2 5aa81393 2020-01-06 stsp * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
4 718b3ab0 2018-03-17 stsp * Permission to use, copy, modify, and distribute this software for any
5 718b3ab0 2018-03-17 stsp * purpose with or without fee is hereby granted, provided that the above
6 718b3ab0 2018-03-17 stsp * copyright notice and this permission notice appear in all copies.
8 718b3ab0 2018-03-17 stsp * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 718b3ab0 2018-03-17 stsp * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 718b3ab0 2018-03-17 stsp * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 718b3ab0 2018-03-17 stsp * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 718b3ab0 2018-03-17 stsp * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 718b3ab0 2018-03-17 stsp * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 718b3ab0 2018-03-17 stsp * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 718b3ab0 2018-03-17 stsp /* An open pack file. */
18 718b3ab0 2018-03-17 stsp struct got_pack {
19 718b3ab0 2018-03-17 stsp char *path_packfile;
21 d7464085 2018-07-09 stsp uint8_t *map;
22 718b3ab0 2018-03-17 stsp size_t filesize;
23 876c234b 2018-09-10 stsp struct got_privsep_child *privsep_child;
24 ab2f42e7 2019-11-10 stsp struct got_delta_cache *delta_cache;
27 876c234b 2018-09-10 stsp const struct got_error *got_pack_stop_privsep_child(struct got_pack *);
28 d7464085 2018-07-09 stsp const struct got_error *got_pack_close(struct got_pack *);
30 668a20f6 2020-03-18 stsp const struct got_error *got_pack_parse_offset_delta(off_t *, size_t *,
31 668a20f6 2020-03-18 stsp struct got_pack *, off_t, int);
32 668a20f6 2020-03-18 stsp const struct got_error *got_pack_resolve_delta_chain(struct got_delta_chain *,
33 668a20f6 2020-03-18 stsp struct got_packidx *, struct got_pack *, off_t, size_t, int, size_t,
34 668a20f6 2020-03-18 stsp unsigned int);
35 668a20f6 2020-03-18 stsp const struct got_error *got_pack_parse_object_type_and_size(uint8_t *,
36 668a20f6 2020-03-18 stsp uint64_t *, size_t *, struct got_pack *, off_t);
38 1510f469 2018-09-09 stsp #define GOT_PACK_PREFIX "pack-"
39 1510f469 2018-09-09 stsp #define GOT_PACKFILE_SUFFIX ".pack"
40 1510f469 2018-09-09 stsp #define GOT_PACKIDX_SUFFIX ".idx"
41 1510f469 2018-09-09 stsp #define GOT_PACKFILE_NAMELEN (strlen(GOT_PACK_PREFIX) + \
42 1510f469 2018-09-09 stsp SHA1_DIGEST_STRING_LENGTH - 1 + \
43 1510f469 2018-09-09 stsp strlen(GOT_PACKFILE_SUFFIX))
44 1510f469 2018-09-09 stsp #define GOT_PACKIDX_NAMELEN (strlen(GOT_PACK_PREFIX) + \
45 1510f469 2018-09-09 stsp SHA1_DIGEST_STRING_LENGTH - 1 + \
46 1510f469 2018-09-09 stsp strlen(GOT_PACKIDX_SUFFIX))
48 718b3ab0 2018-03-17 stsp /* See Documentation/technical/pack-format.txt in Git. */
50 718b3ab0 2018-03-17 stsp struct got_packidx_trailer {
51 718b3ab0 2018-03-17 stsp u_int8_t packfile_sha1[SHA1_DIGEST_LENGTH];
52 718b3ab0 2018-03-17 stsp u_int8_t packidx_sha1[SHA1_DIGEST_LENGTH];
53 718b3ab0 2018-03-17 stsp } __attribute__((__packed__));
55 57b35b75 2018-06-22 stsp struct got_packidx_object_id {
56 57b35b75 2018-06-22 stsp u_int8_t sha1[SHA1_DIGEST_LENGTH];
57 57b35b75 2018-06-22 stsp } __attribute__((__packed__));
59 718b3ab0 2018-03-17 stsp /* Ignore pack index version 1 which is no longer written by Git. */
60 718b3ab0 2018-03-17 stsp #define GOT_PACKIDX_VERSION 2
62 718b3ab0 2018-03-17 stsp struct got_packidx_v2_hdr {
63 57b35b75 2018-06-22 stsp uint32_t *magic; /* big endian */
64 718b3ab0 2018-03-17 stsp #define GOT_PACKIDX_V2_MAGIC 0xff744f63 /* "\377t0c" */
65 57b35b75 2018-06-22 stsp uint32_t *version;
68 718b3ab0 2018-03-17 stsp * Each entry N in the fanout table contains the number of objects in
69 718b3ab0 2018-03-17 stsp * the packfile whose SHA1 begins with a byte less than or equal to N.
