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>
19 #include <sys/queue.h>
22 #include <sys/socket.h>
39 #include "got_error.h"
40 #include "got_object.h"
41 #include "got_repository.h"
42 #include "got_opentemp.h"
45 #include "got_lib_sha1.h"
46 #include "got_lib_delta.h"
47 #include "got_lib_inflate.h"
48 #include "got_lib_object.h"
49 #include "got_lib_object_parse.h"
50 #include "got_lib_object_cache.h"
51 #include "got_lib_pack.h"
52 #include "got_lib_privsep.h"
53 #include "got_lib_repository.h"
56 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
59 struct got_object_id *
60 got_object_id_dup(struct got_object_id *id1)
62 struct got_object_id *id2;
64 id2 = malloc(sizeof(*id2));
67 memcpy(id2, id1, sizeof(*id2));
72 got_object_id_cmp(const struct got_object_id *id1,
73 const struct got_object_id *id2)
75 return memcmp(id1->sha1, id2->sha1, SHA1_DIGEST_LENGTH);
78 const struct got_error *
79 got_object_qid_alloc_partial(struct got_object_qid **qid)
81 *qid = malloc(sizeof(**qid));
83 return got_error_from_errno("malloc");
89 const struct got_error *
90 got_object_id_str(char **outbuf, struct got_object_id *id)
92 static const size_t len = SHA1_DIGEST_STRING_LENGTH;
94 *outbuf = malloc(len);
96 return got_error_from_errno("malloc");
98 if (got_sha1_digest_to_str(id->sha1, *outbuf, len) == NULL) {
101 return got_error(GOT_ERR_BAD_OBJ_ID_STR);
108 got_object_close(struct got_object *obj)
110 if (obj->refcnt > 0) {
116 if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
117 struct got_delta *delta;
118 while (!STAILQ_EMPTY(&obj->deltas.entries)) {
119 delta = STAILQ_FIRST(&obj->deltas.entries);
120 STAILQ_REMOVE_HEAD(&obj->deltas.entries, entry);
127 const struct got_error *
128 got_object_raw_close(struct got_raw_object *obj)
130 const struct got_error *err = NULL;
132 if (obj->refcnt > 0) {
138 if (obj->f == NULL) {
140 if (munmap(obj->data, obj->hdrlen + obj->size) == -1)
141 err = got_error_from_errno("munmap");
142 if (close(obj->fd) == -1 && err == NULL)
143 err = got_error_from_errno("close");
147 if (fclose(obj->f) == EOF && err == NULL)
148 err = got_error_from_errno("fclose");
155 got_object_qid_free(struct got_object_qid *qid)
161 got_object_id_queue_free(struct got_object_id_queue *ids)
163 struct got_object_qid *qid;
165 while (!STAILQ_EMPTY(ids)) {
166 qid = STAILQ_FIRST(ids);
167 STAILQ_REMOVE_HEAD(ids, entry);
168 got_object_qid_free(qid);
172 const struct got_error *
173 got_object_parse_header(struct got_object **obj, char *buf, size_t len)
175 const char *obj_labels[] = {
176 GOT_OBJ_LABEL_COMMIT,
181 const int obj_types[] = {
194 end = memchr(buf, '\0', len);
196 return got_error(GOT_ERR_BAD_OBJ_HDR);
198 for (i = 0; i < nitems(obj_labels); i++) {
199 const char *label = obj_labels[i];
200 size_t label_len = strlen(label);
203 if (len <= label_len || buf + label_len >= end ||
204 strncmp(buf, label, label_len) != 0)
208 size = strtonum(buf + label_len, 0, LONG_MAX, &errstr);
210 return got_error(GOT_ERR_BAD_OBJ_HDR);
215 return got_error(GOT_ERR_BAD_OBJ_HDR);
217 *obj = calloc(1, sizeof(**obj));
219 return got_error_from_errno("calloc");
221 (*obj)->hdrlen = end - buf + 1;
226 const struct got_error *
227 got_object_read_header(struct got_object **obj, int fd)
229 const struct got_error *err;
230 struct got_inflate_buf zb;
232 const size_t zbsize = 64;
233 size_t outlen, totlen;
238 buf = malloc(zbsize);
240 return got_error_from_errno("malloc");
243 err = got_inflate_init(&zb, buf, zbsize, NULL);
249 err = got_inflate_read_fd(&zb, fd, &outlen, NULL);
255 if (memchr(zb.outbuf, '\0', outlen) == NULL) {
258 newbuf = recallocarray(buf, nbuf - 1, nbuf, zbsize);
259 if (newbuf == NULL) {
260 err = got_error_from_errno("recallocarray");
264 zb.outbuf = newbuf + totlen;
265 zb.outlen = (nbuf * zbsize) - totlen;
267 } while (memchr(zb.outbuf, '\0', outlen) == NULL);
