tog man page: tweak documentation of vertical/horizontal split for clarity
[got-portable.git] / lib / commit_graph.c
blob1286c9b89a7ba9696c5d6c9a9de0a2f40bf29cd7
1 /*
2 * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
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.
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.
17 #include <sys/types.h>
18 #include <sys/stat.h>
20 #include <limits.h>
21 #include <stdio.h>
22 #include <stdlib.h>
23 #include <string.h>
24 #include <zlib.h>
25 #include <ctype.h>
27 #include "got_compat.h"
29 #include "got_error.h"
30 #include "got_object.h"
31 #include "got_cancel.h"
32 #include "got_commit_graph.h"
33 #include "got_path.h"
35 #include "got_lib_delta.h"
36 #include "got_lib_inflate.h"
37 #include "got_lib_object.h"
38 #include "got_lib_object_idset.h"
40 struct got_commit_graph_node {
41 struct got_object_id id;
43 /* Used only during iteration. */
44 time_t timestamp;
45 TAILQ_ENTRY(got_commit_graph_node) entry;
48 TAILQ_HEAD(got_commit_graph_iter_list, got_commit_graph_node);
50 struct got_commit_graph_branch_tip {
51 struct got_object_id *commit_id;
52 struct got_commit_object *commit;
53 struct got_commit_graph_node *new_node;
56 struct got_commit_graph {
57 /* The set of all commits we have traversed. */
58 struct got_object_idset *node_ids;
60 int flags;
61 #define GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL 0x01
64 * A set of object IDs of known parent commits which we have not yet
65 * traversed. Each commit ID in this set represents a branch in commit
66 * history: Either the first-parent branch of the head node, or another
67 * branch corresponding to a traversed merge commit for which we have
68 * not traversed a branch point commit yet.
70 * Whenever we add a commit with a matching ID to the graph, we remove
71 * its corresponding element from this set, and add new elements for
72 * each of that commit's parent commits which were not traversed yet.
74 * When API users ask us to fetch more commits, we fetch commits from
75 * all currently open branches. This allows API users to process
76 * commits in linear order even though the history contains branches.
78 struct got_object_idset *open_branches;
80 /* Array of branch tips for fetch_commits_from_open_branches(). */
81 struct got_commit_graph_branch_tip *tips;
82 int ntips;
84 /* Path of tree entry of interest to the API user. */
85 char *path;
88 * Nodes which will be passed to the API user next, sorted by
89 * commit timestmap.
91 struct got_commit_graph_iter_list iter_list;
94 static const struct got_error *
95 detect_changed_path(int *changed, struct got_commit_object *commit,
96 struct got_object_id *commit_id, const char *path,
97 struct got_repository *repo)
99 const struct got_error *err = NULL;
100 struct got_commit_object *pcommit = NULL;
101 struct got_tree_object *tree = NULL, *ptree = NULL;
102 struct got_object_qid *pid;
104 if (got_path_is_root_dir(path)) {
105 *changed = 1;
106 return NULL;
109 *changed = 0;
111 pid = STAILQ_FIRST(&commit->parent_ids);
112 if (pid == NULL) {
113 struct got_object_id *obj_id;
114 err = got_object_id_by_path(&obj_id, repo, commit, path);
115 if (err) {
116 if (err->code == GOT_ERR_NO_TREE_ENTRY)
117 err = NULL;
118 } else
119 *changed = 1; /* The path was created in this commit. */
120 free(obj_id);
121 return err;
124 err = got_object_open_as_tree(&tree, repo, commit->tree_id);
125 if (err)
126 return err;
128 err = got_object_open_as_commit(&pcommit, repo, &pid->id);
129 if (err)
130 goto done;
132 err = got_object_open_as_tree(&ptree, repo, pcommit->tree_id);
133 if (err)
134 goto done;
136 err = got_object_tree_path_changed(changed, tree, ptree, path, repo);
137 done:
138 if (tree)
139 got_object_tree_close(tree);
140 if (ptree)
141 got_object_tree_close(ptree);
142 if (pcommit)
143 got_object_commit_close(pcommit);
144 return err;
147 static void
148 add_node_to_iter_list(struct got_commit_graph *graph,
