Package no.priv.garshol.duke

Examples of no.priv.garshol.duke.ConfigurationImpl$HighComparator


    props.add(email);
    Property phone = new PropertyImpl("PHONE", comp, 0.4, 0.51);
    props.add(phone);
    phone.setLookupBehaviour(Property.Lookup.TRUE);

    ConfigurationImpl config = new ConfigurationImpl();
    config.setThreshold(0.85);
    config.setProperties(props);
    config.validate();

    Collection<Property> lookups = config.getLookupProperties();
    assertEquals(3, lookups.size());
    assertTrue(lookups.contains(name));
    assertTrue(lookups.contains(email));
    assertTrue(lookups.contains(phone));
  }
View Full Code Here


    Property email = new PropertyImpl("EMAIL", comp, 0.48, 0.8);
    props.add(email);
    Property phone = new PropertyImpl("PHONE", comp, 0.48, 0.51);
    props.add(phone);

    ConfigurationImpl config = new ConfigurationImpl();
    config.setThreshold(0.85);
    config.setProperties(props);
    config.validate();

    Collection<Property> lookups = config.getLookupProperties();
    assertEquals(2, lookups.size());
    assertTrue(lookups.contains(name));
    assertTrue(lookups.contains(email));
  }
View Full Code Here

    props.add(email);
    Property phone = new PropertyImpl("PHONE", comp, 0.4, 0.51);
    props.add(phone);
    phone.setLookupBehaviour(Property.Lookup.REQUIRED);

    ConfigurationImpl config = new ConfigurationImpl();
    config.setThreshold(0.85);
    config.setProperties(props);
    config.validate();

    Collection<Property> lookups = config.getLookupProperties();
    assertEquals(3, lookups.size());
    assertTrue(lookups.contains(name));
    assertTrue(lookups.contains(email));
    assertTrue(lookups.contains(phone));
  }
View Full Code Here

    List<Property> props = new ArrayList();
    props.add(new PropertyImpl("ID"));
    props.add(new PropertyImpl("NAME", lev, 0.3, 0.8));
    props.add(new PropertyImpl("EMAIL", lev, 0.3, 0.8));
   
    config1 = new ConfigurationImpl();
    ((ConfigurationImpl) config1).setProperties(props);
    ((ConfigurationImpl) config1).setThreshold(0.85);
  }
View Full Code Here

    List<Property> props = new ArrayList();
    props.add(new PropertyImpl("ID"));
    props.add(new PropertyImpl("NAME", lev, 0.2, 0.9));
    props.add(new PropertyImpl("EMAIL", lev, 0.2, 0.9));
   
    Configuration other = new ConfigurationImpl();
    ((ConfigurationImpl) other).setProperties(props);
    ((ConfigurationImpl) other).setThreshold(0.75);
    GeneticConfiguration g_other = new GeneticConfiguration(other);

    // proceed to mate
View Full Code Here

  public void setup() throws CorruptIndexException, IOException {
    ExactComparator comp = new ExactComparator();
    List<Property> props = new ArrayList();
    props.add(new PropertyImpl("ID"));
    props.add(new PropertyImpl("NAME", comp, 0.3, 0.8));
    ConfigurationImpl config = new ConfigurationImpl();
    config.setProperties(props);
    config.setThreshold(0.45);
    db = new LuceneDatabase();
    db.setConfiguration(config);
  }
View Full Code Here

    List<Property> props = new ArrayList();
    props.add(new PropertyImpl("ID"));
    props.add(new PropertyImpl("NAME", comp, 0.3, 0.8));
    props.add(new PropertyImpl("EMAIL", comp, 0.3, 0.8));

    config = new ConfigurationImpl();
    config.setProperties(props);
    config.setThreshold(0.85);
    config.setMaybeThreshold(0.8);
    db = createDatabase(config);
  }
View Full Code Here

    List<Property> props = new ArrayList();
    props.add(new PropertyImpl("ID"));
    props.add(new PropertyImpl("NAME", lev, 0.3, 0.8));
    props.add(new PropertyImpl("EMAIL", lev, 0.3, 0.8));
   
    Configuration config = new ConfigurationImpl();
    ((ConfigurationImpl) config).setProperties(props);
    ((ConfigurationImpl) config).setThreshold(0.85);
    ((ConfigurationImpl) config).setMaybeThreshold(0.7);

    // --- write and reload
    File outfile = tmpdir.newFile("config.xml");
    ConfigWriter.write(config, outfile.getAbsolutePath());
    config = ConfigLoader.load(outfile.getAbsolutePath());

    // --- verify loaded correctly   
    assertTrue(config.getDataSources().isEmpty());
    assertTrue(config.getDataSources(1).isEmpty());
    assertTrue(config.getDataSources(2).isEmpty());
    assertEquals(config.getThreshold(), 0.85);
    assertEquals(config.getMaybeThreshold(), 0.7);
    assertEquals(3, config.getProperties().size());

    Property prop = config.getPropertyByName("ID");
    assertTrue("ID property lost", prop.isIdProperty());

    prop = config.getPropertyByName("NAME");
    assertEquals(lev.getClass(), prop.getComparator().getClass());
    assertEquals(0.3, prop.getLowProbability());
    assertEquals(0.8, prop.getHighProbability());

    prop = config.getPropertyByName("EMAIL");
    assertEquals(lev.getClass(), prop.getComparator().getClass());
    assertEquals(0.3, prop.getLowProbability());
    assertEquals(0.8, prop.getHighProbability());
  }
View Full Code Here

