de.innovationgate.wgpublisher.lucene

Class LuceneManager

java.lang.Object

de.innovationgate.wgpublisher.lucene.LuceneManager

All Implemented Interfaces:

de.innovationgate.webgate.api.WGContentEventListener, de.innovationgate.webgate.api.WGDatabaseConnectListener, java.util.EventListener

public class LuceneManager
extends java.lang.Object
implements de.innovationgate.webgate.api.WGContentEventListener, de.innovationgate.webgate.api.WGDatabaseConnectListener

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
class	LuceneManager.Indexer

Field Summary

Fields

Modifier and Type	Field and Description
static java.lang.String	DOCTYPE_ATTACHMENT
static java.lang.String	DOCTYPE_CONTENT
static java.lang.String	INDEXFIELD_ALLATTACHMENTS
static java.lang.String	INDEXFIELD_ALLCONTENT
static java.lang.String	INDEXFIELD_CONTENTKEY
static java.lang.String	INDEXFIELD_DBKEY
static java.lang.String	INDEXFIELD_DOCTYPE
static java.lang.String	INDEXFIELD_FILENAME
static java.lang.String	INDEXFIELD_ISVIRTUALCONTENT
static org.apache.log4j.Logger	LOG
static int	MAX_CONCURRENT_SEARCHES
static java.lang.String	NATIVE_QUERYOPTION_DOCTYPE
static java.lang.String	NATIVE_QUERYOPTION_INCLUDEVIRTUALCONTENT if this option is given as native query option - virtual content will be included in search results per default virtual content is excluded
static java.lang.String	QUERYOPTION_HIGHLIGHT Boolean option to enable highlighting support
static java.lang.String	QUERYOPTION_SEARCHDBKEYS comma separated list of db_keys to search in if search scope is SEARCHSCOPE_DB_LIST
static java.lang.String	QUERYOPTION_SEARCHSCOPE search_scope for lucene search Set to String SEARCHSCOPE_DOMAIN, SEARCHSCOPE_WGA, SEARCHSCOPE_DB, SEARCHSCOPE_DB_LIST
static java.lang.String	RELATION_PREFIX
static java.lang.String	RELATIONGROUP_PREFIX
static java.lang.String	SEARCHSCOPE_DB
static java.lang.String	SEARCHSCOPE_DB_LIST
static java.lang.String	SEARCHSCOPE_DOMAIN values for QUERYOPTION_SEARCHSCOPE
static java.lang.String	SEARCHSCOPE_WGA
static java.lang.String	SYSPROP_DISABLE_AUTO_INDEX_OPTIMIZATION
static java.lang.String	SYSPROP_USE_CONSTANT_IDF_SIMILARITY
static java.lang.String	TAGINFO_SIMPLIFIEDQUERY
static java.lang.String	TAGINFO_UNSPECIFICQUERY
static java.util.List<java.lang.String>	TECHNICAL_ITEMS
static java.lang.String	VIRTUALMETA_AREA
static java.lang.String	VIRTUALMETA_CONTENTTYPE
static java.lang.String	VIRTUALMETA_DBKEY
static java.lang.String	VIRTUALMETA_KEY
static java.lang.String	VIRTUALMETA_PAGEPUBLISHED
static java.lang.String	VIRTUALMETA_PARENT
static java.lang.String	VIRTUALMETA_PATH
static java.lang.String[]	VIRTUALMETAS

