2 5d56da81 2019-01-13 stsp * Copyright (c) 2018, 2019 Stefan Sperling <stsp@openbsd.org>
4 d71d75ad 2017-11-05 stsp * Permission to use, copy, modify, and distribute this software for any
5 d71d75ad 2017-11-05 stsp * purpose with or without fee is hereby granted, provided that the above
6 d71d75ad 2017-11-05 stsp * copyright notice and this permission notice appear in all copies.
8 d71d75ad 2017-11-05 stsp * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 d71d75ad 2017-11-05 stsp * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 d71d75ad 2017-11-05 stsp * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 d71d75ad 2017-11-05 stsp * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 d71d75ad 2017-11-05 stsp * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 d71d75ad 2017-11-05 stsp * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 d71d75ad 2017-11-05 stsp * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 2178c42e 2018-04-22 stsp #include <sys/types.h>
18 0ffeb3c2 2017-11-26 stsp #include <sys/stat.h>
19 d1cda826 2017-11-06 stsp #include <sys/queue.h>
20 f8b19efd 2021-10-13 stsp #include <sys/tree.h>
21 2178c42e 2018-04-22 stsp #include <sys/uio.h>
22 64a8571e 2022-01-07 stsp #include <sys/mman.h>
24 a1fd68d8 2018-01-12 stsp #include <errno.h>
25 2178c42e 2018-04-22 stsp #include <fcntl.h>
26 d71d75ad 2017-11-05 stsp #include <stdio.h>
27 ab9a70b2 2017-11-06 stsp #include <stdlib.h>
28 ab9a70b2 2017-11-06 stsp #include <string.h>
29 d71d75ad 2017-11-05 stsp #include <sha1.h>
30 e83f12a6 2023-02-12 op #include <sha2.h>
31 81a12da5 2020-09-09 naddy #include <unistd.h>
32 ab9a70b2 2017-11-06 stsp #include <zlib.h>
33 e40622f4 2020-07-23 stsp #include <libgen.h>
34 ab9a70b2 2017-11-06 stsp #include <limits.h>
35 2178c42e 2018-04-22 stsp #include <imsg.h>
37 ab9a70b2 2017-11-06 stsp #include "got_error.h"
38 d71d75ad 2017-11-05 stsp #include "got_object.h"
39 ab9a70b2 2017-11-06 stsp #include "got_repository.h"
40 511a516b 2018-05-19 stsp #include "got_opentemp.h"
41 324d37e7 2019-05-11 stsp #include "got_path.h"
43 4123af3c 2023-02-12 op #include "got_lib_hash.h"
44 718b3ab0 2018-03-17 stsp #include "got_lib_delta.h"
45 63581804 2018-07-09 stsp #include "got_lib_inflate.h"
46 718b3ab0 2018-03-17 stsp #include "got_lib_object.h"
47 6bef87be 2018-09-11 stsp #include "got_lib_object_idcache.h"
48 6bef87be 2018-09-11 stsp #include "got_lib_object_cache.h"
49 ad242220 2018-09-08 stsp #include "got_lib_object_parse.h"
50 15a94983 2018-12-23 stsp #include "got_lib_pack.h"
51 7bb0daa1 2018-06-21 stsp #include "got_lib_repository.h"
54 ab9a70b2 2017-11-06 stsp #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
57 0ab4c957 2022-06-13 stsp #ifndef nitems
58 0ab4c957 2022-06-13 stsp #define nitems(_a) (sizeof((_a)) / sizeof((_a)[0]))
61 3235492e 2018-04-01 stsp struct got_object_id *
62 3235492e 2018-04-01 stsp got_object_get_id(struct got_object *obj)
64 6402fb3c 2018-09-15 stsp return &obj->id;
67 bacc9935 2018-05-20 stsp const struct got_error *
68 bacc9935 2018-05-20 stsp got_object_get_id_str(char **outbuf, struct got_object *obj)
70 bacc9935 2018-05-20 stsp return got_object_id_str(outbuf, &obj->id);
73 15a94983 2018-12-23 stsp const struct got_error *
74 15a94983 2018-12-23 stsp got_object_get_type(int *type, struct got_repository *repo,
75 15a94983 2018-12-23 stsp struct got_object_id *id)
77 15a94983 2018-12-23 stsp const struct got_error *err = NULL;
78 15a94983 2018-12-23 stsp struct got_object *obj;
80 15a94983 2018-12-23 stsp err = got_object_open(&obj, repo, id);
84 b107e67f 2018-01-19 stsp switch (obj->type) {
85 a1fd68d8 2018-01-12 stsp case GOT_OBJ_TYPE_COMMIT:
86 a1fd68d8 2018-01-12 stsp case GOT_OBJ_TYPE_TREE:
87 a1fd68d8 2018-01-12 stsp case GOT_OBJ_TYPE_BLOB:
88 b107e67f 2018-01-19 stsp case GOT_OBJ_TYPE_TAG:
89 15a94983 2018-12-23 stsp *type = obj->type;
92 15a94983 2018-12-23 stsp err = got_error(GOT_ERR_OBJ_TYPE);
96 15a94983 2018-12-23 stsp got_object_close(obj);
100 90bdb554 2019-04-11 stsp const struct got_error *
