Andrew's git - gitweb.git/blob - Documentation/technical/pack-format.txt

   1GIT pack format
   2===============
   3
   4= pack-*.pack file has the following format:
   5
   6   - The header appears at the beginning and consists of the following:
   7
   8     4-byte signature
   9     4-byte version number (network byte order)
  10     4-byte number of objects contained in the pack (network byte order)
  11
  12     Observation: we cannot have more than 4G versions ;-) and
  13     more than 4G objects in a pack.
  14
  15   - The header is followed by number of object entries, each of
  16     which looks like this:
  17
  18     (undeltified representation)
  19     n-byte type and length (4-bit type, (n-1)*7+4-bit length)
  20     compressed data
  21
  22     (deltified representation)
  23     n-byte type and length (4-bit type, (n-1)*7+4-bit length)
  24     20-byte base object name
  25     compressed delta data
  26
  27     Observation: length of each object is encoded in a variable
  28     length format and is not constrained to 32-bit or anything.
  29
  30  - The trailer records 20-byte SHA1 checksum of all of the above.
  31
  32= pack-*.idx file has the following format:
  33
  34  - The header consists of 256 4-byte network byte order
  35    integers.  N-th entry of this table records the number of
  36    objects in the corresponding pack, the first byte of whose
  37    object name are smaller than N.  This is called the
  38    'first-level fan-out' table.
  39
  40    Observation: we would need to extend this to an array of
  41    8-byte integers to go beyond 4G objects per pack, but it is
  42    not strictly necessary.
  43
  44  - The header is followed by sorted 28-byte entries, one entry
  45    per object in the pack.  Each entry is:
  46
  47    4-byte network byte order integer, recording where the
  48    object is stored in the packfile as the offset from the
  49    beginning.
  50
  51    20-byte object name.
  52
  53    Observation: we would definitely need to extend this to
  54    8-byte integer plus 20-byte object name to handle a packfile
  55    that is larger than 4GB.
  56
  57  - The file is concluded with a trailer:
  58
  59    A copy of the 20-byte SHA1 checksum at the end of
  60    corresponding packfile.
  61
  62    20-byte SHA1-checksum of all of the above.
  63
  64Pack Idx file:
  65
  66        idx
  67            +--------------------------------+
  68            | fanout[0] = 2                  |-.
  69            +--------------------------------+ |
  70            | fanout[1]                      | |
  71            +--------------------------------+ |
  72            | fanout[2]                      | |
  73            ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ |
  74            | fanout[255]                    | |
  75            +--------------------------------+ |
  76main        | offset                         | |
  77index       | object name 00XXXXXXXXXXXXXXXX | |
  78table       +--------------------------------+ | 
  79            | offset                         | |
  80            | object name 00XXXXXXXXXXXXXXXX | |
  81            +--------------------------------+ |
  82          .-| offset                         |<+
  83          | | object name 01XXXXXXXXXXXXXXXX |
  84          | +--------------------------------+
  85          | | offset                         |
  86          | | object name 01XXXXXXXXXXXXXXXX |
  87          | ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  88          | | offset                         |
  89          | | object name FFXXXXXXXXXXXXXXXX |
  90          | +--------------------------------+
  91trailer   | | packfile checksum              |
  92          | +--------------------------------+
  93          | | idxfile checksum               |
  94          | +--------------------------------+
  95          .-------.      
  96                  |
  97Pack file entry: <+
  98
  99     packed object header:
 100        1-byte type (upper 4-bit)
 101               size0 (lower 4-bit) 
 102        n-byte sizeN (as long as MSB is set, each 7-bit)
 103                size0..sizeN form 4+7+7+..+7 bit integer, size0
 104                is the most significant part.
 105     packed object data:
 106        If it is not DELTA, then deflated bytes (the size above
 107                is the size before compression).
 108        If it is DELTA, then
 109          20-byte base object name SHA1 (the size above is the
 110                size of the delta data that follows).
 111          delta data, deflated.