| 1 |
49291
|
michele.ar
|
package eu.dnetlib.data.mdstore.plugins;
|
| 2 |
|
|
|
| 3 |
|
|
import java.net.URI;
|
| 4 |
|
|
import java.net.URISyntaxException;
|
| 5 |
|
|
|
| 6 |
|
|
import org.apache.commons.logging.Log;
|
| 7 |
|
|
import org.apache.commons.logging.LogFactory;
|
| 8 |
60864
|
michele.ar
|
import org.bson.Document;
|
| 9 |
49291
|
michele.ar
|
|
| 10 |
60864
|
michele.ar
|
import com.mongodb.BasicDBObject;
|
| 11 |
|
|
import com.mongodb.client.MongoCollection;
|
| 12 |
|
|
|
| 13 |
51020
|
michele.ar
|
import eu.dnetlib.data.mdstore.plugins.objects.MdRecord;
|
| 14 |
51852
|
michele.ar
|
import eu.dnetlib.data.utils.HttpFetcher;
|
| 15 |
49291
|
michele.ar
|
|
| 16 |
51020
|
michele.ar
|
public abstract class GenericDoiMdstorePlugin extends MdRecordPlugin {
|
| 17 |
49291
|
michele.ar
|
|
| 18 |
|
|
private static final Log log = LogFactory.getLog(GenericDoiMdstorePlugin.class);
|
| 19 |
|
|
|
| 20 |
|
|
@Override
|
| 21 |
60388
|
michele.ar
|
protected final boolean updateRecord(final String recordId, final MdRecord record) {
|
| 22 |
51020
|
michele.ar
|
for (final String doi : record.getDois()) {
|
| 23 |
|
|
log.debug(" Record " + record.getId() + " has doi " + doi);
|
| 24 |
60864
|
michele.ar
|
final String response = obtainData(doi);
|
| 25 |
51020
|
michele.ar
|
if ((response != null) && updateDocument(record, response)) { return true; }
|
| 26 |
49291
|
michele.ar
|
}
|
| 27 |
51020
|
michele.ar
|
return false;
|
| 28 |
49291
|
michele.ar
|
}
|
| 29 |
|
|
|
| 30 |
51020
|
michele.ar
|
abstract protected boolean updateDocument(MdRecord doc, String response);
|
| 31 |
49291
|
michele.ar
|
|
| 32 |
|
|
abstract protected URI prepareURI(String doi) throws URISyntaxException;
|
| 33 |
|
|
|
| 34 |
60864
|
michele.ar
|
abstract protected MongoCollection<Document> getCacheCollection();
|
| 35 |
|
|
|
| 36 |
|
|
private String obtainData(final String doi) {
|
| 37 |
|
|
final MongoCollection<Document> cacheColl = getCacheCollection();
|
| 38 |
|
|
|
| 39 |
|
|
final Document row = cacheColl.find(new BasicDBObject("doi", doi)).first();
|
| 40 |
|
|
if (row != null) {
|
| 41 |
|
|
return row.get("data") != null ? row.get("data").toString() : null;
|
| 42 |
|
|
} else {
|
| 43 |
|
|
final String res = download(doi);
|
| 44 |
|
|
|
| 45 |
|
|
final Document doc = new Document();
|
| 46 |
|
|
doc.append("doi", doi);
|
| 47 |
|
|
doc.append("data", res);
|
| 48 |
|
|
cacheColl.insertOne(doc);
|
| 49 |
|
|
|
| 50 |
|
|
return res;
|
| 51 |
|
|
}
|
| 52 |
|
|
}
|
| 53 |
|
|
|
| 54 |
49291
|
michele.ar
|
private String download(final String doi) {
|
| 55 |
|
|
try {
|
| 56 |
51852
|
michele.ar
|
return HttpFetcher.fetch(prepareURI(doi));
|
| 57 |
49291
|
michele.ar
|
} catch (final URISyntaxException e) {
|
| 58 |
|
|
log.error("Error resolving doi: " + doi, e);
|
| 59 |
|
|
return null;
|
| 60 |
|
|
}
|
| 61 |
|
|
}
|
| 62 |
|
|
|
| 63 |
|
|
}
|