Lucene查询时连接返回零结果，求排查解决方案

问题描述 #

尝试使用Lucene的Query Time Join功能连接两个独立索引，两个索引均有效，共享Long类型的关联ID字段，但连接查询始终返回0结果。测试代码如下：

package org.test;

import java.io.File;
import java.nio.file.FileSystems;

import org.apache.lucene.analysis.Analyzer;
import org.apache.lucene.analysis.standard.StandardAnalyzer;
import org.apache.lucene.document.Document;
import org.apache.lucene.document.Field;
import org.apache.lucene.document.FieldType;
import org.apache.lucene.document.NumericDocValuesField;
import org.apache.lucene.document.StoredField;
import org.apache.lucene.document.TextField;
import org.apache.lucene.index.DirectoryReader;
import org.apache.lucene.index.IndexReader;
import org.apache.lucene.index.IndexWriter;
import org.apache.lucene.index.IndexWriterConfig;
import org.apache.lucene.index.Term;
import org.apache.lucene.search.BooleanClause;
import org.apache.lucene.search.BooleanQuery;
import org.apache.lucene.search.IndexSearcher;
import org.apache.lucene.search.Query;
import org.apache.lucene.search.TermQuery;
import org.apache.lucene.search.TopDocs;
import org.apache.lucene.search.join.JoinUtil;
import org.apache.lucene.search.join.ScoreMode;
import org.apache.lucene.store.Directory;
import org.apache.lucene.store.FSDirectory;
import org.apache.lucene.store.NIOFSDirectory;

public class LuceneJoinTest {

    public static void main(String[] args) {
        try {
            buildIndex("C:\\Temp\\index\\fromIndex", "fromFieldId", 50);
            buildIndex("C:\\Temp\\index\\toIndex", "toFieldId", 10);
            searchIndex();

        } catch (Throwable t) {
            t.printStackTrace();
        }
    }

    public static void buildIndex(String path, String fieldName, int docCount) throws Exception {

        System.out.println("buildIndex:" + path);
        Analyzer la = new StandardAnalyzer();
        IndexWriterConfig config = new IndexWriterConfig(la);
        config.setRAMBufferSizeMB(50);
        Directory d = new NIOFSDirectory(FileSystems.getDefault().getPath(path));
        IndexWriter writer = new IndexWriter(d, config);

        for (int i = 1; i <= docCount; i++) {
            System.out.println("Created Document:" + i);
            Integer id = Integer.valueOf(i);
            Document doc = new Document();

            FieldType storedOmitNorms = new FieldType(TextField.TYPE_STORED);
            storedOmitNorms.setOmitNorms(false);
            doc.add(new Field("id", id.toString(), storedOmitNorms));
            doc.add(new NumericDocValuesField(fieldName, id.longValue()));
            doc.add(new StoredField(fieldName, id.longValue()));

            FieldType fieldNotStoredOmitNorms = new FieldType(TextField.TYPE_NOT_STORED);
            fieldNotStoredOmitNorms.setOmitNorms(false);
            doc.add(new Field("content", "one two three...", fieldNotStoredOmitNorms));

            writer.addDocument(doc);
        }

        writer.close();
    }

    public static void searchIndex() throws Exception {

        Directory fromIndex = FSDirectory.open(new File("C:\\Temp\\index\\fromIndex").toPath());
        IndexReader fromReader = DirectoryReader.open(fromIndex);
        IndexSearcher fromSearcher = new IndexSearcher(fromReader);

        // Test the FROM INDEX
        BooleanQuery.Builder fromQueryBuilder = new BooleanQuery.Builder();
        Term fromTerm = new Term("content", "two");
        fromQueryBuilder.add(new TermQuery(fromTerm), BooleanClause.Occur.MUST);
        Query fromQuery = fromQueryBuilder.build();

        TopDocs fromResults = fromSearcher.search(fromQuery, 100);
        System.out.println("fromSearcher results:" + fromResults.totalHits);

        // Test the TO INDEX
        Directory toIndex = FSDirectory.open(new File("C:\\Temp\\index\\toIndex").toPath());
        IndexReader toReader = DirectoryReader.open(toIndex);
        IndexSearcher toSearcher = new IndexSearcher(toReader);

        BooleanQuery.Builder toQueryBuilder = new BooleanQuery.Builder();
        Term toTerm = new Term("content", "three");
        toQueryBuilder.add(new TermQuery(toTerm), BooleanClause.Occur.MUST);
        Query toQuery = toQueryBuilder.build();

        TopDocs toResults = toSearcher.search(toQuery, 100);
        System.out.println("toSearcher results:" + toResults.totalHits);

        // Now test with a JOIN...

