4 #include "commit-graph.h"
5 #include "repository.h"
6 #include "object-store.h"
13 #include "gpg-interface.h"
14 #include "mergesort.h"
15 #include "commit-slab.h"
16 #include "prio-queue.h"
17 #include "sha1-lookup.h"
18 #include "wt-status.h"
21 static struct commit_extra_header *read_commit_extra_header_lines(const char *buf, size_t len, const char **);
23 int save_commit_buffer = 1;
25 const char *commit_type = "commit";
27 struct commit *lookup_commit_reference_gently(struct repository *r,
28 const struct object_id *oid, int quiet)
30 struct object *obj = deref_tag(r,
36 return object_as_type(r, obj, OBJ_COMMIT, quiet);
39 struct commit *lookup_commit_reference(struct repository *r, const struct object_id *oid)
41 return lookup_commit_reference_gently(r, oid, 0);
44 struct commit *lookup_commit_or_die(const struct object_id *oid, const char *ref_name)
46 struct commit *c = lookup_commit_reference(the_repository, oid);
48 die(_("could not parse %s"), ref_name);
49 if (!oideq(oid, &c->object.oid)) {
50 warning(_("%s %s is not a commit!"),
51 ref_name, oid_to_hex(oid));
56 struct commit *lookup_commit(struct repository *r, const struct object_id *oid)
58 struct object *obj = lookup_object(r, oid->hash);
60 return create_object(r, oid->hash,
61 alloc_commit_node(r));
62 return object_as_type(r, obj, OBJ_COMMIT, 0);
65 struct commit *lookup_commit_reference_by_name(const char *name)
68 struct commit *commit;
70 if (get_oid_committish(name, &oid))
72 commit = lookup_commit_reference(the_repository, &oid);
73 if (parse_commit(commit))
78 static timestamp_t parse_commit_date(const char *buf, const char *tail)
84 if (memcmp(buf, "author", 6))
86 while (buf < tail && *buf++ != '\n')
90 if (memcmp(buf, "committer", 9))
92 while (buf < tail && *buf++ != '>')
97 while (buf < tail && *buf++ != '\n')
101 /* dateptr < buf && buf[-1] == '\n', so parsing will stop at buf-1 */
102 return parse_timestamp(dateptr, NULL, 10);
105 static const unsigned char *commit_graft_sha1_access(size_t index, void *table)
107 struct commit_graft **commit_graft_table = table;
108 return commit_graft_table[index]->oid.hash;
111 static int commit_graft_pos(struct repository *r, const unsigned char *sha1)
113 return sha1_pos(sha1, r->parsed_objects->grafts,
114 r->parsed_objects->grafts_nr,
115 commit_graft_sha1_access);
118 int register_commit_graft(struct repository *r, struct commit_graft *graft,
121 int pos = commit_graft_pos(r, graft->oid.hash);
127 free(r->parsed_objects->grafts[pos]);
128 r->parsed_objects->grafts[pos] = graft;
133 ALLOC_GROW(r->parsed_objects->grafts,
134 r->parsed_objects->grafts_nr + 1,
135 r->parsed_objects->grafts_alloc);
136 r->parsed_objects->grafts_nr++;
137 if (pos < r->parsed_objects->grafts_nr)
138 memmove(r->parsed_objects->grafts + pos + 1,
139 r->parsed_objects->grafts + pos,
140 (r->parsed_objects->grafts_nr - pos - 1) *
141 sizeof(*r->parsed_objects->grafts));
142 r->parsed_objects->grafts[pos] = graft;
146 struct commit_graft *read_graft_line(struct strbuf *line)
148 /* The format is just "Commit Parent1 Parent2 ...\n" */
150 const char *tail = NULL;
151 struct commit_graft *graft = NULL;
152 struct object_id dummy_oid, *oid;
155 if (!line->len || line->buf[0] == '#')
158 * phase 0 verifies line, counts hashes in line and allocates graft
159 * phase 1 fills graft
161 for (phase = 0; phase < 2; phase++) {
162 oid = graft ? &graft->oid : &dummy_oid;
163 if (parse_oid_hex(line->buf, oid, &tail))
165 for (i = 0; *tail != '\0'; i++) {
166 oid = graft ? &graft->parent[i] : &dummy_oid;
167 if (!isspace(*tail++) || parse_oid_hex(tail, oid, &tail))
171 graft = xmalloc(st_add(sizeof(*graft),
172 st_mult(sizeof(struct object_id), i)));
173 graft->nr_parent = i;
179 error("bad graft data: %s", line->buf);
184 static int read_graft_file(struct repository *r, const char *graft_file)
186 FILE *fp = fopen_or_warn(graft_file, "r");
187 struct strbuf buf = STRBUF_INIT;
190 if (advice_graft_file_deprecated)