269 err = got_object_parse_header(obj, buf, totlen);
272 got_inflate_end(&zb);
276 struct got_commit_object *
277 got_object_commit_alloc_partial(void)
279 struct got_commit_object *commit;
281 commit = calloc(1, sizeof(*commit));
284 commit->tree_id = malloc(sizeof(*commit->tree_id));
285 if (commit->tree_id == NULL) {
290 STAILQ_INIT(&commit->parent_ids);
295 const struct got_error *
296 got_object_commit_add_parent(struct got_commit_object *commit,
299 const struct got_error *err = NULL;
300 struct got_object_qid *qid;
302 err = got_object_qid_alloc_partial(&qid);
306 if (!got_parse_sha1_digest(qid->id.sha1, id_str)) {
307 err = got_error(GOT_ERR_BAD_OBJ_DATA);
308 got_object_qid_free(qid);
312 STAILQ_INSERT_TAIL(&commit->parent_ids, qid, entry);
318 static const struct got_error *
319 parse_gmtoff(time_t *gmtoff, const char *tzstr)
322 const char *p = tzstr;
330 return got_error(GOT_ERR_BAD_OBJ_DATA);
332 if (!isdigit(*p) && !isdigit(*(p + 1)))
333 return got_error(GOT_ERR_BAD_OBJ_DATA);
334 h = (((*p - '0') * 10) + (*(p + 1) - '0'));
337 if (!isdigit(*p) && !isdigit(*(p + 1)))
338 return got_error(GOT_ERR_BAD_OBJ_DATA);
339 m = ((*p - '0') * 10) + (*(p + 1) - '0');
341 *gmtoff = (h * 60 * 60 + m * 60) * sign;
345 static const struct got_error *
346 parse_commit_time(time_t *time, time_t *gmtoff, char *committer)
348 const struct got_error *err = NULL;
352 /* Parse and strip off trailing timezone indicator string. */
353 space = strrchr(committer, ' ');
355 return got_error(GOT_ERR_BAD_OBJ_DATA);
356 tzstr = strdup(space + 1);
358 return got_error_from_errno("strdup");
359 err = parse_gmtoff(gmtoff, tzstr);
362 if (err->code != GOT_ERR_BAD_OBJ_DATA)
364 /* Old versions of Git omitted the timestamp. */
371 /* Timestamp is separated from committer name + email by space. */
372 space = strrchr(committer, ' ');
374 return got_error(GOT_ERR_BAD_OBJ_DATA);
376 /* Timestamp parsed here is expressed as UNIX timestamp (UTC). */
377 *time = strtonum(space + 1, 0, INT64_MAX, &errstr);
379 return got_error(GOT_ERR_BAD_OBJ_DATA);
381 /* Strip off parsed time information, leaving just author and email. */
388 got_object_commit_close(struct got_commit_object *commit)
390 if (commit->refcnt > 0) {
392 if (commit->refcnt > 0)
396 got_object_id_queue_free(&commit->parent_ids);
397 free(commit->tree_id);
398 free(commit->author);
399 free(commit->committer);
400 free(commit->logmsg);
404 struct got_object_id *
405 got_object_commit_get_tree_id(struct got_commit_object *commit)
407 return commit->tree_id;
411 got_object_commit_get_nparents(struct got_commit_object *commit)
413 return commit->nparents;
416 const struct got_object_id_queue *
417 got_object_commit_get_parent_ids(struct got_commit_object *commit)
419 return &commit->parent_ids;
423 got_object_commit_get_author(struct got_commit_object *commit)
425 return commit->author;
429 got_object_commit_get_author_time(struct got_commit_object *commit)
431 return commit->author_time;
434 time_t got_object_commit_get_author_gmtoff(struct got_commit_object *commit)
436 return commit->author_gmtoff;
440 got_object_commit_get_committer(struct got_commit_object *commit)
442 return commit->committer;
446 got_object_commit_get_committer_time(struct got_commit_object *commit)
448 return commit->committer_time;
452 got_object_commit_get_committer_gmtoff(struct got_commit_object *commit)
454 return commit->committer_gmtoff;
457 const struct got_error *
458 got_object_commit_get_logmsg(char **logmsg, struct got_commit_object *commit)
460 const struct got_error *err = NULL;
465 len = strlen(commit->logmsg);
466 *logmsg = malloc(len + 2); /* leave room for a trailing \n and \0 */
468 return got_error_from_errno("malloc");
471 * Strip out unusual headers. Headers are separated from the commit
472 * message body by a single empty line.
