Package io.trino.plugin.deltalake.transactionlog

Class TransactionLogAccess

java.lang.Object

io.trino.plugin.deltalake.transactionlog.TransactionLogAccess

public class TransactionLogAccess
extends Object

Constructor Summary

Constructors

Constructor	Description
TransactionLogAccess(TypeManager typeManager, CheckpointSchemaManager checkpointSchemaManager, DeltaLakeConfig deltaLakeConfig, io.trino.plugin.hive.FileFormatDataSourceStats fileFormatDataSourceStats, TrinoFileSystemFactory fileSystemFactory, io.trino.plugin.hive.parquet.ParquetReaderConfig parquetReaderConfig)

Method Summary

Modifier and Type	Method	Description
static String	canonicalizeColumnName(String columnName)
static List<DeltaLakeColumnMetadata>	columnsWithStats(MetadataEntry metadataEntry, ProtocolEntry protocolEntry, TypeManager typeManager)
static com.google.common.collect.ImmutableList<DeltaLakeColumnMetadata>	columnsWithStats(List<DeltaLakeColumnMetadata> schema, List<String> partitionColumns)
void	flushCache()
Stream<AddFileEntry>	getActiveFiles(ConnectorSession session, TableSnapshot tableSnapshot, MetadataEntry metadataEntry, ProtocolEntry protocolEntry, TupleDomain<DeltaLakeColumnHandle> partitionConstraint, Predicate<String> addStatsMinMaxColumnFilter)
Stream<AddFileEntry>	getActiveFiles(ConnectorSession session, TableSnapshot tableSnapshot, MetadataEntry metadataEntry, ProtocolEntry protocolEntry, TupleDomain<DeltaLakeColumnHandle> partitionConstraint, Set<DeltaLakeColumnHandle> projectedColumns)
Stream<CommitInfoEntry>	getCommitInfoEntries(ConnectorSession session, TableSnapshot tableSnapshot)
io.airlift.jmx.CacheStatsMBean	getDataFileMetadataCacheStats()
Stream<DeltaLakeTransactionLogEntry>	getJsonEntries(TrinoFileSystem fileSystem, String transactionLogDir, List<Long> forVersions)
io.airlift.jmx.CacheStatsMBean	getMetadataCacheStats()
MetadataEntry	getMetadataEntry(ConnectorSession session, TableSnapshot tableSnapshot)
List<Long>	getPastTableVersions(TrinoFileSystem fileSystem, String transactionLogDir, Instant startAt, long lastVersion)	Returns a stream of transaction log versions between startAt point in time and lastVersion.
Stream<ProtocolEntry>	getProtocolEntries(ConnectorSession session, TableSnapshot tableSnapshot)
ProtocolEntry	getProtocolEntry(ConnectorSession session, TableSnapshot tableSnapshot)
Stream<RemoveFileEntry>	getRemoveEntries(ConnectorSession session, TableSnapshot tableSnapshot)
Map<Class<?>,Object>	getTransactionLogEntries(ConnectorSession session, TableSnapshot tableSnapshot, Set<CheckpointEntryIterator.EntryType> entryTypes, Function<Stream<DeltaLakeTransactionLogEntry>,Stream<Object>> entryMapper)
void	invalidateCache(SchemaTableName schemaTableName, Optional<String> tableLocation)
TableSnapshot	loadSnapshot(ConnectorSession session, SchemaTableName table, String tableLocation)
static <T> Map<CanonicalColumnName,T>	toCanonicalNameKeyedMap(Map<String,T> map, Map<String,CanonicalColumnName> canonicalColumnNames)
static <T> Map<String,T>	toOriginalNameKeyedMap(Map<CanonicalColumnName,T> map)	Deprecated.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