191 advise(_("Support for <GIT_DIR>/info/grafts is deprecated\n"
192 "and will be removed in a future Git version.\n"
194 "Please use \"git replace --convert-graft-file\"\n"
195 "to convert the grafts into replace refs.\n"
197 "Turn this message off by running\n"
198 "\"git config advice.graftFileDeprecated false\""));
199 while (!strbuf_getwholeline(&buf, fp, '\n')) {
200 /* The format is just "Commit Parent1 Parent2 ...\n" */
201 struct commit_graft *graft = read_graft_line(&buf);
204 if (register_commit_graft(r, graft, 1))
205 error("duplicate graft data: %s", buf.buf);
208 strbuf_release(&buf);
212 void prepare_commit_graft(struct repository *r)
216 if (r->parsed_objects->commit_graft_prepared)
218 if (!startup_info->have_repository)
221 graft_file = get_graft_file(r);
222 read_graft_file(r, graft_file);
223 /* make sure shallows are read */
224 is_repository_shallow(r);
225 r->parsed_objects->commit_graft_prepared = 1;
228 struct commit_graft *lookup_commit_graft(struct repository *r, const struct object_id *oid)
231 prepare_commit_graft(r);
232 pos = commit_graft_pos(r, oid->hash);
235 return r->parsed_objects->grafts[pos];
238 int for_each_commit_graft(each_commit_graft_fn fn, void *cb_data)
241 for (i = ret = 0; i < the_repository->parsed_objects->grafts_nr && !ret; i++)
242 ret = fn(the_repository->parsed_objects->grafts[i], cb_data);
246 int unregister_shallow(const struct object_id *oid)
248 int pos = commit_graft_pos(the_repository, oid->hash);
251 if (pos + 1 < the_repository->parsed_objects->grafts_nr)
252 MOVE_ARRAY(the_repository->parsed_objects->grafts + pos,
253 the_repository->parsed_objects->grafts + pos + 1,
254 the_repository->parsed_objects->grafts_nr - pos - 1);
255 the_repository->parsed_objects->grafts_nr--;
259 struct commit_buffer {
263 define_commit_slab(buffer_slab, struct commit_buffer);
265 struct buffer_slab *allocate_commit_buffer_slab(void)
267 struct buffer_slab *bs = xmalloc(sizeof(*bs));
268 init_buffer_slab(bs);
272 void free_commit_buffer_slab(struct buffer_slab *bs)
274 clear_buffer_slab(bs);
278 void set_commit_buffer(struct repository *r, struct commit *commit, void *buffer, unsigned long size)
280 struct commit_buffer *v = buffer_slab_at(
281 r->parsed_objects->buffer_slab, commit);
286 const void *get_cached_commit_buffer(struct repository *r, const struct commit *commit, unsigned long *sizep)
288 struct commit_buffer *v = buffer_slab_peek(
289 r->parsed_objects->buffer_slab, commit);
300 const void *repo_get_commit_buffer(struct repository *r,
301 const struct commit *commit,
302 unsigned long *sizep)
304 const void *ret = get_cached_commit_buffer(r, commit, sizep);
306 enum object_type type;
308 ret = repo_read_object_file(r, &commit->object.oid, &type, &size);
310 die("cannot read commit object %s",
311 oid_to_hex(&commit->object.oid));
312 if (type != OBJ_COMMIT)
313 die("expected commit for %s, got %s",
314 oid_to_hex(&commit->object.oid), type_name(type));
321 void unuse_commit_buffer(const struct commit *commit, const void *buffer)
323 struct commit_buffer *v = buffer_slab_peek(
324 the_repository->parsed_objects->buffer_slab, commit);
325 if (!(v && v->buffer == buffer))
326 free((void *)buffer);
329 void free_commit_buffer(struct commit *commit)
331 struct commit_buffer *v = buffer_slab_peek(
332 the_repository->parsed_objects->buffer_slab, commit);
334 FREE_AND_NULL(v->buffer);
339 struct tree *get_commit_tree(const struct commit *commit)
341 if (commit->maybe_tree || !commit->object.parsed)
342 return commit->maybe_tree;
344 if (commit->graph_pos == COMMIT_NOT_FROM_GRAPH)
345 BUG("commit has NULL tree, but was not loaded from commit-graph");
347 return get_commit_tree_in_graph(the_repository, commit);
350 struct object_id *get_commit_tree_oid(const struct commit *commit)
352 return &get_commit_tree(commit)->object.oid;
355 void release_commit_memory(struct commit *c)
357 c->maybe_tree = NULL;
359 free_commit_buffer(c);
360 free_commit_list(c->parents);
