au.org.ala.checklist.lucene
Class CBCreateLuceneIndex

java.lang.Object
  au.org.ala.checklist.lucene.CBCreateLuceneIndex

Direct Known Subclasses:

DwcaNameIndexer

public class CBCreateLuceneIndex
extends java.lang.Object

Creates the Lucene index based on the cb_names export generated by ChecklistBankExporter

Author:

Natasha

Nested Class Summary

static class

CBCreateLuceneIndex.IndexField

Field Summary

protected java.lang.String	afdFile
protected java.lang.String	alaConcepts
protected java.lang.String	alaSynonyms
protected java.lang.String	apniFile
protected java.lang.String	cbExportFile
protected java.lang.String	colFile
protected javax.sql.DataSource	dataSource
protected java.lang.String	extraALAConcepts
protected java.lang.String	irmngDwcaDirectory
protected java.lang.String	irmngFile
protected java.lang.String	lexFile
protected org.apache.commons.logging.Log	log
protected java.lang.String	taxonConeptName

Constructor Summary

CBCreateLuceneIndex()

Method Summary

void	addAdditionalName(java.lang.String lsid, java.lang.String scientificName, java.lang.String author, LinnaeanRankClassification cl)
void	commit()
void	commit(boolean close, boolean merge)
org.apache.lucene.document.Document	createALAIndexDocument(java.lang.String name, java.lang.String id, java.lang.String lsid, java.lang.String author, LinnaeanRankClassification cl)
org.apache.lucene.document.Document	createALAIndexDocument(java.lang.String name, java.lang.String id, java.lang.String lsid, java.lang.String author, java.lang.String rank, java.lang.String rankId, java.lang.String left, java.lang.String right, LinnaeanRankClassification cl)
protected org.apache.lucene.document.Document	createALAIndexDocument(java.lang.String name, java.lang.String id, java.lang.String lsid, java.lang.String rank, java.lang.String rankString, java.lang.String kingdom, java.lang.String kid, java.lang.String phylum, java.lang.String pid, java.lang.String clazz, java.lang.String cid, java.lang.String order, java.lang.String oid, java.lang.String family, java.lang.String fid, java.lang.String genus, java.lang.String gid, java.lang.String species, java.lang.String sid, java.lang.String left, java.lang.String right, java.lang.String acceptedConcept, java.lang.String specificEpithet, java.lang.String infraspecificEpithet, java.lang.String author, float boost)
protected org.apache.lucene.document.Document	createALASynonymDocument(java.lang.String scientificName, java.lang.String author, java.lang.String id, java.lang.String lsid, java.lang.String nameLsid, java.lang.String acceptedLsid, java.lang.String acceptedId, float boost, java.lang.String synonymType)
void	createIndex(java.lang.String exportsDir, java.lang.String indexDir, boolean generateSciNames, boolean generateCommonNames) Creates the index from the specified checklist bank names usage export file into the specified index directory.
void	createIndex(java.lang.String exportsDir, java.lang.String indexDir, java.lang.String acceptedFile, java.lang.String synonymFile, java.lang.String irmngDwca, boolean generateSciNames, boolean generateCommonNames)
protected org.apache.lucene.index.IndexWriter	createIndexWriter(java.io.File directory, org.apache.lucene.analysis.Analyzer analyzer, boolean replace) Creates an index writer in the specified directory.
void	createIrmngIndex(java.lang.String exportsDir, java.lang.String indexDir)
void	deleteName(java.lang.String lsid) Deletes the entry that has the supplied lsid.
protected org.apache.lucene.document.Document	getCommonNameDocument(java.lang.String cn, java.lang.String sn, java.lang.String lsid, float boost)
protected org.apache.lucene.document.Document	getCommonNameDocument(java.lang.String cn, java.lang.String sn, java.lang.String lsid, float boost, boolean checkAccepted)
java.lang.String	getIndexDirectory()
protected void	indexIrmngDwcA(org.apache.lucene.index.IndexWriter iw, java.lang.String archiveDirectory)
void	init()
static void	main(java.lang.String[] args) Generates the Lucene index required for the name matching API.
void	setIndexDirectory(java.lang.String indexDirectory)

Methods inherited from class java.lang.Object

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

Field Detail

cbExportFile

protected java.lang.String cbExportFile

lexFile

protected java.lang.String lexFile

irmngFile

protected java.lang.String irmngFile

colFile

protected java.lang.String colFile

extraALAConcepts

protected java.lang.String extraALAConcepts

alaConcepts

protected java.lang.String alaConcepts

alaSynonyms

protected java.lang.String alaSynonyms

irmngDwcaDirectory

protected java.lang.String irmngDwcaDirectory

afdFile

protected java.lang.String afdFile

apniFile

protected java.lang.String apniFile

taxonConeptName

protected java.lang.String taxonConeptName

log

protected org.apache.commons.logging.Log log

dataSource

protected javax.sql.DataSource dataSource

Constructor Detail

CBCreateLuceneIndex

public CBCreateLuceneIndex()

Method Detail

init

public void init()
          throws java.lang.Exception

Throws:

java.lang.Exception

createIndex

public void createIndex(java.lang.String exportsDir,
                        java.lang.String indexDir,
                        boolean generateSciNames,
                        boolean generateCommonNames)
                 throws java.lang.Exception

Creates the index from the specified checklist bank names usage export file into the specified index directory.

