|
1
|
package eu.dnetlib.data.mapreduce.hbase.dedup.experiment;
|
|
2
|
|
|
3
|
import java.util.List;
|
|
4
|
|
|
5
|
import com.google.common.base.Splitter;
|
|
6
|
import org.apache.commons.lang.StringUtils;
|
|
7
|
import org.dom4j.Element;
|
|
8
|
|
|
9
|
/**
|
|
10
|
* Created by claudio on 25/03/16.
|
|
11
|
*/
|
|
12
|
public class SubjectParser {
|
|
13
|
|
|
14
|
public static final String REGEX_SUBJECT = "^(info:eu-repo)\\/(classification)\\/([a-zA-Z]*)\\/(.*)$";
|
|
15
|
private static final int MIN_LENGTH = 5;
|
|
16
|
|
|
17
|
public SubjectsMap parse(final org.dom4j.Document doc) {
|
|
18
|
|
|
19
|
final List subjectNodes = doc.selectNodes("//*[local-name() = 'subject']");
|
|
20
|
final SubjectsMap subjectMap = new SubjectsMap();
|
|
21
|
|
|
22
|
for(int i = 0; i<subjectNodes.size(); i++) {
|
|
23
|
final Element e = (Element) subjectNodes.get(i);
|
|
24
|
final String subject = e.getText();
|
|
25
|
|
|
26
|
final String type = guessType(subject);
|
|
27
|
if (!subjectMap.containsKey(type)) {
|
|
28
|
subjectMap.put(type, new Subjects());
|
|
29
|
}
|
|
30
|
|
|
31
|
if (StringUtils.isNotBlank(type)) {
|
|
32
|
if ("keyword".equals(type)) {
|
|
33
|
final Splitter splitter = Splitter.on(",").trimResults().omitEmptyStrings();
|
|
34
|
for (String token : splitter.split(subject)) {
|
|
35
|
final String value = token.replaceAll("[^a-zA-Z ]", "").toLowerCase();
|
|
36
|
if (value.length() >= MIN_LENGTH) {
|
|
37
|
subjectMap.get(type).add(value);
|
|
38
|
}
|
|
39
|
}
|
|
40
|
} else {
|
|
41
|
String token = subject.replaceFirst(REGEX_SUBJECT, "$4");
|
|
42
|
|
|
43
|
if (StringUtils.isNotBlank(token)) {
|
|
44
|
final String value = token.replaceAll("[^a-zA-Z ]", "").toLowerCase();
|
|
45
|
if (value.length() >= MIN_LENGTH) {
|
|
46
|
subjectMap.get(type).add(value);
|
|
47
|
}
|
|
48
|
}
|
|
49
|
}
|
|
50
|
}
|
|
51
|
}
|
|
52
|
|
|
53
|
return subjectMap;
|
|
54
|
}
|
|
55
|
|
|
56
|
private String guessType(final String subject) {
|
|
57
|
if (subject.startsWith("info:eu-repo")) {
|
|
58
|
final String s = subject.replaceAll(REGEX_SUBJECT, "$3");
|
|
59
|
return s;
|
|
60
|
} else {
|
|
61
|
return "keyword";
|
|
62
|
}
|
|
63
|
}
|
|
64
|
}
|