361 /* TODO: what about commit->util? */
363 c->object.parsed = 0;
366 const void *detach_commit_buffer(struct commit *commit, unsigned long *sizep)
368 struct commit_buffer *v = buffer_slab_peek(
369 the_repository->parsed_objects->buffer_slab, commit);
386 int parse_commit_buffer(struct repository *r, struct commit *item, const void *buffer, unsigned long size, int check_graph)
388 const char *tail = buffer;
389 const char *bufptr = buffer;
390 struct object_id parent;
391 struct commit_list **pptr;
392 struct commit_graft *graft;
393 const int tree_entry_len = the_hash_algo->hexsz + 5;
394 const int parent_entry_len = the_hash_algo->hexsz + 7;
396 if (item->object.parsed)
398 item->object.parsed = 1;
400 if (tail <= bufptr + tree_entry_len + 1 || memcmp(bufptr, "tree ", 5) ||
401 bufptr[tree_entry_len] != '\n')
402 return error("bogus commit object %s", oid_to_hex(&item->object.oid));
403 if (get_oid_hex(bufptr + 5, &parent) < 0)
404 return error("bad tree pointer in commit %s",
405 oid_to_hex(&item->object.oid));
406 item->maybe_tree = lookup_tree(r, &parent);
407 bufptr += tree_entry_len + 1; /* "tree " + "hex sha1" + "\n" */
408 pptr = &item->parents;
410 graft = lookup_commit_graft(r, &item->object.oid);
411 while (bufptr + parent_entry_len < tail && !memcmp(bufptr, "parent ", 7)) {
412 struct commit *new_parent;
414 if (tail <= bufptr + parent_entry_len + 1 ||
415 get_oid_hex(bufptr + 7, &parent) ||
416 bufptr[parent_entry_len] != '\n')
417 return error("bad parents in commit %s", oid_to_hex(&item->object.oid));
418 bufptr += parent_entry_len + 1;
420 * The clone is shallow if nr_parent < 0, and we must
421 * not traverse its real parents even when we unhide them.
423 if (graft && (graft->nr_parent < 0 || grafts_replace_parents))
425 new_parent = lookup_commit(r, &parent);
427 pptr = &commit_list_insert(new_parent, pptr)->next;
431 struct commit *new_parent;
432 for (i = 0; i < graft->nr_parent; i++) {
433 new_parent = lookup_commit(r,
437 pptr = &commit_list_insert(new_parent, pptr)->next;
440 item->date = parse_commit_date(bufptr, tail);
443 load_commit_graph_info(the_repository, item);
448 int repo_parse_commit_internal(struct repository *r,
450 int quiet_on_missing,
451 int use_commit_graph)
453 enum object_type type;
460 if (item->object.parsed)
462 if (use_commit_graph && parse_commit_in_graph(r, item))
464 buffer = repo_read_object_file(r, &item->object.oid, &type, &size);
466 return quiet_on_missing ? -1 :
467 error("Could not read %s",
468 oid_to_hex(&item->object.oid));
469 if (type != OBJ_COMMIT) {
471 return error("Object %s not a commit",
472 oid_to_hex(&item->object.oid));
475 ret = parse_commit_buffer(r, item, buffer, size, 0);
476 if (save_commit_buffer && !ret) {
477 set_commit_buffer(r, item, buffer, size);
484 int repo_parse_commit_gently(struct repository *r,
485 struct commit *item, int quiet_on_missing)
487 return repo_parse_commit_internal(r, item, quiet_on_missing, 1);
490 void parse_commit_or_die(struct commit *item)
492 if (parse_commit(item))
493 die("unable to parse commit %s",
494 item ? oid_to_hex(&item->object.oid) : "(null)");
497 int find_commit_subject(const char *commit_buffer, const char **subject)
500 const char *p = commit_buffer;
502 while (*p && (*p != '\n' || p[1] != '\n'))
505 p = skip_blank_lines(p + 2);
506 eol = strchrnul(p, '\n');
515 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
517 struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
518 new_list->item = item;
519 new_list->next = *list_p;
524 unsigned commit_list_count(const struct commit_list *l)
527 for (; l; l = l->next )
532 struct commit_list *copy_commit_list(struct commit_list *list)
534 struct commit_list *head = NULL;
535 struct commit_list **pp = &head;
537 pp = commit_list_append(list->item, pp);
543 void free_commit_list(struct commit_list *list)
549 struct commit_list * commit_list_insert_by_date(struct commit *item, struct commit_list **list)
551 struct commit_list **pp = list;