Method Summary

Methods

Modifier and Type	Method and Description
void	closeOpenedResultSets() close opened resultsets for current thread
void	configurationHasChanged(java.util.Set newConnectedDBKeys) wga-configuration has changed, read new configuration and do necessary index updates
void	contentCreated(de.innovationgate.webgate.api.WGContentEvent contentEvent)
void	contentHasBeenDeleted(de.innovationgate.webgate.api.WGContentEvent event)
void	contentHasBeenMoved(de.innovationgate.webgate.api.WGContentEvent event)
void	contentHasBeenSaved(de.innovationgate.webgate.api.WGContentEvent event)
boolean	contentSaved(de.innovationgate.webgate.api.WGContentEvent contentEvent)
void	contentStatusChanged(de.innovationgate.webgate.api.WGContentEvent event)
org.apache.lucene.search.highlight.Highlighter	createHighlighter(java.lang.String itemOrMeta, org.apache.lucene.search.Query query, org.apache.lucene.search.highlight.Formatter formatter)
org.apache.lucene.analysis.TokenStream	createTokenStream(java.io.Reader reader, de.innovationgate.webgate.api.WGContent content)
org.apache.lucene.analysis.TokenStream	createTokenStream(java.lang.String text, de.innovationgate.webgate.api.WGContent content)
void	databaseConnected(de.innovationgate.webgate.api.WGDatabaseEvent event)
void	databaseConnectionError(de.innovationgate.webgate.api.WGDatabaseEvent event)
void	destroy()
org.apache.lucene.search.Explanation	explain(org.apache.lucene.search.Query query, int doc)
java.util.Map<java.lang.String,java.util.Queue<de.innovationgate.wgpublisher.lucene.LuceneManager.IndexingRequest>>	getAdditionRequestsMap()
int	getBooleanQueryMaxClauseCount()
de.innovationgate.webgate.api.WGDatabaseRevision	getDatabaseUpdateStatus(java.lang.String dbKey)
java.util.Map<java.lang.String,java.util.Queue<de.innovationgate.wgpublisher.lucene.LuceneManager.IndexingRequest>>	getDeletionRequestsMap()
org.apache.lucene.document.Document	getDocument(int i)
java.io.File	getIndexDirectory()
LuceneManager.Indexer	getIndexer()
long	getIndexInterval()
int	getMaxDocsPerDBSession()
long	getRemainingAdditionRequests()
long	getRemainingDbUpdateRequests()
long	getRemainingDeletionRequests()
boolean	indexIsEnabled(java.lang.String dbKey)
boolean	isIndexerRunning()
boolean	isRebuildingIndex()
void	optimizeIndex()
void	optimizeIndex(org.apache.log4j.Logger log)
void	performCustomDBIndexDeletions(de.innovationgate.webgate.api.WGDatabase db) Deprecated. pending feature - F00003426
void	rebuildIndex() disables the timer task, deletes the index_dir and reinit index
void	rebuildIndex(java.lang.String dbkey) reindex given dbkey
static LuceneManager	retrieve(WGACore core, de.innovationgate.wga.config.LuceneManagerConfiguration config)
org.apache.lucene.analysis.Analyzer	retrieveAnalyzer(de.innovationgate.webgate.api.WGContent content)
de.innovationgate.wga.common.beans.LuceneIndexFileRule	retrieveFileRule(de.innovationgate.webgate.api.WGDatabase db, java.lang.String filename)
de.innovationgate.wga.common.beans.LuceneIndexItemRule	retrieveItemRule(de.innovationgate.webgate.api.WGDatabase db, java.lang.String itemName)
de.innovationgate.wga.common.beans.LuceneConfiguration	retrieveLuceneConfig(java.lang.String dbKey) get LuceneConfiguration from config file
de.innovationgate.webgate.api.WGResultSet	search(de.innovationgate.webgate.api.WGDatabase db, java.util.List<java.lang.String> fields, java.lang.String phrase, java.util.Map parameters, WGA wga)
de.innovationgate.webgate.api.WGResultSet	search(de.innovationgate.webgate.api.WGDatabase db, java.lang.String phrase, java.util.Map parameters, WGA wga)
void	setBooleanQueryMaxClauseCount(int booleanQueryMaxClauseCount)
void	setIndexInterval(long indexInterval)
void	setMaxDocsPerDBSession(int maxDocsPerDBSession)
void	startup() if wga starts up it notifies lucene by this method
void	updateIndex()

Methods inherited from class java.lang.Object

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

Field Detail

LOG

public static final org.apache.log4j.Logger LOG

VIRTUALMETA_PATH

public static final java.lang.String VIRTUALMETA_PATH

See Also:

Constant Field Values

VIRTUALMETA_PARENT

public static final java.lang.String VIRTUALMETA_PARENT

See Also:

Constant Field Values

VIRTUALMETA_AREA

public static final java.lang.String VIRTUALMETA_AREA

See Also:

Constant Field Values

VIRTUALMETA_CONTENTTYPE

public static final java.lang.String VIRTUALMETA_CONTENTTYPE

See Also:

Constant Field Values

VIRTUALMETA_KEY

public static final java.lang.String VIRTUALMETA_KEY

See Also:

Constant Field Values

VIRTUALMETA_DBKEY

public static final java.lang.String VIRTUALMETA_DBKEY

See Also:

Constant Field Values

VIRTUALMETA_PAGEPUBLISHED

public static final java.lang.String VIRTUALMETA_PAGEPUBLISHED

See Also:

