Module range_decoder 

Range decoder primitives for the Opus codec.

This module implements the bit-exact range decoder described in RFC 6716 §4.1 (docs/audio/opus/rfc6716-opus.txt). The implementation is clean-room: every routine is transcribed from the prose and pseudocode equations in the RFC; no external library source was consulted.

The range decoder is the SHARED entropy primitive that both the SILK and CELT layers of Opus invoke for every coded symbol. The [oxideav-celt] crate carries its own copy of the same primitive; each crate owns its copy until a shared low-level primitive crate exists in the workspace. The two copies are independent clean-room transcriptions of the same RFC sections and are expected to be behaviourally identical.

The following routines are wired up:

• Initialization (§4.1.1).
• Symbol-update internal helper (§4.1.2).
• Renormalization (§4.1.2.1).
• RangeDecoder::decode_bin for power-of-two ft symbols (§4.1.3.1).
• RangeDecoder::dec_bit_logp (§4.1.3.2).
• RangeDecoder::dec_icdf for inverse-CDF table decoding (§4.1.3.3).
• RangeDecoder::dec_bits for raw bits (§4.1.4).
• RangeDecoder::dec_uint for uniformly-distributed integers (§4.1.5).
• RangeDecoder::tell for whole-bit accounting (§4.1.6.1).
• RangeDecoder::tell_frac for 1/8th-bit-precision accounting (§4.1.6.2).
• RangeDecoder::ec_decode / RangeDecoder::ec_dec_update for the generic two-step symbol path (§4.1.2). These are the building blocks for custom symbol decoders that an inverse-CDF table cannot express directly — notably the CELT §4.3.2.1 coarse-energy Laplace decoder and the §4.3.3 allocation interpolation search, both of which decode against a frequency model computed at run time rather than a fixed icdf[] table.

Structs

RangeDecoder

Bit-exact CELT/SILK range decoder state per RFC 6716 §4.1.