474 src = commit->logmsg;
476 while (*src != '\0' && *src != '\n') {
477 int copy_header = 1, eol = 0;
478 if (strncmp(src, GOT_COMMIT_LABEL_TREE,
479 strlen(GOT_COMMIT_LABEL_TREE)) != 0 &&
480 strncmp(src, GOT_COMMIT_LABEL_AUTHOR,
481 strlen(GOT_COMMIT_LABEL_AUTHOR)) != 0 &&
482 strncmp(src, GOT_COMMIT_LABEL_PARENT,
483 strlen(GOT_COMMIT_LABEL_PARENT)) != 0 &&
484 strncmp(src, GOT_COMMIT_LABEL_COMMITTER,
485 strlen(GOT_COMMIT_LABEL_COMMITTER)) != 0)
488 while (*src != '\0' && !eol) {
500 if (strlcat(*logmsg, src, len + 1) >= len + 1) {
501 err = got_error(GOT_ERR_NO_SPACE);
505 /* Trim redundant trailing whitespace. */
506 len = strlen(*logmsg);
507 while (len > 1 && isspace((unsigned char)(*logmsg)[len - 2]) &&
508 isspace((unsigned char)(*logmsg)[len - 1])) {
509 (*logmsg)[len - 1] = '\0';
513 /* Append a trailing newline if missing. */
514 if (len > 0 && (*logmsg)[len - 1] != '\n') {
515 (*logmsg)[len] = '\n';
516 (*logmsg)[len + 1] = '\0';
527 got_object_commit_get_logmsg_raw(struct got_commit_object *commit)
529 return commit->logmsg;
532 const struct got_error *
533 got_object_parse_commit(struct got_commit_object **commit, char *buf,
536 const struct got_error *err = NULL;
539 ssize_t remain = (ssize_t)len;
542 return got_error(GOT_ERR_BAD_OBJ_DATA);
544 *commit = got_object_commit_alloc_partial();
546 return got_error_from_errno("got_object_commit_alloc_partial");
548 label_len = strlen(GOT_COMMIT_LABEL_TREE);
549 if (strncmp(s, GOT_COMMIT_LABEL_TREE, label_len) == 0) {
551 if (remain < SHA1_DIGEST_STRING_LENGTH) {
552 err = got_error(GOT_ERR_BAD_OBJ_DATA);
556 if (!got_parse_sha1_digest((*commit)->tree_id->sha1, s)) {
557 err = got_error(GOT_ERR_BAD_OBJ_DATA);
560 remain -= SHA1_DIGEST_STRING_LENGTH;
561 s += SHA1_DIGEST_STRING_LENGTH;
563 err = got_error(GOT_ERR_BAD_OBJ_DATA);
567 label_len = strlen(GOT_COMMIT_LABEL_PARENT);
568 while (strncmp(s, GOT_COMMIT_LABEL_PARENT, label_len) == 0) {
570 if (remain < SHA1_DIGEST_STRING_LENGTH) {
571 err = got_error(GOT_ERR_BAD_OBJ_DATA);
575 err = got_object_commit_add_parent(*commit, s);
579 remain -= SHA1_DIGEST_STRING_LENGTH;
580 s += SHA1_DIGEST_STRING_LENGTH;
583 label_len = strlen(GOT_COMMIT_LABEL_AUTHOR);
584 if (strncmp(s, GOT_COMMIT_LABEL_AUTHOR, label_len) == 0) {
590 err = got_error(GOT_ERR_BAD_OBJ_DATA);
594 p = memchr(s, '\n', remain);
596 err = got_error(GOT_ERR_BAD_OBJ_DATA);
601 err = parse_commit_time(&(*commit)->author_time,
602 &(*commit)->author_gmtoff, s);
605 (*commit)->author = strdup(s);
606 if ((*commit)->author == NULL) {
607 err = got_error_from_errno("strdup");
614 label_len = strlen(GOT_COMMIT_LABEL_COMMITTER);
615 if (strncmp(s, GOT_COMMIT_LABEL_COMMITTER, label_len) == 0) {
621 err = got_error(GOT_ERR_BAD_OBJ_DATA);
625 p = memchr(s, '\n', remain);
627 err = got_error(GOT_ERR_BAD_OBJ_DATA);
632 err = parse_commit_time(&(*commit)->committer_time,
633 &(*commit)->committer_gmtoff, s);
