org.apache.parquet.column

Interface ColumnWriteStore

All Superinterfaces:

AutoCloseable

All Known Implementing Classes:

ColumnWriteStoreV1, ColumnWriteStoreV2

public interface ColumnWriteStore
extends AutoCloseable

Container which can construct writers for multiple columns to be stored together.

Method Summary

Modifier and Type	Method and Description
void	close() Close the related output stream and release any resources
void	endRecord() called to notify of record boundaries
void	flush() when we are done writing to flush to the underlying storage
long	getAllocatedSize() used for information
long	getBufferedSize() used to flush row groups to disk
ColumnWriter	getColumnWriter(ColumnDescriptor path)
default boolean	isColumnFlushNeeded() Returns whether flushing the possibly cached values (or nulls) to the underlying column writers is necessary, because the pages might be closed after the next invocation of endRecord().
String	memUsageString() used for debugging purpose

Method Detail

getColumnWriter

ColumnWriter getColumnWriter(ColumnDescriptor path)

Parameters:

path - the column for which to create a writer

Returns:

the column writer for the given column

flush

void flush()

when we are done writing to flush to the underlying storage

endRecord

void endRecord()

called to notify of record boundaries

getAllocatedSize

long getAllocatedSize()

used for information

Returns:

approximate size used in memory

getBufferedSize

long getBufferedSize()

used to flush row groups to disk

Returns:

approximate size of the buffered encoded binary data

memUsageString

String memUsageString()

used for debugging purpose

Returns:

a formated string representing memory usage per column

close

void close()

Close the related output stream and release any resources

Specified by:

close in interface AutoCloseable

isColumnFlushNeeded

default boolean isColumnFlushNeeded()

Returns whether flushing the possibly cached values (or nulls) to the underlying column writers is necessary, because the pages might be closed after the next invocation of endRecord().

Returns:

true if all the values shall be written to the underlying column writers before calling endRecord()