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>
38 #include "got_error.h"
39 #include "got_object.h"
40 #include "got_repository.h"
41 #include "got_opentemp.h"
44 #include "got_lib_sha1.h"
45 #include "got_lib_delta.h"
46 #include "got_lib_inflate.h"
47 #include "got_lib_object.h"
48 #include "got_lib_object_parse.h"
49 #include "got_lib_object_cache.h"
50 #include "got_lib_pack.h"
51 #include "got_lib_privsep.h"
52 #include "got_lib_repository.h"
55 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
58 struct got_object_id *
59 got_object_id_dup(struct got_object_id *id1)
61 struct got_object_id *id2;
63 id2 = malloc(sizeof(*id2));
66 memcpy(id2, id1, sizeof(*id2));
71 got_object_id_cmp(const struct got_object_id *id1,
72 const struct got_object_id *id2)
74 return memcmp(id1->sha1, id2->sha1, SHA1_DIGEST_LENGTH);
77 const struct got_error *
78 got_object_qid_alloc_partial(struct got_object_qid **qid)
80 const struct got_error *err = NULL;
82 *qid = malloc(sizeof(**qid));
84 return got_error_from_errno("malloc");
86 (*qid)->id = malloc(sizeof(*((*qid)->id)));
87 if ((*qid)->id == NULL) {
88 err = got_error_from_errno("malloc");
89 got_object_qid_free(*qid);
98 const struct got_error *
99 got_object_id_str(char **outbuf, struct got_object_id *id)
101 static const size_t len = SHA1_DIGEST_STRING_LENGTH;
103 *outbuf = malloc(len);
105 return got_error_from_errno("malloc");
107 if (got_sha1_digest_to_str(id->sha1, *outbuf, len) == NULL) {
110 return got_error(GOT_ERR_BAD_OBJ_ID_STR);
117 got_object_close(struct got_object *obj)
119 if (obj->refcnt > 0) {
125 if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
126 struct got_delta *delta;
127 while (!STAILQ_EMPTY(&obj->deltas.entries)) {
128 delta = STAILQ_FIRST(&obj->deltas.entries);
129 STAILQ_REMOVE_HEAD(&obj->deltas.entries, entry);
136 const struct got_error *
137 got_object_raw_close(struct got_raw_object *obj)
139 const struct got_error *err = NULL;
141 if (obj->refcnt > 0) {
148 if (obj->f != NULL && fclose(obj->f) == EOF && err == NULL)
149 err = got_error_from_errno("fclose");
156 got_object_qid_free(struct got_object_qid *qid)
163 got_object_id_queue_free(struct got_object_id_queue *ids)
165 struct got_object_qid *qid;
167 while (!STAILQ_EMPTY(ids)) {
168 qid = STAILQ_FIRST(ids);
169 STAILQ_REMOVE_HEAD(ids, entry);
170 got_object_qid_free(qid);
174 const struct got_error *
175 got_object_parse_header(struct got_object **obj, char *buf, size_t len)
177 const char *obj_labels[] = {
178 GOT_OBJ_LABEL_COMMIT,
183 const int obj_types[] = {
190 size_t size = 0, hdrlen = 0;
195 hdrlen = strnlen(buf, len) + 1 /* '\0' */;
197 return got_error(GOT_ERR_BAD_OBJ_HDR);
199 for (i = 0; i < nitems(obj_labels); i++) {
200 const char *label = obj_labels[i];
201 size_t label_len = strlen(label);
204 if (strncmp(buf, label, label_len) != 0)
208 if (len <= label_len)
209 return got_error(GOT_ERR_BAD_OBJ_HDR);
210 size = strtonum(buf + label_len, 0, LONG_MAX, &errstr);
212 return got_error(GOT_ERR_BAD_OBJ_HDR);
217 return got_error(GOT_ERR_BAD_OBJ_HDR);
219 *obj = calloc(1, sizeof(**obj));
221 return got_error_from_errno("calloc");