636 (*commit)->committer = strdup(s);
637 if ((*commit)->committer == NULL) {
638 err = got_error_from_errno("strdup");
645 (*commit)->logmsg = strndup(s, remain);
646 if ((*commit)->logmsg == NULL) {
647 err = got_error_from_errno("strndup");
652 got_object_commit_close(*commit);
659 got_object_tree_close(struct got_tree_object *tree)
661 if (tree->refcnt > 0) {
663 if (tree->refcnt > 0)
671 static const struct got_error *
672 parse_tree_entry(struct got_parsed_tree_entry *pte, size_t *elen, char *buf,
679 *elen = strnlen(buf, maxlen) + 1;
681 return got_error(GOT_ERR_BAD_OBJ_DATA);
683 space = memchr(buf, ' ', *elen);
684 if (space == NULL || space <= buf)
685 return got_error(GOT_ERR_BAD_OBJ_DATA);
690 if (*p < '0' && *p > '7')
691 return got_error(GOT_ERR_BAD_OBJ_DATA);
693 pte->mode |= *p - '0';
697 if (*elen > maxlen || maxlen - *elen < SHA1_DIGEST_LENGTH)
698 return got_error(GOT_ERR_BAD_OBJ_DATA);
700 pte->name = space + 1;
701 pte->namelen = strlen(pte->name);
704 *elen += SHA1_DIGEST_LENGTH;
709 pte_cmp(const void *pa, const void *pb)
711 const struct got_parsed_tree_entry *a = pa, *b = pb;
713 return got_path_cmp(a->name, b->name, a->namelen, b->namelen);
716 const struct got_error *
717 got_object_parse_tree(struct got_parsed_tree_entry **entries, int *nentries,
718 uint8_t *buf, size_t len)
720 const struct got_error *err = NULL;
721 size_t remain = len, totalloc;
722 const size_t nalloc = 16;
723 struct got_parsed_tree_entry *pte;
728 return NULL; /* tree is empty */
730 *entries = calloc(nalloc, sizeof(**entries));
731 if (*entries == NULL)
732 return got_error_from_errno("calloc");
738 if (*nentries >= totalloc) {
739 pte = recallocarray(*entries, totalloc,
740 totalloc + nalloc, sizeof(**entries));
742 err = got_error_from_errno("recallocarray");
749 pte = &(*entries)[*nentries];
750 err = parse_tree_entry(pte, &elen, buf, remain);
759 err = got_error(GOT_ERR_BAD_OBJ_DATA);
764 mergesort(*entries, *nentries, sizeof(**entries), pte_cmp);
766 for (i = 0; i < *nentries - 1; i++) {
767 struct got_parsed_tree_entry *prev = &(*entries)[i];
768 pte = &(*entries)[i + 1];
769 if (got_path_cmp(prev->name, pte->name,
770 prev->namelen, pte->namelen) == 0) {
771 err = got_error(GOT_ERR_TREE_DUP_ENTRY);
786 got_object_tag_close(struct got_tag_object *tag)
788 if (tag->refcnt > 0) {
800 const struct got_error *
801 got_object_parse_tag(struct got_tag_object **tag, uint8_t *buf, size_t len)
803 const struct got_error *err = NULL;
809 return got_error(GOT_ERR_BAD_OBJ_DATA);
811 *tag = calloc(1, sizeof(**tag));
813 return got_error_from_errno("calloc");
815 label_len = strlen(GOT_TAG_LABEL_OBJECT);
816 if (strncmp(s, GOT_TAG_LABEL_OBJECT, label_len) == 0) {
818 if (remain < SHA1_DIGEST_STRING_LENGTH) {
819 err = got_error(GOT_ERR_BAD_OBJ_DATA);
823 if (!got_parse_sha1_digest((*tag)->id.sha1, s)) {
824 err = got_error(GOT_ERR_BAD_OBJ_DATA);
