Struct HopDataPool 

pub struct HopDataPool { /* private fields */ }

HOP data pool: parity-db metadata + content-addressed blob files.

Implementations

impl HopDataPool

pub fn new( max_size: u64, max_user_size: u64, retention_secs: u64, data_dir: PathBuf, rate_limit_cfg: RateLimitConfig, metrics: HopMetrics, ) -> Result<Self, HopError>

Open or create the metadata DB, rebuild counter caches by iterating it, and remove orphan .blob files in the same pass.

pub fn insert( &self, data: Vec<u8>, recipients: RecipientVec, sender_id: SenderId, signer: MultiSigner, signature: MultiSignature, submit_timestamp: u64, ) -> Result<HopHash, HopError>

Insert data into the pool.

Returns the hash of the data.

pub fn get(&self, hash: &HopHash) -> Option<Vec<u8>>

Get data from the pool by content hash.

pub fn get_with_auth( &self, hash: &HopHash, ) -> Option<(Vec<u8>, MultiSigner, MultiSignature, u64)>

Get data alongside the submitter’s MultiSigner, hop_submit signature, and submit timestamp.

Used by the promoter so the unsigned promotion extrinsic can carry the user’s submit-time signature for runtime-side verification.

pub fn claim( &self, hash: &HopHash, signature: &[u8], ) -> Result<Vec<u8>, HopError>

Claim data from the pool (read-only). Verifies the signature against recipient public keys. Returns the data if the signature matches a recipient.

This does NOT mark the recipient as claimed — call ack after receiving the data to confirm receipt.

Returns AlreadyClaimed if the recipient has already acked (data may be deleted).

pub fn ack(&self, hash: &HopHash, signature: &[u8]) -> Result<(), HopError>

Acknowledge receipt of claimed data. Marks the recipient as claimed and triggers cleanup when all recipients have acked.

Idempotent: acking a recipient that already acked returns Ok(()).

pub fn status(&self) -> PoolStatus

Get pool status.

pub fn cleanup_expired(&self, promotion_buffer_secs: u64) -> u64

Remove expired entries, release their quotas, return total bytes freed.

Uses expiry_index to enumerate expired hashes in O(K), not O(N) over the meta column. Entries are processed in batches so a long outage can’t buffer the whole expired set in RAM before any progress lands.

promotion_buffer_secs does not affect what is cleaned up; as in Self::get_promotable the caller owns the promotion window. Here it only scopes the promotion-backlog gauge, snapshot after the sweep.

pub fn get_promotable( &self, current_block: HopBlockNumber, buffer_secs: u64, limit: usize, ) -> Vec<HopHash>

Return hashes of entries within buffer_secs of expiry that have not yet been promoted. Returns up to limit hashes. Use Self::get to read blob data when needed. The maintenance task runs periodically, so remaining entries are picked up next cycle.

Uses expiry_index to walk only entries inside the promotion window; the meta column is touched only for candidates that pass the window filter, so cost scales with the window size, not the pool.

pub fn mark_promoted(&self, hash: &HopHash)

Mark an entry as promoted to permanent on-chain storage.

pub fn record_promotion_attempt( &self, hash: &HopHash, current_block: HopBlockNumber, check_interval_blocks: u32, )

Record a promotion attempt: bumps the per-entry attempt counter and schedules the next eligible block via exponential back-off. The maintenance task will skip the entry until then. Once MAX_PROMOTION_ATTEMPTS is reached the entry is left to expire.

Called on both an Err from submit_local (the tx pool rejected us) and an Ok followed by a runtime check that the data is not yet on-chain (the tx was accepted into the pool but never included). The backoff schedule is identical for both cases.