552 struct commit_list *p;
553 while ((p = *pp) != NULL) {
554 if (p->item->date < item->date) {
559 return commit_list_insert(item, pp);
562 static int commit_list_compare_by_date(const void *a, const void *b)
564 timestamp_t a_date = ((const struct commit_list *)a)->item->date;
565 timestamp_t b_date = ((const struct commit_list *)b)->item->date;
573 static void *commit_list_get_next(const void *a)
575 return ((const struct commit_list *)a)->next;
578 static void commit_list_set_next(void *a, void *next)
580 ((struct commit_list *)a)->next = next;
583 void commit_list_sort_by_date(struct commit_list **list)
585 *list = llist_mergesort(*list, commit_list_get_next, commit_list_set_next,
586 commit_list_compare_by_date);
589 struct commit *pop_most_recent_commit(struct commit_list **list,
592 struct commit *ret = pop_commit(list);
593 struct commit_list *parents = ret->parents;
596 struct commit *commit = parents->item;
597 if (!parse_commit(commit) && !(commit->object.flags & mark)) {
598 commit->object.flags |= mark;
599 commit_list_insert_by_date(commit, list);
601 parents = parents->next;
606 static void clear_commit_marks_1(struct commit_list **plist,
607 struct commit *commit, unsigned int mark)
610 struct commit_list *parents;
612 if (!(mark & commit->object.flags))
615 commit->object.flags &= ~mark;
617 parents = commit->parents;
621 while ((parents = parents->next))
622 commit_list_insert(parents->item, plist);
624 commit = commit->parents->item;
628 void clear_commit_marks_many(int nr, struct commit **commit, unsigned int mark)
630 struct commit_list *list = NULL;
633 clear_commit_marks_1(&list, *commit, mark);
637 clear_commit_marks_1(&list, pop_commit(&list), mark);
640 void clear_commit_marks(struct commit *commit, unsigned int mark)
642 clear_commit_marks_many(1, &commit, mark);
645 struct commit *pop_commit(struct commit_list **stack)
647 struct commit_list *top = *stack;
648 struct commit *item = top ? top->item : NULL;
658 * Topological sort support
661 /* count number of children that have not been emitted */
662 define_commit_slab(indegree_slab, int);
664 /* record author-date for each commit object */
665 define_commit_slab(author_date_slab, timestamp_t);
667 static void record_author_date(struct author_date_slab *author_date,
668 struct commit *commit)
670 const char *buffer = get_commit_buffer(commit, NULL);
671 struct ident_split ident;
672 const char *ident_line;
677 ident_line = find_commit_header(buffer, "author", &ident_len);
679 goto fail_exit; /* no author line */
680 if (split_ident_line(&ident, ident_line, ident_len) ||
681 !ident.date_begin || !ident.date_end)
682 goto fail_exit; /* malformed "author" line */
684 date = parse_timestamp(ident.date_begin, &date_end, 10);
685 if (date_end != ident.date_end)
686 goto fail_exit; /* malformed date */
687 *(author_date_slab_at(author_date, commit)) = date;
690 unuse_commit_buffer(commit, buffer);
693 static int compare_commits_by_author_date(const void *a_, const void *b_,
696 const struct commit *a = a_, *b = b_;
697 struct author_date_slab *author_date = cb_data;
698 timestamp_t a_date = *(author_date_slab_at(author_date, a));
699 timestamp_t b_date = *(author_date_slab_at(author_date, b));
701 /* newer commits with larger date first */
704 else if (a_date > b_date)
709 int compare_commits_by_gen_then_commit_date(const void *a_, const void *b_, void *unused)
711 const struct commit *a = a_, *b = b_;
713 /* newer commits first */
714 if (a->generation < b->generation)
716 else if (a->generation > b->generation)
719 /* use date as a heuristic when generations are equal */
720 if (a->date < b->date)
722 else if (a->date > b->date)
727 int compare_commits_by_commit_date(const void *a_, const void *b_, void *unused)
729 const struct commit *a = a_, *b = b_;
730 /* newer commits with larger date first */
731 if (a->date < b->date)
733 else if (a->date > b->date)
739 * Performs an in-place topological sort on the list supplied.
