|
1
|
package eu.dnetlib.dli.resolver;
|
|
2
|
|
|
3
|
import com.google.common.collect.Lists;
|
|
4
|
import eu.dnetlib.dli.resolver.model.CompletionStatus;
|
|
5
|
import eu.dnetlib.dli.resolver.model.DLIObjectProvenance;
|
|
6
|
import eu.dnetlib.dli.resolver.model.DLIResolvedObject;
|
|
7
|
|
|
8
|
import eu.dnetlib.pid.resolver.model.ObjectType;
|
|
9
|
import org.apache.commons.logging.Log;
|
|
10
|
import org.apache.commons.logging.LogFactory;
|
|
11
|
|
|
12
|
import javax.xml.stream.XMLInputFactory;
|
|
13
|
import javax.xml.stream.XMLStreamConstants;
|
|
14
|
import javax.xml.stream.XMLStreamException;
|
|
15
|
import javax.xml.stream.XMLStreamReader;
|
|
16
|
import java.io.ByteArrayInputStream;
|
|
17
|
import java.util.Stack;
|
|
18
|
|
|
19
|
/**
|
|
20
|
* @author sandro
|
|
21
|
*/
|
|
22
|
public class RCSBParser {
|
|
23
|
|
|
24
|
/**
|
|
25
|
* The Constant CROSSREF_NS_PREFIX.
|
|
26
|
*/
|
|
27
|
public static final String RCBS_NS_PREFIX = "r3d100010327";
|
|
28
|
/**
|
|
29
|
* The Constant log.
|
|
30
|
*/
|
|
31
|
private static final Log log = LogFactory.getLog(RCSBParser.class); // NOPMD by marko on 11/24/08 5:02 PM
|
|
32
|
|
|
33
|
/**
|
|
34
|
* Parses the record.
|
|
35
|
*
|
|
36
|
* @param record the record
|
|
37
|
* @return the DLI object
|
|
38
|
*/
|
|
39
|
public DLIResolvedObject parseRecord(final String record) {
|
|
40
|
if (record == null) return null;
|
|
41
|
try {
|
|
42
|
XMLInputFactory factory = XMLInputFactory.newInstance();
|
|
43
|
final ByteArrayInputStream stream = new ByteArrayInputStream(record.getBytes());
|
|
44
|
XMLStreamReader parser = factory.createXMLStreamReader(stream);
|
|
45
|
DLIResolvedObject object = new DLIResolvedObject();
|
|
46
|
Stack<String> elementStack = new Stack<>();
|
|
47
|
elementStack.push("/");
|
|
48
|
while (parser.hasNext()) {
|
|
49
|
int event = parser.next();
|
|
50
|
if (event == XMLStreamConstants.END_ELEMENT) {
|
|
51
|
|
|
52
|
elementStack.pop();
|
|
53
|
} else if (event == XMLStreamConstants.START_ELEMENT) {
|
|
54
|
final String localName = parser.getLocalName();
|
|
55
|
elementStack.push(localName);
|
|
56
|
if ("dimStructure.structureId".equals(localName)) {
|
|
57
|
log.debug("found open tag structureId");
|
|
58
|
parser.next();
|
|
59
|
if (parser.hasText()) {
|
|
60
|
object.setPid(parser.getText().trim());
|
|
61
|
object.setPidType("pdb");
|
|
62
|
}
|
|
63
|
} else if ("dimStructure.structureTitle".equals(localName)) {
|
|
64
|
log.debug("found open tag structureTitle");
|
|
65
|
parser.next();
|
|
66
|
if (parser.hasText()) {
|
|
67
|
String title = parser.getText().trim();
|
|
68
|
object.setTitles(Lists.newArrayList(title));
|
|
69
|
}
|
|
70
|
} else if ("dimStructure.structureAuthor".equals(localName)) {
|
|
71
|
log.debug("found open tag structureAuthor");
|
|
72
|
parser.next();
|
|
73
|
if (parser.hasText()) {
|
|
74
|
String author = parser.getText().trim();
|
|
75
|
String[] splittedAuthrs = author.split("#");
|
|
76
|
if (splittedAuthrs != null) {
|
|
77
|
object.setAuthors(Lists.newArrayList(splittedAuthrs));
|
|
78
|
}
|
|
79
|
}
|
|
80
|
} else if ("dimStructure.releaseDate".equals(localName)) {
|
|
81
|
log.debug("found open tag releaseDate");
|
|
82
|
parser.next();
|
|
83
|
if (parser.hasText()) {
|
|
84
|
String date = parser.getText().trim();
|
|
85
|
object.setDate(date);
|
|
86
|
}
|
|
87
|
}
|
|
88
|
|
|
89
|
}
|
|
90
|
}
|
|
91
|
if (object.getPid() != null && object.getPidType() != null) {
|
|
92
|
final DLIObjectProvenance provenance = new DLIObjectProvenance();
|
|
93
|
DLIPIDResolver.setDatasourceProvenance(provenance, RCBS_NS_PREFIX);
|
|
94
|
object.setDatasourceProvenance(Lists.newArrayList(provenance));
|
|
95
|
object.fixContribution(provenance);
|
|
96
|
object.setType(ObjectType.dataset);
|
|
97
|
object.setCompletionStatus(CompletionStatus.complete.toString());
|
|
98
|
return object;
|
|
99
|
}
|
|
100
|
return null;
|
|
101
|
|
|
102
|
} catch (XMLStreamException e) {
|
|
103
|
log.error("error during parsing record " + record, e);
|
|
104
|
return null;
|
|
105
|
}
|
|
106
|
|
|
107
|
}
|
|
108
|
|
|
109
|
}
|