Project

General

Profile

1
package eu.dnetlib.data.mapreduce.dedup;
2

    
3
import java.util.LinkedList;
4
import java.util.List;
5
import java.util.Map;
6
import java.util.PriorityQueue;
7
import java.util.Queue;
8
import java.util.UUID;
9

    
10
import org.apache.commons.io.IOUtils;
11
import org.junit.Before;
12
import org.junit.Test;
13
import org.springframework.core.io.ClassPathResource;
14

    
15
import com.google.common.collect.Lists;
16
import com.google.common.collect.Maps;
17

    
18
import eu.dnetlib.pace.clustering.NGramUtils;
19
import eu.dnetlib.pace.config.Type;
20
import eu.dnetlib.pace.model.Field;
21
import eu.dnetlib.pace.model.FieldListImpl;
22
import eu.dnetlib.pace.model.FieldValueImpl;
23
import eu.dnetlib.pace.model.MapDocument;
24
import eu.dnetlib.pace.model.MapDocumentComparator;
25

    
26
public class TitleOrderingTest {
27

    
28
	private List<MapDocument> results = Lists.newArrayList();
29

    
30
	@Before
31
	public void setUp() throws Exception {
32

    
33
		final List<String> lines = IOUtils.readLines(new ClassPathResource("eu/dnetlib/data/mapreduce/dedup/titles.txt").getInputStream());
34
		for (final String title : lines) {
35
			final Map<String, Field> fieldMap = Maps.newHashMap();
36
			final FieldListImpl list = new FieldListImpl();
37
			list.add(new FieldValueImpl(Type.String, "title", title));
38
			fieldMap.put("title", list);
39
			results.add(new MapDocument("id-" + UUID.randomUUID(), fieldMap));
40
		}
41
	}
42

    
43
	@Test
44
	public void test() {
45

    
46
		final Queue<MapDocument> queue = new PriorityQueue<MapDocument>(100, new MapDocumentComparator("title"));
47

    
48
		queue.addAll(results);
49

    
50
		final Queue<MapDocument> queue2 = simplifyQueue(queue);
51

    
52
		while (!queue2.isEmpty()) {
53
			final MapDocument doc = queue2.remove();
54
			System.out.println(doc.values("title").stringValue());
55
		}
56
	}
57

    
58
	private Queue<MapDocument> simplifyQueue(final Queue<MapDocument> queue) {
59
		final Queue<MapDocument> q = new LinkedList<MapDocument>();
60

    
61
		String fieldRef = "";
62
		final List<MapDocument> tempResults = Lists.newArrayList();
63

    
64
		while (!queue.isEmpty()) {
65
			final MapDocument result = queue.remove();
66

    
67
			if (!result.values("title").isEmpty()) {
68
				final String field = NGramUtils.cleanupForOrdering(result.values("title").stringValue());
69
				if (field.equals(fieldRef)) {
70
					tempResults.add(result);
71
				} else {
72
					if (tempResults.size() < 5) {
73
						q.addAll(tempResults);
74
					} else {
75
						System.out.println("Skipped field: " + fieldRef + " - size: " + tempResults.size());
76
					}
77
					tempResults.clear();
78
					tempResults.add(result);
79
					fieldRef = field;
80
				}
81
			}
82
		}
83
		if (tempResults.size() < 5) {
84
			q.addAll(tempResults);
85
		} else {
86
			System.out.println("Skipped field: " + fieldRef + " - size: " + tempResults.size());
87
		}
88

    
89
		return q;
90
	}
91

    
92
}
(2-2/2)