741 void sort_in_topological_order(struct commit_list **list, enum rev_sort_order sort_order)
743 struct commit_list *next, *orig = *list;
744 struct commit_list **pptr;
745 struct indegree_slab indegree;
746 struct prio_queue queue;
747 struct commit *commit;
748 struct author_date_slab author_date;
754 init_indegree_slab(&indegree);
755 memset(&queue, '\0', sizeof(queue));
757 switch (sort_order) {
758 default: /* REV_SORT_IN_GRAPH_ORDER */
759 queue.compare = NULL;
761 case REV_SORT_BY_COMMIT_DATE:
762 queue.compare = compare_commits_by_commit_date;
764 case REV_SORT_BY_AUTHOR_DATE:
765 init_author_date_slab(&author_date);
766 queue.compare = compare_commits_by_author_date;
767 queue.cb_data = &author_date;
771 /* Mark them and clear the indegree */
772 for (next = orig; next; next = next->next) {
773 struct commit *commit = next->item;
774 *(indegree_slab_at(&indegree, commit)) = 1;
775 /* also record the author dates, if needed */
776 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
777 record_author_date(&author_date, commit);
780 /* update the indegree */
781 for (next = orig; next; next = next->next) {
782 struct commit_list *parents = next->item->parents;
784 struct commit *parent = parents->item;
785 int *pi = indegree_slab_at(&indegree, parent);
789 parents = parents->next;
796 * tips are nodes not reachable from any other node in the list
798 * the tips serve as a starting set for the work queue.
800 for (next = orig; next; next = next->next) {
801 struct commit *commit = next->item;
803 if (*(indegree_slab_at(&indegree, commit)) == 1)
804 prio_queue_put(&queue, commit);
808 * This is unfortunate; the initial tips need to be shown
809 * in the order given from the revision traversal machinery.
811 if (sort_order == REV_SORT_IN_GRAPH_ORDER)
812 prio_queue_reverse(&queue);
814 /* We no longer need the commit list */
815 free_commit_list(orig);
819 while ((commit = prio_queue_get(&queue)) != NULL) {
820 struct commit_list *parents;
822 for (parents = commit->parents; parents ; parents = parents->next) {
823 struct commit *parent = parents->item;
824 int *pi = indegree_slab_at(&indegree, parent);
830 * parents are only enqueued for emission
831 * when all their children have been emitted thereby
832 * guaranteeing topological order.
835 prio_queue_put(&queue, parent);
838 * all children of commit have already been
839 * emitted. we can emit it now.
841 *(indegree_slab_at(&indegree, commit)) = 0;
843 pptr = &commit_list_insert(commit, pptr)->next;
846 clear_indegree_slab(&indegree);
847 clear_prio_queue(&queue);
848 if (sort_order == REV_SORT_BY_AUTHOR_DATE)
849 clear_author_date_slab(&author_date);
852 static const char gpg_sig_header[] = "gpgsig";
853 static const int gpg_sig_header_len = sizeof(gpg_sig_header) - 1;
855 static int do_sign_commit(struct strbuf *buf, const char *keyid)
857 struct strbuf sig = STRBUF_INIT;
861 /* find the end of the header */
862 eoh = strstr(buf->buf, "\n\n");
866 inspos = eoh - buf->buf + 1;
868 if (!keyid || !*keyid)
869 keyid = get_signing_key();
870 if (sign_buffer(buf, &sig, keyid)) {
871 strbuf_release(&sig);
875 for (copypos = 0; sig.buf[copypos]; ) {
876 const char *bol = sig.buf + copypos;
877 const char *eol = strchrnul(bol, '\n');
878 int len = (eol - bol) + !!*eol;
881 strbuf_insert(buf, inspos, gpg_sig_header, gpg_sig_header_len);
882 inspos += gpg_sig_header_len;
884 strbuf_insert(buf, inspos++, " ", 1);
885 strbuf_insert(buf, inspos, bol, len);
889 strbuf_release(&sig);
893 int parse_signed_commit(const struct commit *commit,
894 struct strbuf *payload, struct strbuf *signature)
898 const char *buffer = get_commit_buffer(commit, &size);
899 int in_signature, saw_signature = -1;
900 const char *line, *tail;
903 tail = buffer + size;
906 while (line < tail) {
907 const char *sig = NULL;
908 const char *next = memchr(line, '\n', tail - line);
910 next = next ? next + 1 : tail;
911 if (in_signature && line[0] == ' ')
913 else if (starts_with(line, gpg_sig_header) &&
914 line[gpg_sig_header_len] == ' ')
915 sig = line + gpg_sig_header_len + 1;
917 strbuf_add(signature, sig, next - sig);
922 /* dump the whole remainder of the buffer */
924 strbuf_add(payload, line, next - line);
929 unuse_commit_buffer(commit, buffer);
930 return saw_signature;
933 int remove_signature(struct strbuf *buf)
935 const char *line = buf->buf;
936 const char *tail = buf->buf + buf->len;
937 int in_signature = 0;
938 const char *sig_start = NULL;
939 const char *sig_end = NULL;
941 while (line < tail) {
942 const char *next = memchr(line, '\n', tail - line);
943 next = next ? next + 1 : tail;
945 if (in_signature && line[0] == ' ')
947 else if (starts_with(line, gpg_sig_header) &&
948 line[gpg_sig_header_len] == ' ') {
954 /* dump the whole remainder of the buffer */
962 strbuf_remove(buf, sig_start - buf->buf, sig_end - sig_start);
964 return sig_start != NULL;
967 static void handle_signed_tag(struct commit *parent, struct commit_extra_header ***tail)
969 struct merge_remote_desc *desc;
970 struct commit_extra_header *mergetag;
972 unsigned long size, len;
973 enum object_type type;
975 desc = merge_remote_util(parent);
976 if (!desc || !desc->obj)
978 buf = read_object_file(&desc->obj->oid, &type, &size);
979 if (!buf || type != OBJ_TAG)
981 len = parse_signature(buf, size);
985 * We could verify this signature and either omit the tag when
986 * it does not validate, but the integrator may not have the
987 * public key of the signer of the tag he is merging, while a
988 * later auditor may have it while auditing, so let's not run
989 * verify-signed-buffer here for now...
