org.apache.tika.batch

Class BatchProcess

java.lang.Object

org.apache.tika.batch.BatchProcess

All Implemented Interfaces:

Callable<ParallelFileProcessingResult>

public class BatchProcess
extends Object
implements Callable<ParallelFileProcessingResult>

This is the main processor class for a single process. This class can only be run once.

It requires a FileResourceCrawler and FileResourceConsumers, and it can also support a StatusReporter and an Interrupter.

This is designed to shutdown if a parser has timed out or if there is an OutOfMemoryError. Consider using BatchProcessDriverCLI as a daemon/watchdog that monitors and can restart this batch process;

Note that this classs redirects stderr to stdout so that it can communicate without interference with the parent process on stderr.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	BatchProcess.BATCH_CONSTANTS

Constructor Summary

Constructors

Constructor and Description
BatchProcess(FileResourceCrawler fileResourceCrawler, ConsumersManager consumersManager, StatusReporter reporter, Interrupter interrupter)

Method Summary

Modifier and Type	Method and Description
ParallelFileProcessingResult	call() Runs main execution loop.
void	setMaxAliveTimeSeconds(int maxAliveTimeSeconds) The maximum amount of time that this process can be alive.
void	setPauseOnEarlyTerminationMillis(long pauseOnEarlyTerminationMillis) If there is an early termination via an interrupt or too many timed out consumers or because a consumer or other Runnable threw a Throwable, pause this long before killing the consumers and other threads.
void	setTimeoutCheckPulseMillis(long timeoutCheckPulseMillis)
void	setTimeoutThresholdMillis(long timeoutThresholdMillis) The amount of time allowed before a consumer should be timed out.

Methods inherited from class java.lang.Object

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

Constructor Detail

BatchProcess

public BatchProcess(FileResourceCrawler fileResourceCrawler,
                    ConsumersManager consumersManager,
                    StatusReporter reporter,
                    Interrupter interrupter)

Method Detail

call

public ParallelFileProcessingResult call()
                                  throws InterruptedException

Runs main execution loop.

Redirects stdout to stderr to keep clean communications over stdout with parent process

Specified by:

call in interface Callable<ParallelFileProcessingResult>

Returns:

result of the processing

Throws:

InterruptedException

setPauseOnEarlyTerminationMillis

public void setPauseOnEarlyTerminationMillis(long pauseOnEarlyTerminationMillis)

If there is an early termination via an interrupt or too many timed out consumers or because a consumer or other Runnable threw a Throwable, pause this long before killing the consumers and other threads. Typically makes sense for this to be the same or slightly larger than timeoutThresholdMillis

Parameters:

pauseOnEarlyTerminationMillis - how long to pause if there is an early termination

setTimeoutThresholdMillis

public void setTimeoutThresholdMillis(long timeoutThresholdMillis)

The amount of time allowed before a consumer should be timed out.

Parameters:

timeoutThresholdMillis - threshold in milliseconds before declaring a consumer timed out

setTimeoutCheckPulseMillis

public void setTimeoutCheckPulseMillis(long timeoutCheckPulseMillis)

setMaxAliveTimeSeconds

public void setMaxAliveTimeSeconds(int maxAliveTimeSeconds)

The maximum amount of time that this process can be alive. To avoid memory leaks, it is sometimes beneficial to shutdown (and restart) the process periodically.

If the value is < 0, the process will run until completion, interruption or exception.

Parameters:

maxAliveTimeSeconds - maximum amount of time in seconds to remain alive