223 (*obj)->hdrlen = hdrlen;
228 const struct got_error *
229 got_object_read_header(struct got_object **obj, int fd)
231 const struct got_error *err;
232 struct got_inflate_buf zb;
234 const size_t zbsize = 64;
235 size_t outlen, totlen;
240 buf = malloc(zbsize);
242 return got_error_from_errno("malloc");
244 err = got_inflate_init(&zb, buf, zbsize, NULL);
250 err = got_inflate_read_fd(&zb, fd, &outlen, NULL);
256 if (memchr(zb.outbuf, '\0', outlen) == NULL) {
259 newbuf = recallocarray(buf, nbuf - 1, nbuf, zbsize);
260 if (newbuf == NULL) {
261 err = got_error_from_errno("recallocarray");
265 zb.outbuf = newbuf + totlen;
266 zb.outlen = (nbuf * zbsize) - totlen;
268 } while (memchr(zb.outbuf, '\0', outlen) == NULL);
270 err = got_object_parse_header(obj, buf, totlen);
273 got_inflate_end(&zb);
277 struct got_commit_object *
278 got_object_commit_alloc_partial(void)
280 struct got_commit_object *commit;
282 commit = calloc(1, sizeof(*commit));
285 commit->tree_id = malloc(sizeof(*commit->tree_id));
286 if (commit->tree_id == NULL) {
291 STAILQ_INIT(&commit->parent_ids);
296 const struct got_error *
297 got_object_commit_add_parent(struct got_commit_object *commit,
300 const struct got_error *err = NULL;
301 struct got_object_qid *qid;
303 err = got_object_qid_alloc_partial(&qid);
307 if (!got_parse_sha1_digest(qid->id->sha1, id_str)) {
308 err = got_error(GOT_ERR_BAD_OBJ_DATA);
309 got_object_qid_free(qid);
313 STAILQ_INSERT_TAIL(&commit->parent_ids, qid, entry);
319 static const struct got_error *
320 parse_gmtoff(time_t *gmtoff, const char *tzstr)
323 const char *p = tzstr;
331 return got_error(GOT_ERR_BAD_OBJ_DATA);
333 if (!isdigit(*p) && !isdigit(*(p + 1)))
334 return got_error(GOT_ERR_BAD_OBJ_DATA);
335 h = (((*p - '0') * 10) + (*(p + 1) - '0'));
338 if (!isdigit(*p) && !isdigit(*(p + 1)))
339 return got_error(GOT_ERR_BAD_OBJ_DATA);
340 m = ((*p - '0') * 10) + (*(p + 1) - '0');
342 *gmtoff = (h * 60 * 60 + m * 60) * sign;
346 static const struct got_error *
347 parse_commit_time(time_t *time, time_t *gmtoff, char *committer)
349 const struct got_error *err = NULL;
353 /* Parse and strip off trailing timezone indicator string. */
354 space = strrchr(committer, ' ');
356 return got_error(GOT_ERR_BAD_OBJ_DATA);
357 tzstr = strdup(space + 1);
359 return got_error_from_errno("strdup");
360 err = parse_gmtoff(gmtoff, tzstr);
363 if (err->code != GOT_ERR_BAD_OBJ_DATA)
365 /* Old versions of Git omitted the timestamp. */
372 /* Timestamp is separated from committer name + email by space. */
373 space = strrchr(committer, ' ');
375 return got_error(GOT_ERR_BAD_OBJ_DATA);
377 /* Timestamp parsed here is expressed as UNIX timestamp (UTC). */
378 *time = strtonum(space + 1, 0, INT64_MAX, &errstr);
380 return got_error(GOT_ERR_BAD_OBJ_DATA);
382 /* Strip off parsed time information, leaving just author and email. */
389 got_object_commit_close(struct got_commit_object *commit)
391 if (commit->refcnt > 0) {
393 if (commit->refcnt > 0)
397 got_object_id_queue_free(&commit->parent_ids);
398 free(commit->tree_id);
399 free(commit->author);
400 free(commit->committer);
401 free(commit->logmsg);
