Revision 51020
Added by Michele Artini about 6 years ago
| EnrichLabsPlugin.java | ||
|---|---|---|
| 1 | 1 |
package eu.dnetlib.data.mdstore.plugins; |
| 2 | 2 |
|
| 3 |
import java.io.StringReader;
|
|
| 3 |
import java.util.HashMap;
|
|
| 4 | 4 |
import java.util.Map; |
| 5 | 5 |
import java.util.Set; |
| 6 | 6 |
import java.util.regex.Matcher; |
| 7 | 7 |
import java.util.regex.Pattern; |
| 8 | 8 |
import java.util.stream.Collectors; |
| 9 | 9 |
|
| 10 |
import org.apache.commons.lang3.math.NumberUtils; |
|
| 11 | 10 |
import org.apache.commons.logging.Log; |
| 12 | 11 |
import org.apache.commons.logging.LogFactory; |
| 13 |
import org.dom4j.Document; |
|
| 14 |
import org.dom4j.DocumentException; |
|
| 15 |
import org.dom4j.Element; |
|
| 16 |
import org.dom4j.Node; |
|
| 17 |
import org.dom4j.QName; |
|
| 18 |
import org.dom4j.io.SAXReader; |
|
| 19 | 12 |
import org.springframework.beans.factory.annotation.Autowired; |
| 20 | 13 |
|
| 21 | 14 |
import com.google.common.collect.Sets; |
| 22 |
import com.mongodb.BasicDBObject; |
|
| 23 |
import com.mongodb.DBObject; |
|
| 24 |
import com.mongodb.client.MongoCollection; |
|
| 25 | 15 |
|
| 26 |
import eu.dnetlib.data.db.Affiliation; |
|
| 27 | 16 |
import eu.dnetlib.data.db.AffiliationsDao; |
| 28 | 17 |
import eu.dnetlib.data.db.Group; |
| 29 | 18 |
import eu.dnetlib.data.db.Person; |
| 30 |
import eu.dnetlib.data.mdstore.modular.mongodb.MongoMDStore; |
|
| 31 |
import eu.dnetlib.rmi.data.MDStoreServiceException; |
|
| 32 |
import eu.dnetlib.xml.ISTIUtilityFunction; |
|
| 19 |
import eu.dnetlib.data.mdstore.plugins.objects.Affiliation; |
|
| 20 |
import eu.dnetlib.data.mdstore.plugins.objects.CnrPerson; |
|
| 21 |
import eu.dnetlib.data.mdstore.plugins.objects.MdRecord; |
|
| 22 |
import eu.dnetlib.data.utils.IstiConstants; |
|
| 23 |
import eu.dnetlib.data.utils.XsltFunctions; |
|
| 33 | 24 |
|
| 34 |
public class EnrichLabsPlugin extends AbstractIstiMDStorePlugin {
|
|
| 25 |
public class EnrichLabsPlugin extends MdRecordPlugin {
|
|
| 35 | 26 |
|
| 36 | 27 |
private static final Log log = LogFactory.getLog(EnrichLabsPlugin.class); |
| 37 | 28 |
|
| ... | ... | |
| 40 | 31 |
@Autowired |
| 41 | 32 |
private AffiliationsDao dao; |
| 42 | 33 |
|
| 34 |
private Map<String, Person> affiliations = new HashMap<>(); |
|
| 35 |
|
|
| 43 | 36 |
@Override |
| 44 |
public void process(final MongoMDStore store, final Map<String, String> params) throws MDStoreServiceException {
|
|
| 45 |
log.warn("********************************************");
|
|
| 46 |
log.warn("* ENRICH LABS");
|
|
| 47 |
log.warn("********************************************");
|
|
| 48 |
|
|
| 49 |
// CODE -> year -> Lab |
|
| 50 |
final Map<String, Person> affiliations = dao.listPersonsWithAffiliations() |
|
| 37 |
protected void reconfigure() {
|
|
| 38 |
affiliations = dao.listPersonsWithAffiliations() |
|
| 51 | 39 |
.stream() |
| 52 | 40 |
.collect(Collectors.toMap(Person::getId, a -> a)); |
| 41 |
} |
|
| 53 | 42 |
|
| 54 |
final MongoCollection<DBObject> pubsCollection = store.getCollection(); |
|
| 43 |
@Override |
|
| 44 |
protected boolean updateRecord(final MdRecord doc) {
|
|
| 45 |
final int year = doc.getDate(); |
|
| 55 | 46 |
|
| 56 |
for (final DBObject obj : pubsCollection.find()) {
|
|
| 57 |
final String id = obj.get("id").toString();
|
|
| 58 |
log.debug("**********************************************");
|
|
| 59 |
log.debug("Record " + id);
|
|
| 47 |
for (final CnrPerson cp : doc.getCnrPersons()) {
|
|
| 48 |
cp.getAffiliations().clear(); |
|
| 60 | 49 |
|
| 61 |
try {
|
|
| 62 |
final Document doc = (new SAXReader()).read(new StringReader(obj.get("body").toString()));
|
|
| 50 |
if (year >= IstiConstants.PIMPA_START_YEAR) {
|
|
| 51 |
final String infoId = cp.getId(); |
|
| 52 |
final Pattern pattern = Pattern.compile("info:cnr-pdr\\/author\\/(.+)\\/(.+)\\/(.+)");
|
|
| 53 |
final Matcher matcher = pattern.matcher(infoId); |
|
| 54 |
if (matcher.find()) {
|
|
| 55 |
final String code = matcher.group(1); |
|
| 56 |
final String fullname = XsltFunctions.capitalize(String.format("%s %s", matcher.group(3), matcher.group(2)));
|
|
| 63 | 57 |
|
