edu.cornell.mannlib.vitro.webapp.searchindex

Class SearchIndexerImpl

java.lang.Object

edu.cornell.mannlib.vitro.webapp.searchindex.SearchIndexerImpl

All Implemented Interfaces:

Application.Component, Application.Module, SearchIndexer

public class SearchIndexerImpl
extends Object
implements SearchIndexer

An implementation of the SearchIndexer interface. This implementation uses a single-threaded task queue to permit indexing to run one at a time in a "background" thread. The task queue is controlled by a scheduler that allows us to suspend incoming tasks (pause). A thread pool is available so the tasks can create small units of work to be run in parallel. Each task should block until all of its work units are complete, to preserve the pattern of running one task at a time. The number of threads in the thread pool is specified in the application setup file.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static interface	SearchIndexerImpl.IndexerConfig Interface for tasks to access the Indexer config
static class	SearchIndexerImpl.ListenerList A simple thread-safe list of event listeners.
static interface	SearchIndexerImpl.Task
static class	SearchIndexerImpl.WorkerThreadPool A thread pool for handling many small units of work submitted by a task.

Nested classes/interfaces inherited from interface edu.cornell.mannlib.vitro.webapp.modules.searchIndexer.SearchIndexer

SearchIndexer.Event, SearchIndexer.Listener

Nested classes/interfaces inherited from interface edu.cornell.mannlib.vitro.webapp.modules.Application.Component

Application.Component.LifecycleState

Constructor Summary

Constructors

Constructor and Description
SearchIndexerImpl()

Method Summary

Modifier and Type	Method and Description
void	addListener(SearchIndexer.Listener listener) Add this listener, allowing it to receive events from the indexer.
SearchIndexerStatus	getStatus() What is the current status of the indexer? Still valid after shutdown().
void	pause() Stop processing new tasks.
void	rebuildIndex() Remove all of the existing documents in the index, and schedule updates for all of the individuals in the model.
void	removeListener(SearchIndexer.Listener listener) Remove this listener, meaning that it will no longer receive events from the indexer.
void	scheduleUpdatesForStatements(List<org.apache.jena.rdf.model.Statement> changes) Update any search documents that are affected by these statements.
void	scheduleUpdatesForUris(Collection<String> uris) Update the search documents for these URIs.
void	setThreadPoolSize(String size)
void	shutdown(Application application) Stop processing and release resources.
void	startup(Application application, ComponentStartupStatus ss) Start processing.
void	unpause() Resume processing new tasks.
void	validate()

Methods inherited from class java.lang.Object

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

Constructor Detail

SearchIndexerImpl

public SearchIndexerImpl()

Method Detail

setThreadPoolSize

public void setThreadPoolSize(String size)

validate

public void validate()
              throws Exception

Throws:

Exception

startup

public void startup(Application application,
                    ComponentStartupStatus ss)

Description copied from interface: SearchIndexer

Start processing. If unpaused, schedule any queued tasks.

Specified by:

startup in interface Application.Component

Specified by:

startup in interface SearchIndexer

shutdown

public void shutdown(Application application)

Description copied from interface: SearchIndexer

Stop processing and release resources. This call should block until the dependent threads are stopped. Repeated calls have no effect.

Specified by:

shutdown in interface Application.Component

Specified by:

shutdown in interface SearchIndexer

pause

public void pause()

Description copied from interface: SearchIndexer

Stop processing new tasks. Requests will be queued until a call to unpause(). Fires a PAUSED event to listeners. This call has no effect if already paused, or if called after shutdown.

Specified by:

pause in interface SearchIndexer

unpause

public void unpause()

Description copied from interface: SearchIndexer

Resume processing new tasks. Any requests that were received since the call to pause() will now be scheduled for processing. Fires an UNPAUSED event to listeners. If startup has not been called, the unpaused state will be recorded, but tasks will still await the call to startup. This call has no effect if not paused, or if called after shutdown.

Specified by:

unpause in interface SearchIndexer

getStatus

public SearchIndexerStatus getStatus()

Description copied from interface: SearchIndexer

What is the current status of the indexer? Still valid after shutdown().

Specified by:

getStatus in interface SearchIndexer

scheduleUpdatesForStatements

public void scheduleUpdatesForStatements(List<org.apache.jena.rdf.model.Statement> changes)

Description copied from interface: SearchIndexer

Update any search documents that are affected by these statements. These statements are a mixture of additions and deletions. In either case, we feed them to the URI finders to see what individuals might have been affected by the change. We accumulate a batch of affected URIs, removing duplicates if they occur, and then submit them for updates. If called before startup or while paused, the task will be queued. If called after shutdown, this has no effect.

Specified by:

scheduleUpdatesForStatements in interface SearchIndexer

Parameters:

changes - if null or empty, this call has no effect.

scheduleUpdatesForUris

public void scheduleUpdatesForUris(Collection<String> uris)

Description copied from interface: SearchIndexer

Update the search documents for these URIs. For each URI that belongs in the index, a new search document is built, replacing any document that may already exist for that URI. For each URI that does not belong in the index, any existing document is removed. A URI belongs in the index if it refers to an existing individual in the model, and is not excluded. If called before startup or while paused, the task will be queued. If called after shutdown, this has no effect.

Specified by:

scheduleUpdatesForUris in interface SearchIndexer

Parameters:

uris - if null or empty, this call has no effect.

rebuildIndex

public void rebuildIndex()

Description copied from interface: SearchIndexer

Remove all of the existing documents in the index, and schedule updates for all of the individuals in the model. If a rebuild is already pending or in progress, this method has no effect. If called before startup or while paused, the task will be queued. If called after shutdown, this has no effect.

Specified by:

rebuildIndex in interface SearchIndexer

addListener

public void addListener(SearchIndexer.Listener listener)

Description copied from interface: SearchIndexer

Add this listener, allowing it to receive events from the indexer. The listener can safely assume that the SearchIndexer is not paused. If the SearchIndexer is indeed paused when the listener is added, then a PAUSE event will immediately be passed to the listener. If this listener has already been added, or if called after shutdown, this method has no effect.

Specified by:

addListener in interface SearchIndexer

Parameters:

listener - if null, this method has no effect.

removeListener

public void removeListener(SearchIndexer.Listener listener)

Description copied from interface: SearchIndexer

Remove this listener, meaning that it will no longer receive events from the indexer. If this listener is not active, or if called after shutdown, this method has no effect.

Specified by:

removeListener in interface SearchIndexer

Parameters:

listener - if null, this method has no effect.