Java Record类代码示例

import no.priv.garshol.duke.Record; //导入依赖的package包/类
@Override
public RecordIterator getRecords() {
    it = new DukeStitcherRecordIterator<T>();

    // Sequential implementation
    // for (T item : items) {
    // it.addRecord(convertToRecord(item));
    // }

    // Parallel implementation
    Collection<Record> records =
            items.parallelStream().map(item -> convertToRecord(item)).collect(Collectors.toList());
    it.addRecords(records);

    return it.start();
}
 

import no.priv.garshol.duke.Record; //导入依赖的package包/类
@Test
public void testMultiValue() throws IOException {
	// First, index up the record
	HashMap<String, Collection<String>> props = new HashMap<String, Collection<String>>();
	props.put("ID", Collections.singleton("abc"));
	Collection<String> list = new ArrayList<String>();
	list.add("b");
	list.add("c");
	props.put("NAME", list);
	Record r = new RecordImpl(props);
	db.index(r);
	db.commit();

	// Then, retrieve it and verify that it's correct
	r = db.findRecordById("abc");
	assertEquals("abc", r.getValue("ID"));
	list = r.getValues("NAME");
	assertEquals(2, list.size());
	assertTrue(list.contains("b"));
	assertTrue(list.contains("c"));
}
 

import no.priv.garshol.duke.Record; //导入依赖的package包/类
@Test
public void testOneRowMapColumnTwice() {
  perform("insert into testdata values (1, 'smith, john')");

  RegexpCleaner givencleaner = new RegexpCleaner();
  givencleaner.setRegexp(", (.+)");
  RegexpCleaner familycleaner = new RegexpCleaner();
  familycleaner.setRegexp("^([^,]+), ");
  
  source.addColumn(new Column("ID", null, null, null));
  source.addColumn(new Column("NAME", "GIVENNAME", null, givencleaner));
  source.addColumn(new Column("NAME", "FAMILYNAME", null, familycleaner));
      
  RecordIterator it = source.getRecords();
  assertTrue(it.hasNext());

  Record r = it.next();
  assertEquals("1", r.getValue("ID"));
  assertEquals("john", r.getValue("GIVENNAME"));
  assertEquals("smith", r.getValue("FAMILYNAME"));

  assertFalse(it.hasNext());
}
 

import no.priv.garshol.duke.Record; //导入依赖的package包/类
@Test
public void testBadCoordinate() throws IOException {
  Collection<Record> records = new ArrayList();
  records.add(TestUtils.makeRecord("ID", "1",
                                   "LOCATION", "59.948011,11.042239"));
  records.add(TestUtils.makeRecord("ID", "2",
                                   "LOCATION", "159.948053,11.042276"));

  try {
    processor.deduplicate(records);
  } catch (com.spatial4j.core.exception.InvalidShapeException e) {
    // this is not a legal coordinate, because a latitude of 159 degrees
    // makes no sense
    return;
  }

  fail("Invalid coordinate accepted.");
}
 

import no.priv.garshol.duke.Record; //导入依赖的package包/类
@Test
public void testMultiValue() throws IOException {
  // First, index up the record
  HashMap props = new HashMap();
  props.put("ID", Collections.singleton("abc"));
  Collection<String> list = new ArrayList();
  list.add("b");
  list.add("c");
  props.put("NAME", list);
  Record r = new RecordImpl(props);
  db.index(r);
  db.commit();

  // Then, retrieve it and verify that it's correct
  r = db.findRecordById("abc");
  assertEquals("abc", r.getValue("ID"));
  list = r.getValues("NAME");
  assertEquals(2, list.size());
  assertTrue(list.contains("b"));
  assertTrue(list.contains("c"));
}
 

import no.priv.garshol.duke.Record; //导入依赖的package包/类
@Test
public void testOneMatchOneMiss2() throws IOException {
  config.setMaybeThreshold(0.0);
  source1.add(TestUtils.makeRecord("ID", "1", "NAME", "aaaaa", "EMAIL", "bbbbb"));
  source2.add(TestUtils.makeRecord("ID", "2", "NAME", "aaaaa", "EMAIL", "bbbbb"));
  source2.add(TestUtils.makeRecord("ID", "3", "NAME", "xxxxx", "EMAIL", "bbbbb"));
  processor.link();

  assertEquals("bad record count", 2, listener.getRecordCount());
  List<TestUtils.Pair> matches = listener.getMatches();
  assertEquals("bad number of matches", 1, matches.size());
  assertEquals("bad number of missed matches", 1, listener.getNoMatchCount());

