Struct Writer 

pub struct Writer { /* private fields */ }

Implementations

impl Writer

pub fn create( path: &Path, asset_uuid: [u8; 16], timescale: u32, compress: bool, ) -> Result<Self>

Create a new CAVS-1 file at path. compress enables per-chunk zstd (chunks that don’t shrink are stored raw regardless).

pub fn set_zstd_level(&mut self, level: i32)

zstd level for chunk storage/wire compression (default 3).

pub fn sign_with(&mut self, secret: &[u8; 32])

Sign the packed content with this Ed25519 secret key. The signature (over the canonical content message, see cavs_hash::content_signature_message) and the public key are embedded as sig.ed25519 / sig.pubkey meta entries at finish().

pub fn add_chunks_parallel( &mut self, data: &[u8], ranges: &[Range<usize>], ) -> Result<Vec<u32>>

Add a batch of chunk payloads given as byte ranges of data, preparing the expensive per-chunk work (BLAKE3 identity + zstd compression) on all cores. Ingest order, dedup decisions and the stored byte stream are exactly those of calling Writer::add_chunk on each range in order — the resulting file is byte-identical; only wall-clock changes (3–7× faster on real payloads, more at high zstd levels).

pub fn add_chunk(&mut self, raw: &[u8]) -> Result<u32>

Add one chunk payload. Returns its chunk-table index. Duplicate payloads (same BLAKE3) are not stored again.

pub fn add_track(&mut self, track: TrackRecord) -> Result<()>

pub fn add_segment(&mut self, segment: SegmentRecord) -> Result<()>

pub fn pin_dict(&mut self, chunk_index: u32) -> Result<()>

Mark a chunk as part of the global dictionary (privileged reuse: init segments, bootstrap assets, shared headers…).

pub fn set_meta(&mut self, key: &str, value: &str)

pub fn chunk_count(&self) -> u32

pub fn finish(self) -> Result<PackStats>

Write all tables, the section directory and the final superblock.