Collaboration diagram for duckdb::TupleDataCollection:

[legend]

Public Member Functions
	TupleDataCollection (BufferManager &buffer_manager, shared_ptr< TupleDataLayout > layout_ptr, MemoryTag tag, shared_ptr< ArenaAllocator > stl_allocator=nullptr)
	Constructs a TupleDataCollection with the specified layout.

	TupleDataCollection (ClientContext &context, shared_ptr< TupleDataLayout > layout_ptr, MemoryTag tag, shared_ptr< ArenaAllocator > stl_allocator=nullptr)

unique_ptr< TupleDataCollection >	CreateUnique () const
	Create a TupleDataCollection with the same layout.

shared_ptr< TupleDataLayout >	GetLayoutPtr () const
	Get the layout (shared pointer) of the rows.

const TupleDataLayout &	GetLayout () const
	The layout of the stored rows.

idx_t	TuplesPerBlock () const
	How many tuples fit per block.

const idx_t &	Count () const
	The number of rows stored in the tuple data collection.

idx_t	ChunkCount () const
	The number of chunks stored in the tuple data collection.

idx_t	SizeInBytes () const
	The size (in bytes) of the blocks held by this tuple data collection.

void	Unpin ()
	Unpins all held pins.

void	SetPartitionIndex (idx_t index)
	Sets the partition index of this tuple data collection.

vector< data_ptr_t >	GetRowBlockPointers () const
	Gets the pointers to the start of every block.

void	DestroyChunks (idx_t chunk_idx_begin, idx_t chunk_idx_end)
	Destroy the blocks corresponding to the chunk indices.

void	InitializeAppend (TupleDataAppendState &append_state, TupleDataPinProperties properties=TupleDataPinProperties::UNPIN_AFTER_DONE)
	Initializes an Append state - useful for optimizing many appends made to the same tuple data collection.

void	InitializeAppend (TupleDataAppendState &append_state, vector< column_t > column_ids, TupleDataPinProperties properties=TupleDataPinProperties::UNPIN_AFTER_DONE)
	Initializes an Append state - useful for optimizing many appends made to the same tuple data collection.

void	InitializeAppend (TupleDataPinState &pin_state, TupleDataPinProperties=TupleDataPinProperties::UNPIN_AFTER_DONE)

void	InitializeChunkState (TupleDataChunkState &chunk_state, vector< column_t > column_ids={})

void	Append (DataChunk &new_chunk, const SelectionVector &append_sel= *FlatVector::IncrementalSelectionVector(), idx_t append_count=DConstants::INVALID_INDEX)
	Append a DataChunk directly to this TupleDataCollection - calls InitializeAppend and Append internally.

void	Append (DataChunk &new_chunk, vector< column_t > column_ids, const SelectionVector &append_sel= *FlatVector::IncrementalSelectionVector(), const idx_t append_count=DConstants::INVALID_INDEX)
	Append a DataChunk directly to this TupleDataCollection - calls InitializeAppend and Append internally.

void	Append (TupleDataAppendState &append_state, DataChunk &new_chunk, const SelectionVector &append_sel= *FlatVector::IncrementalSelectionVector(), const idx_t append_count=DConstants::INVALID_INDEX)
	Append a DataChunk to this TupleDataCollection using the specified Append state.

void	Append (TupleDataPinState &pin_state, TupleDataChunkState &chunk_state, DataChunk &new_chunk, const SelectionVector &append_sel= *FlatVector::IncrementalSelectionVector(), const idx_t append_count=DConstants::INVALID_INDEX)
	Append a DataChunk to this TupleDataCollection using the specified pin and Chunk states.

void	AppendUnified (TupleDataPinState &pin_state, TupleDataChunkState &chunk_state, DataChunk &new_chunk, const SelectionVector &append_sel= *FlatVector::IncrementalSelectionVector(), const idx_t append_count=DConstants::INVALID_INDEX)

void	Build (TupleDataPinState &pin_state, TupleDataChunkState &chunk_state, const idx_t append_offset, const idx_t append_count)
	Builds out the buffer space for the specified Chunk state.

void	Scatter (TupleDataChunkState &chunk_state, const DataChunk &new_chunk, const SelectionVector &append_sel, const idx_t append_count) const
	Scatters the given DataChunk to the rows in the specified Chunk state.

