duckdb::CompressionFunction Class Reference

Collaboration diagram for duckdb::CompressionFunction:

Public Member Functions
	CompressionFunction (CompressionType type, PhysicalType data_type, compression_init_analyze_t init_analyze, compression_analyze_t analyze, compression_final_analyze_t final_analyze, compression_init_compression_t init_compression, compression_compress_data_t compress, compression_compress_finalize_t compress_finalize, compression_init_segment_scan_t init_scan, compression_scan_vector_t scan_vector, compression_scan_partial_t scan_partial, compression_fetch_row_t fetch_row, compression_skip_t skip, compression_init_segment_t init_segment=nullptr, compression_init_append_t init_append=nullptr, compression_append_t append=nullptr, compression_finalize_append_t finalize_append=nullptr, compression_revert_append_t revert_append=nullptr, compression_serialize_state_t serialize_state=nullptr, compression_deserialize_state_t deserialize_state=nullptr, compression_visit_block_ids_t visit_block_ids=nullptr, compression_init_prefetch_t init_prefetch=nullptr, compression_select_t select=nullptr, compression_filter_t filter=nullptr)

Public Attributes
CompressionType	type
	Compression type.
PhysicalType	data_type
	The data type this function can compress.
compression_init_analyze_t	init_analyze
compression_analyze_t	analyze
compression_final_analyze_t	final_analyze
compression_init_compression_t	init_compression
compression_compress_data_t	compress
	compress is called several times (once per vector in the row group)
compression_compress_finalize_t	compress_finalize
	compress_finalize is called after
compression_init_prefetch_t	init_prefetch
	Initialize prefetch state with required I/O data to scan this segment.
compression_init_segment_scan_t	init_scan
	init_scan is called to set up the scan state
compression_scan_vector_t	scan_vector
	scan_vector scans an entire vector using the scan state
compression_scan_partial_t	scan_partial
compression_select_t	select
	scan a subset of a vector
compression_filter_t	filter
	Scan and apply a filter to a vector while scanning.
compression_fetch_row_t	fetch_row
compression_skip_t	skip
	Skip forward in the compressed segment.
compression_init_segment_t	init_segment
	This only really needs to be defined for uncompressed segments.
compression_init_append_t	init_append
	Initialize the append state (optional)
compression_append_t	append
	Append to the compressed segment (optional)
compression_finalize_append_t	finalize_append
	Finalize an append to the segment.
compression_revert_append_t	revert_append
	Revert append (optional)
compression_serialize_state_t	serialize_state
	This is only necessary if the segment state has information that must be written to disk in the metadata.
compression_deserialize_state_t	deserialize_state
	Deserialize the segment state to the metadata (optional)
compression_visit_block_ids_t	visit_block_ids
	Iterate over any extra block ids used by the compression algorithm (optional)
compression_get_segment_info_t	get_segment_info = nullptr
	Get stringified segment information directly from reading the column segment.
CompressionValidity	validity = CompressionValidity::REQUIRES_VALIDITY

Constructor & Destructor Documentation

CompressionFunction()

duckdb::CompressionFunction::CompressionFunction ( CompressionType  type, PhysicalType  data_type, compression_init_analyze_t  init_analyze, compression_analyze_t  analyze, compression_final_analyze_t  final_analyze, compression_init_compression_t  init_compression, compression_compress_data_t  compress, compression_compress_finalize_t  compress_finalize, compression_init_segment_scan_t  init_scan, compression_scan_vector_t  scan_vector, compression_scan_partial_t  scan_partial, compression_fetch_row_t  fetch_row, compression_skip_t  skip, compression_init_segment_t  init_segment = nullptr, compression_init_append_t  init_append = nullptr, compression_append_t  append = nullptr, compression_finalize_append_t  finalize_append = nullptr, compression_revert_append_t  revert_append = nullptr, compression_serialize_state_t  serialize_state = nullptr, compression_deserialize_state_t  deserialize_state = nullptr, compression_visit_block_ids_t  visit_block_ids = nullptr, compression_init_prefetch_t  init_prefetch = nullptr, compression_select_t  select = nullptr, compression_filter_t  filter = nullptr  )

inline

        : type(type), data_type(data_type), init_analyze(init_analyze), analyze(analyze), final_analyze(final_analyze),
          init_compression(init_compression), compress(compress), compress_finalize(compress_finalize),
          init_prefetch(init_prefetch), init_scan(init_scan), scan_vector(scan_vector), scan_partial(scan_partial),
          select(select), filter(filter), fetch_row(fetch_row), skip(skip), init_segment(init_segment),
          init_append(init_append), append(append), finalize_append(finalize_append), revert_append(revert_append),
          serialize_state(serialize_state), deserialize_state(deserialize_state), visit_block_ids(visit_block_ids) {
    }

Member Data Documentation

init_analyze

compression_init_analyze_t duckdb::CompressionFunction::init_analyze

Analyze step: determine which compression function is the most effective init_analyze is called once to set up the analyze state

analyze

compression_analyze_t duckdb::CompressionFunction::analyze

analyze is called several times (once per vector in the row group) analyze should return true, unless compression is no longer possible with this compression method in that case false should be returned

final_analyze

compression_final_analyze_t duckdb::CompressionFunction::final_analyze

final_analyze should return the score of the compression function ideally this is the exact number of bytes required to store the data this is not required/enforced: it can be an estimate as well also this function can return DConstants::INVALID_INDEX to skip this compression method

init_compression

compression_init_compression_t duckdb::CompressionFunction::init_compression

Compression step: actually compress the data init_compression is called once to set up the comperssion state

scan_partial

compression_scan_partial_t duckdb::CompressionFunction::scan_partial

scan_partial scans a subset of a vector this can request > vector_size as well this is used if a vector crosses segment boundaries, or for child columns of lists

fetch_row

compression_fetch_row_t duckdb::CompressionFunction::fetch_row

fetch an individual row from the compressed vector used for index lookups

init_segment

compression_init_segment_t duckdb::CompressionFunction::init_segment

This only really needs to be defined for uncompressed segments.

Initialize a compressed segment (optional)

serialize_state

compression_serialize_state_t duckdb::CompressionFunction::serialize_state

This is only necessary if the segment state has information that must be written to disk in the metadata.

Serialize the segment state to the metadata (optional)

get_segment_info

compression_get_segment_info_t duckdb::CompressionFunction::get_segment_info = nullptr

Get stringified segment information directly from reading the column segment.

This is only necessary if you want to convey more information about the segment in the 'pragma_storage_info' result

validity

CompressionValidity duckdb::CompressionFunction::validity = CompressionValidity::REQUIRES_VALIDITY

Whether the validity mask should be separately compressed or this compression function can also be used to decompress the validity

---

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

• external/duckdb/duckdb.hpp