|
1
|
package eu.dnetlib.data.mdstore.plugins;
|
|
2
|
|
|
3
|
import java.io.StringReader;
|
|
4
|
import java.util.Map;
|
|
5
|
|
|
6
|
import org.apache.commons.logging.Log;
|
|
7
|
import org.apache.commons.logging.LogFactory;
|
|
8
|
import org.dom4j.Document;
|
|
9
|
import org.dom4j.DocumentException;
|
|
10
|
import org.dom4j.io.SAXReader;
|
|
11
|
|
|
12
|
import com.mongodb.BasicDBObject;
|
|
13
|
import com.mongodb.DBObject;
|
|
14
|
import com.mongodb.client.MongoCollection;
|
|
15
|
|
|
16
|
import eu.dnetlib.data.mdstore.modular.mongodb.MongoMDStore;
|
|
17
|
import eu.dnetlib.rmi.data.MDStoreServiceException;
|
|
18
|
|
|
19
|
public class FilterMdRecordsPlugin extends AbstractIstiMDStorePlugin {
|
|
20
|
|
|
21
|
private static final Log log = LogFactory.getLog(FilterMdRecordsPlugin.class);
|
|
22
|
|
|
23
|
@Override
|
|
24
|
public final void process(final MongoMDStore store, final Map<String, String> params) throws MDStoreServiceException {
|
|
25
|
|
|
26
|
final String xpath = params.get("xpath");
|
|
27
|
|
|
28
|
final MongoCollection<DBObject> collPubs = store.getCollection();
|
|
29
|
|
|
30
|
long valid = 0;
|
|
31
|
long skipped = 0;
|
|
32
|
|
|
33
|
final SAXReader reader = new SAXReader();
|
|
34
|
for (final DBObject obj : collPubs.find()) {
|
|
35
|
try {
|
|
36
|
final String recordId = obj.get("id").toString();
|
|
37
|
final Document doc = reader.read(new StringReader(obj.get("body").toString()));
|
|
38
|
if (doc.selectNodes(xpath).isEmpty()) {
|
|
39
|
collPubs.deleteOne(new BasicDBObject("id", recordId));
|
|
40
|
skipped++;
|
|
41
|
} else {
|
|
42
|
valid++;
|
|
43
|
}
|
|
44
|
} catch (final DocumentException e) {
|
|
45
|
log.warn("Problem parsing a mdstore record", e);
|
|
46
|
}
|
|
47
|
}
|
|
48
|
|
|
49
|
log.info("***** Filtering records *****");
|
|
50
|
log.info("* xpath: " + xpath);
|
|
51
|
log.info("* valid: " + valid);
|
|
52
|
log.info("* skipped: " + skipped);
|
|
53
|
log.info("*****************************");
|
|
54
|
|
|
55
|
touch(store);
|
|
56
|
}
|
|
57
|
|
|
58
|
}
|