70 718b3ab0 2018-03-17 stsp * The last entry (index 255) contains the number of objects in the
71 718b3ab0 2018-03-17 stsp * pack file whose first SHA1 byte is <= 0xff, and thus records the
72 718b3ab0 2018-03-17 stsp * total number of objects in the pack file. All pointer variables
73 718b3ab0 2018-03-17 stsp * below point to tables with a corresponding number of entries.
75 57b35b75 2018-06-22 stsp uint32_t *fanout_table; /* values are big endian */
76 57b35b75 2018-06-22 stsp #define GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS (0xff + 1)
78 718b3ab0 2018-03-17 stsp /* Sorted SHA1 checksums for each object in the pack file. */
79 57b35b75 2018-06-22 stsp struct got_packidx_object_id *sorted_ids;
81 718b3ab0 2018-03-17 stsp /* CRC32 of the packed representation of each object. */
82 718b3ab0 2018-03-17 stsp uint32_t *crc32;
84 718b3ab0 2018-03-17 stsp /* Offset into the pack file for each object. */
85 718b3ab0 2018-03-17 stsp uint32_t *offsets; /* values are big endian */
86 718b3ab0 2018-03-17 stsp #define GOT_PACKIDX_OFFSET_VAL_MASK 0x7fffffff
87 718b3ab0 2018-03-17 stsp #define GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX 0x80000000
89 718b3ab0 2018-03-17 stsp /* Large offsets table is empty for pack files < 2 GB. */
90 718b3ab0 2018-03-17 stsp uint64_t *large_offsets; /* values are big endian */
92 57b35b75 2018-06-22 stsp struct got_packidx_trailer *trailer;
95 6fd11751 2018-06-04 stsp /* An open pack index file. */
96 6fd11751 2018-06-04 stsp struct got_packidx {
97 6fd11751 2018-06-04 stsp char *path_packidx; /* actual on-disk path */
99 57b35b75 2018-06-22 stsp uint8_t *map;
100 57b35b75 2018-06-22 stsp size_t len;
101 5e6be232 2019-11-08 stsp size_t nlargeobj;
102 57b35b75 2018-06-22 stsp struct got_packidx_v2_hdr hdr; /* convenient pointers into map */
105 718b3ab0 2018-03-17 stsp struct got_packfile_hdr {
106 718b3ab0 2018-03-17 stsp uint32_t signature;
107 718b3ab0 2018-03-17 stsp #define GOT_PACKFILE_SIGNATURE 0x5041434b /* 'P' 'A' 'C' 'K' */
108 718b3ab0 2018-03-17 stsp uint32_t version; /* big endian */
109 718b3ab0 2018-03-17 stsp #define GOT_PACKFILE_VERSION 2
110 718b3ab0 2018-03-17 stsp uint32_t nobjects; /* big endian */
113 718b3ab0 2018-03-17 stsp struct got_packfile_obj_hdr {
115 718b3ab0 2018-03-17 stsp * The object size field uses a variable length encoding:
116 718b3ab0 2018-03-17 stsp * size0...sizeN form a 4+7+7+...+7 bit integer, where size0 is the
117 718b3ab0 2018-03-17 stsp * least significant part and sizeN is the most significant part.
118 718b3ab0 2018-03-17 stsp * If the MSB of a size byte is set, an additional size byte follows.
119 718b3ab0 2018-03-17 stsp * Of the 7 remaining bits of size0, the first 3 bits indicate the
120 718b3ab0 2018-03-17 stsp * object's type, and the remaining 4 bits contribute to the size.
122 718b3ab0 2018-03-17 stsp uint8_t *size; /* variable length */
123 718b3ab0 2018-03-17 stsp #define GOT_PACK_OBJ_SIZE_MORE 0x80
124 718b3ab0 2018-03-17 stsp #define GOT_PACK_OBJ_SIZE0_TYPE_MASK 0x70 /* See struct got_object->type */
125 718b3ab0 2018-03-17 stsp #define GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT 4
126 718b3ab0 2018-03-17 stsp #define GOT_PACK_OBJ_SIZE0_VAL_MASK 0x0f
127 718b3ab0 2018-03-17 stsp #define GOT_PACK_OBJ_SIZE_VAL_MASK 0x7f
130 718b3ab0 2018-03-17 stsp /* If object is not a DELTA type. */
131 718b3ab0 2018-03-17 stsp struct got_packfile_object_data {
132 718b3ab0 2018-03-17 stsp uint8_t *data; /* compressed */
135 718b3ab0 2018-03-17 stsp /* If object is of type GOT_OBJ_TYPE_REF_DELTA. */
136 718b3ab0 2018-03-17 stsp struct got_packfile_object_data_ref_delta {
137 718b3ab0 2018-03-17 stsp uint8_t sha1[SHA1_DIGEST_LENGTH];
138 718b3ab0 2018-03-17 stsp uint8_t *delta_data; /* compressed */
141 718b3ab0 2018-03-17 stsp /* If object is of type GOT_OBJ_TYPE_OFFSET_DELTA. */
142 718b3ab0 2018-03-17 stsp struct got_packfile_object_data_offset_delta {
144 718b3ab0 2018-03-17 stsp * This offset is interpreted as a negative offset from
145 718b3ab0 2018-03-17 stsp * the got_packfile_obj_hdr corresponding to this object.
