Mercurial Hosting > luan
annotate src/goodjava/lucene/api/LuceneIndexWriter.java @ 1525:f848d40b3b07
lucene.api add boosts
| author | Franklin Schmidt <fschmidt@gmail.com> |
|---|---|
| date | Sat, 18 Jul 2020 20:41:47 -0600 |
| parents | 7d145095cc0b |
| children | 3bd4d7963456 |
| rev | line source |
|---|---|
| 1460 | 1 package goodjava.lucene.api; |
| 2 | |
| 3 import java.io.IOException; | |
| 4 import java.util.Map; | |
| 5 import java.util.HashMap; | |
| 6 import java.util.List; | |
| 7 import org.apache.lucene.analysis.Analyzer; | |
| 8 import org.apache.lucene.document.Document; | |
| 9 import org.apache.lucene.document.Field; | |
| 10 import org.apache.lucene.document.StoredField; | |
| 11 import org.apache.lucene.document.StringField; | |
| 12 import org.apache.lucene.document.TextField; | |
| 13 import org.apache.lucene.document.IntField; | |
| 14 import org.apache.lucene.document.LongField; | |
| 15 import org.apache.lucene.document.DoubleField; | |
| 16 import org.apache.lucene.document.FloatField; | |
| 17 import org.apache.lucene.index.IndexWriter; | |
| 18 import org.apache.lucene.index.IndexWriterConfig; | |
| 19 import org.apache.lucene.index.Term; | |
| 20 import org.apache.lucene.index.DirectoryReader; | |
| 21 import org.apache.lucene.index.IndexReader; | |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1465
diff
changeset
|
22 import org.apache.lucene.index.CheckIndex; |
| 1460 | 23 import org.apache.lucene.search.Query; |
| 24 import org.apache.lucene.search.IndexSearcher; | |
| 25 import org.apache.lucene.store.Directory; | |
| 26 import org.apache.lucene.util.Version; | |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1465
diff
changeset
|
27 import goodjava.logging.Logger; |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1465
diff
changeset
|
28 import goodjava.logging.LoggerFactory; |
| 1460 | 29 |
| 30 | |
| 31 public final class LuceneIndexWriter implements GoodIndexWriter { | |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1465
diff
changeset
|
32 private static final Logger logger = LoggerFactory.getLogger(LuceneIndexWriter.class); |
| 1465 | 33 private final FieldAnalyzer fieldAnalyzer = new FieldAnalyzer(); |
| 34 public final Version luceneVersion; | |
| 1460 | 35 public final IndexWriterConfig luceneConfig; |
| 1465 | 36 public final IndexWriter luceneWriter; |
| 1460 | 37 public final GoodIndexWriterConfig goodConfig; |
| 38 private final Map<String,Boolean> indexedMap = new HashMap<String,Boolean>(); | |
| 39 | |
| 1465 | 40 public LuceneIndexWriter(Version luceneVersion,Directory dir,GoodIndexWriterConfig goodConfig) throws IOException { |
| 41 this.luceneVersion = luceneVersion; | |
| 42 this.luceneConfig = new IndexWriterConfig(luceneVersion,fieldAnalyzer); | |
| 43 this.luceneWriter = new IndexWriter(dir,luceneConfig); | |
| 1460 | 44 this.goodConfig = goodConfig; |
| 1465 | 45 luceneWriter.commit(); // commit index creation |
| 1460 | 46 } |
| 47 | |
| 48 public void close() throws IOException { | |
| 49 luceneWriter.close(); | |
| 50 } | |
| 51 | |
| 52 public void commit() throws IOException { | |
| 53 luceneWriter.commit(); | |
| 54 } | |
| 55 | |
| 56 public void rollback() throws IOException { | |
| 57 luceneWriter.rollback(); | |
| 58 } | |
| 59 | |
| 60 public void deleteAll() throws IOException { | |
| 61 luceneWriter.deleteAll(); | |
| 62 } | |
| 63 | |
| 64 public void deleteDocuments(Query query) throws IOException { | |
| 65 luceneWriter.deleteDocuments(query); | |
| 66 } | |
| 67 | |
