com.pervasive.datarush.operators.io

Interface DataFormat

All Known Implementing Classes:

ARFFDataFormat, AvroFormat, DelimitedTextFormat, FixedTextFormat, LogDataFormat, ORCFormat, ParquetFormat

public interface DataFormat

Describes the record format of external data, such as in a file. A DataFormat object provides the necessary information for reading and writing external data, converting it to and from records in a dataflow graph.

Many formats are predefined in the library; an implementation is only required if a new format needs to be defined. Normally, it is not necessary to work directly with formats. Instead, operators are provided which hide the DataFormat object and present a view more appropriate to the specific format. Examples of this technique are the ReadDelimitedText and WriteDelimitedText operators.

See Also:

ReadSource, WriteSink

Nested Class Summary

Nested Classes

Modifier and Type	Interface and Description
static interface	DataFormat.DataFormatter A formatter for converting record data to binary or text format.
static interface	DataFormat.DataParser A parser for record data in binary or text format.

Method Summary

Modifier and Type	Method and Description
DataFormat.DataParser	createParser(ParsingOptions options) Create a new parser for the format using the specified parsing options.
DataFormat.DataFormatter	createWriter(FormattingOptions options) Create a new writer for the format using the specified formatting options.
FileMetadata	getMetadata() Gets the metadata associated with the format.
RecordTokenType	getType() Gets the record type associated with the format.
boolean	isSplittable() Indicates if the format supports parsing of subsections of a file.
FileMetadata	readMetadata(FileClient fileClient, ByteSource source) Reads the metadata associated with the format.
void	setMetadata(FileMetadata metadata) Sets the metadata associated with the format.
void	writeMetadata(FileMetadata metadata, FileClient fileClient, ByteSink target) Writes the provided metadata associated with the format.

Method Detail

getType

RecordTokenType getType()

Gets the record type associated with the format. Records produced by the associated parser or consumed by the associated formatter will be of this type.

For many formats, this may be derived from a schema object describing the format layout.

Returns:

the format's record type

getMetadata

FileMetadata getMetadata()

Gets the metadata associated with the format. Records produces by the associated parser or consumed by the associated formatter will use this metadata.

Returns:

the format's metadata

setMetadata

void setMetadata(FileMetadata metadata)

Sets the metadata associated with the format.

readMetadata

FileMetadata readMetadata(FileClient fileClient,
                          ByteSource source)

Reads the metadata associated with the format.

Parameters:

fileClient - client used to read file

source - location of the files

writeMetadata

void writeMetadata(FileMetadata metadata,
                   FileClient fileClient,
                   ByteSink target)

Writes the provided metadata associated with the format.

Parameters:

metadata - the metadata to write

fileClient - client used to write file

source - location of the files

createParser

DataFormat.DataParser createParser(ParsingOptions options)

Create a new parser for the format using the specified parsing options.

Parameters:

options - parsing options to use

Returns:

a new parser for reading external data

createWriter

DataFormat.DataFormatter createWriter(FormattingOptions options)

Create a new writer for the format using the specified formatting options.

Parameters:

options - formatting options to use

Returns:

a new formatter for writing external data

isSplittable

boolean isSplittable()

Indicates if the format supports parsing of subsections of a file.

A format should only return true if it can, at least in some situations, support this sort of parsing. If a format requires reading the entire file, it must return false.

If a format is not splittable, a file in the format cannot be parsed in parallel; however, individual files can still be parsed independently in parallel, as when reading the contents of a directory or using a file globbing pattern.

Returns:

true if the format supports parsing only a portion of the file, false otherwise