Project

General

Profile

« Previous | Next » 

Revision 57270

Added by Enrico Ottonello about 5 years ago

added firstname and surname in case of authors without orcid id, extracted from fullname, needed for propagation

View differences:

modules/dnet-mapreduce-jobs/trunk/src/test/java/eu/dnetlib/data/mapreduce/actions/OrcidToActionsTest.java
28 28 
import eu.dnetlib.data.mapreduce.hbase.dataimport.DumpToActionsUtility;
29 29 
import eu.dnetlib.data.mapreduce.hbase.dataimport.OrcidToActions;
30 30 
import eu.dnetlib.data.proto.OafProtos.Oaf;
31 
import eu.dnetlib.miscutils.collections.Pair;
32 
import eu.dnetlib.pace.model.Person;
31 33 

  		




  32
  34
  
    
public class OrcidToActionsTest {


  		




  33
  35
  
    

  		




  ......




  36
  38
  
    
	private Reporter reporter;


  		




  37
  39
  
    

  		




  38
  40
  
    
	private final String generatedJson =


  		




  39
  
  
    
			"{\"kind\": \"entity\",\"entity\": {\"type\": \"result\",\"result\": {\"metadata\": {\"title\": [{\"value\": \"Factors influencing accuracy of referral and the likelihood of false positive referral by optometrists in Bradford, United Kingdom.\",\"qualifier\": {\"classid\": \"main title\",\"classname\": \"main title\",\"schemeid\": \"dnet:dataCite_title\",\"schemename\": \"dnet:dataCite_title\"}}],\"relevantdate\": [{\"value\": \"2015-11-01\",\"qualifier\": {\"classid\": \"issued\",\"classname\": \"issued\",\"schemeid\": \"dnet:dataCite_date\",\"schemename\": \"dnet:dataCite_date\"}}],\"dateofacceptance\": {\"value\": \"2015-11-01\"},\"resulttype\": {\"classid\": \"publication\",\"classname\": \"publication\",\"schemeid\": \"dnet:result_typologies\",\"schemename\": \"dnet:result_typologies\"},\"resourcetype\": {\"classid\": \"journal-article\",\"classname\": \"journal-article\",\"schemeid\": \"dnet:dataCite_resource\",\"schemename\": \"dnet:dataCite_resource\"},\"source\": [{\"value\": \"Europe PubMed Central\"}],\"author\": [{\"fullname\": \"Christopher Davey\",\"name\": \"Christopher\",\"surname\": \"Davey\",\"rank\": 1,\"pid\": [{\"key\": \"ORCID\",\"value\": \"0000-0002-9050-379X\"}]},{\"fullname\": \"Scally AJ\",\"rank\": 2},{\"fullname\": \"Green C\",\"rank\": 3},{\"fullname\": \"Mitchell ES\",\"rank\": 4},{\"fullname\": \"Elliott DB\",\"rank\": 5}]},\"instance\": [{\"accessright\": {\"classid\": \"UNKNOWN\",\"classname\": \"UNKNOWN\",\"schemeid\": \"dnet:access_modes\",\"schemename\": \"dnet:access_modes\"},\"instancetype\": {\"classid\": \"0001\",\"classname\": \"Article\",\"schemeid\": \"dnet:publication_resource\",\"schemename\": \"dnet:publication_resource\"},\"hostedby\": {\"key\": \"10|openaire____::55045bd2a65019fd8e6741a755395c8c\",\"value\": \"Unknown Repository\"},\"url\": [\"http://europepmc.org/abstract/med/26614021\"],\"collectedfrom\": {\"key\": \"10|openaire____::806360c771262b4d6770e7cdf04b5c5a\",\"value\": \"ORCID\"},\"dateofacceptance\": {\"value\": \"2015-11-01\"}}]},\"originalId\": [\"24099977\"],\"collectedfrom\": [{\"key\": \"10|openaire____::806360c771262b4d6770e7cdf04b5c5a\",\"value\": \"ORCID\"}],\"pid\": [{\"value\": \"26614021\",\"qualifier\": {\"classid\": \"pmid\",\"classname\": \"pmid\",\"schemeid\": \"dnet:pid_types\",\"schemename\": \"dnet:pid_types\"}}],\"dateofcollection\": \"2018-10-22\",\"id\": \"50|orcid____::57b8b813748049402c0900d3d17f1c23\",\"dateoftransformation\": \"2019-06-21T11:16:10+02:00\"},\"dataInfo\": {\"inferred\": false,\"deletedbyinference\": false,\"trust\": \"0.9\",\"provenanceaction\": {\"classid\": \"sysimport:actionset:orcidworks-no-doi\",\"classname\": \"sysimport:actionset:orcidworks-no-doi\",\"schemeid\": \"dnet:provenanceActions\",\"schemename\": \"dnet:provenanceActions\"}},\"lastupdatetimestamp\": 1561108570991}\n"