101 90bdb554 2019-04-11 stsp got_object_get_path(char **path, struct got_object_id *id,
102 90bdb554 2019-04-11 stsp struct got_repository *repo)
104 ab9a70b2 2017-11-06 stsp const struct got_error *err = NULL;
105 7a132809 2018-07-23 stsp char *hex = NULL;
106 41d2888b 2019-08-11 stsp char *path_objects;
108 e6b1056e 2018-04-22 stsp *path = NULL;
110 41d2888b 2019-08-11 stsp path_objects = got_repo_get_path_objects(repo);
111 ab9a70b2 2017-11-06 stsp if (path_objects == NULL)
112 638f9024 2019-05-13 stsp return got_error_from_errno("got_repo_get_path_objects");
114 ef0981d5 2018-02-12 stsp err = got_object_id_str(&hex, id);
118 d1cda826 2017-11-06 stsp if (asprintf(path, "%s/%.2x/%s", path_objects,
119 a41a7d4a 2023-02-12 op id->hash[0], hex + 2) == -1)
120 638f9024 2019-05-13 stsp err = got_error_from_errno("asprintf");
124 d1cda826 2017-11-06 stsp free(path_objects);
125 d1cda826 2017-11-06 stsp return err;
128 762d73f4 2021-04-10 stsp const struct got_error *
129 762d73f4 2021-04-10 stsp got_object_open_loose_fd(int *fd, struct got_object_id *id,
130 4796fb13 2018-12-23 stsp struct got_repository *repo)
132 d1cda826 2017-11-06 stsp const struct got_error *err = NULL;
133 a1fd68d8 2018-01-12 stsp char *path;
135 90bdb554 2019-04-11 stsp err = got_object_get_path(&path, id, repo);
137 d1cda826 2017-11-06 stsp return err;
138 8bd0cdad 2021-12-31 stsp *fd = open(path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
139 d5003b79 2018-04-22 stsp if (*fd == -1) {
140 e82b1d81 2019-07-27 stsp err = got_error_from_errno2("open", path);
144 4558fcd4 2018-01-14 stsp free(path);
145 59d1e4a0 2021-03-10 stsp return err;
148 59d1e4a0 2021-03-10 stsp const struct got_error *
149 6dfa2fd3 2018-02-12 stsp got_object_open_by_id_str(struct got_object **obj, struct got_repository *repo,
150 6dfa2fd3 2018-02-12 stsp const char *id_str)
152 6dfa2fd3 2018-02-12 stsp struct got_object_id id;
154 2d9874c2 2023-02-12 op memset(&id, 0, sizeof(id));
155 2d9874c2 2023-02-12 op id.algo = got_repo_get_object_format(repo);
156 2d9874c2 2023-02-12 op if (!got_parse_hash_digest(id.hash, id_str, id.algo))
157 6dd1ece6 2019-11-10 stsp return got_error_path(id_str, GOT_ERR_BAD_OBJ_ID_STR);
159 6dfa2fd3 2018-02-12 stsp return got_object_open(obj, repo, &id);
162 15a94983 2018-12-23 stsp const struct got_error *
163 15a94983 2018-12-23 stsp got_object_resolve_id_str(struct got_object_id **id,
164 15a94983 2018-12-23 stsp struct got_repository *repo, const char *id_str)
166 15a94983 2018-12-23 stsp const struct got_error *err = NULL;
167 15a94983 2018-12-23 stsp struct got_object *obj;
169 15a94983 2018-12-23 stsp err = got_object_open_by_id_str(&obj, repo, id_str);
171 15a94983 2018-12-23 stsp return err;
173 15a94983 2018-12-23 stsp *id = got_object_id_dup(got_object_get_id(obj));
174 15a94983 2018-12-23 stsp got_object_close(obj);
175 15a94983 2018-12-23 stsp if (*id == NULL)
176 638f9024 2019-05-13 stsp return got_error_from_errno("got_object_id_dup");
178 15a94983 2018-12-23 stsp return NULL;
181 e32baab7 2018-11-05 stsp const struct got_error *
182 dbc6a6b6 2018-07-12 stsp got_object_qid_alloc(struct got_object_qid **qid, struct got_object_id *id)
184 dbc6a6b6 2018-07-12 stsp *qid = calloc(1, sizeof(**qid));
185 dbc6a6b6 2018-07-12 stsp if (*qid == NULL)
186 638f9024 2019-05-13 stsp return got_error_from_errno("calloc");
188 d7b5a0e8 2022-04-20 stsp memcpy(&(*qid)->id, id, sizeof((*qid)->id));
189 9ca9aafb 2021-06-18 stsp return NULL;
192 9ca9aafb 2021-06-18 stsp const struct got_error *
193 9ca9aafb 2021-06-18 stsp got_object_id_queue_copy(const struct got_object_id_queue *src,
194 9ca9aafb 2021-06-18 stsp struct got_object_id_queue *dest)
196 9ca9aafb 2021-06-18 stsp const struct got_error *err;
197 9ca9aafb 2021-06-18 stsp struct got_object_qid *qid;
199 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(qid, src, entry) {
200 9ca9aafb 2021-06-18 stsp struct got_object_qid *new;
202 9ca9aafb 2021-06-18 stsp * Deep-copy the object ID only. Let the caller deal
203 9ca9aafb 2021-06-18 stsp * with setting up the new->data pointer if needed.