149 struct got_commit_graph_node *node, time_t committer_time)
151 struct got_commit_graph_node *n, *next;
153 node->timestamp = committer_time;
155 n = TAILQ_FIRST(&graph->iter_list);
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;
162 n = next;
164 TAILQ_INSERT_TAIL(&graph->iter_list, node, entry);
167 static const struct got_error *
168 add_node(struct got_commit_graph_node **new_node,
169 struct got_commit_graph *graph, struct got_object_id *commit_id,
170 struct got_repository *repo)
172 const struct got_error *err = NULL;
173 struct got_commit_graph_node *node;
175 *new_node = NULL;
177 node = calloc(1, sizeof(*node));
178 if (node == NULL)
179 return got_error_from_errno("calloc");
181 memcpy(&node->id, commit_id, sizeof(node->id));
182 err = got_object_idset_add(graph->node_ids, &node->id, NULL);
183 if (err)
184 free(node);
185 else
186 *new_node = node;
187 return err;
191 * Ask got-read-pack to traverse first-parent history until a commit is
192 * encountered which modified graph->path, or until the pack file runs
193 * out of relevant commits. This is faster than sending an individual
194 * request for each commit stored in the pack file.
196 static const struct got_error *
197 packed_first_parent_traversal(int *ncommits_traversed,
198 struct got_commit_graph *graph, struct got_object_id *commit_id,
199 struct got_repository *repo)
201 const struct got_error *err = NULL;
202 struct got_object_id_queue traversed_commits;
203 struct got_object_qid *qid;
205 STAILQ_INIT(&traversed_commits);
206 *ncommits_traversed = 0;
208 err = got_traverse_packed_commits(&traversed_commits,
209 commit_id, graph->path, repo);
210 if (err)
211 return err;
213 /* Add all traversed commits to the graph... */
214 STAILQ_FOREACH(qid, &traversed_commits, entry) {
215 struct got_commit_graph_node *node;
217 if (got_object_idset_contains(graph->open_branches, &qid->id))
218 continue;
219 if (got_object_idset_contains(graph->node_ids, &qid->id))
220 continue;
222 (*ncommits_traversed)++;
224 /* ... except the last commit is the new branch tip. */
225 if (STAILQ_NEXT(qid, entry) == NULL) {
226 err = got_object_idset_add(graph->open_branches,
227 &qid->id, NULL);
228 break;
231 err = add_node(&node, graph, &qid->id, repo);
232 if (err)
233 break;
236 got_object_id_queue_free(&traversed_commits);
237 return err;
240 static const struct got_error *
241 close_branch(struct got_commit_graph *graph, struct got_object_id *commit_id)
243 const struct got_error *err;
245 err = got_object_idset_remove(NULL, graph->open_branches, commit_id);
246 if (err && err->code != GOT_ERR_NO_OBJ)
247 return err;
248 return NULL;
251 static const struct got_error *
252 advance_branch(struct got_commit_graph *graph, struct got_object_id *commit_id,
253 struct got_commit_object *commit, struct got_repository *repo)
255 const struct got_error *err;
256 struct got_object_qid *qid;
258 err = close_branch(graph, commit_id);
259 if (err)
260 return err;
262 if (graph->flags & GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL) {
263 qid = STAILQ_FIRST(&commit->parent_ids);
264 if (qid == NULL ||
265 got_object_idset_contains(graph->open_branches, &qid->id))
266 return NULL;
268 * The root directory always changes by definition, and when
269 * logging the root we want to traverse consecutive commits
270 * even if they point at the same tree.
271 * But if we are looking for a specific path then we can avoid
272 * fetching packed commits which did not modify the path and
273 * only fetch their IDs. This speeds up 'got blame'.
275 if (!got_path_is_root_dir(graph->path) &&
276 (commit->flags & GOT_COMMIT_FLAG_PACKED)) {
277 int ncommits = 0;
278 err = packed_first_parent_traversal(&ncommits,
279 graph, &qid->id, repo);
280 if (err || ncommits > 0)
281 return err;
283 return got_object_idset_add(graph->open_branches,
284 &qid->id, NULL);
288 * If we are graphing commits for a specific path, skip branches
289 * which do not contribute any content to this path.
