Struct ScanRegion

pub(crate) struct ScanRegion {
    version: Arc<Version>,
    access_layer: AccessLayerRef,
    request: ScanRequest,
    cache_strategy: CacheStrategy,
    parallel_scan_channel_size: usize,
    ignore_inverted_index: bool,
    ignore_fulltext_index: bool,
    ignore_bloom_filter: bool,
    start_time: Option<Instant>,
}

Helper to scans a region by ScanRequest.

ScanRegion collects SSTs and memtables to scan without actually reading them. It creates a Scanner to actually scan these targets in Scanner::scan().

classDiagram class ScanRegion { -VersionRef version -ScanRequest request ~scanner() Scanner ~seq_scan() SeqScan } class Scanner { <> SeqScan UnorderedScan +scan() SendableRecordBatchStream } class SeqScan { -ScanInput input +build() SendableRecordBatchStream } class UnorderedScan { -ScanInput input +build() SendableRecordBatchStream } class ScanInput { -ProjectionMapper mapper -Option~TimeRange~ time_range -Option~Predicate~ predicate -Vec~MemtableRef~ memtables -Vec~FileHandle~ files } class ProjectionMapper { ~output_schema() SchemaRef ~convert(Batch) RecordBatch } ScanRegion -- Scanner ScanRegion o-- ScanRequest Scanner o-- SeqScan Scanner o-- UnorderedScan SeqScan o-- ScanInput UnorderedScan o-- ScanInput Scanner -- SendableRecordBatchStream ScanInput o-- ProjectionMapper SeqScan -- SendableRecordBatchStream UnorderedScan -- SendableRecordBatchStream

Fields

version: Arc<Version>

Version of the region at scan.

access_layer: AccessLayerRef

Access layer of the region.

request: ScanRequest

Scan request.

cache_strategy: CacheStrategy

Cache.

parallel_scan_channel_size: usize

Capacity of the channel to send data from parallel scan tasks to the main task.

ignore_inverted_index: bool

Whether to ignore inverted index.

ignore_fulltext_index: bool

Whether to ignore fulltext index.

ignore_bloom_filter: bool

Whether to ignore bloom filter.

start_time: Option<Instant>

Start time of the scan task.

Implementations

impl ScanRegion

pub(crate) fn new( version: Arc<Version>, access_layer: AccessLayerRef, request: ScanRequest, cache_strategy: CacheStrategy, ) -> ScanRegion

Creates a ScanRegion.

pub(crate) fn with_parallel_scan_channel_size( self, parallel_scan_channel_size: usize, ) -> Self

Sets parallel scan task channel size.

pub(crate) fn with_ignore_inverted_index(self, ignore: bool) -> Self

Sets whether to ignore inverted index.

pub(crate) fn with_ignore_fulltext_index(self, ignore: bool) -> Self

Sets whether to ignore fulltext index.

pub(crate) fn with_ignore_bloom_filter(self, ignore: bool) -> Self

Sets whether to ignore bloom filter.

pub(crate) fn with_start_time(self, now: Instant) -> Self

pub(crate) fn scanner(self) -> Result<Scanner>

Returns a Scanner to scan the region.

pub(crate) fn region_scanner(self) -> Result<RegionScannerRef>

Returns a [RegionScanner] to scan the region.

pub(crate) fn seq_scan(self) -> Result<SeqScan>

Scan sequentially.

pub(crate) fn unordered_scan(self) -> Result<UnorderedScan>

Unordered scan.

fn use_unordered_scan(&self) -> bool

Returns true if the region can use unordered scan for current request.

fn scan_input(self, filter_deleted: bool) -> Result<ScanInput>

Creates a scan input.

fn build_time_range_predicate(&self) -> TimestampRange

Build time range predicate from filters.

fn maybe_remove_field_filters(&mut self)

Remove field filters if the merge mode is MergeMode::LastNonNull.

fn build_invereted_index_applier(&self) -> Option<Arc<InvertedIndexApplier>>

Use the latest schema to build the inverted index applier.

fn build_bloom_filter_applier(&self) -> Option<Arc<BloomFilterIndexApplier>>

Use the latest schema to build the bloom filter index applier.

fn build_fulltext_index_applier(&self) -> Option<Arc<FulltextIndexApplier>>

Use the latest schema to build the fulltext index applier.