205 5e91dae4 2022-08-30 stsp err = got_object_qid_alloc(&new, &qid->id);
207 9ca9aafb 2021-06-18 stsp got_object_id_queue_free(dest);
208 9ca9aafb 2021-06-18 stsp return err;
210 dbdddfee 2021-06-23 naddy STAILQ_INSERT_TAIL(dest, new, entry);
213 dbc6a6b6 2018-07-12 stsp return NULL;
217 56e0773d 2019-11-28 stsp got_object_tree_get_nentries(struct got_tree_object *tree)
219 56e0773d 2019-11-28 stsp return tree->nentries;
222 56e0773d 2019-11-28 stsp struct got_tree_entry *
223 56e0773d 2019-11-28 stsp got_object_tree_get_first_entry(struct got_tree_object *tree)
225 56e0773d 2019-11-28 stsp return got_object_tree_get_entry(tree, 0);
228 56e0773d 2019-11-28 stsp struct got_tree_entry *
229 56e0773d 2019-11-28 stsp got_object_tree_get_last_entry(struct got_tree_object *tree)
231 56e0773d 2019-11-28 stsp return got_object_tree_get_entry(tree, tree->nentries - 1);
234 56e0773d 2019-11-28 stsp struct got_tree_entry *
235 56e0773d 2019-11-28 stsp got_object_tree_get_entry(struct got_tree_object *tree, int i)
237 56e0773d 2019-11-28 stsp if (i < 0 || i >= tree->nentries)
238 56e0773d 2019-11-28 stsp return NULL;
239 56e0773d 2019-11-28 stsp return &tree->entries[i];
243 56e0773d 2019-11-28 stsp got_tree_entry_get_mode(struct got_tree_entry *te)
245 56e0773d 2019-11-28 stsp return te->mode;
248 56e0773d 2019-11-28 stsp const char *
249 56e0773d 2019-11-28 stsp got_tree_entry_get_name(struct got_tree_entry *te)
251 56e0773d 2019-11-28 stsp return &te->name[0];
254 56e0773d 2019-11-28 stsp struct got_object_id *
255 56e0773d 2019-11-28 stsp got_tree_entry_get_id(struct got_tree_entry *te)
257 56e0773d 2019-11-28 stsp return &te->id;
260 0d6c6ee3 2020-05-20 stsp const struct got_error *
261 af57b12a 2020-07-23 stsp got_object_blob_read_to_str(char **s, struct got_blob_object *blob)
263 0d6c6ee3 2020-05-20 stsp const struct got_error *err = NULL;
264 32596e16 2020-07-23 stsp size_t len, totlen, hdrlen, offset;
268 659dc16e 2020-07-23 stsp hdrlen = got_object_blob_get_hdrlen(blob);
269 659dc16e 2020-07-23 stsp totlen = 0;
270 32596e16 2020-07-23 stsp offset = 0;
274 659dc16e 2020-07-23 stsp err = got_object_blob_read_block(&len, blob);
276 af57b12a 2020-07-23 stsp return err;
278 659dc16e 2020-07-23 stsp if (len == 0)
281 659dc16e 2020-07-23 stsp totlen += len - hdrlen;
282 af57b12a 2020-07-23 stsp p = realloc(*s, totlen + 1);
283 659dc16e 2020-07-23 stsp if (p == NULL) {
284 659dc16e 2020-07-23 stsp err = got_error_from_errno("realloc");
287 af57b12a 2020-07-23 stsp return err;
290 659dc16e 2020-07-23 stsp /* Skip blob object header first time around. */
291 af57b12a 2020-07-23 stsp memcpy(*s + offset,
292 f8f7c882 2020-07-23 stsp got_object_blob_get_read_buf(blob) + hdrlen, len - hdrlen);
293 659dc16e 2020-07-23 stsp hdrlen = 0;
294 32596e16 2020-07-23 stsp offset = totlen;
295 659dc16e 2020-07-23 stsp } while (len > 0);
297 af57b12a 2020-07-23 stsp (*s)[totlen] = '\0';
298 af57b12a 2020-07-23 stsp return NULL;
301 af57b12a 2020-07-23 stsp const struct got_error *
302 af57b12a 2020-07-23 stsp got_tree_entry_get_symlink_target(char **link_target, struct got_tree_entry *te,
303 af57b12a 2020-07-23 stsp struct got_repository *repo)
305 af57b12a 2020-07-23 stsp const struct got_error *err = NULL;
306 af57b12a 2020-07-23 stsp struct got_blob_object *blob = NULL;
307 eb81bc23 2022-06-28 tracey int fd = -1;
309 af57b12a 2020-07-23 stsp *link_target = NULL;
311 af57b12a 2020-07-23 stsp if (!got_object_tree_entry_is_symlink(te))
312 af57b12a 2020-07-23 stsp return got_error(GOT_ERR_TREE_ENTRY_TYPE);
314 eb81bc23 2022-06-28 tracey fd = got_opentempfd();
315 eb81bc23 2022-06-28 tracey if (fd == -1) {
316 eb81bc23 2022-06-28 tracey err = got_error_from_errno("got_opentempfd");
317 eb81bc23 2022-06-28 tracey goto done;
320 eb81bc23 2022-06-28 tracey err = got_object_open_as_blob(&blob, repo,
321 eb81bc23 2022-06-28 tracey got_tree_entry_get_id(te), PATH_MAX, fd);
323 eb81bc23 2022-06-28 tracey goto done;
325 af57b12a 2020-07-23 stsp err = got_object_blob_read_to_str(link_target, blob);
327 eb81bc23 2022-06-28 tracey if (fd != -1 && close(fd) == -1 && err == NULL)
328 eb81bc23 2022-06-28 tracey err = got_error_from_errno("close");
329 eb81bc23 2022-06-28 tracey if (blob)
330 eb81bc23 2022-06-28 tracey got_object_blob_close(blob);
332 659dc16e 2020-07-23 stsp free(*link_target);
333 659dc16e 2020-07-23 stsp *link_target = NULL;
335 0d6c6ee3 2020-05-20 stsp return err;
339 56e0773d 2019-11-28 stsp got_tree_entry_get_index(struct got_tree_entry *te)
341 56e0773d 2019-11-28 stsp return te->idx;
344 56e0773d 2019-11-28 stsp struct got_tree_entry *
345 56e0773d 2019-11-28 stsp got_tree_entry_get_next(struct got_tree_object *tree,
346 56e0773d 2019-11-28 stsp struct got_tree_entry *te)
