Package org.alfresco.repo.search.impl.lucene.index

Class IndexInfo

java.lang.Object

org.alfresco.repo.search.impl.lucene.index.IndexInfo

All Implemented Interfaces:

IndexMonitor

public class IndexInfo
extends Object
implements IndexMonitor

The information that makes up an index. IndexInfoVersion Repeated information of the form

1. Index Type.
2. sub-directory name.
3. Status
   1. Indexes, sub indexes, and overlays must be committed. Status is ACTIVE, MERGING, COMPLETING_INDEX
   2. Delta: Transaction status
   3. Overlay: Transaction status

Merges always take place to new indexes so we can detect merge failure or partial merges. Or we do not know what has merged. Incomplete delete merging does not matter - the overlay would still exist and be treated as such. So a document may be deleted in the index as well as in the applied overlay. It is still correctly deleted. NOTE: Public methods lock as required, the private methods assume that the appropriate locks have been obtained. TODO: Write element status into individual directories. This would be enough for recovery if both index files are lost or corrupted. TODO: Tidy up index status at start up or after some time. How long would you leave a merge to run?

The index structure is duplicated to two files. If one is currupted the second is used.

TODO:

1. make the index sharing configurable
2. use a thread pool for deletions, merging and index deletions
3. something to control the maximum number of overlays to limit the number of things layered together for searching
4. look at lucene locking again post 2.0, to see if it is improved
5. clean up old data files (that are not old index entries) - should be a config option

Author:

Andy Hind

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static interface	IndexInfo.LockWork<Result>

Field Summary

Fields

Modifier and Type	Field	Description
static String	MAIN_READER

Method Summary

Modifier and Type	Method	Description
void	addApplicationListener(org.springframework.context.ApplicationListener listener)	Registers a listener for events on this index.
void	closeDelta(String id)	Make sure the writer and reader for TX data are closed.
void	closeDeltaIndexReader(String id)	Manage closing and unregistering an index reader.
void	closeDeltaIndexWriter(String id)	Manage closing and unregistering an index writer .
void	delete(String deltaId)
static void	destroy()
void	destroyInstance()
<R> R	doReadOnly(IndexInfo.LockWork<R> lockWork)
long	getActualSize()	Gets the actual size of the index in bytes.
Set<String>	getDeletions(String id)	Get the deletions for a given index (there is no check if they should be applied that is up to the calling layer)
org.apache.lucene.index.IndexReader	getDeltaIndexReader(String id)	This method should only be called from one thread as it is bound to a transaction.
org.apache.lucene.index.IndexWriter	getDeltaIndexWriter(String id, org.apache.lucene.analysis.Analyzer analyzer)	Manage getting a lucene index writer for transactional data - looks after registration and checking there is no active reader.
static IndexInfo	getIndexInfo(File file, LuceneConfig config)	Get the IndexInfo object based in the given directory.
org.apache.lucene.index.IndexReader	getMainIndexReferenceCountingReadOnlyIndexReader()	Get the main reader for committed index data
org.apache.lucene.index.IndexReader	getMainIndexReferenceCountingReadOnlyIndexReader(String id, Set<String> deletions, Set<String> containerDeletions, boolean deleteOnlyNodes)	Get the main index reader augmented with the specified TX data As above but we add the TX data
int	getNumberOfDocuments()	Gets the number of documents in the index.
int	getNumberOfFields()	Gets the number of fields known to the index.
int	getNumberOfIndexedFields()	Gets the number of indexed fields.
String	getRelativePath()	Gets the relative path of the index directory.
Map<String,Integer>	getStatusSnapshot()	Gets a snapshot of the statuses of the individual entries in this index.
long	getUsedSize()	Gets the size used on disk by the index directory.
static void	main(String[] args)	Helper to print out index information
void	setPreparedState(String id, Set<String> toDelete, Set<String> containersToDelete, long documents, boolean deleteNodesOnly)	Set the aux data for the index entry for a transactional unit of work.
void	setStatus(String id, TransactionStatus state, Set<org.apache.lucene.index.Term> toDelete, Set<org.apache.lucene.index.Term> read)
String	toString()

Methods inherited from class java.lang.Object

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

Field Detail

MAIN_READER

public static final String MAIN_READER

See Also:

Constant Field Values

Method Detail

destroy

public static void destroy()

destroyInstance

public void destroyInstance()

delete

public void delete(String deltaId)

getIndexInfo

public static IndexInfo getIndexInfo(File file,
                                     LuceneConfig config)
                              throws IndexerException

Get the IndexInfo object based in the given directory. There is only one object per directory per JVM.

Parameters:

file - File

config - LuceneConfig

Returns:

IndexInfo

Throws:

IndexerException

getDeltaIndexReader

public org.apache.lucene.index.IndexReader getDeltaIndexReader(String id)
                                                        throws IOException

This method should only be called from one thread as it is bound to a transaction.

Parameters:

id - String

Returns:

IndexReader

Throws:

IOException

getDeltaIndexWriter

public org.apache.lucene.index.IndexWriter getDeltaIndexWriter(String id,
                                                               org.apache.lucene.analysis.Analyzer analyzer)
                                                        throws IOException

Manage getting a lucene index writer for transactional data - looks after registration and checking there is no active reader.

