
503 
504 	for (i = 0; i < arg.ntips; i++) {
505 		struct got_object_id *commit_id;
506 		struct got_commit_object *commit;
507 		struct got_commit_graph_node *new_node;
508 		int branch_done, changed;
509 
510 		commit_id = arg.tips[i].commit_id;
511 		commit = arg.tips[i].commit;
512 		new_node = arg.tips[i].new_node;
513 		branch_done = arg.tips[i].branch_done;
514 		changed = arg.tips[i].changed;
515 
516 		if (branch_done)
517 			err = close_branch(graph, commit_id);
518 		else
519 			err = advance_branch(graph, new_node, commit_id,
520 			    commit, repo);
521 		if (err)
522 			break;
523 		if (changed && *changed_id == NULL)
524 			*changed_id = commit_id;
525 	}
526 done:
527 	for (i = 0; i < arg.ntips; i++)
528 		got_object_commit_close(arg.tips[i].commit);
529 	(*nfetched) = arg.ntips;
530 	return err;
531 }
532 
533 const struct got_error *
534 got_commit_graph_fetch_commits(struct got_commit_graph *graph, int limit,
535     struct got_repository *repo)
536 {
537 	const struct got_error *err;
538 	int nfetched = 0, ncommits;
539 	struct got_object_id *changed_id = NULL;
540 
541 	while (nfetched < limit) {
542 		err = fetch_commits_from_open_branches(&ncommits,
543 		    &changed_id, graph, repo);
544 		if (err)
545 			return err;
546 		if (ncommits == 0)
547 			break;
548 		if (changed_id)
549 			nfetched += ncommits;
550 	}
551 
552 	return NULL;
553 }
554 
555 static const struct got_error *
556 free_node_iter(struct got_object_id *id, void *data, void *arg)
557 {
558 	struct got_commit_graph_node *node = data;
559 	free(node);
560 	return NULL;
561 }
562 
563 void
564 got_commit_graph_close(struct got_commit_graph *graph)
565 {
566 	got_object_idset_free(graph->open_branches);
567 	got_object_idset_for_each(graph->node_ids, free_node_iter, NULL);
568 	got_object_idset_free(graph->node_ids);
569 	free(graph->tips);
570 	free(graph->path);
571 	free(graph);
572 }
573 
574 const struct got_error *
575 got_commit_graph_iter_start(struct got_commit_graph *graph,
576     struct got_object_id *id, struct got_repository *repo)
577 {
578 	const struct got_error *err = NULL;
579 	struct got_commit_graph_node *start_node;
580 	struct got_commit_object *commit;
581 	int changed;
582 
583 	start_node = got_object_idset_get(graph->node_ids, id);
584 	if (start_node == NULL)
585 		return got_error_no_obj(id);
586 
587 	err = got_object_open_as_commit(&commit, repo, &start_node->id);
588 	if (err)
589 		return err;
590 
591 	err = detect_changed_path(&changed, commit, &start_node->id,
592 	    graph->path, repo);
593 	if (err) {
594 		got_object_commit_close(commit);
595 		return err;
596 	}
597 
598 	if (!changed) {
599 		/* Locate first commit which changed graph->path. */
600 		struct got_object_id *changed_id = NULL;
601 		while (changed_id == NULL) {
602 			int ncommits;
603 			err = fetch_commits_from_open_branches(&ncommits,
604 			    &changed_id, graph, repo);
605 			if (err) {
606 				got_object_commit_close(commit);
607 				return err;
608 			}
609 		}
610 		start_node = got_object_idset_get(graph->node_ids, changed_id);
611 	}
612 	got_object_commit_close(commit);
613 
614 	graph->iter_node = start_node;
615 	return NULL;
616 }
617 
618 const struct got_error *
619 got_commit_graph_iter_next(struct got_object_id **id,
620     struct got_commit_graph *graph)
621 {
622 	*id = NULL;
623 
624 	if (graph->iter_node == NULL) {
625 		/* We are done iterating, or iteration was not started. */
626 		return got_error(GOT_ERR_ITER_COMPLETED);
627 	}
628 
629 	if (graph->iter_node ==
630 	    TAILQ_LAST(&graph->iter_list, got_commit_graph_iter_list) &&
631 	    got_object_idset_num_elements(graph->open_branches) == 0) {
632 		/* We are done iterating. */
633 		*id = &graph->iter_node->id;
634 		graph->iter_node = NULL;
635 		return NULL;
636 	}
637 
638 	if (TAILQ_NEXT(graph->iter_node, entry) == NULL)
639 		return got_error(GOT_ERR_ITER_NEED_MORE);
640 
641 	*id = &graph->iter_node->id;
642 	graph->iter_node = TAILQ_NEXT(graph->iter_node, entry);
643 	return NULL;
644 }