|
1
|
package eu.dnetlib.data.mdstore.utils;
|
|
2
|
|
|
3
|
import java.io.ByteArrayInputStream;
|
|
4
|
import java.util.Date;
|
|
5
|
import java.util.Stack;
|
|
6
|
import java.util.function.Function;
|
|
7
|
import javax.xml.stream.XMLInputFactory;
|
|
8
|
import javax.xml.stream.XMLStreamConstants;
|
|
9
|
import javax.xml.stream.XMLStreamException;
|
|
10
|
import javax.xml.stream.XMLStreamReader;
|
|
11
|
|
|
12
|
import eu.dnetlib.data.mdstore.model.MDStoreRecord;
|
|
13
|
import org.apache.commons.lang3.StringUtils;
|
|
14
|
|
|
15
|
/**
|
|
16
|
* Created by claudio on 20/03/2017.
|
|
17
|
*/
|
|
18
|
public class RecordParser implements Function<String, MDStoreRecord> {
|
|
19
|
|
|
20
|
public static final String OBJ_IDENTIFIER = "objIdentifier";
|
|
21
|
public static final String IDENTIFIER = "identifier";
|
|
22
|
|
|
23
|
private Long timestamp;
|
|
24
|
|
|
25
|
private XMLInputFactory factory;
|
|
26
|
|
|
27
|
public RecordParser() {
|
|
28
|
factory = XMLInputFactory.newInstance();
|
|
29
|
timestamp = null;
|
|
30
|
}
|
|
31
|
|
|
32
|
public RecordParser(final long timestamp) {
|
|
33
|
factory = XMLInputFactory.newInstance();
|
|
34
|
this.timestamp = timestamp;
|
|
35
|
}
|
|
36
|
|
|
37
|
@Override
|
|
38
|
public MDStoreRecord apply(final String s) {
|
|
39
|
try {
|
|
40
|
final XMLStreamReader parser = factory.createXMLStreamReader(new ByteArrayInputStream(s.getBytes()));
|
|
41
|
final Stack<String> elementStack = new Stack<>();
|
|
42
|
elementStack.push("/");
|
|
43
|
|
|
44
|
final MDStoreRecord res = MDStoreRecord.create()
|
|
45
|
.setBody(s)
|
|
46
|
.setTimestamp(timestamp != null ? timestamp : new Date().getTime());
|
|
47
|
|
|
48
|
//System.out.println(res.getTimestamp());
|
|
49
|
|
|
50
|
while (parser.hasNext()) {
|
|
51
|
int event = parser.next();
|
|
52
|
|
|
53
|
if (event == XMLStreamConstants.END_ELEMENT) {
|
|
54
|
elementStack.pop();
|
|
55
|
} else if (event == XMLStreamConstants.START_ELEMENT) {
|
|
56
|
final String localName = parser.getLocalName();
|
|
57
|
elementStack.push(localName);
|
|
58
|
|
|
59
|
if (OBJ_IDENTIFIER.equals(localName)) {
|
|
60
|
parser.next();
|
|
61
|
res.setId(parser.getText().trim());
|
|
62
|
|
|
63
|
} else if (IDENTIFIER.equals(localName)) {
|
|
64
|
if (StringUtils.isBlank(res.getOriginalId())) {
|
|
65
|
parser.next();
|
|
66
|
res.setOriginalId(parser.getText().trim());
|
|
67
|
}
|
|
68
|
}
|
|
69
|
if (StringUtils.isNotBlank(res.getId()) && StringUtils.isNotBlank(res.getOriginalId())) {
|
|
70
|
return res;
|
|
71
|
}
|
|
72
|
}
|
|
73
|
}
|
|
74
|
return res;
|
|
75
|
} catch (XMLStreamException e) {
|
|
76
|
throw new IllegalStateException(e);
|
|
77
|
}
|
|
78
|
}
|
|
79
|
}
|