    List<Property> props = new ArrayList();
    props.add(new PropertyImpl("ID"));
    props.add(new PropertyImpl("NAME", lev, 0.3, 0.8));
    props.add(new PropertyImpl("EMAIL", lev, 0.3, 0.8));
   
    Configuration config = new ConfigurationImpl();
    ((ConfigurationImpl) config).setProperties(props);
    ((ConfigurationImpl) config).setThreshold(0.85);
    ((ConfigurationImpl) config).setMaybeThreshold(0.7);

    CSVDataSource csv = new CSVDataSource();
    csv.setInputFile("test.csv");
    csv.addColumn(new Column("id", "ID", null, null));
    csv.addColumn(new Column("name", "NAME", null, null));
    Column emailCol = new Column("email", "EMAIL", null, null);
    emailCol.setSplitOn(";");
    csv.addColumn(emailCol);
    ((ConfigurationImpl) config).addDataSource(0, csv);
   
    // --- write and reload
    File outfile = tmpdir.newFile("config.xml");           
    ConfigWriter.write(config, outfile.getAbsolutePath());
    config = ConfigLoader.load(outfile.getAbsolutePath());
   
    // --- verify loaded correctly   
    assertEquals(1, config.getDataSources().size());

    csv = (CSVDataSource) config.getDataSources().iterator().next();
    assertTrue(csv.getInputFile().endsWith("test.csv"));
    assertEquals(3, csv.getColumns().size());
    Collection<Column> csvEmailColList = csv.getColumn("email");
    Column csvEmailCol = (Column) csvEmailColList.toArray()[0];
    assertTrue(csvEmailCol.isSplit());
    // FIXME: check the columns (kind of hard given lack of ordering)
   
    assertTrue(config.getDataSources(1).isEmpty());
    assertTrue(config.getDataSources(2).isEmpty());
    assertEquals(config.getThreshold(), 0.85);
    assertEquals(config.getMaybeThreshold(), 0.7);
    assertEquals(3, config.getProperties().size());

    Property prop = config.getPropertyByName("ID");
    assertTrue("ID property lost", prop.isIdProperty());

    prop = config.getPropertyByName("NAME");
    assertEquals(lev.getClass(), prop.getComparator().getClass());
    assertEquals(0.3, prop.getLowProbability());
    assertEquals(0.8, prop.getHighProbability());

    prop = config.getPropertyByName("EMAIL");
    assertEquals(lev.getClass(), prop.getComparator().getClass());
    assertEquals(0.3, prop.getLowProbability());
    assertEquals(0.8, prop.getHighProbability());
  }
View Full Code Here

    List<Property> props = new ArrayList();
    props.add(new PropertyImpl("ID"));
    props.add(new PropertyImpl("NAME", lev, 0.3, 0.8));
    props.add(new PropertyImpl("EMAIL", lev, 0.3, 0.8));
   
    Configuration config = new ConfigurationImpl();
    ((ConfigurationImpl) config).setProperties(props);
    ((ConfigurationImpl) config).setThreshold(0.85);
    ((ConfigurationImpl) config).setMaybeThreshold(0.7);

    JDBCDataSource jdbc = new JDBCDataSource();
    jdbc.setDriverClass("klass");
    jdbc.setConnectionString("konnection");
    jdbc.setUserName("user");
    jdbc.setPassword("secret");
    jdbc.setQuery("select");
    jdbc.addColumn(new Column("id", "ID", null, null));
    jdbc.addColumn(new Column("name", "NAME", null, null));
    jdbc.addColumn(new Column("email", "EMAIL", null, null));
    ((ConfigurationImpl) config).addDataSource(0, jdbc);
   
    // --- write and reload
    File outfile = tmpdir.newFile("config.xml");           
    ConfigWriter.write(config, outfile.getAbsolutePath());
    config = ConfigLoader.load(outfile.getAbsolutePath());
   
    // --- verify loaded correctly   
    assertEquals(1, config.getDataSources().size());

    jdbc = (JDBCDataSource) config.getDataSources().iterator().next();
    assertEquals("klass", jdbc.getDriverClass());
    assertEquals("konnection", jdbc.getConnectionString());
    assertEquals("user", jdbc.getUserName());
    assertEquals("secret", jdbc.getPassword());
    assertEquals("select", jdbc.getQuery());
    assertEquals(3, jdbc.getColumns().size());
    // FIXME: check the columns (kind of hard given lack of ordering)
   
    assertTrue(config.getDataSources(1).isEmpty());
    assertTrue(config.getDataSources(2).isEmpty());
    assertEquals(config.getThreshold(), 0.85);
    assertEquals(config.getMaybeThreshold(), 0.7);
    assertEquals(3, config.getProperties().size());

    Property prop = config.getPropertyByName("ID");
    assertTrue("ID property lost", prop.isIdProperty());

    prop = config.getPropertyByName("NAME");
    assertEquals(lev.getClass(), prop.getComparator().getClass());
    assertEquals(0.3, prop.getLowProbability());
    assertEquals(0.8, prop.getHighProbability());

    prop = config.getPropertyByName("EMAIL");
    assertEquals(lev.getClass(), prop.getComparator().getClass());
    assertEquals(0.3, prop.getLowProbability());
    assertEquals(0.8, prop.getHighProbability());
  }
View Full Code Here

TOP

Related Classes of no.priv.garshol.duke.ConfigurationImpl$HighComparator

Copyright © 2018 www.massapicom. All rights reserved.
All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.