object-store.hon commit log -G: ignore binary files (e0e7cb8)
   1#ifndef OBJECT_STORE_H
   2#define OBJECT_STORE_H
   3
   4#include "cache.h"
   5#include "oidmap.h"
   6#include "list.h"
   7#include "sha1-array.h"
   8#include "strbuf.h"
   9
  10struct alternate_object_database {
  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
  27        /*
  28         * Path to the alternative object store. If this is a relative path,
  29         * it is relative to the current working directory.
  30         */
  31        char path[FLEX_ARRAY];
  32};
  33void prepare_alt_odb(struct repository *r);
  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
  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;
  78        int index;              /* for builtin/pack-objects.c */
  79        unsigned pack_local:1,
  80                 pack_keep:1,
  81                 pack_keep_in_core:1,
  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
  91struct multi_pack_index;
  92
  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;
 102
 103        struct alternate_object_database *alt_odb_list;
 104        struct alternate_object_database **alt_odb_tail;
 105
 106        /*
 107         * Objects that should be substituted by other objects
 108         * (see git-replace(1)).
 109         */
 110        struct oidmap *replace_map;
 111
 112        struct commit_graph *commit_graph;
 113        unsigned commit_graph_attempted : 1; /* if loading has been attempted */
 114
 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
 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;
 131
 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
 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
 146        /*
 147         * Whether packed_git has already been populated with this repository's
 148         * packs.
 149         */
 150        unsigned packed_git_initialized : 1;
 151};
 152
 153struct raw_object_store *raw_object_store_new(void);
 154void raw_object_store_clear(struct raw_object_store *o);
 155
 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 */
 160void sha1_file_name(struct repository *r, struct strbuf *buf, const unsigned char *sha1);
 161
 162void *map_sha1_file(struct repository *r, const unsigned char *sha1, unsigned long *size);
 163
 164extern void *read_object_file_extended(const struct object_id *oid,
 165                                       enum object_type *type,
 166                                       unsigned long *size, int lookup_replace);
 167static inline void *read_object_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
 168{
 169        return read_object_file_extended(oid, type, size, 1);
 170}
 171
 172/* Read and unpack an object file into memory, write memory to an object file */
 173int oid_object_info(struct repository *r, const struct object_id *, unsigned long *);
 174
 175extern int hash_object_file(const void *buf, unsigned long len,
 176                            const char *type, struct object_id *oid);
 177
 178extern int write_object_file(const void *buf, unsigned long len,
 179                             const char *type, struct object_id *oid);
 180
 181extern int hash_object_file_literally(const void *buf, unsigned long len,
 182                                      const char *type, struct object_id *oid,
 183                                      unsigned flags);
 184
 185extern int pretend_object_file(void *, unsigned long, enum object_type,
 186                               struct object_id *oid);
 187
 188extern int force_object_loose(const struct object_id *oid, time_t mtime);
 189
 190/*
 191 * Open the loose object at path, check its hash, and return the contents,
 192 * type, and size. If the object is a blob, then "contents" may return NULL,
 193 * to allow streaming of large blobs.
 194 *
 195 * Returns 0 on success, negative on error (details may be written to stderr).
 196 */
 197int read_loose_object(const char *path,
 198                      const struct object_id *expected_oid,
 199                      enum object_type *type,
 200                      unsigned long *size,
 201                      void **contents);
 202
 203/*
 204 * Convenience for sha1_object_info_extended() with a NULL struct
 205 * object_info. OBJECT_INFO_SKIP_CACHED is automatically set; pass
 206 * nonzero flags to also set other flags.
 207 */
 208extern int has_sha1_file_with_flags(const unsigned char *sha1, int flags);
 209static inline int has_sha1_file(const unsigned char *sha1)
 210{
 211        return has_sha1_file_with_flags(sha1, 0);
 212}
 213
 214/* Same as the above, except for struct object_id. */
 215extern int has_object_file(const struct object_id *oid);
 216extern int has_object_file_with_flags(const struct object_id *oid, int flags);
 217
 218/*
 219 * Return true iff an alternate object database has a loose object
 220 * with the specified name.  This function does not respect replace
 221 * references.
 222 */
 223extern int has_loose_object_nonlocal(const struct object_id *);
 224
 225extern void assert_oid_type(const struct object_id *oid, enum object_type expect);
 226
 227struct object_info {
 228        /* Request */
 229        enum object_type *typep;
 230        unsigned long *sizep;
 231        off_t *disk_sizep;
 232        unsigned char *delta_base_sha1;
 233        struct strbuf *type_name;
 234        void **contentp;
 235
 236        /* Response */
 237        enum {
 238                OI_CACHED,
 239                OI_LOOSE,
 240                OI_PACKED,
 241                OI_DBCACHED
 242        } whence;
 243        union {
 244                /*
 245                 * struct {
 246                 *      ... Nothing to expose in this case
 247                 * } cached;
 248                 * struct {
 249                 *      ... Nothing to expose in this case
 250                 * } loose;
 251                 */
 252                struct {
 253                        struct packed_git *pack;
 254                        off_t offset;
 255                        unsigned int is_delta;
 256                } packed;
 257        } u;
 258};
 259
 260/*
 261 * Initializer for a "struct object_info" that wants no items. You may
 262 * also memset() the memory to all-zeroes.