348 56e0773d 2019-11-28 stsp return got_object_tree_get_entry(tree, te->idx + 1);
351 56e0773d 2019-11-28 stsp struct got_tree_entry *
352 56e0773d 2019-11-28 stsp got_tree_entry_get_prev(struct got_tree_object *tree,
353 56e0773d 2019-11-28 stsp struct got_tree_entry *te)
355 56e0773d 2019-11-28 stsp return got_object_tree_get_entry(tree, te->idx - 1);
358 a19581a2 2018-06-21 stsp const struct got_error *
359 68482ea3 2017-11-27 stsp got_object_blob_close(struct got_blob_object *blob)
361 fb43ecf1 2019-02-11 stsp const struct got_error *err = NULL;
362 15c8b0e6 2018-04-24 stsp free(blob->read_buf);
363 56b63ca4 2021-01-22 stsp if (blob->f && fclose(blob->f) == EOF)
364 638f9024 2019-05-13 stsp err = got_error_from_errno("fclose");
365 ac544f8c 2019-01-13 stsp free(blob->data);
366 68482ea3 2017-11-27 stsp free(blob);
367 fb43ecf1 2019-02-11 stsp return err;
371 8ba819a3 2020-07-23 stsp got_object_blob_rewind(struct got_blob_object *blob)
373 8ba819a3 2020-07-23 stsp if (blob->f)
374 8ba819a3 2020-07-23 stsp rewind(blob->f);
378 f934cf2c 2018-02-12 stsp got_object_blob_id_str(struct got_blob_object *blob, char *buf, size_t size)
380 bbc740ac 2023-02-04 op return got_object_id_hex(&blob->id, buf, size);
384 f934cf2c 2018-02-12 stsp got_object_blob_get_hdrlen(struct got_blob_object *blob)
386 f934cf2c 2018-02-12 stsp return blob->hdrlen;
389 f934cf2c 2018-02-12 stsp const uint8_t *
390 f934cf2c 2018-02-12 stsp got_object_blob_get_read_buf(struct got_blob_object *blob)
392 f934cf2c 2018-02-12 stsp return blob->read_buf;
395 68482ea3 2017-11-27 stsp const struct got_error *
396 eb651edf 2018-02-11 stsp got_object_blob_read_block(size_t *outlenp, struct got_blob_object *blob)
400 eb651edf 2018-02-11 stsp n = fread(blob->read_buf, 1, blob->blocksize, blob->f);
401 eb651edf 2018-02-11 stsp if (n == 0 && ferror(blob->f))
402 eb651edf 2018-02-11 stsp return got_ferror(blob->f, GOT_ERR_IO);
403 eb651edf 2018-02-11 stsp *outlenp = n;
404 35e9ba5d 2018-06-21 stsp return NULL;
407 35e9ba5d 2018-06-21 stsp const struct got_error *
408 0d569390 2023-01-04 op got_object_blob_is_binary(int *binary, struct got_blob_object *blob)
410 0d569390 2023-01-04 op const struct got_error *err;
411 0d569390 2023-01-04 op size_t hdrlen, len;
414 0d569390 2023-01-04 op hdrlen = got_object_blob_get_hdrlen(blob);
416 0d569390 2023-01-04 op if (fseeko(blob->f, hdrlen, SEEK_SET) == -1)
417 0d569390 2023-01-04 op return got_error_from_errno("fseeko");
419 0d569390 2023-01-04 op err = got_object_blob_read_block(&len, blob);
423 0d569390 2023-01-04 op *binary = memchr(blob->read_buf, '\0', len) != NULL;
425 0d569390 2023-01-04 op if (fseeko(blob->f, hdrlen, SEEK_SET) == -1)
426 0d569390 2023-01-04 op return got_error_from_errno("fseeko");
430 389a68d8 2023-01-05 op const struct got_error *
431 389a68d8 2023-01-05 op got_object_blob_getline(char **line, ssize_t *linelen, size_t *linesize,
432 389a68d8 2023-01-05 op struct got_blob_object *blob)
434 389a68d8 2023-01-05 op *linelen = getline(line, linesize, blob->f);
435 389a68d8 2023-01-05 op if (*linelen == -1 && !feof(blob->f))
436 389a68d8 2023-01-05 op return got_error_from_errno("getline");
440 0d569390 2023-01-04 op const struct got_error *
441 be659d10 2020-11-18 stsp got_object_blob_dump_to_file(off_t *filesize, int *nlines,
442 6c4c42e0 2019-06-24 stsp off_t **line_offsets, FILE *outfile, struct got_blob_object *blob)
444 35e9ba5d 2018-06-21 stsp const struct got_error *err = NULL;
445 b6752625 2018-12-24 stsp size_t n, len, hdrlen;
446 84451b3e 2018-07-10 stsp const uint8_t *buf;
448 c33ebc60 2020-11-18 stsp const int alloc_chunksz = 512;
449 c33ebc60 2020-11-18 stsp size_t nalloc = 0;
450 f595d9bd 2019-08-14 stsp off_t off = 0, total_len = 0;
452 6c4c42e0 2019-06-24 stsp if (line_offsets)
453 6c4c42e0 2019-06-24 stsp *line_offsets = NULL;
454 f595d9bd 2019-08-14 stsp if (filesize)
455 f595d9bd 2019-08-14 stsp *filesize = 0;
456 84451b3e 2018-07-10 stsp if (nlines)
457 84451b3e 2018-07-10 stsp *nlines = 0;
459 35e9ba5d 2018-06-21 stsp hdrlen = got_object_blob_get_hdrlen(blob);
461 35e9ba5d 2018-06-21 stsp err = got_object_blob_read_block(&len, blob);
463 35e9ba5d 2018-06-21 stsp return err;
464 35e9ba5d 2018-06-21 stsp if (len == 0)
466 84451b3e 2018-07-10 stsp buf = got_object_blob_get_read_buf(blob);
467 b02560ec 2019-08-19 stsp i = hdrlen;
468 f1cbc3bc 2020-11-18 stsp if (nlines) {
469 f1cbc3bc 2020-11-18 stsp if (line_offsets && *line_offsets == NULL) {
470 78695fb7 2019-08-12 stsp /* Have some data but perhaps no '\n'. */
471 78695fb7 2019-08-12 stsp *nlines = 1;
472 c33ebc60 2020-11-18 stsp nalloc = alloc_chunksz;
473 c33ebc60 2020-11-18 stsp *line_offsets = calloc(nalloc,
474 c33ebc60 2020-11-18 stsp sizeof(**line_offsets));