291 if (commit->nparents > 1 && !got_path_is_root_dir(graph->path)) {
292 struct got_object_id *merged_id, *prev_id = NULL;
293 int branches_differ = 0;
295 err = got_object_id_by_path(&merged_id, repo, commit,
296 graph->path);
297 if (err)
298 return err;
300 STAILQ_FOREACH(qid, &commit->parent_ids, entry) {
301 struct got_object_id *id = NULL;
302 struct got_commit_object *pcommit = NULL;
304 if (got_object_idset_contains(graph->open_branches,
305 &qid->id))
306 continue;
308 err = got_object_open_as_commit(&pcommit, repo,
309 &qid->id);
310 if (err) {
311 free(merged_id);
312 free(prev_id);
313 return err;
315 err = got_object_id_by_path(&id, repo, pcommit,
316 graph->path);
317 got_object_commit_close(pcommit);
318 pcommit = NULL;
319 if (err) {
320 if (err->code == GOT_ERR_NO_TREE_ENTRY) {
321 branches_differ = 1;
322 continue;
324 free(merged_id);
325 free(prev_id);
326 return err;
329 if (prev_id) {
330 if (!branches_differ &&
331 got_object_id_cmp(id, prev_id) != 0)
332 branches_differ = 1;
333 free(prev_id);
335 prev_id = id;
338 * If a branch has created the merged content we can
339 * skip any other branches.
341 if (got_object_id_cmp(merged_id, id) == 0) {
342 err = got_object_idset_add(graph->open_branches,
343 &qid->id, NULL);
344 free(merged_id);
345 free(id);
346 return err;
350 free(prev_id);
351 prev_id = NULL;
352 free(merged_id);
353 merged_id = NULL;
356 * If the path's content is the same on all branches,
357 * follow the first parent only.
359 if (!branches_differ) {
360 qid = STAILQ_FIRST(&commit->parent_ids);
361 if (qid == NULL)
362 return NULL;
363 if (got_object_idset_contains(graph->open_branches,
364 &qid->id))
365 return NULL;
366 if (got_object_idset_contains(graph->node_ids,
367 &qid->id))
368 return NULL; /* parent already traversed */
369 return got_object_idset_add(graph->open_branches,
370 &qid->id, NULL);
374 STAILQ_FOREACH(qid, &commit->parent_ids, entry) {
375 if (got_object_idset_contains(graph->open_branches, &qid->id))
376 continue;
377 if (got_object_idset_contains(graph->node_ids, &qid->id))
378 continue; /* parent already traversed */
379 err = got_object_idset_add(graph->open_branches, &qid->id,
380 NULL);
381 if (err)
382 return err;
385 return NULL;
388 const struct got_error *
389 got_commit_graph_open(struct got_commit_graph **graph,
390 const char *path, int first_parent_traversal)
392 const struct got_error *err = NULL;
394 *graph = calloc(1, sizeof(**graph));
395 if (*graph == NULL)
396 return got_error_from_errno("calloc");
398 TAILQ_INIT(&(*graph)->iter_list);
400 (*graph)->path = strdup(path);
401 if ((*graph)->path == NULL) {
402 err = got_error_from_errno("strdup");
403 goto done;
406 (*graph)->node_ids = got_object_idset_alloc();
407 if ((*graph)->node_ids == NULL) {
408 err = got_error_from_errno("got_object_idset_alloc");
409 goto done;
412 (*graph)->open_branches = got_object_idset_alloc();
413 if ((*graph)->open_branches == NULL) {
414 err = got_error_from_errno("got_object_idset_alloc");
415 goto done;
418 if (first_parent_traversal)
419 (*graph)->flags |= GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL;
420 done:
421 if (err) {
422 got_commit_graph_close(*graph);
423 *graph = NULL;
425 return err;
428 struct add_branch_tip_arg {
429 struct got_commit_graph_branch_tip *tips;
430 int ntips;
431 struct got_repository *repo;
432 struct got_commit_graph *graph;
435 static const struct got_error *
436 add_branch_tip(struct got_object_id *commit_id, void *data, void *arg)