  		




  40
  
  
    
					+


  		




  41
  
  
    
					"";


  		




  
  41
  
    
			"{\"kind\": \"entity\",\"entity\": {\"type\": \"result\",\"result\": {\"metadata\": {\"title\": [{\"value\": \"Cryptogenic fibrosing alveolitis in Bergen Hospital district\",\"qualifier\": {\"classid\": \"main title\",\"classname\": \"main title\",\"schemeid\": \"dnet:dataCite_title\",\"schemename\": \"dnet:dataCite_title\"}}],\"relevantdate\": [{\"value\": \"1999-01-01\",\"qualifier\": {\"classid\": \"issued\",\"classname\": \"issued\",\"schemeid\": \"dnet:dataCite_date\",\"schemename\": \"dnet:dataCite_date\"}}],\"dateofacceptance\": {\"value\": \"1999-01-01\"},\"resulttype\": {\"classid\": \"publication\",\"classname\": \"publication\",\"schemeid\": \"dnet:result_typologies\",\"schemename\": \"dnet:result_typologies\"},\"resourcetype\": {\"classid\": \"conference-abstract\",\"classname\": \"conference-abstract\",\"schemeid\": \"dnet:dataCite_resource\",\"schemename\": \"dnet:dataCite_resource\"},\"source\": [{\"value\": \"University of Southern Denmark\"}],\"author\": [{\"fullname\": \"Christian von Plessen\",\"name\": \"Christian\",\"surname\": \"von Plessen\",\"rank\": 1,\"pid\": [{\"key\": \"ORCID\",\"value\": \"0000-0002-6134-6780\"}]},{\"fullname\": \"Grinde, Ø.\",\"name\": \"Ø.\",\"surname\": \"Grinde\",\"rank\": 2},{\"fullname\": \"Gulsvik, A.\",\"name\": \"A.\",\"surname\": \"Gulsvik\",\"rank\": 3}]},\"instance\": [{\"accessright\": {\"classid\": \"UNKNOWN\",\"classname\": \"UNKNOWN\",\"schemeid\": \"dnet:access_modes\",\"schemename\": \"dnet:access_modes\"},\"instancetype\": {\"classid\": \"0004\",\"classname\": \"Conference object\",\"schemeid\": \"dnet:publication_resource\",\"schemename\": \"dnet:publication_resource\"},\"hostedby\": {\"key\": \"10|openaire____::55045bd2a65019fd8e6741a755395c8c\",\"value\": \"Unknown Repository\"},\"url\": [\"http://findresearcher.sdu.dk/portal/en/publications/cryptogenic-fibrosing-alveolitis-in-bergen-hospital-district(14a8eea6-ddf6-4055-9e7e-a82525ae36c2).html\"],\"collectedfrom\": {\"key\": \"10|openaire____::806360c771262b4d6770e7cdf04b5c5a\",\"value\": \"ORCID\"},\"dateofacceptance\": {\"value\": \"1999-01-01\"}}]},\"originalId\": [\"29721246\"],\"collectedfrom\": [{\"key\": \"10|openaire____::806360c771262b4d6770e7cdf04b5c5a\",\"value\": \"ORCID\"}],\"pid\": [{\"value\": \"14a8eea6-ddf6-4055-9e7e-a82525ae36c2\",\"qualifier\": {\"classid\": \"orcidworkid\",\"classname\": \"orcidworkid\",\"schemeid\": \"dnet:pid_types\",\"schemename\": \"dnet:pid_types\"}}],\"dateofcollection\": \"2018-10-22\",\"id\": \"50|orcid_______::01d661c1f7824215e5150d59b30f7cf7\",\"dateoftransformation\": \"2019-10-04T12:14:24+02:00\"},\"dataInfo\": {\"inferred\": false,\"deletedbyinference\": false,\"trust\": \"0.9\",\"provenanceaction\": {\"classid\": \"sysimport:actionset:orcidworks-no-doi\",\"classname\": \"sysimport:actionset:orcidworks-no-doi\",\"schemeid\": \"dnet:provenanceActions\",\"schemename\": \"dnet:provenanceActions\"}},\"lastupdatetimestamp\": 1570184064208}\n" + 


  		




  
  42
  
    
			"";


  		




  42
  43
  
    

  		




  43
  44
  
    
	@Before


  		




  44
  45
  
    
	public void setup() {


  		




  ......




