crev-recursive-digest-0.2.0 has been yanked.
Algorithm
Given any digest algorithm H
(a Hash function algorithm),
a RecursiveDigest(H, path)
is:
- for a file:
H("F" || file_content)
- for a symlink:
H("S" || symlink_content)
- for a directory:
H("D" || directory_content)
As you can see a one-letter ASCII prefix is used to make it impossible
to create a file that has the same digest as a directory,
etc. The drawback of this approach is that RecursiveDigest(H, path)
of
a simple file is not the same as just a normal digest of it (H(file_content)
) .
file_content
is just the byte content of a file.
symlink_content
is just the path the symlink is pointing to, as bytes.
directory_content
is created by:
- sorting all entries of a directory by name, in ascending order, using a simple byte-sequence comparison
- for all entries concatenating pairs of:
H(entry_name)
RecursiveDigest(H, entry_path)