475 78695fb7 2019-08-12 stsp if (*line_offsets == NULL)
476 845785d4 2020-02-02 tracey return got_error_from_errno("calloc");
478 b02560ec 2019-08-19 stsp /* Skip forward over end of first line. */
479 b02560ec 2019-08-19 stsp while (i < len) {
480 b02560ec 2019-08-19 stsp if (buf[i] == '\n')
485 b02560ec 2019-08-19 stsp /* Scan '\n' offsets in remaining chunk of data. */
486 b02560ec 2019-08-19 stsp while (i < len) {
487 b02560ec 2019-08-19 stsp if (buf[i] != '\n') {
491 f595d9bd 2019-08-14 stsp (*nlines)++;
492 c33ebc60 2020-11-18 stsp if (line_offsets && nalloc < *nlines) {
493 c33ebc60 2020-11-18 stsp size_t n = *nlines + alloc_chunksz;
494 78695fb7 2019-08-12 stsp off_t *o = recallocarray(*line_offsets,
495 c33ebc60 2020-11-18 stsp nalloc, n, sizeof(**line_offsets));
496 78695fb7 2019-08-12 stsp if (o == NULL) {
497 78695fb7 2019-08-12 stsp free(*line_offsets);
498 78695fb7 2019-08-12 stsp *line_offsets = NULL;
499 78695fb7 2019-08-12 stsp return got_error_from_errno(
500 78695fb7 2019-08-12 stsp "recallocarray");
502 78695fb7 2019-08-12 stsp *line_offsets = o;
503 c33ebc60 2020-11-18 stsp nalloc = n;
505 f1cbc3bc 2020-11-18 stsp if (line_offsets) {
506 f1cbc3bc 2020-11-18 stsp off = total_len + i - hdrlen + 1;
507 f1cbc3bc 2020-11-18 stsp (*line_offsets)[*nlines - 1] = off;
512 35e9ba5d 2018-06-21 stsp /* Skip blob object header first time around. */
513 454a6b59 2018-12-24 stsp n = fwrite(buf + hdrlen, 1, len - hdrlen, outfile);
514 b6752625 2018-12-24 stsp if (n != len - hdrlen)
515 b6752625 2018-12-24 stsp return got_ferror(outfile, GOT_ERR_IO);
516 f595d9bd 2019-08-14 stsp total_len += len - hdrlen;
517 35e9ba5d 2018-06-21 stsp hdrlen = 0;
518 35e9ba5d 2018-06-21 stsp } while (len != 0);
520 cbe7f848 2019-02-11 stsp if (fflush(outfile) != 0)
521 638f9024 2019-05-13 stsp return got_error_from_errno("fflush");
522 35e9ba5d 2018-06-21 stsp rewind(outfile);
524 f595d9bd 2019-08-14 stsp if (filesize)
525 f595d9bd 2019-08-14 stsp *filesize = total_len;
527 776d4d29 2018-06-17 stsp return NULL;
530 d24820bf 2019-08-11 stsp const char *
531 d24820bf 2019-08-11 stsp got_object_tag_get_name(struct got_tag_object *tag)
533 d24820bf 2019-08-11 stsp return tag->tag;
537 0bd18d37 2019-02-01 stsp got_object_tag_get_object_type(struct got_tag_object *tag)
539 0bd18d37 2019-02-01 stsp return tag->obj_type;
542 0bd18d37 2019-02-01 stsp struct got_object_id *
543 0bd18d37 2019-02-01 stsp got_object_tag_get_object_id(struct got_tag_object *tag)
545 0bd18d37 2019-02-01 stsp return &tag->id;
549 01073a5d 2019-08-22 stsp got_object_tag_get_tagger_time(struct got_tag_object *tag)
551 01073a5d 2019-08-22 stsp return tag->tagger_time;
555 01073a5d 2019-08-22 stsp got_object_tag_get_tagger_gmtoff(struct got_tag_object *tag)
557 01073a5d 2019-08-22 stsp return tag->tagger_gmtoff;
560 01073a5d 2019-08-22 stsp const char *
561 01073a5d 2019-08-22 stsp got_object_tag_get_tagger(struct got_tag_object *tag)
563 01073a5d 2019-08-22 stsp return tag->tagger;
566 01073a5d 2019-08-22 stsp const char *
567 01073a5d 2019-08-22 stsp got_object_tag_get_message(struct got_tag_object *tag)
569 01073a5d 2019-08-22 stsp return tag->tagmsg;
572 776d4d29 2018-06-17 stsp static struct got_tree_entry *
573 65a9bbe9 2018-09-15 stsp find_entry_by_name(struct got_tree_object *tree, const char *name, size_t len)
577 63da309a 2018-11-07 stsp /* Note that tree entries are sorted in strncmp() order. */
578 56e0773d 2019-11-28 stsp for (i = 0; i < tree->nentries; i++) {
579 56e0773d 2019-11-28 stsp struct got_tree_entry *te = &tree->entries[i];
580 63da309a 2018-11-07 stsp int cmp = strncmp(te->name, name, len);
581 63da309a 2018-11-07 stsp if (cmp < 0)
583 63da309a 2018-11-07 stsp if (cmp > 0)
585 63da309a 2018-11-07 stsp if (te->name[len] == '\0')
588 eb651edf 2018-02-11 stsp return NULL;
591 56e0773d 2019-11-28 stsp struct got_tree_entry *
592 a129376b 2019-03-28 stsp got_object_tree_find_entry(struct got_tree_object *tree, const char *name)
594 a129376b 2019-03-28 stsp return find_entry_by_name(tree, name, strlen(name));
597 776d4d29 2018-06-17 stsp const struct got_error *
598 67b631c9 2021-10-10 stsp got_object_tree_find_path(struct got_object_id **id, mode_t *mode,
599 67b631c9 2021-10-10 stsp struct got_repository *repo, struct got_tree_object *tree,
600 67b631c9 2021-10-10 stsp const char *path)
602 776d4d29 2018-06-17 stsp const struct got_error *err = NULL;
603 67b631c9 2021-10-10 stsp struct got_tree_object *subtree = NULL;
604 db37e2c0 2018-06-21 stsp struct got_tree_entry *te = NULL;
605 65a9bbe9 2018-09-15 stsp const char *seg, *s;
606 b7cd37e5 2018-11-18 stsp size_t seglen;
608 27d434c2 2018-09-15 stsp *id = NULL;
611 5e54fb30 2019-05-31 stsp while (s[0] == '/')
