ncdu file list format v1                    11-08-2007


This document describes the file format of the file list export feature of
ncdu. This document was used as a reference for implementing the export
feature, and may be useful to anyone who wants to read or write these files.



definitions
  byte  = unsigned 8bit integer
  2uint = unsigned 16bit integer
  4uint = unsigned 32bit integer
  8int  = signed 64bit integer
  8uint = unsigned 64bit integer

  All integers are in network (big-endian) byte order.



header
 ssssvnttttttttiiii

  s = "ncdu" (hex: 6e 63 64 75)
  v = version of the file format, always 1 (hex: 01)
  n = NULL-terminated string containing the package name and version of the
      program that created this file. ncdu 1.3 would be "ncdu 1.3\0". Maximum
      of 64 bytes including the terminating NULL byte.
  t = 8int, unix timestamp of when the file was created
  i = 4uint, total number of directory items



directory item
 llfssssssssaaaaaaaaddddn

  l = 2uint, level. 0 for parent, 1 for sub, 2 for sub sub, etc. (see below)
  f = byte, flags: (mostly the same as defined in ncdu.h)
       01  item is a directory
       02  item is a file
       04  error while reading this item
       08  item was excluded because it was on an other filesystem
       10  item was exluded using exclude patterns
  s = 8uint, disk usage of current item (st_blocks * 512)
  a = 8uint, apparent size of current item (st_size)
  n = NULL-terminated string, name of the current item, absolute maximum length
      of 8192 bytes including the terminating NULL-byte.



global layout
  An ncdu datafile always starts with the header, this header also specifies
  the number of directory items. All data after the last item should be
  ignored.

  After the header follows the list of directory items. The first directory
  item should be the parent directory. This is the only item to have a full and
  absolute path as name, all other items should only include the name of the
  item itself. This is also the only item to have a level of 0.

  The second directory item should level set to 1, and should be an item
  located in the parent directory. Any higher level than the previous item
  indicates that the item is located in the previous item, the same level
  indicates this item is located in the same directory as the previous item,
  and a lower level indicates that this item is located in de same directory
  as the parent directory of the previous item.

  The disk usage and apparent size of each directory item is accumulated, it
  should already include the sizes of the subitems.

  Example showing the use of levels and names:

    [header](items=7)
    [directory item](level=0, name=/parent/dir)       /parent/dir
    [directory item](level=1, name=file1)             /parent/dir/file1
    [directory item](level=1, name=dir1)              /parent/dir/dir1
    [directory item](level=2, name=sfile1)            /parent/dir/dir1/sfile1
    [directory item](level=2, name=sfile2)            /parent/dir/dir1/sfile2
    [directory item](level=1, name=file2)             /parent/dir/file2
    [directory item](level=1, name=file3)             /parent/dir/file3