org.apache.solr.common.SolrInputDocument#getFieldValues

Source File: AtomicUpdateDocumentMerger.java From lucene-solr with Apache License 2.0

6 votes

/**
 *
 * @param fullDoc the full doc to  be compared against
 * @param partialDoc the sub document to be tested
 * @return whether partialDoc is derived from fullDoc
 */
public static boolean isDerivedFromDoc(SolrInputDocument fullDoc, SolrInputDocument partialDoc) {
  for(SolrInputField subSif: partialDoc) {
    Collection<Object> fieldValues = fullDoc.getFieldValues(subSif.getName());
    if (fieldValues == null) return false;
    if (fieldValues.size() < subSif.getValueCount()) return false;
    Collection<Object> partialFieldValues = subSif.getValues();
    // filter all derived child docs from partial field values since they fail List#containsAll check (uses SolrInputDocument#equals which fails).
    // If a child doc exists in partialDoc but not in full doc, it will not be filtered, and therefore List#containsAll will return false
    Stream<Object> nonChildDocElements = partialFieldValues.stream().filter(x -> !(isChildDoc(x) &&
        (fieldValues.stream().anyMatch(y ->
            (isChildDoc(x) &&
                isDerivedFromDoc((SolrInputDocument) y, (SolrInputDocument) x)
            )
        )
        )));
    if (!nonChildDocElements.allMatch(fieldValues::contains)) return false;
  }
  return true;
}

Source File: ClassificationUpdateProcessorTest.java From lucene-solr with Apache License 2.0

6 votes

@Test
public void knnMultiClass_boostFieldsMaxOutputClasses2_shouldAssignMax2Classes() throws Exception {
  UpdateRequestProcessor mockProcessor=mock(UpdateRequestProcessor.class);
  prepareTrainedIndexMultiClass();

  AddUpdateCommand update=new AddUpdateCommand(req());
  SolrInputDocument unseenDocument1 = sdoc(ID, "10",
      TITLE, "word4 word4 word4",
      CONTENT, "word2 word2 ",
      AUTHOR, "unseenAuthor");
  update.solrDoc=unseenDocument1;

  ClassificationUpdateProcessorParams params= initParams(ClassificationUpdateProcessorFactory.Algorithm.KNN);
  params.setInputFieldNames(new String[]{TITLE+"^1.5",CONTENT+"^0.5",AUTHOR+"^2.5"});
  params.setMaxPredictedClasses(2);

  updateProcessorToTest=new ClassificationUpdateProcessor(params,mockProcessor,reader,req().getSchema());

  updateProcessorToTest.processAdd(update);

  @SuppressWarnings({"unchecked"})
  ArrayList<Object> assignedClasses = (ArrayList)unseenDocument1.getFieldValues(TRAINING_CLASS);
  assertThat(assignedClasses.size(),is(2));
  assertThat(assignedClasses.get(0),is("class4"));
  assertThat(assignedClasses.get(1),is("class6"));
}

Source File: ClassificationUpdateProcessorTest.java From lucene-solr with Apache License 2.0

6 votes

@Test
public void bayesMultiClass_maxOutputClasses2_shouldAssignMax2Classes() throws Exception {
  UpdateRequestProcessor mockProcessor=mock(UpdateRequestProcessor.class);
  prepareTrainedIndexMultiClass();

  AddUpdateCommand update=new AddUpdateCommand(req());
  SolrInputDocument unseenDocument1 = sdoc(ID, "10",
      TITLE, "word1 word1 word1",
      CONTENT, "word2 word2 ",
      AUTHOR, "unseenAuthor");
  update.solrDoc=unseenDocument1;

  ClassificationUpdateProcessorParams params= initParams(ClassificationUpdateProcessorFactory.Algorithm.BAYES);
  params.setMaxPredictedClasses(2);

  updateProcessorToTest=new ClassificationUpdateProcessor(params,mockProcessor,reader,req().getSchema());
  updateProcessorToTest.processAdd(update);

  @SuppressWarnings({"unchecked"})
  ArrayList<Object> assignedClasses = (ArrayList)unseenDocument1.getFieldValues(TRAINING_CLASS);
  assertThat(assignedClasses.size(),is(2));
  assertThat(assignedClasses.get(0),is("class2"));
  assertThat(assignedClasses.get(1),is("class1"));
}

Source File: CloneFieldUpdateProcessorFactoryTest.java From lucene-solr with Apache License 2.0

6 votes

@Test
public void testSelectorClone() throws Exception {
  SolrInputDocument doc = processAdd("clone-selector",
                                     doc(f("id", "1"),
                                         f("source0_s", "nope, not me"),
                                         f("source1_s", "foo"),
                                         f("source2_s", "bar")));
                                         
  assertEquals("source0_s should have stringValue", "nope, not me", doc.getFieldValue("source0_s"));
  assertEquals("source1_s should have stringValue", "foo", doc.getFieldValue("source1_s"));
  assertEquals("source2_s should have stringValue", "bar", doc.getFieldValue("source2_s"));
  Collection<Object> dest_s = doc.getFieldValues("dest_s");
  assertTrue(dest_s.contains("foo"));
  assertTrue(dest_s.contains("bar"));
  assertFalse(dest_s.contains("nope, not me"));
}

Source File: ClassificationUpdateProcessorTest.java From lucene-solr with Apache License 2.0

6 votes

@Test
public void knnMultiClass_maxOutputClasses2_shouldAssignMax2Classes() throws Exception {
  UpdateRequestProcessor mockProcessor=mock(UpdateRequestProcessor.class);
  prepareTrainedIndexMultiClass();

  AddUpdateCommand update=new AddUpdateCommand(req());
  SolrInputDocument unseenDocument1 = sdoc(ID, "10",
      TITLE, "word1 word1 word1",
      CONTENT, "word2 word2 ",
      AUTHOR, "unseenAuthor");
  update.solrDoc=unseenDocument1;

  ClassificationUpdateProcessorParams params= initParams(ClassificationUpdateProcessorFactory.Algorithm.KNN);
  params.setMaxPredictedClasses(2);

  updateProcessorToTest=new ClassificationUpdateProcessor(params,mockProcessor,reader,req().getSchema());
  updateProcessorToTest.processAdd(update);

  @SuppressWarnings({"unchecked"})
  ArrayList<Object> assignedClasses = (ArrayList)unseenDocument1.getFieldValues(TRAINING_CLASS);
  assertThat(assignedClasses.size(),is(2));
  assertThat(assignedClasses.get(0),is("class2"));
  assertThat(assignedClasses.get(1),is("class1"));
}