Function zstd::dict::from_sample_iterator

pub fn from_sample_iterator<I, R>(
    samples: I,
    max_size: usize,
) -> Result<Vec<u8>>
where
    I: IntoIterator<Item = Result<R>>,
    R: Read,

Train a dictionary from multiple samples.

Unlike from_samples, this does not require having a list of all samples. It also allows running into an error when iterating through the samples.

They will still be copied to a continuous array and fed to from_continuous.

• samples is an iterator of individual samples to train on.
• max_size is the maximum size of the dictionary to generate.

The result is the dictionary data. You can, for example, feed it to CDict::create.

Examples

// Train from a couple of json files.
let dict_buffer = zstd::dict::from_sample_iterator(
    ["file_a.json", "file_b.json"]
        .into_iter()
        .map(|filename| std::fs::File::open(filename)),
    10_000,  // 10kB dictionary
).unwrap();

use std::io::BufRead as _;
// Treat each line from stdin as a separate sample.
let dict_buffer = zstd::dict::from_sample_iterator(
    std::io::stdin().lock().lines().map(|line: std::io::Result<String>| {
        // Transform each line into a `Cursor<Vec<u8>>` so they implement Read.
        line.map(String::into_bytes)
            .map(std::io::Cursor::new)
    }),
    10_000,  // 10kB dictionary
).unwrap();