2 * Copyright (c) 2018, 2019 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>
36 #include "got_error.h"
37 #include "got_object.h"
38 #include "got_repository.h"
39 #include "got_opentemp.h"
42 #include "got_lib_sha1.h"
43 #include "got_lib_delta.h"
44 #include "got_lib_inflate.h"
45 #include "got_lib_object.h"
46 #include "got_lib_object_idcache.h"
47 #include "got_lib_object_cache.h"
48 #include "got_lib_object_parse.h"
49 #include "got_lib_pack.h"
50 #include "got_lib_repository.h"
53 #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
57 #define nitems(_a) (sizeof((_a)) / sizeof((_a)[0]))
60 struct got_object_id *
61 got_object_get_id(struct got_object *obj)
66 const struct got_error *
67 got_object_get_id_str(char **outbuf, struct got_object *obj)
69 return got_object_id_str(outbuf, &obj->id);
72 const struct got_error *
73 got_object_get_type(int *type, struct got_repository *repo,
74 struct got_object_id *id)
76 const struct got_error *err = NULL;
77 struct got_object *obj;
79 err = got_object_open(&obj, repo, id);
84 case GOT_OBJ_TYPE_COMMIT:
85 case GOT_OBJ_TYPE_TREE:
86 case GOT_OBJ_TYPE_BLOB:
87 case GOT_OBJ_TYPE_TAG:
91 err = got_error(GOT_ERR_OBJ_TYPE);
95 got_object_close(obj);
99 const struct got_error *
100 got_object_get_path(char **path, struct got_object_id *id,
101 struct got_repository *repo)
103 const struct got_error *err = NULL;
109 path_objects = got_repo_get_path_objects(repo);
110 if (path_objects == NULL)
111 return got_error_from_errno("got_repo_get_path_objects");
113 err = got_object_id_str(&hex, id);
117 if (asprintf(path, "%s/%.2x/%s", path_objects,
118 id->sha1[0], hex + 2) == -1)
119 err = got_error_from_errno("asprintf");
127 const struct got_error *
128 got_object_open_loose_fd(int *fd, struct got_object_id *id,
129 struct got_repository *repo)
131 const struct got_error *err = NULL;
134 err = got_object_get_path(&path, id, repo);
137 *fd = open(path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
139 err = got_error_from_errno2("open", path);
147 const struct got_error *
148 got_object_open_by_id_str(struct got_object **obj, struct got_repository *repo,
151 struct got_object_id id;
153 if (!got_parse_sha1_digest(id.sha1, id_str))
154 return got_error_path(id_str, GOT_ERR_BAD_OBJ_ID_STR);
156 return got_object_open(obj, repo, &id);
159 const struct got_error *
160 got_object_resolve_id_str(struct got_object_id **id,
161 struct got_repository *repo, const char *id_str)
163 const struct got_error *err = NULL;
164 struct got_object *obj;
166 err = got_object_open_by_id_str(&obj, repo, id_str);
170 *id = got_object_id_dup(got_object_get_id(obj));
171 got_object_close(obj);
173 return got_error_from_errno("got_object_id_dup");
178 const struct got_error *
179 got_object_qid_alloc(struct got_object_qid **qid, struct got_object_id *id)
181 *qid = calloc(1, sizeof(**qid));
183 return got_error_from_errno("calloc");
185 memcpy(&(*qid)->id, id, sizeof((*qid)->id));
189 const struct got_error *
190 got_object_id_queue_copy(const struct got_object_id_queue *src,
191 struct got_object_id_queue *dest)
193 const struct got_error *err;
194 struct got_object_qid *qid;
196 STAILQ_FOREACH(qid, src, entry) {
197 struct got_object_qid *new;
199 * Deep-copy the object ID only. Let the caller deal
200 * with setting up the new->data pointer if needed.
