Struct S3Cleaner 

pub struct S3Cleaner { /* private fields */ }

S3-backed cleaner for a single hyperfile rooted at a given s3://<bucket>/<prefix> URI.

The cleaner does not retain or open the file’s data. It only reads the published inode and walks the metadata B+tree to enumerate referenced segments, then deletes the unreferenced ones via DeleteObjects.

Concurrency: the cleaner is not aware of in-flight writers. If a writer publishes a new checkpoint while collect / prune is running, the result reflects whatever inode was loaded at the start of the call. New segments published after that point are outside the configured [low, high] range and are not affected.

Implementations

impl S3Cleaner

pub async fn new(root_uri: &str) -> Result<Self>

Construct a cleaner for root_uri, using the AWS SDK’s default credential / region resolution (aws_config::load_from_env).

Returns InvalidInput if root_uri is not a valid S3 URI.

pub async fn with_client(client: &Client, root_uri: &str) -> Result<Self>

Construct a cleaner using a caller-supplied Client. Use this when you need to control region / endpoint / credential resolution explicitly (for example, in tests or in a tool that already has a configured client).

Returns InvalidInput if root_uri is not a valid S3 URI.

Trait Implementations

impl Cleaner for S3Cleaner

async fn prune(&self, low: SegmentId, high: SegmentId) -> Result<()>

Prune unreferenced segments in [low, high]. Implemented as collect followed by DeleteObjects of the unreferenced list.

async fn collect( &self, low: SegmentId, high: SegmentId, ) -> Result<(Vec<SegmentId>, Vec<SegmentId>)>

Walk the metadata tree of the checkpoint identified by high (or the latest if high == 0) and return (referenced, unreferenced) segment id lists within [low, high].