Пример #1
0
  protected void setUp() throws IOException {
    String contents1Doc1 =
        "word1 word2 word3 word32 word1 word45 word56 word67 word67 word2 word67 word88 word99 word99 word33";
    String contents2Doc1 = "word1 word2 word3 word32 word1 word45 word56 word67";
    String contents1Doc2 = "word2 word3 word4 word55 word96 word2 word54 word33 wordss";
    String contents2Doc2 = "word2 word3 word4 word55";
    String contents1Doc3 = "word1 word100 word400 word555 word966 word544 word333 wordss";
    String contents2Doc3 = "word1 word100";
    LgteIndexWriter writer = new LgteIndexWriter(pathUnique, true, Model.OkapiBM25Model);
    LgteDocumentWrapper doc1 = new LgteDocumentWrapper();
    doc1.indexText(Globals.DOCUMENT_ID_FIELD, "1");
    doc1.indexText("contents1", contents1Doc1);
    doc1.indexText("contents2", contents2Doc1);
    LgteDocumentWrapper doc2 = new LgteDocumentWrapper();
    doc2.indexText(Globals.DOCUMENT_ID_FIELD, "2");
    doc2.indexText("contents1", contents1Doc2);
    doc2.indexText("contents2", contents2Doc2);
    LgteDocumentWrapper doc3 = new LgteDocumentWrapper();
    doc3.indexText(Globals.DOCUMENT_ID_FIELD, "3");
    doc3.indexText("contents1", contents1Doc3);
    doc3.indexText("contents2", contents2Doc3);

    writer.addDocument(doc1);
    writer.addDocument(doc2);
    writer.addDocument(doc3);
    writer.close();

    // these two writers will replace the previous one
    // The first one will index contents1 and the second will index the contents2
    writer = new LgteIndexWriter(pathMulti1, true);
    doc1 = new LgteDocumentWrapper();
    doc1.indexText(Globals.DOCUMENT_ID_FIELD, "1");
    doc1.indexText("contents1", contents1Doc1);
    doc2 = new LgteDocumentWrapper();
    doc2.indexText(Globals.DOCUMENT_ID_FIELD, "2");
    doc2.indexText("contents1", contents1Doc2);
    doc3 = new LgteDocumentWrapper();
    doc3.indexText(Globals.DOCUMENT_ID_FIELD, "3");
    doc3.indexText("contents1", contents1Doc3);
    writer.addDocument(doc1);
    writer.addDocument(doc2);
    writer.addDocument(doc3);
    writer.close();

    writer = new LgteIndexWriter(pathMulti2, true);
    doc1 = new LgteDocumentWrapper();
    doc1.indexText(Globals.DOCUMENT_ID_FIELD, "1");
    doc1.indexText("contents2", contents2Doc1);
    doc2 = new LgteDocumentWrapper();
    doc2.indexText(Globals.DOCUMENT_ID_FIELD, "2");
    doc2.indexText("contents2", contents2Doc2);
    doc3 = new LgteDocumentWrapper();
    doc3.indexText(Globals.DOCUMENT_ID_FIELD, "3");
    doc3.indexText("contents2", contents2Doc3);
    writer.addDocument(doc1);
    writer.addDocument(doc2);
    writer.addDocument(doc3);
    writer.close();
  }