405 struct got_object_id *
406 got_object_commit_get_tree_id(struct got_commit_object *commit)
408 return commit->tree_id;
412 got_object_commit_get_nparents(struct got_commit_object *commit)
414 return commit->nparents;
417 const struct got_object_id_queue *
418 got_object_commit_get_parent_ids(struct got_commit_object *commit)
420 return &commit->parent_ids;
424 got_object_commit_get_author(struct got_commit_object *commit)
426 return commit->author;
430 got_object_commit_get_author_time(struct got_commit_object *commit)
432 return commit->author_time;
435 time_t got_object_commit_get_author_gmtoff(struct got_commit_object *commit)
437 return commit->author_gmtoff;
441 got_object_commit_get_committer(struct got_commit_object *commit)
443 return commit->committer;
447 got_object_commit_get_committer_time(struct got_commit_object *commit)
449 return commit->committer_time;
453 got_object_commit_get_committer_gmtoff(struct got_commit_object *commit)
455 return commit->committer_gmtoff;
458 const struct got_error *
459 got_object_commit_get_logmsg(char **logmsg, struct got_commit_object *commit)
461 const struct got_error *err = NULL;
466 len = strlen(commit->logmsg);
467 *logmsg = malloc(len + 2); /* leave room for a trailing \n and \0 */
469 return got_error_from_errno("malloc");
472 * Strip out unusual headers. Headers are separated from the commit
473 * message body by a single empty line.
475 src = commit->logmsg;
477 while (*src != '\0' && *src != '\n') {
478 int copy_header = 1, eol = 0;
479 if (strncmp(src, GOT_COMMIT_LABEL_TREE,
480 strlen(GOT_COMMIT_LABEL_TREE)) != 0 &&
481 strncmp(src, GOT_COMMIT_LABEL_AUTHOR,
482 strlen(GOT_COMMIT_LABEL_AUTHOR)) != 0 &&
483 strncmp(src, GOT_COMMIT_LABEL_PARENT,
484 strlen(GOT_COMMIT_LABEL_PARENT)) != 0 &&
485 strncmp(src, GOT_COMMIT_LABEL_COMMITTER,
486 strlen(GOT_COMMIT_LABEL_COMMITTER)) != 0)
489 while (*src != '\0' && !eol) {
501 if (strlcat(*logmsg, src, len + 1) >= len + 1) {
502 err = got_error(GOT_ERR_NO_SPACE);
506 /* Trim redundant trailing whitespace. */
507 len = strlen(*logmsg);
508 while (len > 1 && isspace((unsigned char)(*logmsg)[len - 2]) &&
509 isspace((unsigned char)(*logmsg)[len - 1])) {
510 (*logmsg)[len - 1] = '\0';
514 /* Append a trailing newline if missing. */
515 if (len > 0 && (*logmsg)[len - 1] != '\n') {
516 (*logmsg)[len] = '\n';
517 (*logmsg)[len + 1] = '\0';
528 got_object_commit_get_logmsg_raw(struct got_commit_object *commit)
530 return commit->logmsg;
533 const struct got_error *
534 got_object_parse_commit(struct got_commit_object **commit, char *buf,
537 const struct got_error *err = NULL;
540 ssize_t remain = (ssize_t)len;
543 return got_error(GOT_ERR_BAD_OBJ_DATA);
545 *commit = got_object_commit_alloc_partial();
547 return got_error_from_errno("got_object_commit_alloc_partial");
549 label_len = strlen(GOT_COMMIT_LABEL_TREE);
550 if (strncmp(s, GOT_COMMIT_LABEL_TREE, label_len) == 0) {
552 if (remain < SHA1_DIGEST_STRING_LENGTH) {
553 err = got_error(GOT_ERR_BAD_OBJ_DATA);
557 if (!got_parse_sha1_digest((*commit)->tree_id->sha1, s)) {
558 err = got_error(GOT_ERR_BAD_OBJ_DATA);
561 remain -= SHA1_DIGEST_STRING_LENGTH;
