1
|
package eu.dnetlib.msro.workflows.nodes.objectstore;
|
2
|
|
3
|
import java.io.File;
|
4
|
import java.util.regex.Matcher;
|
5
|
import java.util.regex.Pattern;
|
6
|
|
7
|
import eu.dnetlib.miscutils.functional.xml.DnetXsltFunctions;
|
8
|
import eu.dnetlib.rmi.data.ObjectStoreFile;
|
9
|
import eu.dnetlib.rmi.data.Protocols;
|
10
|
|
11
|
public class ArxivIteratorRepository extends IteratorRepository {
|
12
|
|
13
|
private final Pattern pattern;
|
14
|
|
15
|
public ArxivIteratorRepository(final Iterable<File> input, final String repositoryPrefix, final String oaiPrefix) {
|
16
|
super(input, repositoryPrefix, oaiPrefix);
|
17
|
this.pattern = Pattern.compile("[0-9]");
|
18
|
|
19
|
}
|
20
|
|
21
|
@Override
|
22
|
String generateNextElement() {
|
23
|
try {
|
24
|
final String inputname = this.input.next().getCanonicalPath();
|
25
|
final String[] values = inputname.split("/");
|
26
|
if (values.length > 1) {
|
27
|
final ObjectStoreFile info = new ObjectStoreFile();
|
28
|
String fileName = values[values.length - 1].replace(".pdf", "");
|
29
|
final Matcher matcher = this.pattern.matcher(fileName);
|
30
|
if (matcher.find()) {
|
31
|
if (matcher.start() != 0) {
|
32
|
fileName = fileName.substring(0, matcher.start()) + "/" + fileName.substring(matcher.start());
|
33
|
}
|
34
|
}
|
35
|
info.setDownloadedURL("http://arxiv.org/abs/" + fileName);
|
36
|
info.setAccessProtocol(Protocols.None);
|
37
|
// String value = "oai:arXiv.org:" + fileName;
|
38
|
final String value = this.oaiPrefix + fileName;
|
39
|
final String resultID = this.repositoryPrefix + "::" + DnetXsltFunctions.md5(value);
|
40
|
info.setObjectID(resultID + "::" + DnetXsltFunctions.md5(info.getDownloadedURL()));
|
41
|
info.setMimeType("pdf");
|
42
|
info.setURI(inputname);
|
43
|
return info.toJSON();
|
44
|
}
|
45
|
} catch (final Exception e) {
|
46
|
return null;
|
47
|
}
|
48
|
return null;
|
49
|
}
|
50
|
|
51
|
}
|