Read manifest entries from a manifest file. More...

#include <manifest_reader.h>

Public Member Functions
Result< std::vector< ManifestEntry > >	Entries ()
	Read all manifest entries in the manifest file.

Result< std::vector< ManifestEntry > >	LiveEntries ()
	Read only live (non-deleted) manifest entries.

virtual Result< ManifestEntryStreamPtr >	EntriesStream ()=0
	Lazily read manifest entries.

virtual Result< ManifestEntryStreamPtr >	LiveEntriesStream ()=0
	Lazily read only live (non-deleted) manifest entries.

virtual ManifestReader &	Select (const std::vector< std::string > &columns)=0
	Select specific columns of data file to read from the manifest entries.

virtual ManifestReader &	FilterPartitions (std::shared_ptr< Expression > expr)=0
	Filter manifest entries by partition filter.

virtual ManifestReader &	FilterPartitions (std::shared_ptr< class PartitionSet > partition_set)=0
	Filter manifest entries to a specific set of partitions.

virtual ManifestReader &	FilterRows (std::shared_ptr< Expression > expr)=0
	Filter manifest entries by row-level filter.

virtual ManifestReader &	CaseSensitive (bool case_sensitive)=0
	Set case sensitivity for column name matching.

virtual ManifestReader &	TryDropStats ()=0
	Try to drop stats from returned DataFile objects.

virtual ManifestReader &	SkipCounter (std::shared_ptr< Counter > counter)=0
	Set a counter to increment for each entry skipped by per-entry filters.

Static Public Member Functions
static bool	ShouldDropStats (const std::vector< std::string > &columns)
	Determine whether stats should be dropped based on selected columns.

static Result< std::unique_ptr< ManifestReader > >	Make (const ManifestFile &manifest, std::shared_ptr< FileIO > file_io, std::shared_ptr< Schema > schema, std::shared_ptr< PartitionSpec > spec, bool is_committed=true)
	Creates a reader for a manifest file.

static Result< std::unique_ptr< ManifestReader > >	Make (const ManifestFile &manifest, std::shared_ptr< FileIO > file_io, std::shared_ptr< Schema > schema, const std::unordered_map< int32_t, std::shared_ptr< PartitionSpec > > &specs_by_id, bool is_committed=true)
	Creates a reader for a manifest file using specs keyed by ID.

static Result< std::unique_ptr< ManifestReader > >	Make (std::string_view manifest_location, std::optional< int64_t > manifest_length, std::shared_ptr< FileIO > file_io, std::shared_ptr< Schema > schema, std::shared_ptr< PartitionSpec > spec, std::unique_ptr< InheritableMetadata > inheritable_metadata, std::optional< int64_t > first_row_id=std::nullopt, bool is_committed=true)
	Creates a reader for a manifest file.

static std::vector< std::string >	WithStatsColumns (const std::vector< std::string > &columns)
	Add stats columns to the column list if needed.

Detailed Description

Read manifest entries from a manifest file.

Implementations must override EntriesStream() and LiveEntriesStream(), returning self-contained streams that may outlive the reader. Entries() and LiveEntries() collect these streams into vectors.

Member Function Documentation

◆ Entries()

Result< std::vector< ManifestEntry > > iceberg::ManifestReader::Entries ( )

Read all manifest entries in the manifest file.

Collects EntriesStream() into a vector.

◆ EntriesStream()

virtual Result< ManifestEntryStreamPtr > iceberg::ManifestReader::EntriesStream ( )

pure virtual

Lazily read manifest entries.

The returned stream is fallible and single-pass. It must own all resources required for consumption and must not depend on this reader remaining alive.

◆ FilterPartitions()

virtual ManifestReader & iceberg::ManifestReader::FilterPartitions ( std::shared_ptr< Expression > expr )

pure virtual

Filter manifest entries by partition filter.

Note: Unlike the Java implementation, this method does not combine new expressions with existing ones. Each call replaces the previous partition filter.

◆ FilterRows()

virtual ManifestReader & iceberg::ManifestReader::FilterRows ( std::shared_ptr< Expression > expr )

pure virtual

Filter manifest entries by row-level filter.

