#directory-tree #hash #hashing #single #utility #path

bin+lib dasher

Utility for reducing a whole directory tree to a single hash digest

5 releases

0.3.3 Mar 6, 2023
0.3.2 Mar 3, 2023
0.3.1 Mar 3, 2023
0.3.0 Mar 3, 2023
0.2.0 Mar 3, 2023

#1080 in Filesystem

Download history 5/week @ 2024-02-26 13/week @ 2024-03-11 50/week @ 2024-04-01

63 downloads per month

Apache-2.0 OR MIT

20KB
254 lines

Dasher

A Directory Hasher

dasher is a small utility intended to allow you to condense the "status" of an entire directory tree down to a single hash digest. Thus, you can tell that something has changed in the tree (but not what has changed) if the hash has changed.

dasher currently only uses and supports the SHA3-256 hashing algorithm, but more may be added in future.

Installation

dasher is easiest to install via Cargo

$ cargo install dasher

You can, of course, also clone this repository and cargo build/cargo run the code that way.

Usage

dasher has a very simple CLI, akin to other hashing tools like sha1sum. Simply call dasher with one or more paths and dasher will return the hash for each of them.

$ dasher src test_data
6a181e4113e4c2abf39ada58158772316fe0444d3476c084759148fdd5be7e8c        src
3b5e49ac9126759771d677bdacbc18a63ff94ad4e07718c18347254d7b9c6cb1        test_data

Hashing scheme

The hashing scheme is, in essence, generating a Merkle tree, but with extra steps. Each node in the directory tree has its name hashed, then its contents, then those hashes are concatenated with a separator byte based on the node's type, and that data is hashed again to generate the node's hash. This process is repeated, from the bottom up in the directory tree, until all nodes have been hashed and a final hash for the entire directory can be returned.

For normal files, the node hash is simply:

hash(hash(name) + byte + hash(content))

For directories, the node hash includes arbitrarily many content hashes, one per sub-node:

hash(hash(name) + byte + hash(content_1) + byte + hash(content_2) + ... + byte + hash(content_n))

Finally, for symlinks, the link isn't followed. Instead, the content hash is the hash of the path to the file the link points to.

hash(hash(name) + byte + hash(path))

Traversal of the directory is not recursive --- rather, the process starts with the leaves in the lexicographically "first" directory. For example, in the directory

.git
├── COMMIT_EDITMSG
├── config
├── info
│   └── exclude
└── logs
    ├── HEAD
    └── refs
        ├── heads
        │   ├── add-cli
        │   ├── dh-main
        │   └── main
        └── remotes
            └── origin
                └── main

the first item to be hashed would be info/exclude, followed by the directory hash of the info directory. After that, logs/refs/heads/* would be hashed, then logs/refs/heads/remotes/origin/main, then logs/refs/heads/remotes/origin as a directory, then logs/refs/heads/remotes as a directory, then finally climbing back up to hash logs/refs, since both its sub-directories have been hashed.

In a way, I guess you could consider this as being recursive, but it is not implemented recursively.

License

dasher is licensed under either of

at your option.

Is it any good?

yes.

Dependencies

~1.5MB
~14K SLoC