Constant Field Values

VIRTUALMETAS

public static final java.lang.String[] VIRTUALMETAS

TECHNICAL_ITEMS

public static final java.util.List<java.lang.String> TECHNICAL_ITEMS

QUERYOPTION_SEARCHSCOPE

public static final java.lang.String QUERYOPTION_SEARCHSCOPE

search_scope for lucene search Set to String SEARCHSCOPE_DOMAIN, SEARCHSCOPE_WGA, SEARCHSCOPE_DB, SEARCHSCOPE_DB_LIST

See Also:

Constant Field Values

QUERYOPTION_SEARCHDBKEYS

public static final java.lang.String QUERYOPTION_SEARCHDBKEYS

comma separated list of db_keys to search in if search scope is SEARCHSCOPE_DB_LIST

See Also:

Constant Field Values

QUERYOPTION_HIGHLIGHT

public static final java.lang.String QUERYOPTION_HIGHLIGHT

Boolean option to enable highlighting support

See Also:

Constant Field Values

NATIVE_QUERYOPTION_INCLUDEVIRTUALCONTENT

public static final java.lang.String NATIVE_QUERYOPTION_INCLUDEVIRTUALCONTENT

if this option is given as native query option - virtual content will be included in search results per default virtual content is excluded

See Also:

Constant Field Values

NATIVE_QUERYOPTION_DOCTYPE

public static final java.lang.String NATIVE_QUERYOPTION_DOCTYPE

See Also:

Constant Field Values

SEARCHSCOPE_DOMAIN

public static final java.lang.String SEARCHSCOPE_DOMAIN

values for QUERYOPTION_SEARCHSCOPE

See Also:

Constant Field Values

SEARCHSCOPE_WGA

public static final java.lang.String SEARCHSCOPE_WGA

See Also:

Constant Field Values

SEARCHSCOPE_DB

public static final java.lang.String SEARCHSCOPE_DB

See Also:

Constant Field Values

SEARCHSCOPE_DB_LIST

public static final java.lang.String SEARCHSCOPE_DB_LIST

See Also:

Constant Field Values

INDEXFIELD_DBKEY

public static final java.lang.String INDEXFIELD_DBKEY

See Also:

Constant Field Values

INDEXFIELD_CONTENTKEY

public static final java.lang.String INDEXFIELD_CONTENTKEY

See Also:

Constant Field Values

INDEXFIELD_ISVIRTUALCONTENT

public static final java.lang.String INDEXFIELD_ISVIRTUALCONTENT

See Also:

Constant Field Values

INDEXFIELD_ALLCONTENT

public static final java.lang.String INDEXFIELD_ALLCONTENT

See Also:

Constant Field Values

INDEXFIELD_ALLATTACHMENTS

public static final java.lang.String INDEXFIELD_ALLATTACHMENTS

See Also:

Constant Field Values

SYSPROP_DISABLE_AUTO_INDEX_OPTIMIZATION

public static final java.lang.String SYSPROP_DISABLE_AUTO_INDEX_OPTIMIZATION

See Also:

Constant Field Values

SYSPROP_USE_CONSTANT_IDF_SIMILARITY

public static final java.lang.String SYSPROP_USE_CONSTANT_IDF_SIMILARITY

See Also:

Constant Field Values

INDEXFIELD_FILENAME

public static final java.lang.String INDEXFIELD_FILENAME

See Also:

Constant Field Values

INDEXFIELD_DOCTYPE

public static final java.lang.String INDEXFIELD_DOCTYPE

See Also:

Constant Field Values

DOCTYPE_CONTENT

public static final java.lang.String DOCTYPE_CONTENT

See Also:

Constant Field Values

DOCTYPE_ATTACHMENT

public static final java.lang.String DOCTYPE_ATTACHMENT

See Also:

Constant Field Values

TAGINFO_UNSPECIFICQUERY

public static final java.lang.String TAGINFO_UNSPECIFICQUERY

See Also:

Constant Field Values

TAGINFO_SIMPLIFIEDQUERY

public static final java.lang.String TAGINFO_SIMPLIFIEDQUERY

See Also:

Constant Field Values

RELATION_PREFIX

public static final java.lang.String RELATION_PREFIX

See Also:

Constant Field Values

RELATIONGROUP_PREFIX

public static final java.lang.String RELATIONGROUP_PREFIX

See Also:

Constant Field Values

MAX_CONCURRENT_SEARCHES

public static final int MAX_CONCURRENT_SEARCHES

See Also:

Constant Field Values

Method Detail

retrieve

public static LuceneManager retrieve(WGACore core,
                     de.innovationgate.wga.config.LuceneManagerConfiguration config)
                              throws java.lang.IllegalArgumentException,
                                     java.io.IOException,
                                     org.dom4j.DocumentException,
                                     de.innovationgate.webgate.api.WGSystemException

Throws:

java.lang.IllegalArgumentException

java.io.IOException

org.dom4j.DocumentException

de.innovationgate.webgate.api.WGSystemException

updateIndex

public void updateIndex()
                 throws java.lang.InterruptedException

Throws:

java.lang.InterruptedException

getIndexInterval

public long getIndexInterval()

setIndexInterval

public void setIndexInterval(long indexInterval)

rebuildIndex

public void rebuildIndex()
                  throws java.io.IOException,
                         org.dom4j.DocumentException

disables the timer task, deletes the index_dir and reinit index

Throws:

java.io.IOException

org.dom4j.DocumentException

rebuildIndex

public void rebuildIndex(java.lang.String dbkey)

reindex given dbkey

Parameters:

dbkey -

destroy

public void destroy()

startup

public void startup()

if wga starts up it notifies lucene by this method

configurationHasChanged

public void configurationHasChanged(java.util.Set newConnectedDBKeys)

wga-configuration has changed, read new configuration and do necessary index updates

retrieveLuceneConfig

public de.innovationgate.wga.common.beans.LuceneConfiguration retrieveLuceneConfig(java.lang.String dbKey)

get LuceneConfiguration from config file

Parameters:

dbKey - DatabaseKey

Returns:

LuceneConfiguration, null if not found

optimizeIndex

public void optimizeIndex()

optimizeIndex

public void optimizeIndex(org.apache.log4j.Logger log)

retrieveAnalyzer

public org.apache.lucene.analysis.Analyzer retrieveAnalyzer(de.innovationgate.webgate.api.WGContent content)
                                                     throws de.innovationgate.webgate.api.WGAPIException

Parameters:

content -

Returns:

configured lucene analyzer for content language or default analyzer

Throws:

de.innovationgate.webgate.api.WGAPIException

contentCreated

public void contentCreated(de.innovationgate.webgate.api.WGContentEvent contentEvent)

Specified by:

contentCreated in interface de.innovationgate.webgate.api.WGContentEventListener

contentSaved

public boolean contentSaved(de.innovationgate.webgate.api.WGContentEvent contentEvent)

Specified by:

contentSaved in interface de.innovationgate.webgate.api.WGContentEventListener

contentHasBeenDeleted

public void contentHasBeenDeleted(de.innovationgate.webgate.api.WGContentEvent event)

Specified by:

contentHasBeenDeleted in interface de.innovationgate.webgate.api.WGContentEventListener

contentHasBeenSaved

public void contentHasBeenSaved(de.innovationgate.webgate.api.WGContentEvent event)

Specified by:

contentHasBeenSaved in interface de.innovationgate.webgate.api.WGContentEventListener

search

public de.innovationgate.webgate.api.WGResultSet search(de.innovationgate.webgate.api.WGDatabase db,
                                               java.lang.String phrase,
                                               java.util.Map parameters,
                                               WGA wga)
                                                 throws de.innovationgate.webgate.api.WGQueryException

Throws:

de.innovationgate.webgate.api.WGQueryException

search

public de.innovationgate.webgate.api.WGResultSet search(de.innovationgate.webgate.api.WGDatabase db,
                                               java.util.List<java.lang.String> fields,
                                               java.lang.String phrase,
                                               java.util.Map parameters,
                                               WGA wga)
                                                 throws de.innovationgate.webgate.api.WGQueryException

Throws:

de.innovationgate.webgate.api.WGQueryException

closeOpenedResultSets

public void closeOpenedResultSets()

close opened resultsets for current thread

isRebuildingIndex

public boolean isRebuildingIndex()

indexIsEnabled

public boolean indexIsEnabled(java.lang.String dbKey)

getBooleanQueryMaxClauseCount

public int getBooleanQueryMaxClauseCount()

setBooleanQueryMaxClauseCount

public void setBooleanQueryMaxClauseCount(int booleanQueryMaxClauseCount)

isIndexerRunning

public boolean isIndexerRunning()

Returns:

Returns the indexerIsRunning.