Note: Unlike the Java implementation, this method does not combine new expressions with existing ones. Each call replaces the previous row filter.

◆ LiveEntries()

Result< std::vector< ManifestEntry > > iceberg::ManifestReader::LiveEntries ( )

Read only live (non-deleted) manifest entries.

Collects LiveEntriesStream() into a vector.

◆ LiveEntriesStream()

virtual Result< ManifestEntryStreamPtr > iceberg::ManifestReader::LiveEntriesStream ( )

pure virtual

Lazily read only live (non-deleted) manifest entries.

The returned stream is fallible and single-pass. It must own all resources required for consumption and must not depend on this reader remaining alive.

◆ Make() [1/3]

static Result< std::unique_ptr< ManifestReader > > iceberg::ManifestReader::Make	(	const ManifestFile &	manifest,
		std::shared_ptr< FileIO >	file_io,
		std::shared_ptr< Schema >	schema,
		const std::unordered_map< int32_t, std::shared_ptr< PartitionSpec > > &	specs_by_id,
		bool	is_committed = `true`
	)

static

Creates a reader for a manifest file using specs keyed by ID.

Parameters

manifest	A ManifestFile object containing metadata about the manifest.
file_io	File IO implementation to use.
schema	Schema used to bind the partition type.
specs_by_id	Mapping of partition spec ID to PartitionSpec.
is_committed	Whether the manifest was committed by an older snapshot.

Returns: A Result containing the reader or an error.

◆ Make() [2/3]

static Result< std::unique_ptr< ManifestReader > > iceberg::ManifestReader::Make	(	const ManifestFile &	manifest,
		std::shared_ptr< FileIO >	file_io,
		std::shared_ptr< Schema >	schema,
		std::shared_ptr< PartitionSpec >	spec,
		bool	is_committed = `true`
	)

static

Creates a reader for a manifest file.

Parameters

manifest	A ManifestFile object containing metadata about the manifest.
file_io	File IO implementation to use.
schema	Schema used to bind the partition type.
spec	Partition spec used for this manifest file.
is_committed	Whether the manifest was committed by an older snapshot.

Returns: A Result containing the reader or an error.

◆ Make() [3/3]

static Result< std::unique_ptr< ManifestReader > > iceberg::ManifestReader::Make	(	std::string_view	manifest_location,
		std::optional< int64_t >	manifest_length,
		std::shared_ptr< FileIO >	file_io,
		std::shared_ptr< Schema >	schema,
		std::shared_ptr< PartitionSpec >	spec,
		std::unique_ptr< InheritableMetadata >	inheritable_metadata,
		std::optional< int64_t >	first_row_id = `std::nullopt`,
		bool	is_committed = `true`
	)

static

Creates a reader for a manifest file.

Parameters

manifest_location	Path to the manifest file.
manifest_length	Length of the manifest file.
file_io	File IO implementation to use.
schema	Schema used to bind the partition type.
spec	Partition spec used for this manifest file.
inheritable_metadata	Inheritable metadata.
first_row_id	First row ID to use for the manifest entries.
is_committed	Whether the manifest was committed by an older snapshot.

Returns: A Result containing the reader or an error.

◆ Select()

virtual ManifestReader & iceberg::ManifestReader::Select ( const std::vector< std::string > & columns )

pure virtual

Select specific columns of data file to read from the manifest entries.

Note: Column names should match the names in DataFile schema. Unmatched names will be ignored.; An empty list selects no user columns. A list containing * selects all columns. If this method is not called, all columns are selected.

◆ ShouldDropStats()

static bool iceberg::ManifestReader::ShouldDropStats ( const std::vector< std::string > & columns )

static

Determine whether stats should be dropped based on selected columns.

Returns true if the selected columns do not include any stats columns, or only include record_count (which is a primitive, not a large map).

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

iceberg/manifest/manifest_reader.h

Public Member Functions

Static Public Member Functions

Detailed Description

Member Function Documentation

◆ Entries()

◆ EntriesStream()

◆ FilterPartitions()

◆ FilterRows()

◆ LiveEntries()

◆ LiveEntriesStream()

◆ Make() [1/3]

◆ Make() [2/3]

◆ Make() [3/3]

◆ Select()

◆ ShouldDropStats()