991 * if (verify_signed_buffer(buf, len, buf + len, size - len, ...))
992 * warn("warning: signed tag unverified.");
994 mergetag = xcalloc(1, sizeof(*mergetag));
995 mergetag->key = xstrdup("mergetag");
996 mergetag->value = buf;
997 mergetag->len = size;
1000 *tail = &mergetag->next;
1007 int check_commit_signature(const struct commit *commit, struct signature_check *sigc)
1009 struct strbuf payload = STRBUF_INIT;
1010 struct strbuf signature = STRBUF_INIT;
1015 if (parse_signed_commit(commit, &payload, &signature) <= 0)
1017 ret = check_signature(payload.buf, payload.len, signature.buf,
1018 signature.len, sigc);
1021 strbuf_release(&payload);
1022 strbuf_release(&signature);
1029 void append_merge_tag_headers(struct commit_list *parents,
1030 struct commit_extra_header ***tail)
1033 struct commit *parent = parents->item;
1034 handle_signed_tag(parent, tail);
1035 parents = parents->next;
1039 static void add_extra_header(struct strbuf *buffer,
1040 struct commit_extra_header *extra)
1042 strbuf_addstr(buffer, extra->key);
1044 strbuf_add_lines(buffer, " ", extra->value, extra->len);
1046 strbuf_addch(buffer, '\n');
1049 struct commit_extra_header *read_commit_extra_headers(struct commit *commit,
1050 const char **exclude)
1052 struct commit_extra_header *extra = NULL;
1054 const char *buffer = get_commit_buffer(commit, &size);
1055 extra = read_commit_extra_header_lines(buffer, size, exclude);
1056 unuse_commit_buffer(commit, buffer);
1060 int for_each_mergetag(each_mergetag_fn fn, struct commit *commit, void *data)
1062 struct commit_extra_header *extra, *to_free;
1065 to_free = read_commit_extra_headers(commit, NULL);
1066 for (extra = to_free; !res && extra; extra = extra->next) {
1067 if (strcmp(extra->key, "mergetag"))
1068 continue; /* not a merge tag */
1069 res = fn(commit, extra, data);
1071 free_commit_extra_headers(to_free);
1075 static inline int standard_header_field(const char *field, size_t len)
1077 return ((len == 4 && !memcmp(field, "tree", 4)) ||
1078 (len == 6 && !memcmp(field, "parent", 6)) ||
1079 (len == 6 && !memcmp(field, "author", 6)) ||
1080 (len == 9 && !memcmp(field, "committer", 9)) ||
1081 (len == 8 && !memcmp(field, "encoding", 8)));
1084 static int excluded_header_field(const char *field, size_t len, const char **exclude)
1090 size_t xlen = strlen(*exclude);
1091 if (len == xlen && !memcmp(field, *exclude, xlen))
1098 static struct commit_extra_header *read_commit_extra_header_lines(
1099 const char *buffer, size_t size,
1100 const char **exclude)
1102 struct commit_extra_header *extra = NULL, **tail = &extra, *it = NULL;
1103 const char *line, *next, *eof, *eob;
1104 struct strbuf buf = STRBUF_INIT;
1106 for (line = buffer, eob = line + size;
1107 line < eob && *line != '\n';
1109 next = memchr(line, '\n', eob - line);
1110 next = next ? next + 1 : eob;
1114 strbuf_add(&buf, line + 1, next - (line + 1));
1118 it->value = strbuf_detach(&buf, &it->len);
1122 eof = memchr(line, ' ', next - line);
1125 else if (standard_header_field(line, eof - line) ||
1126 excluded_header_field(line, eof - line, exclude))
1129 it = xcalloc(1, sizeof(*it));
1130 it->key = xmemdupz(line, eof-line);
1134 strbuf_add(&buf, eof + 1, next - (eof + 1));
1137 it->value = strbuf_detach(&buf, &it->len);
1141 void free_commit_extra_headers(struct commit_extra_header *extra)
1144 struct commit_extra_header *next = extra->next;
1152 int commit_tree(const char *msg, size_t msg_len, const struct object_id *tree,
1153 struct commit_list *parents, struct object_id *ret,
1154 const char *author, const char *sign_commit)
1156 struct commit_extra_header *extra = NULL, **tail = &extra;
1159 append_merge_tag_headers(parents, &tail);
1160 result = commit_tree_extended(msg, msg_len, tree, parents, ret,
1161 author, sign_commit, extra);
1162 free_commit_extra_headers(extra);
1166 static int find_invalid_utf8(const char *buf, int len)
1169 static const unsigned int max_codepoint[] = {
1170 0x7f, 0x7ff, 0xffff, 0x10ffff
1174 unsigned char c = *buf++;
1175 int bytes, bad_offset;
1176 unsigned int codepoint;
1177 unsigned int min_val, max_val;
1182 /* Simple US-ASCII? No worries. */
1186 bad_offset = offset-1;
1189 * Count how many more high bits set: that's how
1190 * many more bytes this sequence should have.