202 err = got_object_qid_alloc(&new, &qid->id);
204 got_object_id_queue_free(dest);
207 STAILQ_INSERT_TAIL(dest, new, entry);
214 got_object_tree_get_nentries(struct got_tree_object *tree)
216 return tree->nentries;
219 struct got_tree_entry *
220 got_object_tree_get_first_entry(struct got_tree_object *tree)
222 return got_object_tree_get_entry(tree, 0);
225 struct got_tree_entry *
226 got_object_tree_get_last_entry(struct got_tree_object *tree)
228 return got_object_tree_get_entry(tree, tree->nentries - 1);
231 struct got_tree_entry *
232 got_object_tree_get_entry(struct got_tree_object *tree, int i)
234 if (i < 0 || i >= tree->nentries)
236 return &tree->entries[i];
240 got_tree_entry_get_mode(struct got_tree_entry *te)
246 got_tree_entry_get_name(struct got_tree_entry *te)
251 struct got_object_id *
252 got_tree_entry_get_id(struct got_tree_entry *te)
257 const struct got_error *
258 got_object_blob_read_to_str(char **s, struct got_blob_object *blob)
260 const struct got_error *err = NULL;
261 size_t len, totlen, hdrlen, offset;
265 hdrlen = got_object_blob_get_hdrlen(blob);
271 err = got_object_blob_read_block(&len, blob);
278 totlen += len - hdrlen;
279 p = realloc(*s, totlen + 1);
281 err = got_error_from_errno("realloc");
287 /* Skip blob object header first time around. */
289 got_object_blob_get_read_buf(blob) + hdrlen, len - hdrlen);
298 const struct got_error *
299 got_tree_entry_get_symlink_target(char **link_target, struct got_tree_entry *te,
300 struct got_repository *repo)
302 const struct got_error *err = NULL;
303 struct got_blob_object *blob = NULL;
308 if (!got_object_tree_entry_is_symlink(te))
309 return got_error(GOT_ERR_TREE_ENTRY_TYPE);
311 fd = got_opentempfd();
313 err = got_error_from_errno("got_opentempfd");
317 err = got_object_open_as_blob(&blob, repo,
318 got_tree_entry_get_id(te), PATH_MAX, fd);
322 err = got_object_blob_read_to_str(link_target, blob);
324 if (fd != -1 && close(fd) == -1 && err == NULL)
325 err = got_error_from_errno("close");
327 got_object_blob_close(blob);
336 got_tree_entry_get_index(struct got_tree_entry *te)
341 struct got_tree_entry *
342 got_tree_entry_get_next(struct got_tree_object *tree,
343 struct got_tree_entry *te)
345 return got_object_tree_get_entry(tree, te->idx + 1);
348 struct got_tree_entry *
349 got_tree_entry_get_prev(struct got_tree_object *tree,
350 struct got_tree_entry *te)
352 return got_object_tree_get_entry(tree, te->idx - 1);
355 const struct got_error *
356 got_object_blob_close(struct got_blob_object *blob)
358 const struct got_error *err = NULL;
359 free(blob->read_buf);
360 if (blob->f && fclose(blob->f) == EOF)
361 err = got_error_from_errno("fclose");
368 got_object_blob_rewind(struct got_blob_object *blob)
375 got_object_blob_id_str(struct got_blob_object *blob, char *buf, size_t size)
377 return got_sha1_digest_to_str(blob->id.sha1, buf, size);
381 got_object_blob_get_hdrlen(struct got_blob_object *blob)
387 got_object_blob_get_read_buf(struct got_blob_object *blob)
389 return blob->read_buf;
392 const struct got_error *
393 got_object_blob_read_block(size_t *outlenp, struct got_blob_object *blob)
397 n = fread(blob->read_buf, 1, blob->blocksize, blob->f);
398 if (n == 0 && ferror(blob->f))
399 return got_ferror(blob->f, GOT_ERR_IO);
404 const struct got_error *
405 got_object_blob_is_binary(int *binary, struct got_blob_object *blob)
