object-store: prepare read_object_file to deal with any repo
[git/git.git] / object-store.h
CommitLineData
90c62155
SB
1#ifndef OBJECT_STORE_H
2#define OBJECT_STORE_H
3
ef3ca954 4#include "cache.h"
d88f9fdf 5#include "oidmap.h"
14727b7f
JT
6#include "list.h"
7#include "sha1-array.h"
8#include "strbuf.h"
d88f9fdf 9
031dc927 10struct alternate_object_database {
0d4a1321
SB
11 struct alternate_object_database *next;
12
13 /* see alt_scratch_buf() */
14 struct strbuf scratch;
15 size_t base_len;
16
17 /*
18 * Used to store the results of readdir(3) calls when searching
19 * for unique abbreviated hashes. This cache is never
20 * invalidated, thus it's racy and not necessarily accurate.
21 * That's fine for its purpose; don't use it for tasks requiring
22 * greater accuracy!
23 */
24 char loose_objects_subdir_seen[256];
25 struct oid_array loose_objects_cache;
26
77f012e8
SB
27 /*
28 * Path to the alternative object store. If this is a relative path,
29 * it is relative to the current working directory.
30 */
0d4a1321 31 char path[FLEX_ARRAY];
031dc927 32};
13068bf0 33void prepare_alt_odb(struct repository *r);
0d4a1321
SB
34char *compute_alternate_path(const char *path, struct strbuf *err);
35typedef int alt_odb_fn(struct alternate_object_database *, void *);
36int foreach_alt_odb(alt_odb_fn, void*);
37
38/*
39 * Allocate a "struct alternate_object_database" but do _not_ actually
40 * add it to the list of alternates.
41 */
42struct alternate_object_database *alloc_alt_odb(const char *dir);
43
44/*
45 * Add the directory to the on-disk alternates file; the new entry will also
46 * take effect in the current process.
47 */
48void add_to_alternates_file(const char *dir);
49
50/*
51 * Add the directory to the in-memory list of alternates (along with any
52 * recursive alternates it points to), but do not modify the on-disk alternates
53 * file.
54 */
55void add_to_alternates_memory(const char *dir);
56
57/*
58 * Returns a scratch strbuf pre-filled with the alternate object directory,
59 * including a trailing slash, which can be used to access paths in the
60 * alternate. Always use this over direct access to alt->scratch, as it
61 * cleans up any previous use of the scratch buffer.
62 */
63struct strbuf *alt_scratch_buf(struct alternate_object_database *alt);
64
a80d72db
SB
65struct packed_git {
66 struct packed_git *next;
67 struct list_head mru;
68 struct pack_window *windows;
69 off_t pack_size;
70 const void *index_data;
71 size_t index_size;
72 uint32_t num_objects;
73 uint32_t num_bad_objects;
74 unsigned char *bad_object_sha1;
75 int index_version;
76 time_t mtime;
77 int pack_fd;
43fa44fa 78 int index; /* for builtin/pack-objects.c */
a80d72db
SB
79 unsigned pack_local:1,
80 pack_keep:1,
ed7e5fc3 81 pack_keep_in_core:1,
a80d72db
SB
82 freshened:1,
83 do_not_close:1,
84 pack_promisor:1;
85 unsigned char sha1[20];
86 struct revindex_entry *revindex;
87 /* something like ".git/objects/pack/xxxxx.pack" */
88 char pack_name[FLEX_ARRAY]; /* more */
89};
90
4d80560c
DS
91struct multi_pack_index;
92
90c62155
SB
93struct raw_object_store {
94 /*
95 * Path to the repository's object store.
96 * Cannot be NULL after initialization.
97 */
98 char *objectdir;
99
100 /* Path to extra alternate object database if not NULL */
101 char *alternate_db;
031dc927
SB
102
103 struct alternate_object_database *alt_odb_list;
104 struct alternate_object_database **alt_odb_tail;
a80d72db 105
d88f9fdf
SB
106 /*
107 * Objects that should be substituted by other objects
108 * (see git-replace(1)).
109 */
c1274495 110 struct oidmap *replace_map;
d88f9fdf 111
85277506
JT
112 struct commit_graph *commit_graph;
113 unsigned commit_graph_attempted : 1; /* if loading has been attempted */
114
c4d25228
DS
115 /*
116 * private data
117 *
118 * should only be accessed directly by packfile.c and midx.c
119 */
120 struct multi_pack_index *multi_pack_index;
121
a80d72db
SB
122 /*
123 * private data
124 *
125 * should only be accessed directly by packfile.c
126 */
127
128 struct packed_git *packed_git;
129 /* A most-recently-used ordered version of the packed_git list. */
130 struct list_head packed_git_mru;
5508f693 131
0bff5269
DS
132 /*
133 * A linked list containing all packfiles, starting with those
134 * contained in the multi_pack_index.