438 const struct got_error *err;
439 struct add_branch_tip_arg *a = arg;
440 struct got_commit_graph_node *new_node;
441 struct got_commit_object *commit;
443 err = got_object_open_as_commit(&commit, a->repo, commit_id);
444 if (err)
445 return err;
447 err = add_node(&new_node, a->graph, commit_id, a->repo);
448 if (err)
449 return err;
451 a->tips[a->ntips].commit_id = new_node ? &new_node->id : NULL;
452 a->tips[a->ntips].commit = commit;
453 a->tips[a->ntips].new_node = new_node;
454 a->ntips++;
456 return NULL;
459 static const struct got_error *
460 fetch_commits_from_open_branches(struct got_commit_graph *graph,
461 struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
463 const struct got_error *err;
464 struct add_branch_tip_arg arg;
465 int i, ntips;
467 ntips = got_object_idset_num_elements(graph->open_branches);
468 if (ntips == 0)
469 return NULL;
471 /* (Re-)allocate branch tips array if necessary. */
472 if (graph->ntips < ntips) {
473 struct got_commit_graph_branch_tip *tips;
474 tips = recallocarray(graph->tips, graph->ntips, ntips,
475 sizeof(*tips));
476 if (tips == NULL)
477 return got_error_from_errno("recallocarray");
478 graph->tips = tips;
479 graph->ntips = ntips;
481 arg.tips = graph->tips;
482 arg.ntips = 0; /* add_branch_tip() will increment */
483 arg.repo = repo;
484 arg.graph = graph;
485 err = got_object_idset_for_each(graph->open_branches, add_branch_tip,
486 &arg);
487 if (err)
488 goto done;
490 for (i = 0; i < arg.ntips; i++) {
491 struct got_object_id *commit_id;
492 struct got_commit_object *commit;
493 struct got_commit_graph_node *new_node;
494 int changed;
496 if (cancel_cb) {
497 err = (*cancel_cb)(cancel_arg);
498 if (err)
499 break;
502 commit_id = arg.tips[i].commit_id;
503 commit = arg.tips[i].commit;
504 new_node = arg.tips[i].new_node;
506 err = detect_changed_path(&changed, commit, commit_id,
507 graph->path, repo);
508 if (err) {
509 if (err->code != GOT_ERR_NO_OBJ)
510 break;
512 * History of the path stops here on the current
513 * branch. Keep going on other branches.
515 err = close_branch(graph, commit_id);
516 if (err)
517 break;
518 continue;
520 if (changed)
521 add_node_to_iter_list(graph, new_node,
522 got_object_commit_get_committer_time(commit));
523 err = advance_branch(graph, commit_id, commit, repo);
524 if (err)
525 break;
527 done:
528 for (i = 0; i < arg.ntips; i++)
529 got_object_commit_close(arg.tips[i].commit);
530 return err;
533 void
534 got_commit_graph_close(struct got_commit_graph *graph)
536 if (graph->open_branches)
537 got_object_idset_free(graph->open_branches);
538 if (graph->node_ids)
539 got_object_idset_free(graph->node_ids);
540 free(graph->tips);
541 free(graph->path);
542 free(graph);
545 const struct got_error *
546 got_commit_graph_iter_start(struct got_commit_graph *graph,
547 struct got_object_id *id, struct got_repository *repo,
548 got_cancel_cb cancel_cb, void *cancel_arg)
550 const struct got_error *err = NULL;
552 if (!TAILQ_EMPTY(&graph->iter_list))
553 return got_error(GOT_ERR_ITER_BUSY);
555 err = got_object_idset_add(graph->open_branches, id, NULL);
556 if (err)
557 return err;
559 /* Locate first commit which changed graph->path. */
560 while (TAILQ_EMPTY(&graph->iter_list) &&
561 got_object_idset_num_elements(graph->open_branches) > 0) {
562 err = fetch_commits_from_open_branches(graph, repo,
563 cancel_cb, cancel_arg);
564 if (err)
565 return err;