  122
  123
  
    
			i += 1;


  		




  123
  124
  
    
		}


  		




  124
  125
  
    
	}


  		




  125
  
  
    

  		




  
  126
  
    
	


  		




  126
  127
  
    
	@Test


  		




  127
  128
  
    
	public void testGeneratedJson() throws ParseException {


  		




  128
  129
  
    
		final JsonParser parser = new JsonParser();


  		




  ......




  132
  133
  
    
		final List<JsonObject> pids = DumpToActionsUtility.getArrayObjects(entity, "pid");


  		




  133
  134
  
    
		pids.forEach(pid -> {


  		




  134
  135
  
    
			final JsonObject qualifier = pid.getAsJsonObject("qualifier");


  		




  135
  
  
    
			assertTrue(qualifier.get("classid").getAsString().matches("(ark|arxiv|pmc|pmid|source-work-id|urn)"));


  		




  
  136
  
    
			assertTrue(qualifier.get("classid").getAsString().matches("(ark|arxiv|pmc|pmid|orcidworkid|urn)"));


  		




  136
  137
  
    
		});


  		




  137
  138
  
    
		assertEquals(getStringValue(entity, "type"), "result");


  		




  138
  139
  
    
		assertNotNull(getStringValue(entity, "id"));


  		




  ......




  154
  155
  
    
		});


  		




  155
  156
  
    

  		




  156
  157
  
    
	}


  		




  
  158
  
    
	


  		




  157
  159
  
    
}


  		












  

    
      modules/dnet-mapreduce-jobs/trunk/src/main/java/eu/dnetlib/data/mapreduce/hbase/dataimport/OrcidToActions.java
    
  





  20
  20
  
    
import com.google.gson.JsonArray;


  		




  21
  21
  
    
import com.google.gson.JsonElement;


  		




  22
  22
  
    
import com.google.gson.JsonObject;


  		




  
  23
  
    
import com.googlecode.protobuf.format.JsonFormat;


  		




  23
  24
  
    

  		




  24
  25
  
    
import eu.dnetlib.actionmanager.actions.ActionFactory;


  		




  25
  26
  
    
import eu.dnetlib.actionmanager.actions.AtomicAction;


  		




  ......




  40
  41
  
    
import eu.dnetlib.data.transform.xml.AbstractDNetXsltFunctions;


  		




  41
  42
  
    
import eu.dnetlib.miscutils.collections.Pair;


  		




  42
  43
  
    
import eu.dnetlib.miscutils.datetime.DateUtils;


  		




  
  44
  
    
import eu.dnetlib.pace.model.Person;


  		




  43
  45
  
    

  		




  44
  46
  
    
public class OrcidToActions {


  		




  45
  47
  
    

  		




  ......




  248
  250
  
    

  		




  249
  251
  
    
		actionList.add(factory.createAtomicAction(setName, agent, oaf.getEntity().getId(), "result", "body", oaf.build().toByteArray()));


  		




  250
  252
  
    

  		




  251
  
  
    
		// System.out.println(JsonFormat.printToString(oaf.build()));


  		




  
  253
  
    
//		 System.out.println(JsonFormat.printToString(oaf.build()));


  		




  252
  254
  
    
		return actionList;


  		




  253
  255
  
    

  		




  254
  256
  
    
	}


  		




  ......




  300
  302
  
    
							result.setSurname(surname);


  		




  301
  303
  
    
						}


  		




  302
  304
  
    
					} else {


  		




  
  305
  
    
						String fullname = "";


  		




  303
  306
  
    
						if (StringUtils.isNotBlank(name)) {


  		




  304
  
  
    
							result.setFullname(name);


  		




  
  307
  
    
							fullname = name;


  		




  305
  308
  
    
						} else {


  		




  306
  309
  
    
							if (StringUtils.isNotBlank(surname)) {


  		




  307
  
  
    
								result.setFullname(surname);


  		




  
  310
  
    
								fullname = surname;


  		




  308
  311
  
    
							}


  		




  309
  312
  
    
						}


  		




  
  313
  
    
						Person p = new Person(fullname, false);


  		




  
  314
  
    
						if (p.isAccurate()) {


  		




  
  315
  
    
							result.setName(p.getNormalisedFirstName());


  		




  
  316
  
    
							result.setSurname(p.getNormalisedSurname());


  		




  
  317
  
    
							result.setFullname(p.getNormalisedFullname());


  		




  
  318
  
    
						}


  		




  
  319
  
    
						else {


  		




  
  320
  
    
							result.setFullname(fullname);


  		




  
  321
  
    
						}


  		




  310
  322
  
    
					}


  		




  311
  323
  
    
				}


  		




  312
  324
  
    
				result.setRank(rank);


  		












Also available in:  Unified diff