Parameters:

id - String

analyzer - Analyzer

Returns:

IndexWriter

Throws:

IOException

closeDeltaIndexReader

public void closeDeltaIndexReader(String id)
                           throws IOException

Manage closing and unregistering an index reader.

Parameters:

id - String

Throws:

IOException

closeDeltaIndexWriter

public void closeDeltaIndexWriter(String id)
                           throws IOException

Manage closing and unregistering an index writer .

Parameters:

id - String

Throws:

IOException

closeDelta

public void closeDelta(String id)
                throws IOException

Make sure the writer and reader for TX data are closed.

Parameters:

id - String

Throws:

IOException

getDeletions

public Set<String> getDeletions(String id)
                         throws IOException

Get the deletions for a given index (there is no check if they should be applied that is up to the calling layer)

Parameters:

id - String

Throws:

IOException

setPreparedState

public void setPreparedState(String id,
                             Set<String> toDelete,
                             Set<String> containersToDelete,
                             long documents,
                             boolean deleteNodesOnly)
                      throws IOException

Set the aux data for the index entry for a transactional unit of work.

Parameters:

id - - the tx id

toDelete - - noderefs that should be deleted from previous indexes (not this one)

documents - - the number of docs in the index

deleteNodesOnly - - should deletions on apply to nodes (ie not to containers)

Throws:

IOException

getMainIndexReferenceCountingReadOnlyIndexReader

public org.apache.lucene.index.IndexReader getMainIndexReferenceCountingReadOnlyIndexReader()
                                                                                     throws IOException

Get the main reader for committed index data

Returns:

IndexReader

Throws:

IOException

getMainIndexReferenceCountingReadOnlyIndexReader

public org.apache.lucene.index.IndexReader getMainIndexReferenceCountingReadOnlyIndexReader(String id,
                                                                                            Set<String> deletions,
                                                                                            Set<String> containerDeletions,
                                                                                            boolean deleteOnlyNodes)
                                                                                     throws IOException

Get the main index reader augmented with the specified TX data As above but we add the TX data

Parameters:

id - String

deleteOnlyNodes - boolean

Returns:

IndexReader

Throws:

IOException

setStatus

public void setStatus(String id,
                      TransactionStatus state,
                      Set<org.apache.lucene.index.Term> toDelete,
                      Set<org.apache.lucene.index.Term> read)
               throws IOException

Throws:

IOException

doReadOnly

public <R> R doReadOnly(IndexInfo.LockWork<R> lockWork)

main

public static void main(String[] args)
                 throws Throwable

Helper to print out index information

Parameters:

args - String[]

Throws:

Throwable

toString

public String toString()

Overrides:

toString in class Object

getRelativePath

public String getRelativePath()

Description copied from interface: IndexMonitor

Gets the relative path of the index directory.

Specified by:

getRelativePath in interface IndexMonitor

Returns:

the relative path

getStatusSnapshot

public Map<String,Integer> getStatusSnapshot()

Description copied from interface: IndexMonitor

Gets a snapshot of the statuses of the individual entries in this index.

Specified by:

getStatusSnapshot in interface IndexMonitor

Returns:

a map of entry status names to entry counts

getActualSize

public long getActualSize()
                   throws IOException

Description copied from interface: IndexMonitor

Gets the actual size of the index in bytes.

Specified by:

getActualSize in interface IndexMonitor

Returns:

the actual size in bytes

Throws:

IOException - Signals that an I/O exception has occurred.

getUsedSize

public long getUsedSize()
                 throws IOException

Description copied from interface: IndexMonitor

Gets the size used on disk by the index directory. A large discrepancy from the value returned by IndexMonitor.getActualSize() may indicate that there are unused data files.

Specified by:

getUsedSize in interface IndexMonitor

Returns:

the size on disk in bytes

Throws:

IOException - Signals that an I/O exception has occurred.

getNumberOfDocuments

public int getNumberOfDocuments()
                         throws IOException

Description copied from interface: IndexMonitor

Gets the number of documents in the index.

Specified by:

getNumberOfDocuments in interface IndexMonitor

Returns:

the number of documents

Throws:

IOException - Signals that an I/O exception has occurred.

getNumberOfFields

public int getNumberOfFields()
                      throws IOException

Description copied from interface: IndexMonitor

Gets the number of fields known to the index.

Specified by:

getNumberOfFields in interface IndexMonitor

Returns:

the number of fields

Throws:

IOException - Signals that an I/O exception has occurred.

getNumberOfIndexedFields

public int getNumberOfIndexedFields()
                             throws IOException

Description copied from interface: IndexMonitor

Gets the number of indexed fields.

Specified by:

getNumberOfIndexedFields in interface IndexMonitor

Returns:

the number of indexed fields

Throws:

IOException - Signals that an I/O exception has occurred.

addApplicationListener

public void addApplicationListener(org.springframework.context.ApplicationListener listener)

Description copied from interface: IndexMonitor

Registers a listener for events on this index.

Specified by:

addApplicationListener in interface IndexMonitor

Parameters:

listener - the listener