146 718b3ab0 2018-03-17 stsp * The size provided in the header specifies the amount
147 718b3ab0 2018-03-17 stsp * of compressed delta data that follows.
149 718b3ab0 2018-03-17 stsp * This field uses a variable length encoding of N bytes,
150 718b3ab0 2018-03-17 stsp * where the MSB is always set except for the last byte.
151 718b3ab0 2018-03-17 stsp * The value is encoded as a series of N 7 bit integers,
152 718b3ab0 2018-03-17 stsp * which are concatenated, and if N > 1 the value 2^7 +
153 718b3ab0 2018-03-17 stsp * 2^14 + ... + 2^(7 * (n-1)) is added to the result.
155 718b3ab0 2018-03-17 stsp uint8_t *offset; /* variable length */
156 718b3ab0 2018-03-17 stsp #define GOT_PACK_OBJ_DELTA_OFF_MORE 0x80
157 718b3ab0 2018-03-17 stsp #define GOT_PACK_OBJ_DELTA_OFF_VAL_MASK 0x7f
158 718b3ab0 2018-03-17 stsp uint8_t *delta_data; /* compressed */
161 718b3ab0 2018-03-17 stsp struct got_packfile_obj_data {
163 718b3ab0 2018-03-17 stsp struct got_packfile_object_data data;
164 718b3ab0 2018-03-17 stsp struct got_packfile_object_data_ref_delta ref_delta;
165 718b3ab0 2018-03-17 stsp struct got_packfile_object_data_offset_delta offset_delta;
166 718b3ab0 2018-03-17 stsp } __attribute__((__packed__));
167 718b3ab0 2018-03-17 stsp } __attribute__((__packed__));
169 817c5a18 2018-09-09 stsp const struct got_error *got_packidx_init_hdr(struct got_packidx *, int);
170 6fd11751 2018-06-04 stsp const struct got_error *got_packidx_open(struct got_packidx **,
171 6d5a9006 2020-12-16 yzhong int, const char *, int);
172 4277420a 2019-06-29 stsp const struct got_error *got_packidx_close(struct got_packidx *);
173 509c9973 2021-04-10 stsp const struct got_error *got_packidx_get_packfile_path(char **, struct got_packidx *);
174 1510f469 2018-09-09 stsp int got_packidx_get_object_idx(struct got_packidx *, struct got_object_id *);
175 dd88155e 2019-06-29 stsp const struct got_error *got_packidx_match_id_str_prefix(
176 dd88155e 2019-06-29 stsp struct got_object_id_queue *, struct got_packidx *, const char *);
178 718b3ab0 2018-03-17 stsp const struct got_error *got_packfile_open_object(struct got_object **,
179 2090a03d 2018-09-09 stsp struct got_pack *, struct got_packidx *, int, struct got_object_id *);
180 d582f26c 2020-03-18 stsp const struct got_error *got_pack_get_delta_chain_max_size(uint64_t *,
181 d582f26c 2020-03-18 stsp struct got_delta_chain *, struct got_pack *);
182 85a703fa 2019-01-13 stsp const struct got_error *got_pack_get_max_delta_object_size(uint64_t *,
183 42c69117 2019-11-10 stsp struct got_object *, struct got_pack *);
184 4788f1ce 2020-03-18 stsp const struct got_error *got_pack_dump_delta_chain_to_file(size_t *,
185 4788f1ce 2020-03-18 stsp struct got_delta_chain *, struct got_pack *, FILE *, FILE *, FILE *);
186 668a20f6 2020-03-18 stsp const struct got_error *got_pack_dump_delta_chain_to_mem(uint8_t **, size_t *,
187 668a20f6 2020-03-18 stsp struct got_delta_chain *, struct got_pack *);
188 24140570 2018-09-09 stsp const struct got_error *got_packfile_extract_object(struct got_pack *,
189 3840f4c9 2018-09-12 stsp struct got_object *, FILE *, FILE *, FILE *);
190 718b3ab0 2018-03-17 stsp const struct got_error *got_packfile_extract_object_to_mem(uint8_t **, size_t *,
191 7e212e3d 2018-09-09 stsp struct got_object *, struct got_pack *);
192 1510f469 2018-09-09 stsp struct got_pack *got_repo_get_cached_pack(struct got_repository *, const char *);
