Struct ZstdCodec 

pub struct ZstdCodec {
    pub compression_level: i32,
    pub dict: Option<Vec<u8>>,
}

Zstandard codec with optional pre-trained dictionary support.

ZstdCodec provides both basic Zstd compression and dictionary-accelerated compression. The dictionary approach is particularly effective for RDF triple blocks where subjects, predicates, and objects share common URI prefixes.

Fields

compression_level: i32

Compression level (1-22, default 3)

dict: Option<Vec<u8>>

Optional pre-trained Zstd dictionary bytes

Implementations

impl ZstdCodec

pub fn new(level: i32) -> Self

Create a new ZstdCodec with the given compression level and no dictionary.

Arguments

• level - Compression level clamped to 1-22. Level 3 is the recommended default.

pub fn with_dict(level: i32, dict: Vec<u8>) -> Self

Create a new ZstdCodec with the given compression level and a pre-trained dictionary.

Dictionary-based compression is typically 2-5x more effective on small blocks compared to raw Zstd when the data shares structural patterns with the training samples.

Arguments

• level - Compression level clamped to 1-22
• dict - Pre-trained Zstd dictionary bytes (produced by ZstdCodec::train_dict)

pub fn compress(&self, data: &[u8]) -> Result<Vec<u8>>

Compress data using Zstd, optionally with a pre-trained dictionary.

Returns the compressed bytes. Use decompress with the same dictionary to recover the original data.

pub fn decompress(&self, data: &[u8]) -> Result<Vec<u8>>

Decompress data previously compressed with this codec (same dictionary required).

Errors

Returns TdbError::Other if decompression fails or dictionary mismatch occurs.

pub fn compress_with_stats( &self, data: &[u8], ) -> Result<(Vec<u8>, ZstdCodecStats)>

Compress data and return both the compressed bytes and statistics.

pub fn train_dict(samples: &[Vec<u8>], dict_size: usize) -> Result<Vec<u8>>

Train a Zstd dictionary from a collection of data samples.

This is the key capability for RDF graph compression: by training on a representative sample of RDF triples, the resulting dictionary captures common URI prefixes and literal patterns, yielding significantly better compression for small triple blocks.

Arguments

• samples - Slice of byte vectors representing individual data samples (e.g., serialized triples)
• dict_size - Target dictionary size in bytes. Recommended: 1024-65536 bytes. Larger dictionaries capture more patterns but consume more memory.

Returns

The trained dictionary as a byte vector suitable for passing to ZstdCodec::with_dict.

Errors

Returns TdbError::Other if:

• No samples are provided
• Samples are too small (< 8 bytes each)
• Total sample data exceeds 1GB
• The zstd dictionary trainer fails

pub fn has_dict(&self) -> bool

Check whether this codec has a dictionary loaded.

pub fn dict_size(&self) -> usize

Return the dictionary size in bytes, or 0 if no dictionary is loaded.

Trait Implementations

impl Clone for ZstdCodec

fn clone(&self) -> Self

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ZstdCodec

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for ZstdCodec

fn default() -> Self

Returns the “default value” for a type.