Project

General

Profile

« Previous | Next » 

Revision 43514

Added by Claudio Atzori over 7 years ago

created dataset dedicated entities

View differences:

HBaseTableUtils.java
4 4 
import java.util.Set;
5 5 

  		




  6
  6
  
    
import com.google.common.base.Function;


  		




  7
  
  
    
import com.google.common.base.Predicate;


  		




  8
  7
  
    
import com.google.common.collect.Iterables;


  		




  9
  8
  
    
import com.google.common.collect.Lists;


  		




  10
  9
  
    
import com.google.common.collect.Sets;


  		




  11
  
  
    
import eu.dnetlib.data.proto.DatasourceOrganizationProtos.DatasourceOrganization.Provision;


  		




  12
  
  
    
import eu.dnetlib.data.proto.DedupProtos.Dedup;


  		




  13
  
  
    
import eu.dnetlib.data.proto.DedupSimilarityProtos.DedupSimilarity;


  		




  14
  
  
    
import eu.dnetlib.data.proto.PersonPersonProtos.PersonPerson.CoAuthorship;


  		




  15
  
  
    
import eu.dnetlib.data.proto.PersonResultProtos.PersonResult.Authorship;


  		




  16
  
  
    
import eu.dnetlib.data.proto.ProjectOrganizationProtos.ProjectOrganization.Participation;


  		




  17
  
  
    
import eu.dnetlib.data.proto.ProjectPersonProtos.ProjectPerson.ContactPerson;


  		




  18
  10
  
    
import eu.dnetlib.data.proto.RelTypeProtos.RelType;


  		




  19
  11
  
    
import eu.dnetlib.data.proto.RelTypeProtos.SubRelType;


  		




  20
  
  
    
import eu.dnetlib.data.proto.ResultOrganizationProtos.ResultOrganization.Affiliation;


  		




  21
  
  
    
import eu.dnetlib.data.proto.ResultProjectProtos.ResultProject.Outcome;


  		




  22
  
  
    
import eu.dnetlib.data.proto.ResultResultProtos.ResultResult.Part;


  		




  23
  
  
    
import eu.dnetlib.data.proto.ResultResultProtos.ResultResult.PublicationDataset;


  		




  24
  
  
    
import eu.dnetlib.data.proto.ResultResultProtos.ResultResult.Similarity;


  		




  25
  
  
    
import eu.dnetlib.data.proto.ResultResultProtos.ResultResult.Supplement;


  		




  
  12
  
    
import eu.dnetlib.data.proto.SubRelProtos.*;


  		




  26
  13
  
    
import eu.dnetlib.data.proto.TypeProtos.Type;


  		




  27
  14
  
    

  		




  28
  15
  
    
/**


  		




  ......




  34
  21
  
    

  		




  35
  22
  
    
	private static final String SEPARATOR = "_";


  		




  36
  23
  
    

  		




  37
  
  
    
	private static final Function<Type, String> typeName = new Function<Type, String>() {


  		




  
  24
  
    
	private static final Function<Type, String> typeName = type -> type.toString();


  		




  38
  25
  
    

  		




  39
  
  
    
		@Override


  		




  40
  
  
    
		public String apply(final Type type) {


  		




  41
  
  
    
			return type.toString();


  		




  42
  
  
    
		}


  		




  43
  
  
    
	};


  		




  44
  
  
    

  		




  45
  26
  
    
	public static Set<String> listAllColumns() {


  		




  46
  27
  
    
		final Set<String> union = Sets.union(listEntities(), listRelationships());


  		




  47
  28
  
    
		return Sets.union(union, listDedupRelationships());


  		




  ......




  62
  43
  
    
		cfs.add(RelType.personPerson + SEPARATOR + SubRelType.dedup + SEPARATOR + Dedup.RelName.isMergedIn);


  		




  63
  44
  
    
		cfs.add(RelType.personPerson + SEPARATOR + SubRelType.dedupSimilarity + SEPARATOR + DedupSimilarity.RelName.isSimilarTo);


  		




  64
  45
  
    

  		




  65
  
  
    
		cfs.add(RelType.resultResult + SEPARATOR + SubRelType.dedup + SEPARATOR + Dedup.RelName.merges);


  		




  66
  
  
    
		cfs.add(RelType.resultResult + SEPARATOR + SubRelType.dedup + SEPARATOR + Dedup.RelName.isMergedIn);


  		




  67
  
  
    
		cfs.add(RelType.resultResult + SEPARATOR + SubRelType.dedupSimilarity + SEPARATOR + DedupSimilarity.RelName.isSimilarTo);


  		




  
  46
  
    
		cfs.add(RelType.publicationPublication + SEPARATOR + SubRelType.dedup + SEPARATOR + Dedup.RelName.merges);


  		




