org.apache.flink.connector.file.src

Class AbstractFileSource.AbstractFileSourceBuilder<T,SplitT extends FileSourceSplit,SELF extends AbstractFileSource.AbstractFileSourceBuilder<T,SplitT,SELF>>

java.lang.Object

org.apache.flink.connector.file.src.AbstractFileSource.AbstractFileSourceBuilder<T,SplitT,SELF>

Direct Known Subclasses:

FileSource.FileSourceBuilder

Enclosing class:

AbstractFileSource<T,SplitT extends FileSourceSplit>

protected abstract static class AbstractFileSource.AbstractFileSourceBuilder<T,SplitT extends FileSourceSplit,SELF extends AbstractFileSource.AbstractFileSourceBuilder<T,SplitT,SELF>>
extends Object

The generic base builder. This builder carries a SELF type to make it convenient to extend this for subclasses, using the following pattern.

 public class SubBuilder<T> extends AbstractFileSourceBuilder<T, SubBuilder<T>> {
     ...
 }
 

That way, all return values from builder method defined here are typed to the sub-class type and support fluent chaining.

We don't make the publicly visible builder generic with a SELF type, because it leads to generic signatures that can look complicated and confusing.

Field Summary

Fields

Modifier and Type	Field and Description
protected ContinuousEnumerationSettings	continuousSourceSettings
protected FileEnumerator.Provider	fileEnumerator
protected Path[]	inputPaths
protected BulkFormat<T,SplitT>	readerFormat
protected FileSplitAssigner.Provider	splitAssigner

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	AbstractFileSourceBuilder(Path[] inputPaths, BulkFormat<T,SplitT> readerFormat, FileEnumerator.Provider defaultFileEnumerator, FileSplitAssigner.Provider defaultSplitAssigner)

Method Summary

Modifier and Type	Method and Description
abstract AbstractFileSource<T,SplitT>	build() Creates the file source with the settings applied to this builder.
SELF	monitorContinuously(Duration discoveryInterval) Sets this source to streaming ("continuous monitoring") mode.
SELF	processStaticFileSet() Sets this source to bounded (batch) mode.
SELF	setFileEnumerator(FileEnumerator.Provider fileEnumerator) Configures the FileEnumerator for the source.
SELF	setSplitAssigner(FileSplitAssigner.Provider splitAssigner) Configures the FileSplitAssigner for the source.

Methods inherited from class java.lang.Object

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

Field Detail

inputPaths

protected final Path[] inputPaths

readerFormat

protected final BulkFormat<T,SplitT extends FileSourceSplit> readerFormat

fileEnumerator

protected FileEnumerator.Provider fileEnumerator

splitAssigner

protected FileSplitAssigner.Provider splitAssigner

continuousSourceSettings

@Nullable
protected ContinuousEnumerationSettings continuousSourceSettings

Constructor Detail

AbstractFileSourceBuilder

protected AbstractFileSourceBuilder(Path[] inputPaths,
                                    BulkFormat<T,SplitT> readerFormat,
                                    FileEnumerator.Provider defaultFileEnumerator,
                                    FileSplitAssigner.Provider defaultSplitAssigner)

Method Detail

build

public abstract AbstractFileSource<T,SplitT> build()

Creates the file source with the settings applied to this builder.

monitorContinuously

public SELF monitorContinuously(Duration discoveryInterval)

Sets this source to streaming ("continuous monitoring") mode.

This makes the source a "continuous streaming" source that keeps running, monitoring for new files, and reads these files when they appear and are discovered by the monitoring.

The interval in which the source checks for new files is the discoveryInterval. Shorter intervals mean that files are discovered more quickly, but also imply more frequent listing or directory traversal of the file system / object store.

processStaticFileSet

public SELF processStaticFileSet()

Sets this source to bounded (batch) mode.

In this mode, the source processes the files that are under the given paths when the application is started. Once all files are processed, the source will finish.

This setting is also the default behavior. This method is mainly here to "switch back" to bounded (batch) mode, or to make it explicit in the source construction.

setFileEnumerator

public SELF setFileEnumerator(FileEnumerator.Provider fileEnumerator)

Configures the FileEnumerator for the source. The File Enumerator is responsible for selecting from the input path the set of files that should be processed (and which to filter out). Furthermore, the File Enumerator may split the files further into sub-regions, to enable parallelization beyond the number of files.

setSplitAssigner

public SELF setSplitAssigner(FileSplitAssigner.Provider splitAssigner)

Configures the FileSplitAssigner for the source. The File Split Assigner determines which parallel reader instance gets which FileSourceSplit, and in which order these splits are assigned.