| 68 public void addDocument(Map<String,Object> storedFields) throws IOException { | |
| 69 Document doc = newDocument(storedFields); | |
| 70 luceneWriter.addDocument(doc); | |
| 71 } | |
| 72 | |
| 73 public void updateDocument(String keyFieldName,Map<String,Object> storedFields) throws IOException { | |
| 74 if( !isIndexed(keyFieldName) ) | |
| 75 throw new RuntimeException("can't update using unindexed field "+keyFieldName); | |
| 76 if( fieldAnalyzer.isAdded(keyFieldName) ) | |
| 77 throw new RuntimeException("can't update using analyzeed field "+keyFieldName); | |
| 78 Document doc = newDocument(storedFields); | |
| 79 Object keyValue = storedFields.get(keyFieldName); | |
| 80 if( keyValue==null ) | |
| 81 throw new RuntimeException("no value for field "+keyFieldName); | |
| 82 Term term = LuceneUtils.term(keyFieldName,keyValue); | |
| 83 luceneWriter.updateDocument(term,doc); | |
| 84 } | |
| 85 | |
| 86 private Document newDocument(Map<String,Object> storedFields) { | |
| 87 Document doc = new Document(); | |
|
1525
f848d40b3b07
lucene.api add boosts
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
88 MoreFieldInfo more = goodConfig.getMoreFieldInfo(storedFields); |
|
f848d40b3b07
lucene.api add boosts
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
89 addFields(doc,storedFields,Field.Store.YES,more.boosts); |
|
f848d40b3b07
lucene.api add boosts
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
90 addFields(doc,more.unstoredFields,Field.Store.NO,more.boosts); |
| 1460 | 91 return doc; |
| 92 } | |
| 93 | |
|
1525
f848d40b3b07
lucene.api add boosts
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
94 private void addFields( Document doc, Map<String,Object> fields, Field.Store store, Map<String,Float> boosts ) { |
| 1460 | 95 for( Map.Entry<String,Object> entry : fields.entrySet() ) { |
| 96 String name = entry.getKey(); | |
| 97 Object value = entry.getValue(); | |
|
1525
f848d40b3b07
lucene.api add boosts
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
98 Float boost = boosts.get(name); |
| 1460 | 99 if( value instanceof List ) { |
| 100 for( Object v : (List)value ) { | |
|
1525
f848d40b3b07
lucene.api add boosts
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
101 doc.add( newField(name,v,store,boost) ); |
| 1460 | 102 } |
| 103 } else { | |
|
1525
f848d40b3b07
lucene.api add boosts
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
104 doc.add( newField(name,value,store,boost) ); |
| 1460 | 105 } |
| 106 } | |
| 107 } | |
| 108 | |
|
1525
f848d40b3b07
lucene.api add boosts
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
109 private Field newField( String name, Object value, Field.Store store, Float boost ) { |
|
f848d40b3b07
lucene.api add boosts
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
110 Field field = newField(name,value,store); |
|
f848d40b3b07
lucene.api add boosts
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
111 if( boost != null ) |
|
f848d40b3b07
lucene.api add boosts
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
112 field.setBoost(boost); |
|
f848d40b3b07
lucene.api add boosts
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
113 return field; |
|
f848d40b3b07
lucene.api add boosts
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
114 } |
|
f848d40b3b07
lucene.api add boosts
Franklin Schmidt <fschmidt@gmail.com>
parents:
1476
diff
changeset
|
115 |
| 1460 | 116 private Field newField( String name, Object value, Field.Store store ) { |
| 117 boolean isIndexed = isIndexed(name); | |
| 118 if( store==Field.Store.NO && !isIndexed ) | |