  TestUtils.Pair pair = matches.get(0);
  if (pair.r1.getValue("ID").equals("2")) {
    Record r = pair.r1;
    pair.r1 = pair.r2;
    pair.r2 = r;
  }

  assertEquals("1", pair.r1.getValue("ID"));
  assertEquals("2", pair.r2.getValue("ID"));
}
 

import no.priv.garshol.duke.Record; //导入依赖的package包/类
@Test
public void testTableModePaging() {
  setupTabularConfig1();
  source.setPageSize(1);
  source.setRows(new String[][] {
      {"http://a/1", "Peter", "15"},
      {"http://a/2", "George", "151"}});

  RecordIterator it = source.getRecords();
  assertTrue("data source contains no records", it.hasNext());

  Record r = it.next();
  assertEquals("wrong ID", "http://a/1", r.getValue("ID"));
  assertEquals("wrong NAME", "Peter", r.getValue("NAME"));
  assertEquals("wrong AGE", "15", r.getValue("AGE"));

  assertTrue("data source contains only one record", it.hasNext());

  r = it.next();
  assertEquals("wrong ID", "http://a/2", r.getValue("ID"));
  assertEquals("wrong NAME", "George", r.getValue("NAME"));
  assertEquals("wrong AGE", "151", r.getValue("AGE"));    

  assertFalse("data source contains more than two records", it.hasNext());
  assertEquals("wrong number of pages", 3, source.getPages());
}
 

import no.priv.garshol.duke.Record; //导入依赖的package包/类
@Test
public void testTwoRows() {
  setupTabularConfig1();
  source.setRows(new String[][] {
      {"http://a/1", "Peter", "15"},
      {"http://a/2", "George", "151"}});
  
  RecordIterator it = source.getRecords();
  assertTrue("data source contains no records", it.hasNext());

  Record r = it.next();
  assertEquals("wrong ID", "http://a/1", r.getValue("ID"));
  assertEquals("wrong NAME", "Peter", r.getValue("NAME"));
  assertEquals("wrong AGE", "15", r.getValue("AGE"));

  assertTrue("data source contains only one record", it.hasNext());

  r = it.next();
  assertEquals("wrong ID", "http://a/2", r.getValue("ID"));
  assertEquals("wrong NAME", "George", r.getValue("NAME"));
  assertEquals("wrong AGE", "151", r.getValue("AGE"));    
  
  assertFalse("data source contains more than two records", it.hasNext());
  assertEquals("wrong number of pages", 2, source.getPages());
}
 

import no.priv.garshol.duke.Record; //导入依赖的package包/类
@Test
public void testNoComparator() throws IOException {
  // nulling out comparator
  config.getPropertyByName("EMAIL").setComparator(null);

  // now attempt to match
  Collection<Record> records = new ArrayList();
  records.add(TestUtils.makeRecord("ID", "1",
                                   "NAME", "aaaaa",
                                   "EMAIL", "BBBBB"));
  records.add(TestUtils.makeRecord("ID", "2",
                                   "NAME", "aaaaa",
                                   "EMAIL", "BBBBB"));
  processor.deduplicate(records);

  // this shouldn't produce any matches, because we're not comparing email
  assertEquals(0, listener.getMatches().size());
  assertEquals(2, listener.getRecordCount());
}
 

import no.priv.garshol.duke.Record; //导入依赖的package包/类
@Test
public void testUseColumnTwice() throws IOException {
  source.addColumn(new Column("F1", null, null, null));
  source.addColumn(new Column("F2", "F2a", null,
                              new LowerCaseNormalizeCleaner()));
  source.addColumn(new Column("F2", "F2b", null, null));
  source.addColumn(new Column("F3", null, null, null));

  RecordIterator it = read("F1,F2,F3\na,B,c");

  Record r = it.next();
  assertEquals("a", r.getValue("F1"));
  assertEquals("b", r.getValue("F2a"));
  assertEquals("B", r.getValue("F2b"));
  assertEquals("c", r.getValue("F3"));
}
 

import no.priv.garshol.duke.Record; //导入依赖的package包/类
@Test
public void testMaybe() throws IOException {
  // this corresponds to maybe-threshold not being set at all
  config.setMaybeThreshold(0.0);