  
  47
  
    
		cfs.add(RelType.publicationPublication + SEPARATOR + SubRelType.dedup + SEPARATOR + Dedup.RelName.isMergedIn);


  		




  
  48
  
    
		cfs.add(RelType.publicationPublication + SEPARATOR + SubRelType.dedupSimilarity + SEPARATOR + DedupSimilarity.RelName.isSimilarTo);


  		




  68
  49
  
    

  		




  
  50
  
    
		cfs.add(RelType.publicationDataset + SEPARATOR + SubRelType.dedupSimilarity + SEPARATOR + DedupSimilarity.RelName.isSimilarTo);


  		




  
  51
  
    
		cfs.add(RelType.publicationDataset + SEPARATOR + SubRelType.isRelatedTo + SEPARATOR + IsRelatedTo.RelName.isRelatedTo);


  		




  
  52
  
    

  		




  
  53
  
    
		cfs.add(RelType.datasetDataset + SEPARATOR + SubRelType.dedup + SEPARATOR + Dedup.RelName.merges);


  		




  
  54
  
    
		cfs.add(RelType.datasetDataset + SEPARATOR + SubRelType.dedup + SEPARATOR + Dedup.RelName.isMergedIn);


  		




  
  55
  
    
		cfs.add(RelType.datasetDataset + SEPARATOR + SubRelType.dedupSimilarity + SEPARATOR + DedupSimilarity.RelName.isSimilarTo);


  		




  
  56
  
    

  		




  69
  57
  
    
		return cfs;


  		




  70
  58
  
    
	}


  		




  71
  59
  
    

  		




  72
  60
  
    
	private static Set<String> listEntities(final Collection<String> entityType) {


  		




  73
  
  
    
		return Sets.newHashSet(Iterables.filter(Iterables.transform(Lists.newArrayList(Type.values()), typeName), new Predicate<String>() {


  		




  74
  
  
    

  		




  75
  
  
    
			@Override


  		




  76
  
  
    
			public boolean apply(final String s) {


  		




  77
  
  
    
				return entityType.contains(s);


  		




  78
  
  
    
			}


  		




  
  61
  
    
		return Sets.newHashSet(Iterables.filter(Iterables.transform(Lists.newArrayList(Type.values()), typeName), s -> {


  		




  
  62
  
    
			return entityType.contains(s);


  		




  79
  63
  
    
		}));


  		




  80
  64
  
    
	}


  		




  81
  65
  
    

  		




  ......




  90
  74
  
    

  		




  91
  75
  
    
		cfs.add(RelType.personPerson + SEPARATOR + SubRelType.coauthorship + SEPARATOR + CoAuthorship.RelName.isCoauthorOf);


  		




  92
  76
  
    

  		




  93
  
  
    
		cfs.add(RelType.personResult + SEPARATOR + SubRelType.authorship + SEPARATOR + Authorship.RelName.isAuthorOf);


  		




  94
  
  
    
		cfs.add(RelType.personResult + SEPARATOR + SubRelType.authorship + SEPARATOR + Authorship.RelName.hasAuthor);


  		




  
  77
  
    
		cfs.add(RelType.personPublication + SEPARATOR + SubRelType.authorship + SEPARATOR + Authorship.RelName.isAuthorOf);


  		




  
  78
  
    
		cfs.add(RelType.personPublication + SEPARATOR + SubRelType.authorship + SEPARATOR + Authorship.RelName.hasAuthor);


  		




  95
  79
  
    

  		




  
  80
  
    
		cfs.add(RelType.personDataset + SEPARATOR + SubRelType.authorship + SEPARATOR + Authorship.RelName.isAuthorOf);


  		




  
  81
  
    
		cfs.add(RelType.personDataset + SEPARATOR + SubRelType.authorship + SEPARATOR + Authorship.RelName.hasAuthor);


  		




  
  82
  
    

  		




  96
  83
  
    
		cfs.add(RelType.projectOrganization + SEPARATOR + SubRelType.participation + SEPARATOR + Participation.RelName.hasParticipant);


  		




  97
  84
  
    
		cfs.add(RelType.projectOrganization + SEPARATOR + SubRelType.participation + SEPARATOR + Participation.RelName.isParticipant);


  		




  98
  85
  
    

  		




  99
  86
  
    
		cfs.add(RelType.projectPerson + SEPARATOR + SubRelType.contactPerson + SEPARATOR + ContactPerson.RelName.isContact);


  		




  100
  87
  
    
		cfs.add(RelType.projectPerson + SEPARATOR + SubRelType.contactPerson + SEPARATOR + ContactPerson.RelName.hasContact);


  		




  101
  88
  
    

  		




  102
  
  
    
		cfs.add(RelType.resultProject + SEPARATOR + SubRelType.outcome + SEPARATOR + Outcome.RelName.isProducedBy);