Parameters:

cbExportFile - A cb export file as generated from the ChecklistBankExporter

lexFile -

irmngFile -

indexDir - The directory in which the 2 indices will be created.

Throws:

java.lang.Exception

createIrmngIndex

public void createIrmngIndex(java.lang.String exportsDir,
                             java.lang.String indexDir)
                      throws java.lang.Exception

Throws:

java.lang.Exception

createIndex

public void createIndex(java.lang.String exportsDir,
                        java.lang.String indexDir,
                        java.lang.String acceptedFile,
                        java.lang.String synonymFile,
                        java.lang.String irmngDwca,
                        boolean generateSciNames,
                        boolean generateCommonNames)
                 throws java.lang.Exception

Throws:

java.lang.Exception

createIndexWriter

protected org.apache.lucene.index.IndexWriter createIndexWriter(java.io.File directory,
                                                                org.apache.lucene.analysis.Analyzer analyzer,
                                                                boolean replace)
                                                         throws java.lang.Exception

Creates an index writer in the specified directory. It will create/recreate the target directory

Parameters:

directory -

analyzer -

Returns:

Throws:

java.lang.Exception

addAdditionalName

public void addAdditionalName(java.lang.String lsid,
                              java.lang.String scientificName,
                              java.lang.String author,
                              LinnaeanRankClassification cl)
                       throws java.lang.Exception

Throws:

java.lang.Exception

deleteName

public void deleteName(java.lang.String lsid)
                throws java.lang.Exception

Deletes the entry that has the supplied lsid. It will also delete all the synonyms associated with it

Parameters:

lsid -

Throws:

java.lang.Exception

commit

public void commit()
            throws java.lang.Exception

Throws:

java.lang.Exception

commit

public void commit(boolean close,
                   boolean merge)
            throws java.lang.Exception

Parameters:

merge - whether or not to merge the index

close - whether or not to close the index

Throws:

java.lang.Exception

indexIrmngDwcA

protected void indexIrmngDwcA(org.apache.lucene.index.IndexWriter iw,
                              java.lang.String archiveDirectory)
                       throws java.lang.Exception

Throws:

java.lang.Exception

getCommonNameDocument

protected org.apache.lucene.document.Document getCommonNameDocument(java.lang.String cn,
                                                                    java.lang.String sn,
                                                                    java.lang.String lsid,
                                                                    float boost)

getCommonNameDocument

protected org.apache.lucene.document.Document getCommonNameDocument(java.lang.String cn,
                                                                    java.lang.String sn,
                                                                    java.lang.String lsid,
                                                                    float boost,
                                                                    boolean checkAccepted)

createALAIndexDocument

public org.apache.lucene.document.Document createALAIndexDocument(java.lang.String name,
                                                                  java.lang.String id,
                                                                  java.lang.String lsid,
                                                                  java.lang.String author,
                                                                  LinnaeanRankClassification cl)

createALAIndexDocument

public org.apache.lucene.document.Document createALAIndexDocument(java.lang.String name,
                                                                  java.lang.String id,
                                                                  java.lang.String lsid,
                                                                  java.lang.String author,
                                                                  java.lang.String rank,
                                                                  java.lang.String rankId,
                                                                  java.lang.String left,
                                                                  java.lang.String right,
                                                                  LinnaeanRankClassification cl)

createALASynonymDocument

protected org.apache.lucene.document.Document createALASynonymDocument(java.lang.String scientificName,
                                                                       java.lang.String author,
                                                                       java.lang.String id,
                                                                       java.lang.String lsid,
                                                                       java.lang.String nameLsid,
                                                                       java.lang.String acceptedLsid,
                                                                       java.lang.String acceptedId,
                                                                       float boost,
                                                                       java.lang.String synonymType)

createALAIndexDocument

protected org.apache.lucene.document.Document createALAIndexDocument(java.lang.String name,
                                                                     java.lang.String id,
                                                                     java.lang.String lsid,
                                                                     java.lang.String rank,
                                                                     java.lang.String rankString,
                                                                     java.lang.String kingdom,
                                                                     java.lang.String kid,
                                                                     java.lang.String phylum,
                                                                     java.lang.String pid,
                                                                     java.lang.String clazz,
                                                                     java.lang.String cid,
                                                                     java.lang.String order,
                                                                     java.lang.String oid,
                                                                     java.lang.String family,
                                                                     java.lang.String fid,
                                                                     java.lang.String genus,
                                                                     java.lang.String gid,
                                                                     java.lang.String species,
                                                                     java.lang.String sid,
                                                                     java.lang.String left,
                                                                     java.lang.String right,
                                                                     java.lang.String acceptedConcept,
                                                                     java.lang.String specificEpithet,
                                                                     java.lang.String infraspecificEpithet,
                                                                     java.lang.String author,
                                                                     float boost)

getIndexDirectory

public java.lang.String getIndexDirectory()

setIndexDirectory

public void setIndexDirectory(java.lang.String indexDirectory)

main

public static void main(java.lang.String[] args)
                 throws java.lang.Exception

Generates the Lucene index required for the name matching API. eg au.org.ala.checklist.lucene.CBCreateLuceneIndex "/data/exports" "/data/lucene/namematching" Extra optional args that should appear after the directory names -sn: Only create the indexes necessary for the scientific name lookups -cn: Only create the indexes necessary for the common name lookups

Parameters:

args -

Throws:

java.lang.Exception