568 if (TAILQ_EMPTY(&graph->iter_list)) {
569 const char *path;
570 if (got_path_is_root_dir(graph->path))
571 return got_error_no_obj(id);
572 path = graph->path;
573 while (path[0] == '/')
574 path++;
575 return got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
578 return NULL;
581 const struct got_error *
582 got_commit_graph_iter_next(struct got_object_id **id,
583 struct got_commit_graph *graph, struct got_repository *repo,
584 got_cancel_cb cancel_cb, void *cancel_arg)
586 const struct got_error *err = NULL;
587 struct got_commit_graph_node *node;
589 *id = NULL;
591 node = TAILQ_FIRST(&graph->iter_list);
592 if (node == NULL) {
593 /* We are done iterating, or iteration was not started. */
594 return got_error(GOT_ERR_ITER_COMPLETED);
597 while (TAILQ_NEXT(node, entry) == NULL &&
598 got_object_idset_num_elements(graph->open_branches) > 0) {
599 err = fetch_commits_from_open_branches(graph, repo,
600 cancel_cb, cancel_arg);
601 if (err)
602 return err;
605 *id = &node->id;
606 TAILQ_REMOVE(&graph->iter_list, node, entry);
607 return NULL;
610 const struct got_error *
611 got_commit_graph_find_youngest_common_ancestor(struct got_object_id **yca_id,
612 struct got_object_id *commit_id, struct got_object_id *commit_id2,
613 int first_parent_traversal,
614 struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
616 const struct got_error *err = NULL;
617 struct got_commit_graph *graph = NULL, *graph2 = NULL;
618 int completed = 0, completed2 = 0;
619 struct got_object_idset *commit_ids;
621 *yca_id = NULL;
623 commit_ids = got_object_idset_alloc();
624 if (commit_ids == NULL)
625 return got_error_from_errno("got_object_idset_alloc");
627 err = got_commit_graph_open(&graph, "/", first_parent_traversal);
628 if (err)
629 goto done;
631 err = got_commit_graph_open(&graph2, "/", first_parent_traversal);
632 if (err)
633 goto done;
635 err = got_commit_graph_iter_start(graph, commit_id, repo,
636 cancel_cb, cancel_arg);
637 if (err)
638 goto done;
640 err = got_commit_graph_iter_start(graph2, commit_id2, repo,
641 cancel_cb, cancel_arg);
642 if (err)
643 goto done;
645 for (;;) {
646 struct got_object_id *id = NULL, *id2 = NULL;
648 if (cancel_cb) {
649 err = (*cancel_cb)(cancel_arg);
650 if (err)
651 break;
654 if (!completed) {
655 err = got_commit_graph_iter_next(&id, graph, repo,
656 cancel_cb, cancel_arg);
657 if (err) {
658 if (err->code != GOT_ERR_ITER_COMPLETED)
659 break;
660 err = NULL;
661 completed = 1;
665 if (!completed2) {
666 err = got_commit_graph_iter_next(&id2, graph2, repo,
667 cancel_cb, cancel_arg);
668 if (err) {
669 if (err->code != GOT_ERR_ITER_COMPLETED)
670 break;
671 err = NULL;
672 completed2 = 1;
676 if (id) {
677 if (got_object_idset_contains(commit_ids, id)) {
678 *yca_id = got_object_id_dup(id);
679 if (*yca_id)
680 break;
681 err = got_error_from_errno("got_object_id_dup");
682 break;
685 err = got_object_idset_add(commit_ids, id, NULL);
686 if (err)
687 break;
689 if (id2) {
690 if (got_object_idset_contains(commit_ids, id2)) {
691 *yca_id = got_object_id_dup(id2);
692 if (*yca_id)
693 break;
694 err = got_error_from_errno("got_object_id_dup");
695 break;
698 err = got_object_idset_add(commit_ids, id2, NULL);
699 if (err)
700 break;
703 if (completed && completed2) {
704 err = got_error(GOT_ERR_ANCESTRY);
705 break;
709 done:
710 got_object_idset_free(commit_ids);
711 if (graph)
712 got_commit_graph_close(graph);
713 if (graph2)
714 got_commit_graph_close(graph2);
715 return err;