lucene5.1 fst源码分析(BooleanQuery分析)

.1 测试代码

.1-1 写入代码

public static void writeIndex() throws IOException{

      Directory directory = FSDirectory.open(Paths.get("/Users/waixingren/bigdata-java/mysolr52test/lucenecountdata"));
      Analyzer analyzer = new StandardAnalyzer();
      IndexWriterConfig config = new IndexWriterConfig(analyzer);
      config.setUseCompoundFile(false);
      IndexWriter indexWriter = new IndexWriter(directory, config);

      Document doc1 = newDoc("1", "cubeli", "one of tencent employee");
      Document doc2 = newDoc("2", "liyong", "one of tencent employee");
      Document doc3 = newDoc("3", "yangziying", "one of tencent employee");
      Document doc4 = newDoc("4", "litairan", "one of tencent employee");
      Document doc5 = newDoc("5", "zhanshan", "one of tencent employee");
      Document doc6 = newDoc("6", "cubeli", "one of tencent employee");
      Document doc7 = newDoc("7", "lisi", "one of tencent employee");
      Document doc8 = newDoc("8", "wangwu", "cubeli, one of tencent employee");
      Document doc9 = newDoc("9", "xiaomi", "one of tencent employee");
      Document doc10 = newDoc("10", "taotao,yes he is cubeli son", "one of tencent employee");
      Document doc11 = newDoc("11", "cubeli, cubeli", "cubeli is one of tencent employee, yes cubeli");

      indexWriter.addDocument(doc1);
      indexWriter.addDocument(doc2);
      indexWriter.addDocument(doc3);
      indexWriter.addDocument(doc4);
      indexWriter.addDocument(doc5);
      indexWriter.addDocument(doc6);
      indexWriter.addDocument(doc7);
      indexWriter.addDocument(doc8);
      indexWriter.addDocument(doc9);
      indexWriter.addDocument(doc10);
      indexWriter.addDocument(doc11);
      indexWriter.close();
  }

.1-2 读取代码

        Directory directory = FSDirectory.open(Paths.get("/Users/waixingren/bigdata-java/mysolr52test/lucenecountdata"));
        DirectoryReader reader = DirectoryReader.open(directory);
        IndexSearcher isearcher = new IndexSearcher(reader);
        Query filter1 = new TermQuery(new Term("desc","cubeli"));
        Query filter2 = NumericRangeQuery.newIntRange("id", 6, 11, true, true);//这里会生成一个multitermquery，因为相比于上一个termquery的一个值，这个query会有多个值，比如7，8，9，10

        BooleanQuery bq = new BooleanQuery();
        bq.add(filter1, Occur.MUST);
        bq.add(filter2, Occur.MUST);//两个都是MUST，表示要求交集

        TopDocs topDocs = isearcher.search(bq, Integer.MAX_VALUE);
        ScoreDoc scoreDocs[] = topDocs.scoreDocs;
        for(int i = 0; i < scoreDocs.length; i++){

            Document document = isearcher.doc(scoreDocs[i].doc);
            System.out.println(document.get("id") + ", " + document.get("name") + ", " + document.get("desc"));
        }

.2 源码分析

.2-1createweight

如上图所示，这里会构造一个BooleanWeight，在构造的过程中，使用子query生成多个子weight，添加到一个list中，作为BooleanWeight的一个成员变量weights.

.2-1-1 TermQuery构造weight

这里的seekExact方法，就是先判断这个term是否存在，详细解析见

.2-1-2 MultiTermQueryConstantScoreWrapper构造weight

.2-2 构造scorer

尝试使用第一个weight构造scorer，如果失败，就返回null，例如如果是求交集，那么使用一个scorer是不行的：

所以，要走以下流程构造scorer：

注意，对于multiterm来说，需要遍历所有的term值，例如：

.2-3 求交集/并集/差集

未完待续。。。