  // now lets try some matching
  Collection<Record> records = new ArrayList();
  records.add(TestUtils.makeRecord("ID", "1", "NAME", "aaaaaa",
                                   "EMAIL", "bbbbb"));
  records.add(TestUtils.makeRecord("ID", "2", "NAME", "bbbb",
                                   "EMAIL", "bbbbb"));
  processor.deduplicate(records);
  
  Collection<TestUtils.Pair> matches = listener.getMatches();
  // for (TestUtils.Pair match : matches)
  //   PrintMatchListener.show(match.r1, match.r2, match.conf, "MATCH");
  
  assertEquals("wrong number of records processed",
               2, listener.getRecordCount());
  assertEquals("found matches, but shouldn't have",
               0, matches.size());
  assertEquals("found maybe matches, but shouldn't have",
               0, listener.getMaybeCount());
  assertEquals("wrong number of no-matches",
               2, listener.getNoMatchCount());
}
 

import no.priv.garshol.duke.Record; //导入依赖的package包/类
@Test
public void testTwoRecords() {
  source.addColumn(new Column("?uri", "ID", null, null));
  source.addColumn(new Column("http://b", "PROP", null, null));
  
  RecordIterator it = read("<http://a> <http://b> \"foo\" .\n" +
                           "<http://a> <http://c> \"foo\" .\n" +
                           "<http://a> <" + RDF_TYPE + "> \"http://d\" .\n" +
                           "<http://e> <http://b> \"bar\" .\n" +
                           "<http://e> <http://c> \"foo\" .\n" +
                           "<http://e> <" + RDF_TYPE + "> \"http://f\" .\n");

  Record r = it.next();
  checkAorE(r); // we don't know the order
  assertTrue("second record not found", it.hasNext());
  r = it.next();
  checkAorE(r);
}
 

import no.priv.garshol.duke.Record; //导入依赖的package包/类
/**
 * Creates an association and saves it in the DB.
 *
 * @param r1 first record which represents an item
 * @param r2 second record which represents an item
 * @param confidence associationValue
 */

public void createAssociationBetweenRecords(Record r1, Record r2, double confidence) {

    Date execution = new Date();

    ItemAssocVO<Integer, Integer> itemAssoc = new ItemAssocVO<Integer, Integer>(
            confTanantId,
            new ItemVO(profileDukeGenerator.getTenantId(),
                    r1.getValue("ItemID"),
                    profileDukeGenerator.getItemType()),
            assocType, confidence,
            new ItemVO(profileDukeGenerator.getTenantId(),
                    r2.getValue("ItemID"),
                    profileDukeGenerator.getItemType()),
            sourceType, "ProfileDuke Plugin", viewType, null, execution);

    numberOfCreatedAssociations++;
    itemAssocDAO.insertOrUpdateItemAssoc(itemAssoc);
}
 

import no.priv.garshol.duke.Record; //导入依赖的package包/类
@Test
public void testNoValueForEmptySplit() throws IOException {
  source.addColumn(new Column("F1", null, null, null));
  Column c = new Column("F2", null, null, null);
  c.setSplitOn(";");
  source.addColumn(c);
  source.addColumn(new Column("F3", null, null, null));

  RecordIterator it = read("F1,F2,F3\na,b;;e,c");

  Record r = it.next();
  assertEquals("a", r.getValue("F1"));
  assertEquals("c", r.getValue("F3"));

  Collection<String> values = r.getValues("F2");
  assertEquals(2, values.size());
  assertTrue(values.contains("b"));
  assertTrue(values.contains("e"));
}
 

import no.priv.garshol.duke.Record; //导入依赖的package包/类
public static void htmlCompare(Record r1, Record r2, double confidence,
                                 String heading, List<Property> props) {
  System.out.println("<p>" + heading + " " + confidence + "</p>");

  System.out.println("<table>");
  for (Property p : props) {
    String prop = p.getName();
    if ((r1.getValues(prop) == null || r1.getValues(prop).isEmpty()) &&
        (r2.getValues(prop) == null || r2.getValues(prop).isEmpty()))
      continue;

    System.out.println("<tr><td>" + prop);
    System.out.println("<td>" + value(r1, prop));
    System.out.println("<td>" + value(r2, prop));
  }
  System.out.println("</table>");
}