135 */
136 struct packed_git *all_packs;
137
9a00580d
SB
138 /*
139 * A fast, rough count of the number of objects in the repository.
140 * These two fields are not meant for direct access. Use
141 * approximate_object_count() instead.
142 */
143 unsigned long approximate_object_count;
144 unsigned approximate_object_count_valid : 1;
145
5508f693
SB
146 /*
147 * Whether packed_git has already been populated with this repository's
148 * packs.
149 */
150 unsigned packed_git_initialized : 1;
90c62155
SB
151};
152
153struct raw_object_store *raw_object_store_new(void);
154void raw_object_store_clear(struct raw_object_store *o);
155
cf78ae4f
SB
156/*
157 * Put in `buf` the name of the file in the local object database that
158 * would be used to store a loose object with the specified sha1.
159 */
a68377b5 160void sha1_file_name(struct repository *r, struct strbuf *buf, const unsigned char *sha1);
cf78ae4f 161
bd27f50c 162void *map_sha1_file(struct repository *r, const unsigned char *sha1, unsigned long *size);
e35454fa 163
a3b72c89
SB
164extern void *read_object_file_extended(struct repository *r,
165 const struct object_id *oid,
cbd53a21
SB
166 enum object_type *type,
167 unsigned long *size, int lookup_replace);
afd69dcc
SB
168static inline void *repo_read_object_file(struct repository *r,
169 const struct object_id *oid,
170 enum object_type *type,
171 unsigned long *size)
cbd53a21 172{
afd69dcc 173 return read_object_file_extended(r, oid, type, size, 1);
cbd53a21 174}
afd69dcc
SB
175#ifndef NO_THE_REPOSITORY_COMPATIBILITY_MACROS
176#define read_object_file(oid, type, size) repo_read_object_file(the_repository, oid, type, size)
177#endif
cbd53a21
SB
178
179/* Read and unpack an object file into memory, write memory to an object file */
180int oid_object_info(struct repository *r, const struct object_id *, unsigned long *);
181
182extern int hash_object_file(const void *buf, unsigned long len,
183 const char *type, struct object_id *oid);
184
185extern int write_object_file(const void *buf, unsigned long len,
186 const char *type, struct object_id *oid);
187
188extern int hash_object_file_literally(const void *buf, unsigned long len,
189 const char *type, struct object_id *oid,
190 unsigned flags);
191
192extern int pretend_object_file(void *, unsigned long, enum object_type,
193 struct object_id *oid);
194
195extern int force_object_loose(const struct object_id *oid, time_t mtime);
196
197/*
198 * Open the loose object at path, check its hash, and return the contents,
199 * type, and size. If the object is a blob, then "contents" may return NULL,
200 * to allow streaming of large blobs.
201 *
202 * Returns 0 on success, negative on error (details may be written to stderr).
203 */
204int read_loose_object(const char *path,
205 const struct object_id *expected_oid,
206 enum object_type *type,
207 unsigned long *size,
208 void **contents);
209
210/*
211 * Convenience for sha1_object_info_extended() with a NULL struct
212 * object_info. OBJECT_INFO_SKIP_CACHED is automatically set; pass
213 * nonzero flags to also set other flags.
214 */
215extern int has_sha1_file_with_flags(const unsigned char *sha1, int flags);
216static inline int has_sha1_file(const unsigned char *sha1)
217{
218 return has_sha1_file_with_flags(sha1, 0);
219}
220
221/* Same as the above, except for struct object_id. */
222extern int has_object_file(const struct object_id *oid);
223extern int has_object_file_with_flags(const struct object_id *oid, int flags);
224
225/*
226 * Return true iff an alternate object database has a loose object
227 * with the specified name. This function does not respect replace
228 * references.
229 */
b16b60f7 230extern int has_loose_object_nonlocal(const struct object_id *);
cbd53a21
SB
231
232extern void assert_oid_type(const struct object_id *oid, enum object_type expect);
233
234struct object_info {
235 /* Request */
236 enum object_type *typep;
237 unsigned long *sizep;
238 off_t *disk_sizep;
239 unsigned char *delta_base_sha1;
240 struct strbuf *type_name;
241 void **contentp;
242
243 /* Response */
244 enum {
245 OI_CACHED,
246 OI_LOOSE,
247 OI_PACKED,
248 OI_DBCACHED
249 } whence;
250 union {
251 /*
252 * struct {
253 * ... Nothing to expose in this case
254 * } cached;
255 * struct {
256 * ... Nothing to expose in this case
257 * } loose;
258 */
259 struct {
260 struct packed_git *pack;
261 off_t offset;
262 unsigned int is_delta;
263 } packed;
264 } u;
265};
266
267/*
268 * Initializer for a "struct object_info" that wants no items. You may
269 * also memset() the memory to all-zeroes.