407 const struct got_error *err;
411 hdrlen = got_object_blob_get_hdrlen(blob);
413 if (fseeko(blob->f, hdrlen, SEEK_SET) == -1)
414 return got_error_from_errno("fseeko");
416 err = got_object_blob_read_block(&len, blob);
420 *binary = memchr(blob->read_buf, '\0', len) != NULL;
422 if (fseeko(blob->f, hdrlen, SEEK_SET) == -1)
423 return got_error_from_errno("fseeko");
427 const struct got_error *
428 got_object_blob_getline(char **line, ssize_t *linelen, size_t *linesize,
429 struct got_blob_object *blob)
431 *linelen = getline(line, linesize, blob->f);
432 if (*linelen == -1 && !feof(blob->f))
433 return got_error_from_errno("getline");
437 const struct got_error *
438 got_object_blob_dump_to_file(off_t *filesize, int *nlines,
439 off_t **line_offsets, FILE *outfile, struct got_blob_object *blob)
441 const struct got_error *err = NULL;
442 size_t n, len, hdrlen;
445 const int alloc_chunksz = 512;
447 off_t off = 0, total_len = 0;
450 *line_offsets = NULL;
456 hdrlen = got_object_blob_get_hdrlen(blob);
458 err = got_object_blob_read_block(&len, blob);
463 buf = got_object_blob_get_read_buf(blob);
466 if (line_offsets && *line_offsets == NULL) {
467 /* Have some data but perhaps no '\n'. */
469 nalloc = alloc_chunksz;
470 *line_offsets = calloc(nalloc,
471 sizeof(**line_offsets));
472 if (*line_offsets == NULL)
473 return got_error_from_errno("calloc");
475 /* Skip forward over end of first line. */
482 /* Scan '\n' offsets in remaining chunk of data. */
484 if (buf[i] != '\n') {
489 if (line_offsets && nalloc < *nlines) {
490 size_t n = *nlines + alloc_chunksz;
491 off_t *o = recallocarray(*line_offsets,
492 nalloc, n, sizeof(**line_offsets));
495 *line_offsets = NULL;
496 return got_error_from_errno(
503 off = total_len + i - hdrlen + 1;
504 (*line_offsets)[*nlines - 1] = off;
509 /* Skip blob object header first time around. */
510 n = fwrite(buf + hdrlen, 1, len - hdrlen, outfile);
511 if (n != len - hdrlen)
512 return got_ferror(outfile, GOT_ERR_IO);
513 total_len += len - hdrlen;
517 if (fflush(outfile) != 0)
518 return got_error_from_errno("fflush");
522 *filesize = total_len;
528 got_object_tag_get_name(struct got_tag_object *tag)
534 got_object_tag_get_object_type(struct got_tag_object *tag)
536 return tag->obj_type;
539 struct got_object_id *
540 got_object_tag_get_object_id(struct got_tag_object *tag)
546 got_object_tag_get_tagger_time(struct got_tag_object *tag)
548 return tag->tagger_time;
552 got_object_tag_get_tagger_gmtoff(struct got_tag_object *tag)
554 return tag->tagger_gmtoff;
558 got_object_tag_get_tagger(struct got_tag_object *tag)
564 got_object_tag_get_message(struct got_tag_object *tag)
569 static struct got_tree_entry *
570 find_entry_by_name(struct got_tree_object *tree, const char *name, size_t len)
574 /* Note that tree entries are sorted in strncmp() order. */
575 for (i = 0; i < tree->nentries; i++) {
576 struct got_tree_entry *te = &tree->entries[i];
577 int cmp = strncmp(te->name, name, len);
582 if (te->name[len] == '\0')
588 struct got_tree_entry *
589 got_object_tree_find_entry(struct got_tree_object *tree, const char *name)
591 return find_entry_by_name(tree, name, strlen(name));
594 const struct got_error *
595 got_object_tree_find_path(struct got_object_id **id, mode_t *mode,
596 struct got_repository *repo, struct got_tree_object *tree,
599 const struct got_error *err = NULL;
600 struct got_tree_object *subtree = NULL;
