org.apache.avro.file

Class DataFileWriter<D>

java.lang.Object

org.apache.avro.file.DataFileWriter<D>

All Implemented Interfaces:

Closeable, Flushable, AutoCloseable

public class DataFileWriter<D>
extends Object
implements Closeable, Flushable

Stores in a file a sequence of data conforming to a schema. The schema is stored in the file with the data. Each datum in a file is of the same schema. Data is written with a DatumWriter. Data is grouped into blocks. A synchronization marker is written between blocks, so that files may be split. Blocks may be compressed. Extensible metadata is stored at the end of the file. Files may be appended to.

See Also:

DataFileReader

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	DataFileWriter.AppendWriteException Thrown by append(Object) when an exception occurs while writing a datum to the buffer.

Constructor Summary

Constructors

Constructor and Description
DataFileWriter(DatumWriter<D> dout) Construct a writer, not yet open.

Method Summary

Modifier and Type	Method and Description
void	append(D datum) Append a datum to the file.
void	appendAllFrom(DataFileStream<D> otherFile, boolean recompress) Appends data from another file.
void	appendEncoded(ByteBuffer datum) Expert: Append a pre-encoded datum to the file.
DataFileWriter<D>	appendTo(File file) Open a writer appending to an existing file.
DataFileWriter<D>	appendTo(SeekableInput in, OutputStream out) Open a writer appending to an existing file.
void	close() Flush and close the file.
DataFileWriter<D>	create(Schema schema, File file) Open a new file for data matching a schema with a random sync.
DataFileWriter<D>	create(Schema schema, OutputStream outs) Open a new file for data matching a schema with a random sync.
DataFileWriter<D>	create(Schema schema, OutputStream outs, byte[] sync) Open a new file for data matching a schema with an explicit sync.
void	flush() Calls sync() and then flushes the current state of the file.
void	fSync() If this writer was instantiated using a File or using an Syncable instance, this method flushes all buffers for this writer to disk.
boolean	isFlushOnEveryBlock()
static boolean	isReservedMeta(String key)
DataFileWriter<D>	setCodec(CodecFactory c) Configures this writer to use the given codec.
void	setFlushOnEveryBlock(boolean flushOnEveryBlock) Set whether this writer should flush the block to the stream every time a sync marker is written.
DataFileWriter<D>	setMeta(String key, byte[] value) Set a metadata property.
DataFileWriter<D>	setMeta(String key, long value) Set a metadata property.
DataFileWriter<D>	setMeta(String key, String value) Set a metadata property.
DataFileWriter<D>	setSyncInterval(int syncInterval) Set the synchronization interval for this file, in bytes.
long	sync() Return the current position as a value that may be passed to DataFileReader.seek(long).

Methods inherited from class java.lang.Object

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

Constructor Detail

DataFileWriter

public DataFileWriter(DatumWriter<D> dout)

Construct a writer, not yet open.

Method Detail

setCodec

public DataFileWriter<D> setCodec(CodecFactory c)

Configures this writer to use the given codec. May not be reset after writes have begun.

setSyncInterval

public DataFileWriter<D> setSyncInterval(int syncInterval)

Set the synchronization interval for this file, in bytes. Valid values range from 32 to 2^30 Suggested values are between 2K and 2M The stream is flushed by default at the end of each synchronization interval. If setFlushOnEveryBlock(boolean) is called with param set to false, then the block may not be flushed to the stream after the sync marker is written. In this case, the flush() must be called to flush the stream. Invalid values throw IllegalArgumentException

Parameters:

syncInterval - the approximate number of uncompressed bytes to write in each block

Returns:

this DataFileWriter

create

public DataFileWriter<D> create(Schema schema,
                                File file)
                         throws IOException

Open a new file for data matching a schema with a random sync.

Throws:

IOException

create

public DataFileWriter<D> create(Schema schema,
                                OutputStream outs)
                         throws IOException

Open a new file for data matching a schema with a random sync.

Throws:

IOException

create

public DataFileWriter<D> create(Schema schema,
                                OutputStream outs,
                                byte[] sync)
                         throws IOException

Open a new file for data matching a schema with an explicit sync.

Throws:

IOException

setFlushOnEveryBlock

public void setFlushOnEveryBlock(boolean flushOnEveryBlock)

Set whether this writer should flush the block to the stream every time a sync marker is written. By default, the writer will flush the buffer each time a sync marker is written (if the block size limit is reached or the sync() is called.

Parameters:

flushOnEveryBlock - - If set to false, this writer will not flush the block to the stream until flush() is explicitly called.

isFlushOnEveryBlock

public boolean isFlushOnEveryBlock()

Returns:

- true if this writer flushes the block to the stream every time a sync marker is written. Else returns false.

appendTo

public DataFileWriter<D> appendTo(File file)
                           throws IOException

Open a writer appending to an existing file.

Throws:

IOException

appendTo

public DataFileWriter<D> appendTo(SeekableInput in,
                                  OutputStream out)
                           throws IOException

Open a writer appending to an existing file. Since 1.9.0 this method does not close in.

Parameters:

in - reading the existing file.

out - positioned at the end of the existing file.

Throws:

IOException

setMeta

public DataFileWriter<D> setMeta(String key,
                                 byte[] value)

Set a metadata property.

isReservedMeta

public static boolean isReservedMeta(String key)

setMeta

public DataFileWriter<D> setMeta(String key,
                                 String value)

Set a metadata property.

setMeta

public DataFileWriter<D> setMeta(String key,
                                 long value)

Set a metadata property.

append

public void append(D datum)
            throws IOException

Append a datum to the file.

Throws:

IOException

See Also:

DataFileWriter.AppendWriteException

appendEncoded

public void appendEncoded(ByteBuffer datum)
                   throws IOException

Expert: Append a pre-encoded datum to the file. No validation is performed to check that the encoding conforms to the file's schema. Appending non-conforming data may result in an unreadable file.

Throws:

IOException

appendAllFrom

public void appendAllFrom(DataFileStream<D> otherFile,
                          boolean recompress)
                   throws IOException

Appends data from another file. otherFile must have the same schema. Data blocks will be copied without de-serializing data. If the codecs of the two files are compatible, data blocks are copied directly without decompression. If the codecs are not compatible, blocks from otherFile are uncompressed and then compressed using this file's codec.

If the recompress flag is set all blocks are decompressed and then compressed using this file's codec. This is useful when the two files have compatible compression codecs but different codec options. For example, one might append a file compressed with deflate at compression level 1 to a file with deflate at compression level 7. If recompress is false, blocks will be copied without changing the compression level. If true, they will be converted to the new compression level.

Parameters:

otherFile -

recompress -

Throws:

IOException

sync

public long sync()
          throws IOException

Return the current position as a value that may be passed to DataFileReader.seek(long). Forces the end of the current block, emitting a synchronization marker. By default, this will also flush the block to the stream. If setFlushOnEveryBlock(boolean) is called with param set to false, then this method may not flush the block. In this case, the flush() must be called to flush the stream.

Throws:

IOException

flush

public void flush()
           throws IOException

Calls sync() and then flushes the current state of the file.

Specified by:

flush in interface Flushable

Throws:

IOException

fSync

public void fSync()
           throws IOException

If this writer was instantiated using a File or using an Syncable instance, this method flushes all buffers for this writer to disk. In other cases, this method behaves exactly like flush().

Throws:

IOException

close

public void close()
           throws IOException

Flush and close the file.

Specified by:

close in interface Closeable

Specified by:

close in interface AutoCloseable

Throws:

IOException