| 64 |
final int year = NumberUtils.toInt(doc.valueOf("//*[local-name()='date' and @dateType='Accepted']"), 0);
|
|
| 58 |
final Person p = affiliations.get(code);
|
|
| 65 | 59 |
|
| 66 |
if (year >= Affiliation.START_YEAR) {
|
|
| 67 |
resetAffiliations(doc); |
|
| 68 |
|
|
| 69 |
for (final Object o : doc.selectNodes("//*[local-name()='person']")) {
|
|
| 70 |
final Element node = (Element) o; |
|
| 71 |
final String infoId = node.valueOf("./*[local-name()='infoId']");
|
|
| 72 |
final Pattern pattern = Pattern.compile("info:cnr-pdr\\/author\\/(.+)\\/(.+)\\/(.+)");
|
|
| 73 |
final Matcher matcher = pattern.matcher(infoId); |
|
| 74 |
if (matcher.find()) {
|
|
| 75 |
final String code = matcher.group(1); |
|
| 76 |
final String fullname = ISTIUtilityFunction.capitalize(String.format("%s %s", matcher.group(3), matcher.group(2)));
|
|
| 77 |
|
|
| 78 |
final Person p = affiliations.get(code); |
|
| 79 |
|
|
| 80 |
if (p != null) {
|
|
| 81 |
final Set<Group> groups = p.getAffiliations().get(year); |
|
| 82 |
if (groups != null) {
|
|
| 83 |
node.selectSingleNode("./*[local-name() = 'personName']").setText(p.getName());
|
|
| 84 |
for (final Group g : groups) {
|
|
| 85 |
if (!g.getId().equals(UNKNOWN_GROUP.getId())) {
|
|
| 86 |
final Element affNode = node.addElement(new QName("affiliation", node.getNamespace()));
|
|
| 87 |
affNode.addAttribute("code", code);
|
|
| 88 |
affNode.addAttribute("groupAcronym", g.getId());
|
|
| 89 |
affNode.addAttribute("group", g.getName());
|
|
| 90 |
affNode.addAttribute("type", g.getType());
|
|
| 91 |
} |
|
| 92 |
} |
|
| 93 |
} else {
|
|
| 94 |
log.info("Affiliation not found, infoId: " + infoId + ", year: " + year);
|
|
| 95 |
dao.registerAffiliation(code, UNKNOWN_GROUP.getId(), year); |
|
| 96 |
p.getAffiliations().put(year, Sets.newHashSet(UNKNOWN_GROUP)); |
|
| 60 |
if (p != null) {
|
|
| 61 |
final Set<Group> groups = p.getAffiliations().get(year); |
|
| 62 |
if (groups != null) {
|
|
| 63 |
cp.setName(p.getName()); |
|
| 64 |
for (final Group g : groups) {
|
|
| 65 |
if (!g.getId().equals(UNKNOWN_GROUP.getId())) {
|
|
| 66 |
cp.getAffiliations().add(new Affiliation(code, g.getType(), g.getId(), g.getName())); |
|
| 97 | 67 |
} |
| 98 |
} else {
|
|
| 99 |
log.info("Person not found, infoId: " + infoId + ", year: " + year);
|
|
| 100 |
dao.registerPerson(code, fullname); |
|
| 101 |
dao.registerAffiliation(code, UNKNOWN_GROUP.getId(), year); |
|
| 102 |
|
|
| 103 |
final Person np = new Person(code, fullname); |
|
| 104 |
np.getAffiliations().put(year, Sets.newHashSet(UNKNOWN_GROUP)); |
|
| 105 |
|
|
| 106 |
affiliations.put(code, np); |
|
| 107 | 68 |
} |
| 108 | 69 |
} else {
|
| 109 |
log.warn("Invalid infoId: " + infoId);
|
|
| 70 |
log.info("Affiliation not found, infoId: " + infoId + ", year: " + year);
|
|
| 71 |
dao.registerAffiliation(code, UNKNOWN_GROUP.getId(), year); |
|
| 72 |
p.getAffiliations().put(year, Sets.newHashSet(UNKNOWN_GROUP)); |
|
| 110 | 73 |
} |
| 74 |
} else {
|
|
| 75 |
log.info("Person not found, infoId: " + infoId + ", year: " + year);
|
|
| 76 |
dao.registerPerson(code, fullname); |
|
| 77 |
dao.registerAffiliation(code, UNKNOWN_GROUP.getId(), year); |
|
| 78 |
|
|
| 79 |
final Person np = new Person(code, fullname); |
|
| 80 |
np.getAffiliations().put(year, Sets.newHashSet(UNKNOWN_GROUP)); |
|
| 81 |
|
|
| 82 |
affiliations.put(code, np); |
|
| 111 | 83 |
} |
| 112 |
|
|
| 113 |
pubsCollection.updateOne(new BasicDBObject("id", id), new BasicDBObject("$set", new BasicDBObject("body", doc.asXML())));
|
|
| 84 |
} else {
|
|
| 85 |
log.warn("Invalid infoId: " + infoId);
|
|
| 114 | 86 |
} |
| 115 |
} catch (final DocumentException e) {
|
|
| 116 |
log.warn("Problem parsing a mdstore record");
|
|
| 117 | 87 |
} |
| 118 |
} |
|
| 119 |
} |
|
| 120 | 88 |
|
| 121 |
private void resetAffiliations(final Document doc) {
|
|
| 122 |
for (final Object n : doc.selectNodes("//*[local-name() = 'person']/*[local-name() = 'affiliation']")) {
|
|
| 123 |
((Node) n).detach(); |
|
| 124 | 89 |
} |
| 90 |
return true; |
|
| 125 | 91 |
} |
| 126 | 92 |
|
| 127 | 93 |
} |
Also available in: Unified diff
refactoring using jaxb