void	Scatter (TupleDataChunkState &chunk_state, const Vector &source, const column_t column_id, const SelectionVector &append_sel, const idx_t append_count) const
	Scatters the given Vector to the given column id to the rows in the specified Chunk state.

void	CopyRows (TupleDataChunkState &chunk_state, TupleDataChunkState &input, const SelectionVector &append_sel, const idx_t append_count) const
	Copy rows from input to the built Chunk state.

void	FindHeapPointers (TupleDataChunkState &chunk_state, const idx_t chunk_count) const
	Finds the heap pointers of the rows in the given Chunk state.

void	FinalizePinState (TupleDataPinState &pin_state, TupleDataSegment &segment)
	Finalizes the Pin state, releasing or storing blocks.

void	FinalizePinState (TupleDataPinState &pin_state)
	Finalizes the Pin state, releasing or storing blocks.

void	Combine (TupleDataCollection &other)
	Appends the other TupleDataCollection to this, destroying the other data collection.

void	Combine (unique_ptr< TupleDataCollection > other)
	Appends the other TupleDataCollection to this, destroying the other data collection.

void	Reset ()
	Resets the TupleDataCollection, clearing all data.

void	InitializeChunk (DataChunk &chunk) const
	Initializes a chunk with the correct types that can be used to call Append/Scan.

void	InitializeChunk (DataChunk &chunk, const vector< column_t > &columns) const
	Initializes a chunk with the correct types that can be used to call Append/Scan for the given columns.

void	InitializeScanChunk (const TupleDataScanState &state, DataChunk &chunk) const
	Initializes a chunk with the correct types for a given scan state.

void	InitializeScan (TupleDataScanState &state, TupleDataPinProperties properties=TupleDataPinProperties::UNPIN_AFTER_DONE) const
	Initializes a Scan state for scanning all columns.

void	InitializeScan (TupleDataScanState &state, vector< column_t > column_ids, TupleDataPinProperties properties=TupleDataPinProperties::UNPIN_AFTER_DONE) const
	Initializes a Scan state for scanning a subset of the columns.

void	InitializeScan (TupleDataParallelScanState &state, TupleDataPinProperties properties=TupleDataPinProperties::UNPIN_AFTER_DONE) const
	Initialize a parallel scan over the tuple data collection over all columns.

void	InitializeScan (TupleDataParallelScanState &gstate, vector< column_t > column_ids, TupleDataPinProperties properties=TupleDataPinProperties::UNPIN_AFTER_DONE) const
	Initialize a parallel scan over the tuple data collection over a subset of the columns.

idx_t	FetchChunk (TupleDataScanState &state, idx_t chunk_idx, bool init_heap, optional_ptr< SortKeyPayloadState > sort_key_payload_state=nullptr)
	Grab the chunk state for the given chunk index, returns the count of the chunk.

bool	Scan (TupleDataScanState &state, DataChunk &result)
	Scans a DataChunk from the TupleDataCollection.

bool	Scan (TupleDataParallelScanState &gstate, TupleDataLocalScanState &lstate, DataChunk &result)
	Scans a DataChunk from the TupleDataCollection.

bool	ScanComplete (const TupleDataScanState &state) const
	Whether the last scan has been completed on this TupleDataCollection.

idx_t	Seek (TupleDataScanState &state, const idx_t target_chunk)
	Seeks to the specified chunk index, returning the total row count before it.

void	Gather (Vector &row_locations, const SelectionVector &scan_sel, const idx_t scan_count, DataChunk &result, const SelectionVector &target_sel, vector< unique_ptr< Vector > > &cached_cast_vectors) const
	Gathers a DataChunk from the TupleDataCollection, given the specific row locations (requires full pin)

void	Gather (Vector &row_locations, const SelectionVector &scan_sel, const idx_t scan_count, const vector< column_t > &column_ids, DataChunk &result, const SelectionVector &target_sel, vector< unique_ptr< Vector > > &cached_cast_vectors) const

void	Gather (Vector &row_locations, const SelectionVector &sel, const idx_t scan_count, const column_t column_id, Vector &result, const SelectionVector &target_sel, optional_ptr< Vector > cached_cast_vector) const

string	ToString ()
	Converts this TupleDataCollection to a string representation.

void	Print ()
	Prints the string representation of this TupleDataCollection.

void	VerifyEverythingPinned () const
	Verify that all blocks are pinned.