614 65a9bbe9 2018-09-15 stsp seglen = 0;
615 67b631c9 2021-10-10 stsp subtree = tree;
616 b7cd37e5 2018-11-18 stsp while (*s) {
617 776d4d29 2018-06-17 stsp struct got_tree_object *next_tree;
619 776d4d29 2018-06-17 stsp if (*s != '/') {
626 67b631c9 2021-10-10 stsp te = find_entry_by_name(subtree, seg, seglen);
627 db37e2c0 2018-06-21 stsp if (te == NULL) {
628 b66cd6f3 2020-07-31 stsp err = got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
632 b7cd37e5 2018-11-18 stsp if (*s == '\0')
635 776d4d29 2018-06-17 stsp seg = s + 1;
636 65a9bbe9 2018-09-15 stsp seglen = 0;
639 776d4d29 2018-06-17 stsp err = got_object_open_as_tree(&next_tree, repo,
644 67b631c9 2021-10-10 stsp if (subtree != tree)
645 67b631c9 2021-10-10 stsp got_object_tree_close(subtree);
646 67b631c9 2021-10-10 stsp subtree = next_tree;
651 56e0773d 2019-11-28 stsp *id = got_object_id_dup(&te->id);
652 27d434c2 2018-09-15 stsp if (*id == NULL)
653 638f9024 2019-05-13 stsp return got_error_from_errno("got_object_id_dup");
655 67b631c9 2021-10-10 stsp *mode = te->mode;
657 b66cd6f3 2020-07-31 stsp err = got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
659 67b631c9 2021-10-10 stsp if (subtree && subtree != tree)
660 67b631c9 2021-10-10 stsp got_object_tree_close(subtree);
661 67b631c9 2021-10-10 stsp return err;
664 67b631c9 2021-10-10 stsp const struct got_error *
665 67b631c9 2021-10-10 stsp got_object_id_by_path(struct got_object_id **id, struct got_repository *repo,
666 a44927cc 2022-04-07 stsp struct got_commit_object *commit, const char *path)
668 67b631c9 2021-10-10 stsp const struct got_error *err = NULL;
669 67b631c9 2021-10-10 stsp struct got_tree_object *tree = NULL;
671 67b631c9 2021-10-10 stsp *id = NULL;
673 67b631c9 2021-10-10 stsp /* Handle opening of root of commit's tree. */
674 67b631c9 2021-10-10 stsp if (got_path_is_root_dir(path)) {
675 67b631c9 2021-10-10 stsp *id = got_object_id_dup(commit->tree_id);
676 67b631c9 2021-10-10 stsp if (*id == NULL)
677 67b631c9 2021-10-10 stsp err = got_error_from_errno("got_object_id_dup");
679 67b631c9 2021-10-10 stsp err = got_object_open_as_tree(&tree, repo, commit->tree_id);
682 67b631c9 2021-10-10 stsp err = got_object_tree_find_path(id, NULL, repo, tree, path);
686 776d4d29 2018-06-17 stsp got_object_tree_close(tree);
687 776d4d29 2018-06-17 stsp return err;
691 ac5f2b26 2020-05-05 stsp * Normalize file mode bits to avoid false positive tree entry differences
692 ac5f2b26 2020-05-05 stsp * in case tree entries have unexpected mode bits set.
694 ac5f2b26 2020-05-05 stsp static mode_t
695 ac5f2b26 2020-05-05 stsp normalize_mode_for_comparison(mode_t mode)
698 ac5f2b26 2020-05-05 stsp * For directories, the only relevant bit is the IFDIR bit.
699 ac5f2b26 2020-05-05 stsp * This allows us to detect paths changing from a directory
700 ac5f2b26 2020-05-05 stsp * to a file and vice versa.
702 ac5f2b26 2020-05-05 stsp if (S_ISDIR(mode))
703 ac5f2b26 2020-05-05 stsp return mode & S_IFDIR;
706 40dde666 2020-07-23 stsp * For symlinks, the only relevant bit is the IFLNK bit.
707 40dde666 2020-07-23 stsp * This allows us to detect paths changing from a symlinks
708 40dde666 2020-07-23 stsp * to a file or directory and vice versa.
710 40dde666 2020-07-23 stsp if (S_ISLNK(mode))
711 40dde666 2020-07-23 stsp return mode & S_IFLNK;
713 ac5f2b26 2020-05-05 stsp /* For files, the only change we care about is the executable bit. */
714 ac5f2b26 2020-05-05 stsp return mode & S_IXUSR;
717 07862c20 2018-09-15 stsp const struct got_error *
718 07862c20 2018-09-15 stsp got_object_tree_path_changed(int *changed,
719 07862c20 2018-09-15 stsp struct got_tree_object *tree01, struct got_tree_object *tree02,
720 07862c20 2018-09-15 stsp const char *path, struct got_repository *repo)
722 07862c20 2018-09-15 stsp const struct got_error *err = NULL;
723 07862c20 2018-09-15 stsp struct got_tree_object *tree1 = NULL, *tree2 = NULL;
724 07862c20 2018-09-15 stsp struct got_tree_entry *te1 = NULL, *te2 = NULL;
725 65a9bbe9 2018-09-15 stsp const char *seg, *s;
726 3b7f9878 2018-11-18 stsp size_t seglen;
728 07862c20 2018-09-15 stsp *changed = 0;
730 07862c20 2018-09-15 stsp /* We not do support comparing the root path. */
731 61a7d79f 2020-02-29 stsp if (got_path_is_root_dir(path))
732 63f810e6 2020-02-29 stsp return got_error_path(path, GOT_ERR_BAD_PATH);
734 07862c20 2018-09-15 stsp tree1 = tree01;
735 07862c20 2018-09-15 stsp tree2 = tree02;
737 61a7d79f 2020-02-29 stsp while (*s == '/')
740 65a9bbe9 2018-09-15 stsp seglen = 0;
741 3b7f9878 2018-11-18 stsp while (*s) {
742 07862c20 2018-09-15 stsp struct got_tree_object *next_tree1, *next_tree2;
743 ac5f2b26 2020-05-05 stsp mode_t mode1, mode2;
745 07862c20 2018-09-15 stsp if (*s != '/') {
