Struct re_chunk::builder::ChunkBuilder

pub struct ChunkBuilder {
    id: ChunkId,
    entity_path: EntityPath,
    row_ids: Vec<RowId>,
    timelines: IntMap<Timeline, TimeColumnBuilder>,
    components: IntMap<ComponentDescriptor, Vec<Option<ArrayRef>>>,
}

Helper to incrementally build a Chunk.

Can be created using Chunk::builder.

Fields

id: ChunkId

entity_path: EntityPath

row_ids: Vec<RowId>

timelines: IntMap<Timeline, TimeColumnBuilder>

components: IntMap<ComponentDescriptor, Vec<Option<ArrayRef>>>

Implementations

impl ChunkBuilder

pub fn new(id: ChunkId, entity_path: EntityPath) -> Self

Initializes a new ChunkBuilder.

See also Chunk::builder.

pub fn with_sparse_row( self, row_id: RowId, timepoint: impl Into<TimePoint>, components: impl IntoIterator<Item = (ComponentDescriptor, Option<ArrayRef>)>, ) -> Self

Add a row’s worth of data using the given sparse component data.

pub fn with_row( self, row_id: RowId, timepoint: impl Into<TimePoint>, components: impl IntoIterator<Item = (ComponentDescriptor, ArrayRef)>, ) -> Self

Add a row’s worth of data using the given component data.

pub fn with_archetype( self, row_id: RowId, timepoint: impl Into<TimePoint>, as_components: &dyn AsComponents, ) -> Self

Add a row’s worth of data by destructuring an archetype into component columns.

pub fn with_component_batch( self, row_id: RowId, timepoint: impl Into<TimePoint>, component_batch: &dyn ComponentBatch, ) -> Self

Add a row’s worth of data by serializing a single ComponentBatch.

pub fn with_component_batches<'a>( self, row_id: RowId, timepoint: impl Into<TimePoint>, component_batches: impl IntoIterator<Item = &'a dyn ComponentBatch>, ) -> Self

Add a row’s worth of data by serializing many ComponentBatches.

pub fn with_sparse_component_batches<'a>( self, row_id: RowId, timepoint: impl Into<TimePoint>, component_batches: impl IntoIterator<Item = (ComponentDescriptor, Option<&'a dyn ComponentBatch>)>, ) -> Self

Add a row’s worth of data by serializing many sparse ComponentBatches.

pub fn with_serialized_batch( self, row_id: RowId, timepoint: impl Into<TimePoint>, component_batch: SerializedComponentBatch, ) -> Self

Add a row’s worth of data by serializing a single ComponentBatch.

pub fn with_serialized_batches( self, row_id: RowId, timepoint: impl Into<TimePoint>, component_batches: impl IntoIterator<Item = SerializedComponentBatch>, ) -> Self

Add a row’s worth of data by serializing many ComponentBatches.

pub fn with_sparse_serialized_batches( self, row_id: RowId, timepoint: impl Into<TimePoint>, component_batches: impl IntoIterator<Item = (ComponentDescriptor, Option<SerializedComponentBatch>)>, ) -> Self

Add a row’s worth of data by serializing many sparse ComponentBatches.

pub fn build(self) -> ChunkResult<Chunk>

Builds and returns the final Chunk.

The arrow datatype of each individual column will be guessed by inspecting the data.

If any component column turns out to be fully sparse (i.e. only null values), that column will be stripped out (how could we guess its datatype without any single value to inspect)?

This is generally the desired behavior but, if you want to make sure to keep fully sparse columns (can be useful e.g. for testing purposes), see ChunkBuilder::build_with_datatypes instead.

This returns an error if the chunk fails to sanity_check.

pub fn build_with_datatypes( self, datatypes: &IntMap<ComponentDescriptor, ArrowDatatype>, ) -> ChunkResult<Chunk>

Builds and returns the final Chunk.

The arrow datatype of each individual column will be guessed by inspecting the data.

If any component column turns out to be fully sparse (i.e. only null values), datatypes will be used as a fallback.

If any component column turns out to be fully sparse (i.e. only null values) and doesn’t have an explicit datatype passed in, that column will be stripped out (how could we guess its datatype without any single value to inspect)?

You should rarely want to keep fully sparse columns around outside of testing scenarios. See Self::build.

This returns an error if the chunk fails to sanity_check.