pack-objects: a bit of document about struct object_entry
[git/git.git] / pack-objects.h
CommitLineData
2834bc27
VM
1#ifndef PACK_OBJECTS_H
2#define PACK_OBJECTS_H
3
8d6ccce1
NTND
4/*
5 * basic object info
6 * -----------------
7 * idx.oid is filled up before delta searching starts. idx.crc32 is
8 * only valid after the object is written out and will be used for
9 * generating the index. idx.offset will be both gradually set and
10 * used in writing phase (base objects get offset first, then deltas
11 * refer to them)
12 *
13 * "size" is the uncompressed object size. Compressed size of the raw
14 * data for an object in a pack is not stored anywhere but is computed
15 * and made available when reverse .idx is made.
16 *
17 * "hash" contains a path name hash which is used for sorting the
18 * delta list and also during delta searching. Once prepare_pack()
19 * returns it's no longer needed.
20 *
21 * source pack info
22 * ----------------
23 * The (in_pack, in_pack_offset) tuple contains the location of the
24 * object in the source pack. in_pack_header_size allows quickly
25 * skipping the header and going straight to the zlib stream.
26 *
27 * "type" and "in_pack_type" both describe object type. in_pack_type
28 * may contain a delta type, while type is always the canonical type.
29 *
30 * deltas
31 * ------
32 * Delta links (delta, delta_child and delta_sibling) are created to
33 * reflect that delta graph from the source pack then updated or added
34 * during delta searching phase when we find better deltas.
35 *
36 * delta_child and delta_sibling are last needed in
37 * compute_write_order(). "delta" and "delta_size" must remain valid
38 * at object writing phase in case the delta is not cached.
39 *
40 * If a delta is cached in memory and is compressed, delta_data points
41 * to the data and z_delta_size contains the compressed size. If it's
42 * uncompressed [1], z_delta_size must be zero. delta_size is always
43 * the uncompressed size and must be valid even if the delta is not
44 * cached.
45 *
46 * [1] during try_delta phase we don't bother with compressing because
47 * the delta could be quickly replaced with a better one.
48 */
2834bc27
VM
49struct object_entry {
50 struct pack_idx_entry idx;
51 unsigned long size; /* uncompressed size */
52 struct packed_git *in_pack; /* already in pack */
53 off_t in_pack_offset;
54 struct object_entry *delta; /* delta base object */
55 struct object_entry *delta_child; /* deltified objects who bases me */
56 struct object_entry *delta_sibling; /* other deltified objects who
57 * uses the same base as me
58 */
59 void *delta_data; /* cached delta (uncompressed) */
60 unsigned long delta_size; /* delta data size (uncompressed) */
61 unsigned long z_delta_size; /* delta data size (compressed) */
62 enum object_type type;
63 enum object_type in_pack_type; /* could be delta */
64 uint32_t hash; /* name hint hash */
7cc8f971 65 unsigned int in_pack_pos;
2834bc27
VM
66 unsigned char in_pack_header_size;
67 unsigned preferred_base:1; /*
68 * we do not pack this, but is available
69 * to be used as the base object to delta
70 * objects against.
71 */
72 unsigned no_try_delta:1;
73 unsigned tagged:1; /* near the very tip of refs */
74 unsigned filled:1; /* assigned write-order */
4cf2143e
JK
75
76 /*
77 * State flags for depth-first search used for analyzing delta cycles.
7dbabbbe
JK
78 *
79 * The depth is measured in delta-links to the base (so if A is a delta
80 * against B, then A has a depth of 1, and B a depth of 0).
4cf2143e
JK
81 */
82 enum {
83 DFS_NONE = 0,
84 DFS_ACTIVE,
85 DFS_DONE
86 } dfs_state;
7dbabbbe 87 int depth;
2834bc27
VM
88};
89
90struct packing_data {
91 struct object_entry *objects;
92 uint32_t nr_objects, nr_alloc;
93
94 int32_t *index;
95 uint32_t index_size;
96};
97
98struct object_entry *packlist_alloc(struct packing_data *pdata,
99 const unsigned char *sha1,
100 uint32_t index_pos);
101
102struct object_entry *packlist_find(struct packing_data *pdata,
103 const unsigned char *sha1,
104 uint32_t *index_pos);
105
68fb36eb
VM
106static inline uint32_t pack_name_hash(const char *name)
107{
108 uint32_t c, hash = 0;
109
110 if (!name)
111 return 0;
112
113 /*
114 * This effectively just creates a sortable number from the
115 * last sixteen non-whitespace characters. Last characters
116 * count "most", so things that end in ".c" sort together.
117 */
118 while ((c = *name++) != 0) {
119 if (isspace(c))
120 continue;
121 hash = (hash >> 2) + (c << 24);
122 }
123 return hash;
124}
125
2834bc27 126#endif