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)