Struct FlatConvertFormat

pub(crate) struct FlatConvertFormat {
    metadata: RegionMetadataRef,
    codec: Arc<dyn PrimaryKeyCodec>,
    projected_primary_keys: Vec<(ColumnId, usize, usize)>,
}

Converts a batch that doesn’t have decoded primary key columns into a batch that has decoded primary key columns in flat format.

Fields

metadata: RegionMetadataRef

Metadata of the region.

codec: Arc<dyn PrimaryKeyCodec>

Primary key codec to decode primary keys.

projected_primary_keys: Vec<(ColumnId, usize, usize)>

Projected primary key column information: (column_id, pk_index, column_index in metadata).

Implementations

impl FlatConvertFormat

pub(crate) fn new( metadata: RegionMetadataRef, format_projection: &FormatProjection, codec: Arc<dyn PrimaryKeyCodec>, ) -> Option<Self>

Creates a new FlatConvertFormat.

The format_projection is the projection computed in the FlatReadFormat with the metadata. The codec is the primary key codec of the metadata.

Returns None if there is no primary key.

pub(crate) fn convert(&self, batch: RecordBatch) -> Result<RecordBatch>

Converts a batch to have decoded primary key columns in flat format.

The primary key array in the batch is a dictionary array. We decode each value which is a primary key and reuse the keys array to build a dictionary array for each tag column. The decoded columns are inserted in front of other columns.

fn build_primary_key_column( &self, column_id: ColumnId, pk_index: usize, column_type: &ConcreteDataType, keys: &UInt32Array, decoded_pk_values: &[Option<CompositeValues>], ) -> Result<ArrayRef>

Builds an array for a specific tag column.

It may build a dictionary array if the type is string. Note that the dictionary array may have null values, although keys are not null.