        Query joinQuery = JoinUtil.createJoinQuery("fromFieldId", false, "toFieldId", Long.class, fromQuery,
                fromSearcher, ScoreMode.None);

        TopDocs joinResults = toSearcher.search(joinQuery, 100);
        System.out.println("joinResults:" + joinResults.totalHits);

    }
}

核心问题 #

Lucene的JoinUtil仅支持单个索引内的文档关联（比如父子文档、同索引内的关联字段），不支持直接跨两个独立索引执行Join操作。这是导致查询返回0结果的根本原因。

排查思路 #

1. 确认JoinUtil适用场景：Lucene官方明确说明Query Time Join是为单索引内关联设计的，跨索引Join需手动实现。
2. 验证字段配置：检查两个索引的关联字段是否为NumericDocValuesField（你的配置是正确的，这是Join所需的字段类型）。
3. 单索引验证：将两个索引合并到同一目录，测试JoinUtil是否正常工作，排除字段本身的问题。

解决方案 #

方案1：手动实现跨索引Join #

先查询源索引获取匹配的关联ID集合，再用这些ID作为条件查询目标索引。修改searchIndex方法如下：

import java.util.HashSet;
import java.util.Set;
import org.apache.lucene.search.NumericRangeQuery;

public static void searchIndex() throws Exception {

    Directory fromIndex = FSDirectory.open(new File("C:\\Temp\\index\\fromIndex").toPath());
    IndexReader fromReader = DirectoryReader.open(fromIndex);
    IndexSearcher fromSearcher = new IndexSearcher(fromReader);

    // 源索引查询
    BooleanQuery.Builder fromQueryBuilder = new BooleanQuery.Builder();
    Term fromTerm = new Term("content", "two");
    fromQueryBuilder.add(new TermQuery(fromTerm), BooleanClause.Occur.MUST);
    Query fromQuery = fromQueryBuilder.build();

    TopDocs fromResults = fromSearcher.search(fromQuery, 100);
    System.out.println("fromSearcher results:" + fromResults.totalHits);

    // 提取匹配文档的关联ID
    Set<Long> matchedIds = new HashSet<>();
    for (int i = 0; i < fromResults.scoreDocs.length; i++) {
        int docId = fromResults.scoreDocs[i].doc;
        Document doc = fromSearcher.doc(docId);
        long id = doc.getField("fromFieldId").numericValue().longValue();
        matchedIds.add(id);
    }

    // 目标索引查询
    Directory toIndex = FSDirectory.open(new File("C:\\Temp\\index\\toIndex").toPath());
    IndexReader toReader = DirectoryReader.open(toIndex);
    IndexSearcher toSearcher = new IndexSearcher(toReader);

    BooleanQuery.Builder toQueryBuilder = new BooleanQuery.Builder();
    Term toTerm = new Term("content", "three");
    toQueryBuilder.add(new TermQuery(toTerm), BooleanClause.Occur.MUST);

    // 添加ID匹配条件
    if (!matchedIds.isEmpty()) {
        BooleanQuery.Builder idQueryBuilder = new BooleanQuery.Builder();
        for (Long id : matchedIds) {
            Query idQuery = NumericRangeQuery.newLongRange("toFieldId", id, id, true, true);
            idQueryBuilder.add(idQuery, BooleanClause.Occur.SHOULD);
        }
        toQueryBuilder.add(idQueryBuilder.build(), BooleanClause.Occur.MUST);
    }

    Query finalToQuery = toQueryBuilder.build();
    TopDocs joinResults = toSearcher.search(finalToQuery, 100);
    System.out.println("joinResults:" + joinResults.totalHits);

    // 关闭资源
    fromReader.close();
    toReader.close();
}

方案2：合并索引到同一目录 #

如果业务允许，将两个索引的文档合并到同一个索引，添加docType字段区分来源，再使用JoinUtil进行单索引内Join：

1. 修改buildIndex方法，添加类型标识字段：

   doc.add(new StoredField("docType", path.contains("fromIndex") ? "from" : "to"));
   

2. 合并索引：使用IndexWriter.addIndexes方法将两个索引合并到同一目录。
3. 调整查询：通过docType字段过滤不同类型的文档，再调用JoinUtil执行单索引内Join。

关键注意事项 #

• 关联字段必须是NumericDocValuesField或SortedDocValuesField，确保Join能高效获取字段值。
• 手动跨索引Join时，若匹配ID过多，需注意BooleanQuery的最大子句数限制，可改用分批次查询优化。

内容的提问来源于stack exchange，提问作者BMY