270 */
271#define OBJECT_INFO_INIT {NULL}
272
273/* Invoke lookup_replace_object() on the given hash */
274#define OBJECT_INFO_LOOKUP_REPLACE 1
275/* Allow reading from a loose object file of unknown/bogus type */
276#define OBJECT_INFO_ALLOW_UNKNOWN_TYPE 2
277/* Do not check cached storage */
278#define OBJECT_INFO_SKIP_CACHED 4
279/* Do not retry packed storage after checking packed and loose storage */
280#define OBJECT_INFO_QUICK 8
281/* Do not check loose object */
282#define OBJECT_INFO_IGNORE_LOOSE 16
283
284int oid_object_info_extended(struct repository *r,
285 const struct object_id *,
286 struct object_info *, unsigned flags);
287
0889aae1
JK
288/*
289 * Iterate over the files in the loose-object parts of the object
290 * directory "path", triggering the following callbacks:
291 *
292 * - loose_object is called for each loose object we find.
293 *
294 * - loose_cruft is called for any files that do not appear to be
295 * loose objects. Note that we only look in the loose object
296 * directories "objects/[0-9a-f]{2}/", so we will not report
297 * "objects/foobar" as cruft.
298 *
299 * - loose_subdir is called for each top-level hashed subdirectory
300 * of the object directory (e.g., "$OBJDIR/f0"). It is called
301 * after the objects in the directory are processed.
302 *
303 * Any callback that is NULL will be ignored. Callbacks returning non-zero
304 * will end the iteration.
305 *
306 * In the "buf" variant, "path" is a strbuf which will also be used as a
307 * scratch buffer, but restored to its original contents before
308 * the function returns.
309 */
310typedef int each_loose_object_fn(const struct object_id *oid,
311 const char *path,
312 void *data);
313typedef int each_loose_cruft_fn(const char *basename,
314 const char *path,
315 void *data);
316typedef int each_loose_subdir_fn(unsigned int nr,
317 const char *path,
318 void *data);
319int for_each_file_in_obj_subdir(unsigned int subdir_nr,
320 struct strbuf *path,
321 each_loose_object_fn obj_cb,
322 each_loose_cruft_fn cruft_cb,
323 each_loose_subdir_fn subdir_cb,
324 void *data);
325int for_each_loose_file_in_objdir(const char *path,
326 each_loose_object_fn obj_cb,
327 each_loose_cruft_fn cruft_cb,
328 each_loose_subdir_fn subdir_cb,
329 void *data);
330int for_each_loose_file_in_objdir_buf(struct strbuf *path,
331 each_loose_object_fn obj_cb,
332 each_loose_cruft_fn cruft_cb,
333 each_loose_subdir_fn subdir_cb,
334 void *data);
335
336/* Flags for for_each_*_object() below. */
337enum for_each_object_flags {
338 /* Iterate only over local objects, not alternates. */
339 FOR_EACH_OBJECT_LOCAL_ONLY = (1<<0),
340
341 /* Only iterate over packs obtained from the promisor remote. */
342 FOR_EACH_OBJECT_PROMISOR_ONLY = (1<<1),
343
344 /*
345 * Visit objects within a pack in packfile order rather than .idx order
346 */
347 FOR_EACH_OBJECT_PACK_ORDER = (1<<2),
348};
349
350/*
351 * Iterate over all accessible loose objects without respect to
352 * reachability. By default, this includes both local and alternate objects.
353 * The order in which objects are visited is unspecified.
354 *
355 * Any flags specific to packs are ignored.
356 */
357int for_each_loose_object(each_loose_object_fn, void *,
358 enum for_each_object_flags flags);
359
360/*
361 * Iterate over all accessible packed objects without respect to reachability.
362 * By default, this includes both local and alternate packs.
363 *
364 * Note that some objects may appear twice if they are found in multiple packs.
365 * Each pack is visited in an unspecified order. By default, objects within a
366 * pack are visited in pack-idx order (i.e., sorted by oid).
367 */
368typedef int each_packed_object_fn(const struct object_id *oid,
369 struct packed_git *pack,
370 uint32_t pos,
371 void *data);
372int for_each_object_in_pack(struct packed_git *p,
373 each_packed_object_fn, void *data,
374 enum for_each_object_flags flags);
375int for_each_packed_object(each_packed_object_fn, void *,
376 enum for_each_object_flags flags);
377
90c62155 378#endif /* OBJECT_STORE_H */