Static Public Member Functions
static TupleDataScatterFunction	GetScatterFunction (const LogicalType &type, bool within_collection=false)
	Gets the scatter function for the given type.

static TupleDataGatherFunction	GetGatherFunction (const LogicalType &type)
	Gets the gather function for the given type.

static TupleDataScatterFunction	GetSortKeyScatterFunction (const LogicalType &type, SortKeyType sort_key_type)
	Gets the scatter function for the given sort key type.

static TupleDataGatherFunction	GetSortKeyGatherFunction (const LogicalType &type, SortKeyType sort_key_type)
	Gets the gather function for the given sort key type.

static void	InitializeChunkState (TupleDataChunkState &chunk_state, const vector< LogicalType > &types, vector< column_t > column_ids={})

static void	ToUnifiedFormat (TupleDataChunkState &chunk_state, DataChunk &new_chunk)
	Creates a UnifiedVectorFormat in the given Chunk state for the given DataChunk.

static void	GetVectorData (const TupleDataChunkState &chunk_state, UnifiedVectorFormat result[])
	Gets the UnifiedVectorFormat from the Chunk state as an array.

static void	ResetCachedCastVectors (TupleDataChunkState &chunk_state, const vector< column_t > &column_ids)
	Resets the cached cache vectors (used for ARRAY/LIST casts)

static void	ComputeHeapSizes (TupleDataChunkState &chunk_state, const DataChunk &new_chunk, const SelectionVector &append_sel, const idx_t append_count)
	Computes the heap sizes for the new DataChunk that will be appended.

static void	SortKeyComputeHeapSizes (TupleDataChunkState &chunk_state, const DataChunk &new_chunk, const SelectionVector &append_sel, const idx_t append_count, const SortKeyType sort_key_type)
	Computes the heap sizes for a SortKey layout.

Private Member Functions
void	Initialize ()
	Initializes the TupleDataCollection (called by the constructor)

void	GetAllColumnIDs (vector< column_t > &column_ids)
	Gets all column ids.

void	AddSegment (unsafe_arena_ptr< TupleDataSegment > segment)
	Adds a segment to this TupleDataCollection.

bool	NextScanIndex (TupleDataScanState &scan_state, idx_t &segment_index, idx_t &chunk_index)
	Get the next segment/chunk index for the scan.

void	ScanAtIndex (TupleDataPinState &pin_state, TupleDataChunkState &chunk_state, const vector< column_t > &column_ids, idx_t segment_index, idx_t chunk_index, DataChunk &result)
	Scans the chunk at the given segment/chunk indices.

void	Verify () const
	Verify count/data size of this collection.

Static Private Member Functions
static void	ComputeHeapSizes (Vector &heap_sizes_v, const Vector &source_v, TupleDataVectorFormat &source, const SelectionVector &append_sel, const idx_t append_count)
	Computes the heap sizes for the specific Vector that will be appended.

static void	WithinCollectionComputeHeapSizes (Vector &heap_sizes_v, const Vector &source_v, TupleDataVectorFormat &source_format, const SelectionVector &append_sel, const idx_t append_count, const UnifiedVectorFormat &list_data)
	Computes the heap sizes for the specific Vector that will be appended (within a list)

static void	ComputeFixedWithinCollectionHeapSizes (Vector &heap_sizes_v, const Vector &source_v, TupleDataVectorFormat &source_format, const SelectionVector &append_sel, const idx_t append_count, const UnifiedVectorFormat &list_data)
	Computes the heap sizes for the fixed-size type Vector that will be appended (within a list)

static void	StringWithinCollectionComputeHeapSizes (Vector &heap_sizes_v, const Vector &source_v, TupleDataVectorFormat &source_format, const SelectionVector &append_sel, const idx_t append_count, const UnifiedVectorFormat &list_data)
	Computes the heap sizes for the string Vector that will be appended (within a list)

static void	StructWithinCollectionComputeHeapSizes (Vector &heap_sizes_v, const Vector &source_v, TupleDataVectorFormat &source_format, const SelectionVector &append_sel, const idx_t append_count, const UnifiedVectorFormat &list_data)
	Computes the heap sizes for the struct Vector that will be appended (within a list)

static void	CollectionWithinCollectionComputeHeapSizes (Vector &heap_sizes_v, const Vector &source_v, TupleDataVectorFormat &source_format, const SelectionVector &append_sel, const idx_t append_count, const UnifiedVectorFormat &list_data)
	Computes the heap sizes for the list Vector that will be appended (within a list)

