Struct DictTrainer 

pub struct DictTrainer { /* private fields */ }

Trains an LZ4 dictionary from sample messages using the COVER algorithm.

Example

use lz4rip_encode::DictTrainer;

let mut trainer = DictTrainer::new(2048);
for msg in &[b"hello world" as &[u8], b"hello rust", b"hello lz4"] {
    trainer.add_sample(msg);
}
let dict = trainer.train();
let compressor = lz4rip_encode::DictCompressor::new(&dict);

Implementations

impl DictTrainer

pub fn new(max_dict_size: usize) -> Self

Create a trainer targeting max_dict_size bytes of output.

Typical values: 2048 for small messages, 4096 for larger ones. The dict is capped at 65535 bytes (LZ4 max match distance).

pub fn add_sample(&mut self, data: &[u8])

Add a training sample.

Samples shorter than 4 bytes or longer than max_dict_size are silently skipped.

pub fn sample_count(&self) -> usize

Number of samples added so far.

pub fn total_bytes(&self) -> usize

Total bytes of sample data added so far.

pub fn train(self) -> Vec<u8>

Train a dictionary from the collected samples. Consumes the trainer, freeing all sample data.

Returns a raw byte buffer. If fewer than 2 samples were added, returns an empty vec.