TransactionLogAccess

@Inject
public TransactionLogAccess(TypeManager typeManager,
 CheckpointSchemaManager checkpointSchemaManager,
 DeltaLakeConfig deltaLakeConfig,
 io.trino.plugin.hive.FileFormatDataSourceStats fileFormatDataSourceStats,
 TrinoFileSystemFactory fileSystemFactory,
 io.trino.plugin.hive.parquet.ParquetReaderConfig parquetReaderConfig)

Method Details

getDataFileMetadataCacheStats

public io.airlift.jmx.CacheStatsMBean getDataFileMetadataCacheStats()

getMetadataCacheStats

public io.airlift.jmx.CacheStatsMBean getMetadataCacheStats()

loadSnapshot

public TableSnapshot loadSnapshot(ConnectorSession session,
 SchemaTableName table,
 String tableLocation)
                           throws IOException

Throws:

IOException

flushCache

public void flushCache()

invalidateCache

public void invalidateCache(SchemaTableName schemaTableName,
 Optional<String> tableLocation)

getMetadataEntry

public MetadataEntry getMetadataEntry(ConnectorSession session,
 TableSnapshot tableSnapshot)

getActiveFiles

public Stream<AddFileEntry> getActiveFiles(ConnectorSession session,
 TableSnapshot tableSnapshot,
 MetadataEntry metadataEntry,
 ProtocolEntry protocolEntry,
 TupleDomain<DeltaLakeColumnHandle> partitionConstraint,
 Set<DeltaLakeColumnHandle> projectedColumns)

getActiveFiles

public Stream<AddFileEntry> getActiveFiles(ConnectorSession session,
 TableSnapshot tableSnapshot,
 MetadataEntry metadataEntry,
 ProtocolEntry protocolEntry,
 TupleDomain<DeltaLakeColumnHandle> partitionConstraint,
 Predicate<String> addStatsMinMaxColumnFilter)

columnsWithStats

public static List<DeltaLakeColumnMetadata> columnsWithStats(MetadataEntry metadataEntry,
 ProtocolEntry protocolEntry,
 TypeManager typeManager)

columnsWithStats

public static com.google.common.collect.ImmutableList<DeltaLakeColumnMetadata> columnsWithStats(List<DeltaLakeColumnMetadata> schema,
 List<String> partitionColumns)

getRemoveEntries

public Stream<RemoveFileEntry> getRemoveEntries(ConnectorSession session,
 TableSnapshot tableSnapshot)

getTransactionLogEntries

public Map<Class<?>,Object> getTransactionLogEntries(ConnectorSession session,
 TableSnapshot tableSnapshot,
 Set<CheckpointEntryIterator.EntryType> entryTypes,
 Function<Stream<DeltaLakeTransactionLogEntry>,Stream<Object>> entryMapper)

getProtocolEntry

public ProtocolEntry getProtocolEntry(ConnectorSession session,
 TableSnapshot tableSnapshot)

getProtocolEntries

public Stream<ProtocolEntry> getProtocolEntries(ConnectorSession session,
 TableSnapshot tableSnapshot)

getCommitInfoEntries

public Stream<CommitInfoEntry> getCommitInfoEntries(ConnectorSession session,
 TableSnapshot tableSnapshot)

getJsonEntries

public Stream<DeltaLakeTransactionLogEntry> getJsonEntries(TrinoFileSystem fileSystem,
 String transactionLogDir,
 List<Long> forVersions)

getPastTableVersions

public List<Long> getPastTableVersions(TrinoFileSystem fileSystem,
 String transactionLogDir,
 Instant startAt,
 long lastVersion)

Returns a stream of transaction log versions between startAt point in time and lastVersion.

canonicalizeColumnName

public static String canonicalizeColumnName(String columnName)

toCanonicalNameKeyedMap

public static <T> Map<CanonicalColumnName,T> toCanonicalNameKeyedMap(Map<String,T> map,
 Map<String,CanonicalColumnName> canonicalColumnNames)

toOriginalNameKeyedMap

@Deprecated
public static <T> Map<String,T> toOriginalNameKeyedMap(Map<CanonicalColumnName,T> map)

Deprecated.