testtermvectorsreader.java

Lucene a java open-source SearchEngine Framework

    assertTrue(reader != null);
    TermPositionVector vector = (TermPositionVector) reader.get(0, testFields[0]);
    assertTrue(vector != null);
    String[] terms = vector.getTerms();
    assertTrue(terms != null);
    assertTrue(terms.length == testTerms.length);
    for (int i = 0; i < terms.length; i++) {
      String term = terms[i];
      //System.out.println("Term: " + term);
      assertTrue(term.equals(testTerms[i]));
      int[] positions = vector.getTermPositions(i);
      assertTrue(positions != null);
      assertTrue(positions.length == this.positions[i].length);
      for (int j = 0; j < positions.length; j++) {
        int position = positions[j];
        assertTrue(position == this.positions[i][j]);
      }
      TermVectorOffsetInfo[] offset = vector.getOffsets(i);
      assertTrue(offset != null);
      assertTrue(offset.length == this.offsets[i].length);
      for (int j = 0; j < offset.length; j++) {
        TermVectorOffsetInfo termVectorOffsetInfo = offset[j];
        assertTrue(termVectorOffsetInfo.equals(offsets[i][j]));
      }
    }
  }

  public void testMapper() throws IOException {
    TermVectorsReader reader = new TermVectorsReader(dir, seg, fieldInfos);
    assertTrue(reader != null);
    SortedTermVectorMapper mapper = new SortedTermVectorMapper(new TermVectorEntryFreqSortedComparator());
    reader.get(0, mapper);
    SortedSet set = mapper.getTermVectorEntrySet();
    assertTrue("set is null and it shouldn't be", set != null);
    //three fields, 4 terms, all terms are the same
    assertTrue("set Size: " + set.size() + " is not: " + 4, set.size() == 4);
    //Check offsets and positions
    for (Iterator iterator = set.iterator(); iterator.hasNext();) {
      TermVectorEntry tve = (TermVectorEntry) iterator.next();
      assertTrue("tve is null and it shouldn't be", tve != null);
      assertTrue("tve.getOffsets() is null and it shouldn't be", tve.getOffsets() != null);
      assertTrue("tve.getPositions() is null and it shouldn't be", tve.getPositions() != null);

    }

    mapper = new SortedTermVectorMapper(new TermVectorEntryFreqSortedComparator());
    reader.get(1, mapper);
    set = mapper.getTermVectorEntrySet();
    assertTrue("set is null and it shouldn't be", set != null);
    //three fields, 4 terms, all terms are the same
    assertTrue("set Size: " + set.size() + " is not: " + 4, set.size() == 4);
    //Should have offsets and positions b/c we are munging all the fields together
    for (Iterator iterator = set.iterator(); iterator.hasNext();) {
      TermVectorEntry tve = (TermVectorEntry) iterator.next();
      assertTrue("tve is null and it shouldn't be", tve != null);
      assertTrue("tve.getOffsets() is null and it shouldn't be", tve.getOffsets() != null);
      assertTrue("tve.getPositions() is null and it shouldn't be", tve.getPositions() != null);

    }


    FieldSortedTermVectorMapper fsMapper = new FieldSortedTermVectorMapper(new TermVectorEntryFreqSortedComparator());
    reader.get(0, fsMapper);
    Map map = fsMapper.getFieldToTerms();
    assertTrue("map Size: " + map.size() + " is not: " + testFields.length, map.size() == testFields.length);
    for (Iterator iterator = map.entrySet().iterator(); iterator.hasNext();) {
      Map.Entry entry = (Map.Entry) iterator.next();
      SortedSet sortedSet = (SortedSet) entry.getValue();
      assertTrue("sortedSet Size: " + sortedSet.size() + " is not: " + 4, sortedSet.size() == 4);
      for (Iterator inner = sortedSet.iterator(); inner.hasNext();) {
        TermVectorEntry tve = (TermVectorEntry) inner.next();
        assertTrue("tve is null and it shouldn't be", tve != null);
        //Check offsets and positions.
        assertTrue("tve is null and it shouldn't be", tve != null);
        String field = tve.getField();
        if (field.equals(testFields[0])) {
          //should have offsets

          assertTrue("tve.getOffsets() is null and it shouldn't be", tve.getOffsets() != null);
          assertTrue("tve.getPositions() is null and it shouldn't be", tve.getPositions() != null);
        }
        else if (field.equals(testFields[1])) {
          //should not have offsets