Private Attributes
TaskScheduler &	scheduler
	Task scheduler for parallel destruction.

shared_ptr< ArenaAllocator >	stl_allocator
	Shared allocator for STL allocations.

shared_ptr< TupleDataLayout >	layout_ptr
	The layout of the TupleDataCollection.

const TupleDataLayout &	layout

const MemoryTag	tag
	Memory tag (for keeping track what the allocated memory belongs to)

shared_ptr< TupleDataAllocator >	allocator
	The TupleDataAllocator.

idx_t	count
	The number of entries stored in the TupleDataCollection.

idx_t	data_size
	The size (in bytes) of this TupleDataCollection.

unsafe_arena_vector< unsafe_arena_ptr< TupleDataSegment > >	segments
	The data segments of the TupleDataCollection.

unsafe_arena_vector< TupleDataScatterFunction >	scatter_functions
	The set of scatter functions.

unsafe_arena_vector< TupleDataGatherFunction >	gather_functions
	The set of gather functions.

optional_idx	partition_index
	Partition index (optional, if partitioned)

Friends
class	TupleDataChunkIterator

class	PartitionedTupleData

Detailed Description

TupleDataCollection represents a set of buffer-managed data stored in row format FIXME: rename to RowDataCollection after we phase it out

Member Function Documentation

◆ InitializeAppend()

void duckdb::TupleDataCollection::InitializeAppend	(	TupleDataPinState &	pin_state,
		TupleDataPinProperties	= `TupleDataPinProperties::UNPIN_AFTER_DONE`
	)

Initializes the Pin state of an Append state

Useful for optimizing many appends made to the same tuple data collection

◆ InitializeChunkState() [1/2]

void duckdb::TupleDataCollection::InitializeChunkState	(	TupleDataChunkState &	chunk_state,
		vector< column_t >	column_ids = `{}`
	)

Initializes the Chunk state of an Append state

Useful for optimizing many appends made to the same tuple data collection

◆ InitializeChunkState() [2/2]

static void duckdb::TupleDataCollection::InitializeChunkState	(	TupleDataChunkState &	chunk_state,
		const vector< LogicalType > &	types,
		vector< column_t >	column_ids = `{}`
	)

static

Initializes the Chunk state of an Append state

Useful for optimizing many appends made to the same tuple data collection

◆ AppendUnified()

void duckdb::TupleDataCollection::AppendUnified	(	TupleDataPinState &	pin_state,
		TupleDataChunkState &	chunk_state,
		DataChunk &	new_chunk,
		const SelectionVector &	append_sel = `*FlatVector::IncrementalSelectionVector()`,
		const idx_t	append_count = `DConstants::INVALID_INDEX`
	)

Append a DataChunk to this TupleDataCollection using the specified pin and Chunk states

ToUnifiedFormat has already been called

◆ Gather() [1/2]

void duckdb::TupleDataCollection::Gather	(	Vector &	row_locations,
		const SelectionVector &	scan_sel,
		const idx_t	scan_count,
		const vector< column_t > &	column_ids,
		DataChunk &	result,
		const SelectionVector &	target_sel,
		vector< unique_ptr< Vector > > &	cached_cast_vectors
	)		const

Gathers a DataChunk (only the columns given by column_ids) from the TupleDataCollection, given the specific row locations (requires full pin)

◆ Gather() [2/2]

void duckdb::TupleDataCollection::Gather	(	Vector &	row_locations,
		const SelectionVector &	sel,
		const idx_t	scan_count,
		const column_t	column_id,
		Vector &	result,
		const SelectionVector &	target_sel,
		optional_ptr< Vector >	cached_cast_vector
	)		const

Gathers a Vector (from the given column id) from the TupleDataCollection given the specific row locations (requires full pin)

The documentation for this class was generated from the following file:

external/duckdb/duckdb.cpp

Public Member Functions

Static Public Member Functions

Private Member Functions

Static Private Member Functions

Private Attributes

Friends