  		




  103
  
  
    
		cfs.add(RelType.resultProject + SEPARATOR + SubRelType.outcome + SEPARATOR + Outcome.RelName.produces);


  		




  
  89
  
    
		cfs.add(RelType.publicationProject + SEPARATOR + SubRelType.outcome + SEPARATOR + Outcome.RelName.isProducedBy);


  		




  
  90
  
    
		cfs.add(RelType.publicationProject + SEPARATOR + SubRelType.outcome + SEPARATOR + Outcome.RelName.produces);


  		




  104
  91
  
    

  		




  105
  
  
    
		cfs.add(RelType.resultResult + SEPARATOR + SubRelType.similarity + SEPARATOR + Similarity.RelName.hasAmongTopNSimilarDocuments);


  		




  106
  
  
    
		cfs.add(RelType.resultResult + SEPARATOR + SubRelType.similarity + SEPARATOR + Similarity.RelName.isAmongTopNSimilarDocuments);


  		




  
  92
  
    
		cfs.add(RelType.datasetProject + SEPARATOR + SubRelType.outcome + SEPARATOR + Outcome.RelName.isProducedBy);


  		




  
  93
  
    
		cfs.add(RelType.datasetProject + SEPARATOR + SubRelType.outcome + SEPARATOR + Outcome.RelName.produces);


  		




  107
  94
  
    

  		




  108
  
  
    
		cfs.add(RelType.resultResult + SEPARATOR + SubRelType.supplement + SEPARATOR + Supplement.RelName.isSupplementedBy);


  		




  109
  
  
    
		cfs.add(RelType.resultResult + SEPARATOR + SubRelType.supplement + SEPARATOR + Supplement.RelName.isSupplementTo);


  		




  110
  
  
    
		cfs.add(RelType.resultResult + SEPARATOR + SubRelType.part + SEPARATOR + Part.RelName.isPartOf);


  		




  111
  
  
    
		cfs.add(RelType.resultResult + SEPARATOR + SubRelType.part + SEPARATOR + Part.RelName.hasPart);


  		




  
  95
  
    
		cfs.add(RelType.publicationPublication + SEPARATOR + SubRelType.similarity + SEPARATOR + Similarity.RelName.hasAmongTopNSimilarDocuments);


  		




  
  96
  
    
		cfs.add(RelType.publicationPublication + SEPARATOR + SubRelType.similarity + SEPARATOR + Similarity.RelName.isAmongTopNSimilarDocuments);


  		




  112
  97
  
    

  		




  113
  
  
    
		cfs.add(RelType.resultResult + SEPARATOR + SubRelType.publicationDataset + SEPARATOR + PublicationDataset.RelName.isRelatedTo);


  		




  
  98
  
    
		cfs.add(RelType.publicationPublication + SEPARATOR + SubRelType.supplement + SEPARATOR + Supplement.RelName.isSupplementedBy);


  		




  
  99
  
    
		cfs.add(RelType.publicationPublication + SEPARATOR + SubRelType.supplement + SEPARATOR + Supplement.RelName.isSupplementTo);


  		




  
  100
  
    
		cfs.add(RelType.publicationPublication + SEPARATOR + SubRelType.part + SEPARATOR + Part.RelName.isPartOf);


  		




  
  101
  
    
		cfs.add(RelType.publicationPublication + SEPARATOR + SubRelType.part + SEPARATOR + Part.RelName.hasPart);


  		




  114
  102
  
    

  		




  115
  
  
    
		cfs.add(RelType.resultOrganization + SEPARATOR + SubRelType.affiliation + SEPARATOR + Affiliation.RelName.isAffiliatedWith);


  		




  
  103
  
    
		cfs.add(RelType.publicationDataset + SEPARATOR + SubRelType.isRelatedTo + SEPARATOR + IsRelatedTo.RelName.isRelatedTo);


  		




  116
  104
  
    

  		




  
  105
  
    
		cfs.add(RelType.datasetDataset + SEPARATOR + SubRelType.similarity + SEPARATOR + Similarity.RelName.hasAmongTopNSimilarDocuments);


  		




  
  106
  
    
		cfs.add(RelType.datasetDataset + SEPARATOR + SubRelType.similarity + SEPARATOR + Similarity.RelName.isAmongTopNSimilarDocuments);


  		




  
  107
  
    

  		




  
  108
  
    
		cfs.add(RelType.publicationOrganization + SEPARATOR + SubRelType.affiliation + SEPARATOR + Affiliation.RelName.isAffiliatedWith);


  		




  
  109
  
    

  		




  117
  110
  
    
		return cfs;


  		




  118
  111
  
    
	}


  		




  119
  112
  
    

  		












Also available in:  Unified diff