org.lumongo.server.index.analysis.AnalysisHandler Maven / Gradle / Ivy
The newest version!
package org.lumongo.server.index.analysis;
import org.apache.lucene.analysis.Analyzer;
import org.apache.lucene.analysis.TokenStream;
import org.apache.lucene.analysis.tokenattributes.CharTermAttribute;
import org.apache.lucene.index.IndexReader;
import org.bson.Document;
import org.lumongo.cluster.message.Lumongo;
import org.lumongo.cluster.message.Lumongo.AnalysisRequest;
import org.lumongo.cluster.message.Lumongo.AnalysisRequest.TermSort;
import org.lumongo.server.config.IndexConfig;
import org.lumongo.util.LumongoUtil;
import org.lumongo.util.ResultHelper;
import java.util.List;
/**
* Created by Matt Davis on 6/29/16.
* @author mdavis
*/
public class AnalysisHandler {
private final AnalysisRequest analysisRequest;
private final String indexField;
private final String storedFieldName;
private final Analyzer analyzer;
private final boolean computeDocLevel;
private final boolean summaryLevelEnabled;
private final boolean enabled;
private final int minWordLength;
private final int maxWordLength;
private final AnalysisRequest.SummaryType summaryType;
private Integer minSegmentDocFreqCount;
private Integer maxSegmentDocFreqCount;
private DocFreq docFreq;
private TermFreq summaryTermFreq;
public AnalysisHandler(IndexReader indexReader, Analyzer analyzer, IndexConfig indexConfig, AnalysisRequest analysisRequest) {
this.analysisRequest = analysisRequest;
this.indexField = analysisRequest.getField();
this.storedFieldName = indexConfig.getStoredFieldName(indexField);
this.analyzer = analyzer;
this.summaryType = analysisRequest.getSummaryType();
this.computeDocLevel = analysisRequest.getDocTerms() || analysisRequest.getTokens() || AnalysisRequest.SummaryType.TOP_TERMS_TOP_N.equals(summaryType);
this.summaryLevelEnabled = analysisRequest.getSummaryTerms();
this.enabled = computeDocLevel || summaryLevelEnabled;
this.minWordLength = analysisRequest.getMinWordLen();
this.maxWordLength = analysisRequest.getMaxWordLen();
boolean needDocFreq = (analysisRequest.hasMinSegFreqPerc() || analysisRequest.hasMinSegFreqPerc() || analysisRequest
.hasMinSegFreq() || analysisRequest.hasMaxSegFreq() || TermSort.TFIDF.equals(analysisRequest.getTermSort()));
if (needDocFreq) {
this.docFreq = new DocFreq(indexReader, analysisRequest.getField());
if (analysisRequest.hasMinSegFreqPerc()) {
this.minSegmentDocFreqCount = docFreq.getNumDocsForPercent(analysisRequest.getMinSegFreqPerc());
}
if (analysisRequest.hasMaxSegFreqPerc()) {
this.maxSegmentDocFreqCount = docFreq.getNumDocsForPercent(analysisRequest.getMaxSegFreqPerc());
}
if (analysisRequest.hasMinSegFreq()) {
this.minSegmentDocFreqCount = analysisRequest.getMinSegFreq();
}
if (analysisRequest.hasMaxSegFreq()) {
this.maxSegmentDocFreqCount = analysisRequest.getMaxSegFreq();
}
}
if (summaryLevelEnabled) {
this.summaryTermFreq = new TermFreq(docFreq);
}
}
public static void handleDocument(org.bson.Document doc, List analysisHandlerList, Lumongo.ScoredResult.Builder srBuilder) {
for (AnalysisHandler analysisHandler : analysisHandlerList) {
Lumongo.AnalysisResult analysisResult = analysisHandler.handleDocument(doc);
if (analysisResult != null) {
srBuilder.addAnalysisResult(analysisResult);
}
}
}
public Lumongo.AnalysisResult handleDocument(Document document) {
if (storedFieldName != null && enabled) {
Object storeFieldValues = ResultHelper.getValueFromMongoDocument(document, storedFieldName);
Lumongo.AnalysisResult.Builder analysisResult = Lumongo.AnalysisResult.newBuilder();
analysisResult.setAnalysisRequest(analysisRequest);
TermFreq docTermFreq = null;
if (computeDocLevel || AnalysisRequest.SummaryType.TOP_TERMS_TOP_N.equals(summaryType)) {
docTermFreq = new TermFreq(docFreq);
}
final TermFreq docTermFreqFinal = docTermFreq;
LumongoUtil.handleLists(storeFieldValues, (value) -> {
String content = value.toString();
try (TokenStream tokenStream = analyzer.tokenStream(indexField, content)) {
tokenStream.reset();
while (tokenStream.incrementToken()) {
String token = tokenStream.getAttribute(CharTermAttribute.class).toString();
if (analysisRequest.getTokens()) {
analysisResult.addToken(token);
}
if (minWordLength > 0) {
if (token.length() < minWordLength) {
continue;
}
}
if (maxWordLength > 0) {
if (token.length() > maxWordLength) {
continue;
}
}
if (maxSegmentDocFreqCount != null || minSegmentDocFreqCount != null) {
int termDocFreq = this.docFreq.getDocFreq(token);
if (minSegmentDocFreqCount != null) {
if (termDocFreq < minSegmentDocFreqCount) {
continue;
}
}
if (maxSegmentDocFreqCount != null) {
if (termDocFreq > maxSegmentDocFreqCount) {
continue;
}
}
}
if (computeDocLevel || AnalysisRequest.SummaryType.TOP_TERMS_TOP_N.equals(summaryType)) {
docTermFreqFinal.addTerm(token);
}
if (summaryLevelEnabled && AnalysisRequest.SummaryType.ALL_TERMS_TOP_N.equals(summaryType)) {
summaryTermFreq.addTerm(token);
}
}
}
catch (Exception e) {
throw new RuntimeException(e);
}
});
if (computeDocLevel) {
List termBuilderList = docTermFreq.getTopTerms(analysisRequest.getTopN(), analysisRequest.getTermSort());
if (analysisRequest.getDocTerms()) {
termBuilderList.forEach(analysisResult::addTerms);
return analysisResult.build();
}
if (summaryLevelEnabled && AnalysisRequest.SummaryType.TOP_TERMS_TOP_N.equals(summaryType)) {
termBuilderList.forEach(summaryTermFreq::addTerm);
}
}
return null;
}
return null;
}
public Lumongo.AnalysisResult getSegmentResult() {
if (summaryLevelEnabled) {
Lumongo.AnalysisResult.Builder analysisResult = Lumongo.AnalysisResult.newBuilder();
analysisResult.setAnalysisRequest(analysisRequest);
//return all from segment for now
int segmentTopN = 0;
List termBuilderList = summaryTermFreq.getTopTerms(segmentTopN, analysisRequest.getTermSort());
termBuilderList.forEach(analysisResult::addTerms);
return analysisResult.build();
}
return null;
}
}
© 2015 - 2024 Weber Informatics LLC | Privacy Policy