          assertTrue("tve.getOffsets() is not null and it shouldn't be", tve.getOffsets() == null);
          assertTrue("tve.getPositions() is not null and it shouldn't be", tve.getPositions() == null);
        }
      }
    }
    //Try mapper that ignores offs and positions
    fsMapper = new FieldSortedTermVectorMapper(true, true, new TermVectorEntryFreqSortedComparator());
    reader.get(0, fsMapper);
    map = fsMapper.getFieldToTerms();
    assertTrue("map Size: " + map.size() + " is not: " + testFields.length, map.size() == testFields.length);
    for (Iterator iterator = map.entrySet().iterator(); iterator.hasNext();) {
      Map.Entry entry = (Map.Entry) iterator.next();
      SortedSet sortedSet = (SortedSet) entry.getValue();
      assertTrue("sortedSet Size: " + sortedSet.size() + " is not: " + 4, sortedSet.size() == 4);
      for (Iterator inner = sortedSet.iterator(); inner.hasNext();) {
        TermVectorEntry tve = (TermVectorEntry) inner.next();
        assertTrue("tve is null and it shouldn't be", tve != null);
        //Check offsets and positions.
        assertTrue("tve is null and it shouldn't be", tve != null);
        String field = tve.getField();
        if (field.equals(testFields[0])) {
          //should have offsets

          assertTrue("tve.getOffsets() is null and it shouldn't be", tve.getOffsets() == null);
          assertTrue("tve.getPositions() is null and it shouldn't be", tve.getPositions() == null);
        }
        else if (field.equals(testFields[1])) {
          //should not have offsets

          assertTrue("tve.getOffsets() is not null and it shouldn't be", tve.getOffsets() == null);
          assertTrue("tve.getPositions() is not null and it shouldn't be", tve.getPositions() == null);
        }
      }
    }

    // test setDocumentNumber()
    IndexReader ir = IndexReader.open(dir);
    DocNumAwareMapper docNumAwareMapper = new DocNumAwareMapper();
    assertEquals(-1, docNumAwareMapper.getDocumentNumber());

    ir.getTermFreqVector(0, docNumAwareMapper);
    assertEquals(0, docNumAwareMapper.getDocumentNumber());
    docNumAwareMapper.setDocumentNumber(-1);

    ir.getTermFreqVector(1, docNumAwareMapper);
    assertEquals(1, docNumAwareMapper.getDocumentNumber());
    docNumAwareMapper.setDocumentNumber(-1);

    ir.getTermFreqVector(0, "f1", docNumAwareMapper);
    assertEquals(0, docNumAwareMapper.getDocumentNumber());
    docNumAwareMapper.setDocumentNumber(-1);

    ir.getTermFreqVector(1, "f2", docNumAwareMapper);
    assertEquals(1, docNumAwareMapper.getDocumentNumber());
    docNumAwareMapper.setDocumentNumber(-1);

    ir.getTermFreqVector(0, "f1", docNumAwareMapper);
    assertEquals(0, docNumAwareMapper.getDocumentNumber());

    ir.close();

  }


  /**
   * Make sure exceptions and bad params are handled appropriately
   */
  public void testBadParams() {
    try {
      TermVectorsReader reader = new TermVectorsReader(dir, seg, fieldInfos);
      assertTrue(reader != null);
      //Bad document number, good field number
      reader.get(50, testFields[0]);
      fail();
    } catch (IOException e) {
      // expected exception
    }
    try {
      TermVectorsReader reader = new TermVectorsReader(dir, seg, fieldInfos);
      assertTrue(reader != null);
      //Bad document number, no field
      reader.get(50);
      fail();
    } catch (IOException e) {
      // expected exception
    }
    try {
      TermVectorsReader reader = new TermVectorsReader(dir, seg, fieldInfos);
      assertTrue(reader != null);
      //good document number, bad field number
      TermFreqVector vector = reader.get(0, "f50");
      assertTrue(vector == null);
    } catch (IOException e) {
      fail();
    }
  }


  public static class DocNumAwareMapper extends TermVectorMapper {

    public DocNumAwareMapper() {
    }

    private int documentNumber = -1;

    public void setExpectations(String field, int numTerms, boolean storeOffsets, boolean storePositions) {
      if (documentNumber == -1) {
        throw new RuntimeException("Documentnumber should be set at this point!");
      }
    }

    public void map(String term, int frequency, TermVectorOffsetInfo[] offsets, int[] positions) {
      if (documentNumber == -1) {
        throw new RuntimeException("Documentnumber should be set at this point!");
      }
    }

    public int getDocumentNumber() {
      return documentNumber;
    }

    public void setDocumentNumber(int documentNumber) {
      this.documentNumber = documentNumber;
    }
  }
}