 263 */
 264#define OBJECT_INFO_INIT {NULL}
 265
 266/* Invoke lookup_replace_object() on the given hash */
 267#define OBJECT_INFO_LOOKUP_REPLACE 1
 268/* Allow reading from a loose object file of unknown/bogus type */
 269#define OBJECT_INFO_ALLOW_UNKNOWN_TYPE 2
 270/* Do not check cached storage */
 271#define OBJECT_INFO_SKIP_CACHED 4
 272/* Do not retry packed storage after checking packed and loose storage */
 273#define OBJECT_INFO_QUICK 8
 274/* Do not check loose object */
 275#define OBJECT_INFO_IGNORE_LOOSE 16
 276
 277int oid_object_info_extended(struct repository *r,
 278                             const struct object_id *,
 279                             struct object_info *, unsigned flags);
 280
 281/*
 282 * Iterate over the files in the loose-object parts of the object
 283 * directory "path", triggering the following callbacks:
 284 *
 285 *  - loose_object is called for each loose object we find.
 286 *
 287 *  - loose_cruft is called for any files that do not appear to be
 288 *    loose objects. Note that we only look in the loose object
 289 *    directories "objects/[0-9a-f]{2}/", so we will not report
 290 *    "objects/foobar" as cruft.
 291 *
 292 *  - loose_subdir is called for each top-level hashed subdirectory
 293 *    of the object directory (e.g., "$OBJDIR/f0"). It is called
 294 *    after the objects in the directory are processed.
 295 *
 296 * Any callback that is NULL will be ignored. Callbacks returning non-zero
 297 * will end the iteration.
 298 *
 299 * In the "buf" variant, "path" is a strbuf which will also be used as a
 300 * scratch buffer, but restored to its original contents before
 301 * the function returns.
 302 */
 303typedef int each_loose_object_fn(const struct object_id *oid,
 304                                 const char *path,
 305                                 void *data);
 306typedef int each_loose_cruft_fn(const char *basename,
 307                                const char *path,
 308                                void *data);
 309typedef int each_loose_subdir_fn(unsigned int nr,
 310                                 const char *path,
 311                                 void *data);
 312int for_each_file_in_obj_subdir(unsigned int subdir_nr,
 313                                struct strbuf *path,
 314                                each_loose_object_fn obj_cb,
 315                                each_loose_cruft_fn cruft_cb,
 316                                each_loose_subdir_fn subdir_cb,
 317                                void *data);
 318int for_each_loose_file_in_objdir(const char *path,
 319                                  each_loose_object_fn obj_cb,
 320                                  each_loose_cruft_fn cruft_cb,
 321                                  each_loose_subdir_fn subdir_cb,
 322                                  void *data);
 323int for_each_loose_file_in_objdir_buf(struct strbuf *path,
 324                                      each_loose_object_fn obj_cb,
 325                                      each_loose_cruft_fn cruft_cb,
 326                                      each_loose_subdir_fn subdir_cb,
 327                                      void *data);
 328
 329/* Flags for for_each_*_object() below. */
 330enum for_each_object_flags {
 331        /* Iterate only over local objects, not alternates. */
 332        FOR_EACH_OBJECT_LOCAL_ONLY = (1<<0),
 333
 334        /* Only iterate over packs obtained from the promisor remote. */
 335        FOR_EACH_OBJECT_PROMISOR_ONLY = (1<<1),
 336
 337        /*
 338         * Visit objects within a pack in packfile order rather than .idx order
 339         */
 340        FOR_EACH_OBJECT_PACK_ORDER = (1<<2),
 341};
 342
 343/*
 344 * Iterate over all accessible loose objects without respect to
 345 * reachability. By default, this includes both local and alternate objects.
 346 * The order in which objects are visited is unspecified.
 347 *
 348 * Any flags specific to packs are ignored.
 349 */
 350int for_each_loose_object(each_loose_object_fn, void *,
 351                          enum for_each_object_flags flags);
 352
 353/*
 354 * Iterate over all accessible packed objects without respect to reachability.
 355 * By default, this includes both local and alternate packs.
 356 *
 357 * Note that some objects may appear twice if they are found in multiple packs.
 358 * Each pack is visited in an unspecified order. By default, objects within a
 359 * pack are visited in pack-idx order (i.e., sorted by oid).
 360 */
 361typedef int each_packed_object_fn(const struct object_id *oid,
 362                                  struct packed_git *pack,
 363                                  uint32_t pos,
 364                                  void *data);
 365int for_each_object_in_pack(struct packed_git *p,
 366                            each_packed_object_fn, void *data,
 367                            enum for_each_object_flags flags);
 368int for_each_packed_object(each_packed_object_fn, void *,
 369                           enum for_each_object_flags flags);
 370
 371#endif /* OBJECT_STORE_H */