562 s += SHA1_DIGEST_STRING_LENGTH;
564 err = got_error(GOT_ERR_BAD_OBJ_DATA);
568 label_len = strlen(GOT_COMMIT_LABEL_PARENT);
569 while (strncmp(s, GOT_COMMIT_LABEL_PARENT, label_len) == 0) {
571 if (remain < SHA1_DIGEST_STRING_LENGTH) {
572 err = got_error(GOT_ERR_BAD_OBJ_DATA);
576 err = got_object_commit_add_parent(*commit, s);
580 remain -= SHA1_DIGEST_STRING_LENGTH;
581 s += SHA1_DIGEST_STRING_LENGTH;
584 label_len = strlen(GOT_COMMIT_LABEL_AUTHOR);
585 if (strncmp(s, GOT_COMMIT_LABEL_AUTHOR, label_len) == 0) {
591 err = got_error(GOT_ERR_BAD_OBJ_DATA);
595 p = memchr(s, '\n', remain);
597 err = got_error(GOT_ERR_BAD_OBJ_DATA);
602 err = parse_commit_time(&(*commit)->author_time,
603 &(*commit)->author_gmtoff, s);
606 (*commit)->author = strdup(s);
607 if ((*commit)->author == NULL) {
608 err = got_error_from_errno("strdup");
615 label_len = strlen(GOT_COMMIT_LABEL_COMMITTER);
616 if (strncmp(s, GOT_COMMIT_LABEL_COMMITTER, label_len) == 0) {
622 err = got_error(GOT_ERR_BAD_OBJ_DATA);
626 p = memchr(s, '\n', remain);
628 err = got_error(GOT_ERR_BAD_OBJ_DATA);
633 err = parse_commit_time(&(*commit)->committer_time,
634 &(*commit)->committer_gmtoff, s);
637 (*commit)->committer = strdup(s);
638 if ((*commit)->committer == NULL) {
639 err = got_error_from_errno("strdup");
646 (*commit)->logmsg = strndup(s, remain);
647 if ((*commit)->logmsg == NULL) {
648 err = got_error_from_errno("strndup");
653 got_object_commit_close(*commit);
660 got_object_tree_close(struct got_tree_object *tree)
662 if (tree->refcnt > 0) {
664 if (tree->refcnt > 0)
672 static const struct got_error *
673 parse_tree_entry(struct got_parsed_tree_entry **pte, const char **name,
674 size_t *elen, char *buf,
678 const struct got_error *err = NULL;
683 *pte = malloc(sizeof(**pte));
685 return got_error_from_errno("malloc");
687 *elen = strnlen(buf, maxlen) + 1;
688 if (*elen > maxlen) {
691 return got_error(GOT_ERR_BAD_OBJ_DATA);
694 space = memchr(buf, ' ', *elen);
695 if (space == NULL || space <= buf) {
696 err = got_error(GOT_ERR_BAD_OBJ_DATA);
704 if (*p < '0' && *p > '7') {
705 err = got_error(GOT_ERR_BAD_OBJ_DATA);
709 (*pte)->mode |= *p - '0';
713 if (*elen > maxlen || maxlen - *elen < SHA1_DIGEST_LENGTH) {
714 err = got_error(GOT_ERR_BAD_OBJ_DATA);
720 *elen += SHA1_DIGEST_LENGTH;
729 const struct got_error *
730 got_object_parse_tree(struct got_pathlist_head *entries, int *nentries,
731 uint8_t *buf, size_t len)
733 const struct got_error *err = NULL;
738 return NULL; /* tree is empty */
741 struct got_parsed_tree_entry *pte;
742 struct got_pathlist_entry *new = NULL;
746 err = parse_tree_entry(&pte, &name, &elen, buf, remain);
749 err = got_pathlist_insert(&new, entries, name, pte);
753 err = got_error(GOT_ERR_TREE_DUP_ENTRY);
762 err = got_error(GOT_ERR_BAD_OBJ_DATA);
767 got_object_parsed_tree_entries_free(entries);
774 got_object_parsed_tree_entries_free(struct got_pathlist_head *entries)
776 struct got_pathlist_entry *pe;
778 TAILQ_FOREACH(pe, entries, entry) {
779 struct got_parsed_tree_entry *pte = pe->data;
782 got_pathlist_free(entries);
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);