752 65a9bbe9 2018-09-15 stsp te1 = find_entry_by_name(tree1, seg, seglen);
753 07862c20 2018-09-15 stsp if (te1 == NULL) {
754 07862c20 2018-09-15 stsp err = got_error(GOT_ERR_NO_OBJ);
759 e8bfb8f3 2020-12-18 stsp te2 = find_entry_by_name(tree2, seg, seglen);
762 e8bfb8f3 2020-12-18 stsp mode1 = normalize_mode_for_comparison(te1->mode);
763 e8bfb8f3 2020-12-18 stsp mode2 = normalize_mode_for_comparison(te2->mode);
764 e8bfb8f3 2020-12-18 stsp if (mode1 != mode2) {
765 e8bfb8f3 2020-12-18 stsp *changed = 1;
769 e8bfb8f3 2020-12-18 stsp if (got_object_id_cmp(&te1->id, &te2->id) == 0) {
770 e8bfb8f3 2020-12-18 stsp *changed = 0;
775 3b7f9878 2018-11-18 stsp if (*s == '\0') { /* final path element */
776 07862c20 2018-09-15 stsp *changed = 1;
780 07862c20 2018-09-15 stsp seg = s + 1;
782 65a9bbe9 2018-09-15 stsp seglen = 0;
784 07862c20 2018-09-15 stsp err = got_object_open_as_tree(&next_tree1, repo,
786 07862c20 2018-09-15 stsp te1 = NULL;
789 a31cea73 2018-09-15 stsp if (tree1 != tree01)
790 a31cea73 2018-09-15 stsp got_object_tree_close(tree1);
791 07862c20 2018-09-15 stsp tree1 = next_tree1;
794 e8bfb8f3 2020-12-18 stsp err = got_object_open_as_tree(&next_tree2, repo,
796 e8bfb8f3 2020-12-18 stsp te2 = NULL;
799 e8bfb8f3 2020-12-18 stsp if (tree2 != tree02)
800 e8bfb8f3 2020-12-18 stsp got_object_tree_close(tree2);
801 e8bfb8f3 2020-12-18 stsp tree2 = next_tree2;
802 e8bfb8f3 2020-12-18 stsp } else if (tree2) {
803 e8bfb8f3 2020-12-18 stsp if (tree2 != tree02)
804 e8bfb8f3 2020-12-18 stsp got_object_tree_close(tree2);
805 e8bfb8f3 2020-12-18 stsp tree2 = NULL;
810 a31cea73 2018-09-15 stsp if (tree1 && tree1 != tree01)
811 07862c20 2018-09-15 stsp got_object_tree_close(tree1);
812 a31cea73 2018-09-15 stsp if (tree2 && tree2 != tree02)
813 07862c20 2018-09-15 stsp got_object_tree_close(tree2);
814 77880158 2018-11-04 stsp return err;
817 ed175427 2019-05-09 stsp const struct got_error *
818 ed175427 2019-05-09 stsp got_object_tree_entry_dup(struct got_tree_entry **new_te,
819 ed175427 2019-05-09 stsp struct got_tree_entry *te)
821 ed175427 2019-05-09 stsp const struct got_error *err = NULL;
823 ed175427 2019-05-09 stsp *new_te = calloc(1, sizeof(**new_te));
824 ed175427 2019-05-09 stsp if (*new_te == NULL)
825 638f9024 2019-05-13 stsp return got_error_from_errno("calloc");
827 ed175427 2019-05-09 stsp (*new_te)->mode = te->mode;
828 56e0773d 2019-11-28 stsp memcpy((*new_te)->name, te->name, sizeof((*new_te)->name));
829 56e0773d 2019-11-28 stsp memcpy(&(*new_te)->id, &te->id, sizeof((*new_te)->id));
830 8c4eabf2 2019-05-10 stsp return err;
834 56e0773d 2019-11-28 stsp got_object_tree_entry_is_submodule(struct got_tree_entry *te)
836 63c5ca5d 2019-08-24 stsp return (te->mode & S_IFMT) == (S_IFDIR | S_IFLNK);
840 e40622f4 2020-07-23 stsp got_object_tree_entry_is_symlink(struct got_tree_entry *te)
842 e40622f4 2020-07-23 stsp /* S_IFDIR check avoids confusing symlinks with submodules. */
843 e40622f4 2020-07-23 stsp return ((te->mode & (S_IFDIR | S_IFLNK)) == S_IFLNK);
846 e40622f4 2020-07-23 stsp static const struct got_error *
847 e40622f4 2020-07-23 stsp resolve_symlink(char **link_target, const char *path,
848 a44927cc 2022-04-07 stsp struct got_commit_object *commit, struct got_repository *repo)
850 e40622f4 2020-07-23 stsp const struct got_error *err = NULL;
851 dbdd6209 2020-10-19 stsp char buf[PATH_MAX];
852 e40622f4 2020-07-23 stsp char *name, *parent_path = NULL;
853 e40622f4 2020-07-23 stsp struct got_object_id *tree_obj_id = NULL;
854 e40622f4 2020-07-23 stsp struct got_tree_object *tree = NULL;
855 e40622f4 2020-07-23 stsp struct got_tree_entry *te = NULL;
857 e40622f4 2020-07-23 stsp *link_target = NULL;
859 dbdd6209 2020-10-19 stsp if (strlcpy(buf, path, sizeof(buf)) >= sizeof(buf))
860 dbdd6209 2020-10-19 stsp return got_error(GOT_ERR_NO_SPACE);
862 dbdd6209 2020-10-19 stsp name = basename(buf);
863 e40622f4 2020-07-23 stsp if (name == NULL)
864 e40622f4 2020-07-23 stsp return got_error_from_errno2("basename", path);
866 e40622f4 2020-07-23 stsp err = got_path_dirname(&parent_path, path);
868 e40622f4 2020-07-23 stsp return err;
870 a44927cc 2022-04-07 stsp err = got_object_id_by_path(&tree_obj_id, repo, commit,
871 e40622f4 2020-07-23 stsp parent_path);
873 e40622f4 2020-07-23 stsp if (err->code == GOT_ERR_NO_TREE_ENTRY) {
874 e40622f4 2020-07-23 stsp /* Display the complete path in error message. */
875 e40622f4 2020-07-23 stsp err = got_error_path(path, err->code);
880 e40622f4 2020-07-23 stsp err = got_object_open_as_tree(&tree, repo, tree_obj_id);
884 e40622f4 2020-07-23 stsp te = got_object_tree_find_entry(tree, name);
885 e40622f4 2020-07-23 stsp if (te == NULL) {