1199 * Must be between 1 and 3 more bytes. Longer sequences result in
1200 * codepoints beyond U+10FFFF, which are guaranteed never to exist.
1202 if (bytes < 1 || 3 < bytes)
1205 /* Do we *have* that many bytes? */
1210 * Place the encoded bits at the bottom of the value and compute the
1213 codepoint = (c & 0x7f) >> bytes;
1214 min_val = max_codepoint[bytes-1] + 1;
1215 max_val = max_codepoint[bytes];
1220 /* And verify that they are good continuation bytes */
1223 codepoint |= *buf & 0x3f;
1224 if ((*buf++ & 0xc0) != 0x80)
1228 /* Reject codepoints that are out of range for the sequence length. */
1229 if (codepoint < min_val || codepoint > max_val)
1231 /* Surrogates are only for UTF-16 and cannot be encoded in UTF-8. */
1232 if ((codepoint & 0x1ff800) == 0xd800)
1234 /* U+xxFFFE and U+xxFFFF are guaranteed non-characters. */
1235 if ((codepoint & 0xfffe) == 0xfffe)
1237 /* So are anything in the range U+FDD0..U+FDEF. */
1238 if (codepoint >= 0xfdd0 && codepoint <= 0xfdef)
1245 * This verifies that the buffer is in proper utf8 format.
1247 * If it isn't, it assumes any non-utf8 characters are Latin1,
1248 * and does the conversion.
1250 static int verify_utf8(struct strbuf *buf)
1258 unsigned char replace[2];
1260 bad = find_invalid_utf8(buf->buf + pos, buf->len - pos);
1266 strbuf_remove(buf, pos, 1);
1268 /* We know 'c' must be in the range 128-255 */
1269 replace[0] = 0xc0 + (c >> 6);
1270 replace[1] = 0x80 + (c & 0x3f);
1271 strbuf_insert(buf, pos, replace, 2);
1276 static const char commit_utf8_warn[] =
1277 N_("Warning: commit message did not conform to UTF-8.\n"
1278 "You may want to amend it after fixing the message, or set the config\n"
1279 "variable i18n.commitencoding to the encoding your project uses.\n");
1281 int commit_tree_extended(const char *msg, size_t msg_len,
1282 const struct object_id *tree,
1283 struct commit_list *parents, struct object_id *ret,
1284 const char *author, const char *sign_commit,
1285 struct commit_extra_header *extra)
1288 int encoding_is_utf8;
1289 struct strbuf buffer;
1291 assert_oid_type(tree, OBJ_TREE);
1293 if (memchr(msg, '\0', msg_len))
1294 return error("a NUL byte in commit log message not allowed.");
1296 /* Not having i18n.commitencoding is the same as having utf-8 */
1297 encoding_is_utf8 = is_encoding_utf8(git_commit_encoding);
1299 strbuf_init(&buffer, 8192); /* should avoid reallocs for the headers */
1300 strbuf_addf(&buffer, "tree %s\n", oid_to_hex(tree));
1303 * NOTE! This ordering means that the same exact tree merged with a
1304 * different order of parents will be a _different_ changeset even
1305 * if everything else stays the same.
