Struct RangeCoder 

pub struct RangeCoder { /* private fields */ }

Multi-symbol range coder with table-based Q15 CDF lookup.

Uses a byte-oriented range coder where both encoder and decoder renormalise whenever range < BOT (BOT = 2¹⁶). The encoder emits the top byte of low on each renorm step; the decoder consumes one byte from the bitstream. Both sides use identical arithmetic, so the decoder faithfully tracks the encoder’s interval.

Design

Invariant: range ∈ [BOT, 2³²) after every renormalisation.

Last-symbol optimisation: for the highest-probability symbol (the last entry in a CDF row) the encoder sets range -= step * cum_lo instead of step * (cum_hi - cum_lo). This avoids integer rounding errors that would make low + range > 2³².

Flush: emit exactly 4 bytes of low, which together with the preceding renorm bytes uniquely identify the encoded sequence. The decoder primes its code register with those same 4 leading bytes of the bitstream.

Implementations

impl RangeCoder

pub fn new() -> Self

Create a new range coder in encoder mode.

pub fn init_from_slice(&mut self, data: &[u8]) -> Result<(), CodecError>

Switch to decode mode, priming the code register from data.

Errors

Returns CodecError::InvalidBitstream if data is empty.

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

Flush encoder output and return the byte stream.

Emits the 4 remaining bytes of low, which uniquely identify the terminal interval.

Trait Implementations

impl Clone for RangeCoder

fn clone(&self) -> RangeCoder

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for RangeCoder

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

Formats the value using the given formatter.