| 1 |
42181
|
sandro.lab
|
package eu.dnetlib.oai.parser;
|
| 2 |
|
|
|
| 3 |
|
|
import java.io.StringReader;
|
| 4 |
|
|
import java.util.List;
|
| 5 |
|
|
import java.util.Map.Entry;
|
| 6 |
|
|
|
| 7 |
45455
|
alessia.ba
|
import com.google.common.collect.ArrayListMultimap;
|
| 8 |
|
|
import com.google.common.collect.Iterables;
|
| 9 |
|
|
import com.google.common.collect.Multimap;
|
| 10 |
|
|
import eu.dnetlib.oai.PublisherField;
|
| 11 |
42181
|
sandro.lab
|
import org.apache.commons.logging.Log;
|
| 12 |
|
|
import org.apache.commons.logging.LogFactory;
|
| 13 |
|
|
import org.dom4j.Document;
|
| 14 |
|
|
import org.dom4j.DocumentException;
|
| 15 |
|
|
import org.dom4j.Node;
|
| 16 |
|
|
import org.dom4j.io.SAXReader;
|
| 17 |
|
|
|
| 18 |
|
|
/**
|
| 19 |
|
|
* An instance of this class can parse an XML record and extract the information needed to store the record in a publisher store.
|
| 20 |
|
|
*
|
| 21 |
|
|
* @author alessia
|
| 22 |
|
|
*/
|
| 23 |
|
|
public class PublisherRecordParser {
|
| 24 |
|
|
|
| 25 |
|
|
private static final Log log = LogFactory.getLog(PublisherRecordParser.class); // NOPMD by marko on 11/24/08 5:02 PM
|
| 26 |
45455
|
alessia.ba
|
|
| 27 |
42181
|
sandro.lab
|
/**
|
| 28 |
|
|
* List of the indices of the target store.
|
| 29 |
|
|
*/
|
| 30 |
|
|
private List<PublisherField> storeIndices;
|
| 31 |
|
|
|
| 32 |
45455
|
alessia.ba
|
private final SAXReader saxReader = new SAXReader();
|
| 33 |
42181
|
sandro.lab
|
|
| 34 |
|
|
/**
|
| 35 |
|
|
* Parses the record and returns a map where a key is the name of an index, the value is the value in the record at the xpath specificed
|
| 36 |
|
|
* in this.storeIndices.
|
| 37 |
|
|
*
|
| 38 |
42184
|
michele.ar
|
* @param record
|
| 39 |
|
|
* the XML string to parse.
|
| 40 |
42181
|
sandro.lab
|
* @return a Multimap describing the values to be indexed for this record.
|
| 41 |
|
|
*/
|
| 42 |
|
|
@SuppressWarnings({ "unchecked", "rawtypes" })
|
| 43 |
|
|
public Multimap<String, String> parseRecord(final String record) {
|
| 44 |
|
|
Multimap<String, String> recordProps = ArrayListMultimap.create();
|
| 45 |
|
|
try {
|
| 46 |
45455
|
alessia.ba
|
Document doc = this.saxReader.read(new StringReader(record));
|
| 47 |
|
|
for (PublisherField field : this.storeIndices) {
|
| 48 |
|
|
for (Entry<String, String> indexEntry : field.getSources().entries()) {
|
| 49 |
42181
|
sandro.lab
|
// each xpath can return a list of nodes or strings, depending on the xpath
|
| 50 |
45455
|
alessia.ba
|
List xPathResult = doc.selectNodes(indexEntry.getValue());
|
| 51 |
42181
|
sandro.lab
|
if ((xPathResult != null) && !xPathResult.isEmpty()) {
|
| 52 |
|
|
if (containsStrings(xPathResult)) {
|
| 53 |
|
|
recordProps.putAll(field.getFieldName(), xPathResult);
|
| 54 |
|
|
} else {
|
| 55 |
|
|
if (containsNodes(xPathResult)) {
|
| 56 |
45455
|
alessia.ba
|
recordProps.putAll(field.getFieldName(), Iterables.transform(xPathResult, obj -> {
|
| 57 |
|
|
if (obj == null) return "";
|
| 58 |
|
|
Node node = (Node) obj;
|
| 59 |
|
|
return node.getText();
|
| 60 |
42181
|
sandro.lab
|
}));
|
| 61 |
|
|
}
|
| 62 |
|
|
}
|
| 63 |
|
|
}
|
| 64 |
|
|
}
|
| 65 |
|
|
}
|
| 66 |
|
|
|
| 67 |
45455
|
alessia.ba
|
} catch (DocumentException e) {
|
| 68 |
42181
|
sandro.lab
|
log.fatal("Can't parse record");
|
| 69 |
|
|
recordProps = null;
|
| 70 |
|
|
}
|
| 71 |
|
|
return recordProps;
|
| 72 |
|
|
|
| 73 |
|
|
}
|
| 74 |
|
|
|
| 75 |
|
|
@SuppressWarnings("rawtypes")
|
| 76 |
|
|
private boolean containsStrings(final List objects) {
|
| 77 |
45455
|
alessia.ba
|
Object first = objects.get(0);
|
| 78 |
42181
|
sandro.lab
|
return first instanceof String;
|
| 79 |
|
|
}
|
| 80 |
|
|
|
| 81 |
|
|
@SuppressWarnings("rawtypes")
|
| 82 |
|
|
private boolean containsNodes(final List objects) {
|
| 83 |
45455
|
alessia.ba
|
Object first = objects.get(0);
|
| 84 |
42181
|
sandro.lab
|
return first instanceof Node;
|
| 85 |
|
|
}
|
| 86 |
|
|
|
| 87 |
|
|
public List<PublisherField> getStoreIndices() {
|
| 88 |
45455
|
alessia.ba
|
return storeIndices;
|
| 89 |
42181
|
sandro.lab
|
}
|
| 90 |
|
|
|
| 91 |
|
|
public void setStoreIndices(final List<PublisherField> storeIndices) {
|
| 92 |
|
|
this.storeIndices = storeIndices;
|
| 93 |
|
|
}
|
| 94 |
|
|
|
| 95 |
|
|
public SAXReader getSaxReader() {
|
| 96 |
45455
|
alessia.ba
|
return saxReader;
|
| 97 |
42181
|
sandro.lab
|
}
|
| 98 |
|
|
|
| 99 |
45455
|
alessia.ba
|
public PublisherRecordParser(final List<PublisherField> storeIndices) {
|
| 100 |
|
|
super();
|
| 101 |
|
|
this.storeIndices = storeIndices;
|
| 102 |
|
|
}
|
| 103 |
|
|
|
| 104 |
|
|
public PublisherRecordParser() {
|
| 105 |
|
|
super();
|
| 106 |
|
|
// TODO Auto-generated constructor stub
|
| 107 |
|
|
}
|
| 108 |
|
|
|
| 109 |
|
|
|
| 110 |
42181
|
sandro.lab
|
}
|