Struct BloomFilterIndexApplier

pub struct BloomFilterIndexApplier {
    region_dir: String,
    region_id: RegionId,
    object_store: ObjectStore,
    file_cache: Option<Arc<FileCache>>,
    puffin_manager_factory: PuffinManagerFactory,
    puffin_metadata_cache: Option<PuffinMetadataCacheRef>,
    bloom_filter_index_cache: Option<Arc<IndexCache<(FileId, ColumnId), BloomFilterMeta>>>,
    predicates: HashMap<ColumnId, Vec<InListPredicate>>,
}

BloomFilterIndexApplier applies bloom filter predicates to the SST file.

Fields

region_dir: String

Directory of the region.

region_id: RegionId

ID of the region.

object_store: ObjectStore

Object store to read the index file.

file_cache: Option<Arc<FileCache>>

File cache to read the index file.

puffin_manager_factory: PuffinManagerFactory

Factory to create puffin manager.

puffin_metadata_cache: Option<PuffinMetadataCacheRef>

Cache for puffin metadata.

bloom_filter_index_cache: Option<Arc<IndexCache<(FileId, ColumnId), BloomFilterMeta>>>

Cache for bloom filter index.

predicates: HashMap<ColumnId, Vec<InListPredicate>>

Bloom filter predicates. For each column, the value will be retained only if it contains all predicates.

Implementations

impl BloomFilterIndexApplier

pub fn new( region_dir: String, region_id: RegionId, object_store: ObjectStore, puffin_manager_factory: PuffinManagerFactory, predicates: HashMap<ColumnId, Vec<InListPredicate>>, ) -> Self

Creates a new BloomFilterIndexApplier.

For each column, the value will be retained only if it contains all predicates.

pub fn with_file_cache(self, file_cache: Option<Arc<FileCache>>) -> Self

pub fn with_puffin_metadata_cache( self, puffin_metadata_cache: Option<PuffinMetadataCacheRef>, ) -> Self

pub fn with_bloom_filter_cache( self, bloom_filter_index_cache: Option<Arc<IndexCache<(FileId, ColumnId), BloomFilterMeta>>>, ) -> Self

pub async fn apply( &self, file_id: FileId, file_size_hint: Option<u64>, row_groups: impl Iterator<Item = (usize, bool)>, ) -> Result<Vec<(usize, Vec<Range<usize>>)>>

Applies bloom filter predicates to the provided SST file and returns a list of row group ranges that match the predicates.

The row_groups iterator provides the row group lengths and whether to search in the row group.

async fn blob_reader( &self, file_id: FileId, column_id: ColumnId, file_size_hint: Option<u64>, ) -> Result<Option<<<<FsPuffinManager<Arc<BoundedStager<FileId>>, ObjectStorePuffinFileAccessor> as PuffinManager>::Reader as PuffinReader>::Blob as BlobGuard>::Reader>>

Creates a blob reader from the cached or remote index file.

Returus None if the column does not have an index.

async fn cached_blob_reader( &self, file_id: FileId, column_id: ColumnId, file_size_hint: Option<u64>, ) -> Result<Option<<<<FsPuffinManager<Arc<BoundedStager<FileId>>, ObjectStorePuffinFileAccessor> as PuffinManager>::Reader as PuffinReader>::Blob as BlobGuard>::Reader>>

Creates a blob reader from the cached index file

fn column_blob_name(column_id: ColumnId) -> String

async fn remote_blob_reader( &self, file_id: FileId, column_id: ColumnId, file_size_hint: Option<u64>, ) -> Result<<<<FsPuffinManager<Arc<BoundedStager<FileId>>, ObjectStorePuffinFileAccessor> as PuffinManager>::Reader as PuffinReader>::Blob as BlobGuard>::Reader>

Creates a blob reader from the remote index file

async fn apply_predicates<R: BloomFilterReader + Send + 'static>( &self, reader: R, predicates: &[InListPredicate], output: &mut [(usize, Vec<Range<usize>>)], ) -> Result<(), Error>