databaseConnected

public void databaseConnected(de.innovationgate.webgate.api.WGDatabaseEvent event)

Specified by:

databaseConnected in interface de.innovationgate.webgate.api.WGDatabaseConnectListener

databaseConnectionError

public void databaseConnectionError(de.innovationgate.webgate.api.WGDatabaseEvent event)

Specified by:

databaseConnectionError in interface de.innovationgate.webgate.api.WGDatabaseConnectListener

getMaxDocsPerDBSession

public int getMaxDocsPerDBSession()

setMaxDocsPerDBSession

public void setMaxDocsPerDBSession(int maxDocsPerDBSession)

createHighlighter

public org.apache.lucene.search.highlight.Highlighter createHighlighter(java.lang.String itemOrMeta,
                                                               org.apache.lucene.search.Query query,
                                                               org.apache.lucene.search.highlight.Formatter formatter)

createTokenStream

public org.apache.lucene.analysis.TokenStream createTokenStream(java.lang.String text,
                                                       de.innovationgate.webgate.api.WGContent content)
                                                         throws de.innovationgate.webgate.api.WGAPIException

Throws:

de.innovationgate.webgate.api.WGAPIException

createTokenStream

public org.apache.lucene.analysis.TokenStream createTokenStream(java.io.Reader reader,
                                                       de.innovationgate.webgate.api.WGContent content)
                                                         throws de.innovationgate.webgate.api.WGAPIException

Throws:

de.innovationgate.webgate.api.WGAPIException

performCustomDBIndexDeletions

public void performCustomDBIndexDeletions(de.innovationgate.webgate.api.WGDatabase db)
                                   throws java.io.IOException,
                                          de.innovationgate.webgate.api.WGIllegalArgumentException,
                                          java.lang.InterruptedException

Deprecated. pending feature - F00003426

Parameters:

db -

Throws:

java.io.IOException

de.innovationgate.webgate.api.WGIllegalArgumentException

java.lang.InterruptedException

getIndexDirectory

public java.io.File getIndexDirectory()

contentHasBeenMoved

public void contentHasBeenMoved(de.innovationgate.webgate.api.WGContentEvent event)

Specified by:

contentHasBeenMoved in interface de.innovationgate.webgate.api.WGContentEventListener

getDocument

public org.apache.lucene.document.Document getDocument(int i)
                                                throws org.apache.lucene.index.CorruptIndexException,
                                                       java.io.IOException,
                                                       java.lang.InterruptedException

Throws:

org.apache.lucene.index.CorruptIndexException

java.io.IOException

java.lang.InterruptedException

explain

public org.apache.lucene.search.Explanation explain(org.apache.lucene.search.Query query,
                                           int doc)
                                             throws org.apache.lucene.index.CorruptIndexException,
                                                    java.io.IOException,
                                                    java.lang.InterruptedException

Throws:

org.apache.lucene.index.CorruptIndexException

java.io.IOException

java.lang.InterruptedException

contentStatusChanged

public void contentStatusChanged(de.innovationgate.webgate.api.WGContentEvent event)

Specified by:

contentStatusChanged in interface de.innovationgate.webgate.api.WGContentEventListener

getDeletionRequestsMap

public java.util.Map<java.lang.String,java.util.Queue<de.innovationgate.wgpublisher.lucene.LuceneManager.IndexingRequest>> getDeletionRequestsMap()

getAdditionRequestsMap

public java.util.Map<java.lang.String,java.util.Queue<de.innovationgate.wgpublisher.lucene.LuceneManager.IndexingRequest>> getAdditionRequestsMap()

retrieveFileRule

public de.innovationgate.wga.common.beans.LuceneIndexFileRule retrieveFileRule(de.innovationgate.webgate.api.WGDatabase db,
                                                                      java.lang.String filename)

retrieveItemRule

public de.innovationgate.wga.common.beans.LuceneIndexItemRule retrieveItemRule(de.innovationgate.webgate.api.WGDatabase db,
                                                                      java.lang.String itemName)

getRemainingAdditionRequests

public long getRemainingAdditionRequests()

getRemainingDeletionRequests

public long getRemainingDeletionRequests()

getRemainingDbUpdateRequests

public long getRemainingDbUpdateRequests()

getIndexer

public LuceneManager.Indexer getIndexer()

getDatabaseUpdateStatus

public de.innovationgate.webgate.api.WGDatabaseRevision getDatabaseUpdateStatus(java.lang.String dbKey)