tth-cli

Tiger Tree Hash (TTH) calculator — the fingerprint scheme used by DC++, Gnutella, and eD2K/magnet links (urn:tth:<value>). Includes a parallelized batch CLI originally built for hashing comic archives (.cbr/.cbz), but it works on any file.

Install

npm install -g @rishighan/tth-cli

Usage

Hash one or more files, or recursively scan directories for .cbr/.cbz:

tth-cli path/to/file.cbz
tth-cli path/to/comics/ --out results.csv
tth-cli path/to/comics/ --concurrency 4

• With no --out, results print to stdout as tth,size,path.
• With --out, results are written to a CSV file (tth,size,path header).
• --concurrency sets the number of worker threads (default: number of CPU cores).

As a library:

const { tthOfFile, tthOfBuffer } = require('@rishighan/tth-cli/tth-core');

tthOfFile('/path/to/file.cbz'); // => 'QCX7JC7IAEMJZ7PYXKEQMSJIGX2DQX36OBDPURY'
tthOfBuffer(Buffer.from('...'));

How it works

TTH builds a Merkle tree over 1024-byte blocks of the file:

1. Each 1024-byte block is a leaf, hashed as Tiger192(0x00 || block).
2. Leaves are combined pairwise into internal nodes: Tiger192(0x01 || left || right). If a level has an odd node out, it's promoted unchanged to the next level.
3. The root digest is Base32-encoded (RFC 4648, no padding) to produce the final 39-character TTH string.

This matches the THEX/TTH scheme documented by DC++ and used in magnet links. Verified against the well-known reference TTH for a zero-byte file: LWPNACQDBZRYXW3VHJVCJ64QBZNGHOHHHZWCLNQ.

Performance

The underlying Tiger hash implementation (fb-tiger-hash) is pure-JS BigInt and single-threaded throughput is roughly 4 MB/s. The batch CLI parallelizes across files (not within a single file) using Node's worker_threads, so hashing many files scales close to your core count.

License

MIT