827 remain -= SHA1_DIGEST_STRING_LENGTH;
828 s += SHA1_DIGEST_STRING_LENGTH;
830 err = got_error(GOT_ERR_BAD_OBJ_DATA);
835 err = got_error(GOT_ERR_BAD_OBJ_DATA);
839 label_len = strlen(GOT_TAG_LABEL_TYPE);
840 if (strncmp(s, GOT_TAG_LABEL_TYPE, label_len) == 0) {
843 err = got_error(GOT_ERR_BAD_OBJ_DATA);
847 if (strncmp(s, GOT_OBJ_LABEL_COMMIT,
848 strlen(GOT_OBJ_LABEL_COMMIT)) == 0) {
849 (*tag)->obj_type = GOT_OBJ_TYPE_COMMIT;
850 label_len = strlen(GOT_OBJ_LABEL_COMMIT);
853 } else if (strncmp(s, GOT_OBJ_LABEL_TREE,
854 strlen(GOT_OBJ_LABEL_TREE)) == 0) {
855 (*tag)->obj_type = GOT_OBJ_TYPE_TREE;
856 label_len = strlen(GOT_OBJ_LABEL_TREE);
859 } else if (strncmp(s, GOT_OBJ_LABEL_BLOB,
860 strlen(GOT_OBJ_LABEL_BLOB)) == 0) {
861 (*tag)->obj_type = GOT_OBJ_TYPE_BLOB;
862 label_len = strlen(GOT_OBJ_LABEL_BLOB);
865 } else if (strncmp(s, GOT_OBJ_LABEL_TAG,
866 strlen(GOT_OBJ_LABEL_TAG)) == 0) {
867 (*tag)->obj_type = GOT_OBJ_TYPE_TAG;
868 label_len = strlen(GOT_OBJ_LABEL_TAG);
872 err = got_error(GOT_ERR_BAD_OBJ_DATA);
876 if (remain <= 0 || *s != '\n') {
877 err = got_error(GOT_ERR_BAD_OBJ_DATA);
883 err = got_error(GOT_ERR_BAD_OBJ_DATA);
887 err = got_error(GOT_ERR_BAD_OBJ_DATA);
891 label_len = strlen(GOT_TAG_LABEL_TAG);
892 if (strncmp(s, GOT_TAG_LABEL_TAG, label_len) == 0) {
897 err = got_error(GOT_ERR_BAD_OBJ_DATA);
901 p = memchr(s, '\n', remain);
903 err = got_error(GOT_ERR_BAD_OBJ_DATA);
908 (*tag)->tag = strndup(s, slen);
909 if ((*tag)->tag == NULL) {
910 err = got_error_from_errno("strndup");
916 err = got_error(GOT_ERR_BAD_OBJ_DATA);
920 err = got_error(GOT_ERR_BAD_OBJ_DATA);
924 label_len = strlen(GOT_TAG_LABEL_TAGGER);
925 if (strncmp(s, GOT_TAG_LABEL_TAGGER, label_len) == 0) {
931 err = got_error(GOT_ERR_BAD_OBJ_DATA);
935 p = memchr(s, '\n', remain);
937 err = got_error(GOT_ERR_BAD_OBJ_DATA);
942 err = parse_commit_time(&(*tag)->tagger_time,
943 &(*tag)->tagger_gmtoff, s);
946 (*tag)->tagger = strdup(s);
947 if ((*tag)->tagger == NULL) {
948 err = got_error_from_errno("strdup");
954 err = got_error(GOT_ERR_BAD_OBJ_DATA);
958 /* Some old tags in the Linux git repo have no tagger. */
959 (*tag)->tagger = strdup("");
960 if ((*tag)->tagger == NULL) {
961 err = got_error_from_errno("strdup");
966 (*tag)->tagmsg = strndup(s, remain);
967 if ((*tag)->tagmsg == NULL) {
968 err = got_error_from_errno("strndup");
973 got_object_tag_close(*tag);
979 const struct got_error *
980 got_read_file_to_mem(uint8_t **outbuf, size_t *outlen, FILE *f)
982 const struct got_error *err = NULL;
983 static const size_t blocksize = 512;
984 size_t n, total, remain;
990 buf = malloc(blocksize);
992 return got_error_from_errno("malloc");
999 newbuf = reallocarray(buf, 1, total + blocksize);
1000 if (newbuf == NULL) {
1001 err = got_error_from_errno("reallocarray");
1005 remain += blocksize;
1007 n = fread(buf + total, 1, remain, f);
1010 err = got_ferror(f, GOT_ERR_IO);