886 e40622f4 2020-07-23 stsp err = got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
890 e40622f4 2020-07-23 stsp if (got_object_tree_entry_is_symlink(te)) {
891 e40622f4 2020-07-23 stsp err = got_tree_entry_get_symlink_target(link_target, te, repo);
894 e40622f4 2020-07-23 stsp if (!got_path_is_absolute(*link_target)) {
895 e40622f4 2020-07-23 stsp char *abspath;
896 e40622f4 2020-07-23 stsp if (asprintf(&abspath, "%s/%s", parent_path,
897 e40622f4 2020-07-23 stsp *link_target) == -1) {
898 e40622f4 2020-07-23 stsp err = got_error_from_errno("asprintf");
901 e40622f4 2020-07-23 stsp free(*link_target);
902 e40622f4 2020-07-23 stsp *link_target = malloc(PATH_MAX);
903 e40622f4 2020-07-23 stsp if (*link_target == NULL) {
904 e40622f4 2020-07-23 stsp err = got_error_from_errno("malloc");
907 e40622f4 2020-07-23 stsp err = got_canonpath(abspath, *link_target, PATH_MAX);
908 e40622f4 2020-07-23 stsp free(abspath);
914 b68bd9d2 2022-09-03 op free(parent_path);
915 e40622f4 2020-07-23 stsp free(tree_obj_id);
917 e40622f4 2020-07-23 stsp got_object_tree_close(tree);
919 e40622f4 2020-07-23 stsp free(*link_target);
920 e40622f4 2020-07-23 stsp *link_target = NULL;
922 e40622f4 2020-07-23 stsp return err;
925 ca6e02ac 2020-01-07 stsp const struct got_error *
926 e40622f4 2020-07-23 stsp got_object_resolve_symlinks(char **link_target, const char *path,
927 a44927cc 2022-04-07 stsp struct got_commit_object *commit, struct got_repository *repo)
929 e40622f4 2020-07-23 stsp const struct got_error *err = NULL;
930 e40622f4 2020-07-23 stsp char *next_target = NULL;
931 e40622f4 2020-07-23 stsp int max_recursion = 40; /* matches Git */
933 e40622f4 2020-07-23 stsp *link_target = NULL;
936 e40622f4 2020-07-23 stsp err = resolve_symlink(&next_target,
937 a44927cc 2022-04-07 stsp *link_target ? *link_target : path, commit, repo);
940 e40622f4 2020-07-23 stsp if (next_target) {
941 e40622f4 2020-07-23 stsp free(*link_target);
942 e40622f4 2020-07-23 stsp if (--max_recursion == 0) {
943 e40622f4 2020-07-23 stsp err = got_error_path(path, GOT_ERR_RECURSION);
944 e40622f4 2020-07-23 stsp *link_target = NULL;
947 e40622f4 2020-07-23 stsp *link_target = next_target;
949 e40622f4 2020-07-23 stsp } while (next_target);
951 e40622f4 2020-07-23 stsp return err;
955 34a842a4 2022-09-18 mark got_object_commit_retain(struct got_commit_object *commit)
957 568eae95 2022-09-11 mark commit->refcnt++;
960 13b2bc37 2022-10-23 stsp const struct got_error *
961 13b2bc37 2022-10-23 stsp got_object_raw_alloc(struct got_raw_object **obj, uint8_t *outbuf, int *outfd,
962 60c140ae 2023-01-09 stsp size_t max_in_mem_size, size_t hdrlen, off_t size)
964 13b2bc37 2022-10-23 stsp const struct got_error *err = NULL;
967 1c28a361 2022-10-25 op tot = hdrlen + size;
969 13b2bc37 2022-10-23 stsp *obj = calloc(1, sizeof(**obj));
970 13b2bc37 2022-10-23 stsp if (*obj == NULL) {
971 13b2bc37 2022-10-23 stsp err = got_error_from_errno("calloc");
974 13b2bc37 2022-10-23 stsp (*obj)->fd = -1;
975 13b2bc37 2022-10-23 stsp (*obj)->tempfile_idx = -1;
977 13b2bc37 2022-10-23 stsp if (outbuf) {
978 13b2bc37 2022-10-23 stsp (*obj)->data = outbuf;
980 13b2bc37 2022-10-23 stsp struct stat sb;
981 13b2bc37 2022-10-23 stsp if (fstat(*outfd, &sb) == -1) {
982 13b2bc37 2022-10-23 stsp err = got_error_from_errno("fstat");
986 1c28a361 2022-10-25 op if (sb.st_size != tot) {
987 e98a81e6 2023-01-09 stsp err = got_error_msg(GOT_ERR_BAD_OBJ_HDR,
988 e98a81e6 2023-01-09 stsp "raw object has unexpected size");
991 13b2bc37 2022-10-23 stsp #ifndef GOT_PACK_NO_MMAP
992 60c140ae 2023-01-09 stsp if (tot > 0 && tot <= max_in_mem_size) {
993 1c28a361 2022-10-25 op (*obj)->data = mmap(NULL, tot, PROT_READ,
994 13b2bc37 2022-10-23 stsp MAP_PRIVATE, *outfd, 0);
995 13b2bc37 2022-10-23 stsp if ((*obj)->data == MAP_FAILED) {
996 13b2bc37 2022-10-23 stsp if (errno != ENOMEM) {
997 13b2bc37 2022-10-23 stsp err = got_error_from_errno("mmap");
1000 13b2bc37 2022-10-23 stsp (*obj)->data = NULL;
1002 13b2bc37 2022-10-23 stsp (*obj)->fd = *outfd;
1003 13b2bc37 2022-10-23 stsp *outfd = -1;
1007 13b2bc37 2022-10-23 stsp if (*outfd != -1) {
1008 13b2bc37 2022-10-23 stsp (*obj)->f = fdopen(*outfd, "r");
1009 13b2bc37 2022-10-23 stsp if ((*obj)->f == NULL) {
1010 13b2bc37 2022-10-23 stsp err = got_error_from_errno("fdopen");
1011 13b2bc37 2022-10-23 stsp goto done;
1013 13b2bc37 2022-10-23 stsp *outfd = -1;
1016 13b2bc37 2022-10-23 stsp (*obj)->hdrlen = hdrlen;
1017 13b2bc37 2022-10-23 stsp (*obj)->size = size;
1019 13b2bc37 2022-10-23 stsp if (err) {
1020 13b2bc37 2022-10-23 stsp if (*obj) {
1021 13b2bc37 2022-10-23 stsp got_object_raw_close(*obj);
1022 13b2bc37 2022-10-23 stsp *obj = NULL;
1025 13b2bc37 2022-10-23 stsp (*obj)->refcnt++;
1026 13b2bc37 2022-10-23 stsp return err;
