Package io.trino.parquet

Class ParquetTypeUtils

java.lang.Object

io.trino.parquet.ParquetTypeUtils

public final class ParquetTypeUtils
extends Object

Method Summary

Modifier and Type	Method	Description
static Optional<DecimalType>	createDecimalType(RichColumnDescriptor descriptor)
static org.apache.parquet.io.ColumnIO	getArrayElementColumn(org.apache.parquet.io.ColumnIO columnIO)
static org.apache.parquet.io.MessageColumnIO	getColumnIO(org.apache.parquet.schema.MessageType fileSchema, org.apache.parquet.schema.MessageType requestedSchema)
static List<org.apache.parquet.io.PrimitiveColumnIO>	getColumns(org.apache.parquet.schema.MessageType fileSchema, org.apache.parquet.schema.MessageType requestedSchema)
static Optional<RichColumnDescriptor>	getDescriptor(List<org.apache.parquet.io.PrimitiveColumnIO> columns, List<String> path)
static Map<List<String>,RichColumnDescriptor>	getDescriptors(org.apache.parquet.schema.MessageType fileSchema, org.apache.parquet.schema.MessageType requestedSchema)
static io.airlift.slice.Slice	getLongDecimalValue(byte[] bytes)
static org.apache.parquet.io.GroupColumnIO	getMapKeyValueColumn(org.apache.parquet.io.GroupColumnIO groupColumnIO)
static ParquetEncoding	getParquetEncoding(org.apache.parquet.column.Encoding encoding)
static org.apache.parquet.schema.Type	getParquetTypeByName(String columnName, org.apache.parquet.schema.GroupType groupType)
static long	getShortDecimalValue(byte[] bytes)
static long	getShortDecimalValue(byte[] bytes, int startOffset, int length)
static boolean	isValueNull(boolean required, int definitionLevel, int maxDefinitionLevel)	For optional fields: definitionLevel == maxDefinitionLevel => Value is defined definitionLevel == maxDefinitionLevel - 1 => Value is null definitionLevel < maxDefinitionLevel - 1 => Value does not exist, because one of its optional parent fields is null
static org.apache.parquet.io.ColumnIO	lookupColumnById(org.apache.parquet.io.GroupColumnIO groupColumnIO, int columnId)
static org.apache.parquet.io.ColumnIO	lookupColumnByName(org.apache.parquet.io.GroupColumnIO groupColumnIO, String columnName)	Parquet column names are case-sensitive unlike Hive, which converts all column names to lowercase.

Methods inherited from class java.lang.Object

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

Method Detail

getColumns

public static List<org.apache.parquet.io.PrimitiveColumnIO> getColumns(org.apache.parquet.schema.MessageType fileSchema,
                                                                       org.apache.parquet.schema.MessageType requestedSchema)

getColumnIO

public static org.apache.parquet.io.MessageColumnIO getColumnIO(org.apache.parquet.schema.MessageType fileSchema,
                                                                org.apache.parquet.schema.MessageType requestedSchema)

getMapKeyValueColumn

public static org.apache.parquet.io.GroupColumnIO getMapKeyValueColumn(org.apache.parquet.io.GroupColumnIO groupColumnIO)

getArrayElementColumn

public static org.apache.parquet.io.ColumnIO getArrayElementColumn(org.apache.parquet.io.ColumnIO columnIO)

getDescriptors

public static Map<List<String>,RichColumnDescriptor> getDescriptors(org.apache.parquet.schema.MessageType fileSchema,
                                                                          org.apache.parquet.schema.MessageType requestedSchema)

getDescriptor

public static Optional<RichColumnDescriptor> getDescriptor(List<org.apache.parquet.io.PrimitiveColumnIO> columns,
                                                           List<String> path)

getParquetEncoding

public static ParquetEncoding getParquetEncoding(org.apache.parquet.column.Encoding encoding)

getParquetTypeByName

public static org.apache.parquet.schema.Type getParquetTypeByName(String columnName,
                                                                  org.apache.parquet.schema.GroupType groupType)

lookupColumnByName

public static org.apache.parquet.io.ColumnIO lookupColumnByName(org.apache.parquet.io.GroupColumnIO groupColumnIO,
                                                                String columnName)

Parquet column names are case-sensitive unlike Hive, which converts all column names to lowercase. Therefore, when we look up columns we first check for exact match, and if that fails we look for a case-insensitive match.

lookupColumnById

@Nullable
public static org.apache.parquet.io.ColumnIO lookupColumnById(org.apache.parquet.io.GroupColumnIO groupColumnIO,
                                                              int columnId)

createDecimalType

public static Optional<DecimalType> createDecimalType(RichColumnDescriptor descriptor)

isValueNull

public static boolean isValueNull(boolean required,
                                  int definitionLevel,
                                  int maxDefinitionLevel)

For optional fields: definitionLevel == maxDefinitionLevel => Value is defined definitionLevel == maxDefinitionLevel - 1 => Value is null definitionLevel < maxDefinitionLevel - 1 => Value does not exist, because one of its optional parent fields is null

getShortDecimalValue

public static long getShortDecimalValue(byte[] bytes)

getShortDecimalValue

public static long getShortDecimalValue(byte[] bytes,
                                        int startOffset,
                                        int length)

getLongDecimalValue

public static io.airlift.slice.Slice getLongDecimalValue(byte[] bytes)