| 119 throw new RuntimeException("field '"+name+"' is unstored and unindexed"); | |
| 120 if( value instanceof String ) { | |
| 121 String s = (String)value; | |
| 122 if( !isIndexed ) { | |
| 123 return new StoredField(name,s); | |
| 124 } else if( !fieldAnalyzer.isAdded(name) ) { | |
| 125 return new StringField(name,s,store); | |
| 126 } else { | |
| 127 return new TextField(name,s,store); | |
| 128 } | |
| 129 } else if( value instanceof Integer ) { | |
| 130 int i = (Integer)value; | |
| 131 if( !isIndexed ) { | |
| 132 return new StoredField(name,i); | |
| 133 } else { | |
| 134 return new IntField(name,i,store); | |
| 135 } | |
| 136 } else if( value instanceof Long ) { | |
| 137 long i = (Long)value; | |
| 138 if( !isIndexed ) { | |
| 139 return new StoredField(name,i); | |
| 140 } else { | |
| 141 return new LongField(name,i,store); | |
| 142 } | |
| 143 } else if( value instanceof Double ) { | |
| 144 double i = (Double)value; | |
| 145 if( !isIndexed ) { | |
| 146 return new StoredField(name,i); | |
| 147 } else { | |
| 148 return new DoubleField(name,i,store); | |
| 149 } | |
| 150 } else if( value instanceof Float ) { | |
| 151 float i = (Float)value; | |
| 152 if( !isIndexed ) { | |
| 153 return new StoredField(name,i); | |
| 154 } else { | |
| 155 return new FloatField(name,i,store); | |
| 156 } | |
| 157 } else if( value instanceof byte[] ) { | |
| 158 if( isIndexed ) | |
| 159 throw new RuntimeException("can't index byte field "+name); | |
| 160 byte[] b = (byte[])value; | |
| 161 return new StoredField(name, b); | |
| 162 } else | |
| 163 throw new RuntimeException("invalid value type "+value.getClass()+"' for field '"+name+"'"); | |
| 164 } | |
| 165 | |
| 166 private synchronized boolean isIndexed(String fieldName) { | |
| 167 Boolean b = indexedMap.get(fieldName); | |
| 168 if( b==null ) { | |
| 169 b = goodConfig.isIndexed(fieldName); | |
| 170 indexedMap.put(fieldName,b); | |
| 171 Analyzer analyzer = goodConfig.getAnalyzer(fieldName); | |
| 172 if( analyzer!=null ) | |
| 173 fieldAnalyzer.add(fieldName,analyzer); | |
| 174 } | |
| 175 return b; | |
| 176 } | |
| 177 | |
| 178 | |
| 179 public void reindexDocuments(final String keyFieldName,Query query) throws IOException { | |
| 1465 | 180 IndexReader reader = openReader(); |
| 1460 | 181 final IndexSearcher searcher = new IndexSearcher(reader); |
| 182 searcher.search( query, new GoodCollector(){ | |
| 183 public void collectDoc(int iDoc) throws IOException { | |
| 184 Document doc = searcher.doc(iDoc); | |
| 185 Map<String,Object> storedFields = LuceneUtils.toMap(doc); | |
| 186 updateDocument(keyFieldName,storedFields); | |
| 187 } | |
| 188 }); | |
| 189 reader.close(); | |
| 190 } | |
| 1465 | 191 |
| 192 public IndexReader openReader() throws IOException { | |
| 193 return DirectoryReader.open(luceneWriter.getDirectory()); | |
| 194 } | |
|
1476
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1465
diff
changeset
|
195 |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1465
diff
changeset
|
196 public void check() throws IOException { |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1465
diff
changeset
|
197 CheckIndex.Status status = new CheckIndex(luceneWriter.getDirectory()).checkIndex(); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1465
diff
changeset
|
198 if( !status.clean ) |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1465
diff
changeset
|
199 logger.error("index not clean"); |
|
7d145095cc0b
lucene.logging check
Franklin Schmidt <fschmidt@gmail.com>
parents:
1465
diff
changeset
|
200 } |
| 1460 | 201 } |