601 struct got_tree_entry *te = NULL;
614 struct got_tree_object *next_tree;
623 te = find_entry_by_name(subtree, seg, seglen);
625 err = got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
636 err = got_object_open_as_tree(&next_tree, repo,
642 got_object_tree_close(subtree);
648 *id = got_object_id_dup(&te->id);
650 return got_error_from_errno("got_object_id_dup");
654 err = got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
656 if (subtree && subtree != tree)
657 got_object_tree_close(subtree);
661 const struct got_error *
662 got_object_id_by_path(struct got_object_id **id, struct got_repository *repo,
663 struct got_commit_object *commit, const char *path)
665 const struct got_error *err = NULL;
666 struct got_tree_object *tree = NULL;
670 /* Handle opening of root of commit's tree. */
671 if (got_path_is_root_dir(path)) {
672 *id = got_object_id_dup(commit->tree_id);
674 err = got_error_from_errno("got_object_id_dup");
676 err = got_object_open_as_tree(&tree, repo, commit->tree_id);
679 err = got_object_tree_find_path(id, NULL, repo, tree, path);
683 got_object_tree_close(tree);
688 * Normalize file mode bits to avoid false positive tree entry differences
689 * in case tree entries have unexpected mode bits set.
692 normalize_mode_for_comparison(mode_t mode)
695 * For directories, the only relevant bit is the IFDIR bit.
696 * This allows us to detect paths changing from a directory
697 * to a file and vice versa.
700 return mode & S_IFDIR;
703 * For symlinks, the only relevant bit is the IFLNK bit.
704 * This allows us to detect paths changing from a symlinks
705 * to a file or directory and vice versa.
708 return mode & S_IFLNK;
710 /* For files, the only change we care about is the executable bit. */
711 return mode & S_IXUSR;
714 const struct got_error *
715 got_object_tree_path_changed(int *changed,
716 struct got_tree_object *tree01, struct got_tree_object *tree02,
717 const char *path, struct got_repository *repo)
719 const struct got_error *err = NULL;
720 struct got_tree_object *tree1 = NULL, *tree2 = NULL;
721 struct got_tree_entry *te1 = NULL, *te2 = NULL;
727 /* We not do support comparing the root path. */
728 if (got_path_is_root_dir(path))
729 return got_error_path(path, GOT_ERR_BAD_PATH);
739 struct got_tree_object *next_tree1, *next_tree2;
749 te1 = find_entry_by_name(tree1, seg, seglen);
751 err = got_error(GOT_ERR_NO_OBJ);
756 te2 = find_entry_by_name(tree2, seg, seglen);
759 mode1 = normalize_mode_for_comparison(te1->mode);
760 mode2 = normalize_mode_for_comparison(te2->mode);
761 if (mode1 != mode2) {
766 if (got_object_id_cmp(&te1->id, &te2->id) == 0) {
772 if (*s == '\0') { /* final path element */
781 err = got_object_open_as_tree(&next_tree1, repo,
787 got_object_tree_close(tree1);
791 err = got_object_open_as_tree(&next_tree2, repo,
797 got_object_tree_close(tree2);
801 got_object_tree_close(tree2);
807 if (tree1 && tree1 != tree01)
808 got_object_tree_close(tree1);
809 if (tree2 && tree2 != tree02)
810 got_object_tree_close(tree2);
814 const struct got_error *
815 got_object_tree_entry_dup(struct got_tree_entry **new_te,
816 struct got_tree_entry *te)
818 const struct got_error *err = NULL;
820 *new_te = calloc(1, sizeof(**new_te));
822 return got_error_from_errno("calloc");
824 (*new_te)->mode = te->mode;
825 memcpy((*new_te)->name, te->name, sizeof((*new_te)->name));