1308 struct commit *parent = pop_commit(&parents);
1309 strbuf_addf(&buffer, "parent %s\n",
1310 oid_to_hex(&parent->object.oid));
1313 /* Person/date information */
1315 author = git_author_info(IDENT_STRICT);
1316 strbuf_addf(&buffer, "author %s\n", author);
1317 strbuf_addf(&buffer, "committer %s\n", git_committer_info(IDENT_STRICT));
1318 if (!encoding_is_utf8)
1319 strbuf_addf(&buffer, "encoding %s\n", git_commit_encoding);
1322 add_extra_header(&buffer, extra);
1323 extra = extra->next;
1325 strbuf_addch(&buffer, '\n');
1327 /* And add the comment */
1328 strbuf_add(&buffer, msg, msg_len);
1330 /* And check the encoding */
1331 if (encoding_is_utf8 && !verify_utf8(&buffer))
1332 fprintf(stderr, _(commit_utf8_warn));
1334 if (sign_commit && do_sign_commit(&buffer, sign_commit)) {
1339 result = write_object_file(buffer.buf, buffer.len, commit_type, ret);
1341 strbuf_release(&buffer);
1345 define_commit_slab(merge_desc_slab, struct merge_remote_desc *);
1346 static struct merge_desc_slab merge_desc_slab = COMMIT_SLAB_INIT(1, merge_desc_slab);
1348 struct merge_remote_desc *merge_remote_util(struct commit *commit)
1350 return *merge_desc_slab_at(&merge_desc_slab, commit);
1353 void set_merge_remote_desc(struct commit *commit,
1354 const char *name, struct object *obj)
1356 struct merge_remote_desc *desc;
1357 FLEX_ALLOC_STR(desc, name, name);
1359 *merge_desc_slab_at(&merge_desc_slab, commit) = desc;
1362 struct commit *get_merge_parent(const char *name)
1365 struct commit *commit;
1366 struct object_id oid;
1367 if (get_oid(name, &oid))
1369 obj = parse_object(the_repository, &oid);
1370 commit = (struct commit *)peel_to_type(name, 0, obj, OBJ_COMMIT);
1371 if (commit && !merge_remote_util(commit))
1372 set_merge_remote_desc(commit, name, obj);
1377 * Append a commit to the end of the commit_list.
1379 * next starts by pointing to the variable that holds the head of an
1380 * empty commit_list, and is updated to point to the "next" field of
1381 * the last item on the list as new commits are appended.
1385 * struct commit_list *list;
1386 * struct commit_list **next = &list;
1388 * next = commit_list_append(c1, next);
1389 * next = commit_list_append(c2, next);
1390 * assert(commit_list_count(list) == 2);
1393 struct commit_list **commit_list_append(struct commit *commit,
1394 struct commit_list **next)
1396 struct commit_list *new_commit = xmalloc(sizeof(struct commit_list));
1397 new_commit->item = commit;
1399 new_commit->next = NULL;
1400 return &new_commit->next;
1403 const char *find_commit_header(const char *msg, const char *key, size_t *out_len)
1405 int key_len = strlen(key);
1406 const char *line = msg;
1409 const char *eol = strchrnul(line, '\n');
1414 if (eol - line > key_len &&
1415 !strncmp(line, key, key_len) &&
1416 line[key_len] == ' ') {
1417 *out_len = eol - line - key_len - 1;
1418 return line + key_len + 1;
1420 line = *eol ? eol + 1 : NULL;
1426 * Inspect the given string and determine the true "end" of the log message, in
1427 * order to find where to put a new Signed-off-by: line. Ignored are
1428 * trailing comment lines and blank lines. To support "git commit -s
1429 * --amend" on an existing commit, we also ignore "Conflicts:". To
1430 * support "git commit -v", we truncate at cut lines.
1432 * Returns the number of bytes from the tail to ignore, to be fed as
1433 * the second parameter to append_signoff().
1435 size_t ignore_non_trailer(const char *buf, size_t len)
1439 int in_old_conflicts_block = 0;
1440 size_t cutoff = wt_status_locate_end(buf, len);
1442 while (bol < cutoff) {
1443 const char *next_line = memchr(buf + bol, '\n', len - bol);
1446 next_line = buf + len;
1450 if (buf[bol] == comment_line_char || buf[bol] == '\n') {
1451 /* is this the first of the run of comments? */
1454 /* otherwise, it is just continuing */
1455 } else if (starts_with(buf + bol, "Conflicts:\n")) {
1456 in_old_conflicts_block = 1;
1459 } else if (in_old_conflicts_block && buf[bol] == '\t') {
1460 ; /* a pathname in the conflicts block */
1462 /* the previous was not trailing comment */
1464 in_old_conflicts_block = 0;
1466 bol = next_line - buf;
1468 return boc ? len - boc : len - cutoff;