826 memcpy(&(*new_te)->id, &te->id, sizeof((*new_te)->id));
831 got_object_tree_entry_is_submodule(struct got_tree_entry *te)
833 return (te->mode & S_IFMT) == (S_IFDIR | S_IFLNK);
837 got_object_tree_entry_is_symlink(struct got_tree_entry *te)
839 /* S_IFDIR check avoids confusing symlinks with submodules. */
840 return ((te->mode & (S_IFDIR | S_IFLNK)) == S_IFLNK);
843 static const struct got_error *
844 resolve_symlink(char **link_target, const char *path,
845 struct got_commit_object *commit, struct got_repository *repo)
847 const struct got_error *err = NULL;
849 char *name, *parent_path = NULL;
850 struct got_object_id *tree_obj_id = NULL;
851 struct got_tree_object *tree = NULL;
852 struct got_tree_entry *te = NULL;
856 if (strlcpy(buf, path, sizeof(buf)) >= sizeof(buf))
857 return got_error(GOT_ERR_NO_SPACE);
859 name = basename(buf);
861 return got_error_from_errno2("basename", path);
863 err = got_path_dirname(&parent_path, path);
867 err = got_object_id_by_path(&tree_obj_id, repo, commit,
870 if (err->code == GOT_ERR_NO_TREE_ENTRY) {
871 /* Display the complete path in error message. */
872 err = got_error_path(path, err->code);
877 err = got_object_open_as_tree(&tree, repo, tree_obj_id);
881 te = got_object_tree_find_entry(tree, name);
883 err = got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
887 if (got_object_tree_entry_is_symlink(te)) {
888 err = got_tree_entry_get_symlink_target(link_target, te, repo);
891 if (!got_path_is_absolute(*link_target)) {
893 if (asprintf(&abspath, "%s/%s", parent_path,
894 *link_target) == -1) {
895 err = got_error_from_errno("asprintf");
899 *link_target = malloc(PATH_MAX);
900 if (*link_target == NULL) {
901 err = got_error_from_errno("malloc");
904 err = got_canonpath(abspath, *link_target, PATH_MAX);
914 got_object_tree_close(tree);
922 const struct got_error *
923 got_object_resolve_symlinks(char **link_target, const char *path,
924 struct got_commit_object *commit, struct got_repository *repo)
926 const struct got_error *err = NULL;
927 char *next_target = NULL;
928 int max_recursion = 40; /* matches Git */
933 err = resolve_symlink(&next_target,
934 *link_target ? *link_target : path, commit, repo);
939 if (--max_recursion == 0) {
940 err = got_error_path(path, GOT_ERR_RECURSION);
944 *link_target = next_target;
946 } while (next_target);
952 got_object_commit_retain(struct got_commit_object *commit)
957 const struct got_error *
958 got_object_raw_alloc(struct got_raw_object **obj, uint8_t *outbuf, int *outfd,
959 size_t max_in_mem_size, size_t hdrlen, off_t size)
961 const struct got_error *err = NULL;
966 *obj = calloc(1, sizeof(**obj));
968 err = got_error_from_errno("calloc");
972 (*obj)->tempfile_idx = -1;
975 (*obj)->data = outbuf;
978 if (fstat(*outfd, &sb) == -1) {
979 err = got_error_from_errno("fstat");
983 if (sb.st_size != tot) {
984 err = got_error_msg(GOT_ERR_BAD_OBJ_HDR,
985 "raw object has unexpected size");
988 #ifndef GOT_PACK_NO_MMAP
989 if (tot > 0 && tot <= max_in_mem_size) {
990 (*obj)->data = mmap(NULL, tot, PROT_READ,
991 MAP_PRIVATE, *outfd, 0);
992 if ((*obj)->data == MAP_FAILED) {
993 if (errno != ENOMEM) {
994 err = got_error_from_errno("mmap");
1005 (*obj)->f = fdopen(*outfd, "r");
1006 if ((*obj)->f == NULL) {
1007 err = got_error_from_errno("fdopen");
1013 (*obj)->hdrlen = hdrlen;
1014 (*obj)->size = size;
1018 got_object_raw_close(*obj);