1 |
35869
|
claudio.at
|
package eu.dnetlib.msro.workflows.hadoop.utils;
|
2 |
35797
|
claudio.at
|
|
3 |
|
|
import java.util.Collection;
|
4 |
|
|
import java.util.Set;
|
5 |
|
|
|
6 |
|
|
import com.google.common.base.Function;
|
7 |
|
|
import com.google.common.base.Predicate;
|
8 |
|
|
import com.google.common.collect.Iterables;
|
9 |
|
|
import com.google.common.collect.Lists;
|
10 |
|
|
import com.google.common.collect.Sets;
|
11 |
|
|
|
12 |
|
|
import eu.dnetlib.data.proto.DatasourceOrganizationProtos.DatasourceOrganization.Provision;
|
13 |
|
|
import eu.dnetlib.data.proto.DedupProtos.Dedup;
|
14 |
|
|
import eu.dnetlib.data.proto.DedupSimilarityProtos.DedupSimilarity;
|
15 |
|
|
import eu.dnetlib.data.proto.PersonPersonProtos.PersonPerson.CoAuthorship;
|
16 |
|
|
import eu.dnetlib.data.proto.PersonResultProtos.PersonResult.Authorship;
|
17 |
|
|
import eu.dnetlib.data.proto.ProjectOrganizationProtos.ProjectOrganization.Participation;
|
18 |
|
|
import eu.dnetlib.data.proto.ProjectPersonProtos.ProjectPerson.ContactPerson;
|
19 |
|
|
import eu.dnetlib.data.proto.RelTypeProtos.RelType;
|
20 |
|
|
import eu.dnetlib.data.proto.RelTypeProtos.SubRelType;
|
21 |
|
|
import eu.dnetlib.data.proto.ResultProjectProtos.ResultProject.Outcome;
|
22 |
|
|
import eu.dnetlib.data.proto.ResultResultProtos.ResultResult.PublicationDataset;
|
23 |
|
|
import eu.dnetlib.data.proto.ResultResultProtos.ResultResult.Similarity;
|
24 |
|
|
import eu.dnetlib.data.proto.TypeProtos.Type;
|
25 |
|
|
|
26 |
|
|
/**
|
27 |
|
|
* Common static utility methods to manage the hbase tables
|
28 |
|
|
*
|
29 |
|
|
* @author claudio
|
30 |
|
|
*
|
31 |
|
|
*/
|
32 |
|
|
public class HBaseTableUtils {
|
33 |
|
|
|
34 |
|
|
private static final String _ = "_";
|
35 |
|
|
private static Function<Type, String> typeName = new Function<Type, String>() {
|
36 |
|
|
|
37 |
|
|
@Override
|
38 |
|
|
public String apply(final Type type) {
|
39 |
|
|
return type.toString();
|
40 |
|
|
}
|
41 |
|
|
};
|
42 |
|
|
|
43 |
|
|
public static Set<String> listAllColumns() {
|
44 |
|
|
final Set<String> union = Sets.union(listEntities(), listRelationships());
|
45 |
|
|
return Sets.union(union, listDedupRelationships());
|
46 |
|
|
}
|
47 |
|
|
|
48 |
|
|
public static Set<String> listDedupColumns(final Collection<Type> entityTypes) {
|
49 |
|
|
final Set<String> entities = listEntities(Lists.newArrayList(Iterables.transform(entityTypes, typeName)));
|
50 |
|
|
return Sets.union(entities, listDedupRelationships());
|
51 |
|
|
}
|
52 |
|
|
|
53 |
|
|
private static Set<String> listDedupRelationships() {
|
54 |
|
|
final Set<String> cfs = Sets.newHashSet();
|
55 |
|
|
cfs.add(RelType.organizationOrganization + _ + SubRelType.dedup + _ + Dedup.RelName.merges);
|
56 |
|
|
cfs.add(RelType.organizationOrganization + _ + SubRelType.dedup + _ + Dedup.RelName.isMergedIn);
|
57 |
|
|
cfs.add(RelType.organizationOrganization + _ + SubRelType.dedupSimilarity + _ + DedupSimilarity.RelName.isSimilarTo);
|
58 |
|
|
|
59 |
|
|
cfs.add(RelType.personPerson + _ + SubRelType.dedup + _ + Dedup.RelName.merges);
|
60 |
|
|
cfs.add(RelType.personPerson + _ + SubRelType.dedup + _ + Dedup.RelName.isMergedIn);
|
61 |
|
|
cfs.add(RelType.personPerson + _ + SubRelType.dedupSimilarity + _ + DedupSimilarity.RelName.isSimilarTo);
|
62 |
|
|
|
63 |
|
|
cfs.add(RelType.resultResult + _ + SubRelType.dedup + _ + Dedup.RelName.merges);
|
64 |
|
|
cfs.add(RelType.resultResult + _ + SubRelType.dedup + _ + Dedup.RelName.isMergedIn);
|
65 |
|
|
cfs.add(RelType.resultResult + _ + SubRelType.dedupSimilarity + _ + DedupSimilarity.RelName.isSimilarTo);
|
66 |
|
|
|
67 |
|
|
return cfs;
|
68 |
|
|
}
|
69 |
|
|
|
70 |
|
|
private static Set<String> listEntities(final Collection<String> entityType) {
|
71 |
|
|
return Sets.newHashSet(Iterables.filter(Iterables.transform(Lists.newArrayList(Type.values()), typeName), new Predicate<String>() {
|
72 |
|
|
|
73 |
|
|
@Override
|
74 |
|
|
public boolean apply(final String s) {
|
75 |
|
|
return entityType.contains(s);
|
76 |
|
|
}
|
77 |
|
|
}));
|
78 |
|
|
}
|
79 |
|
|
|
80 |
|
|
public static Set<String> listEntities() {
|
81 |
|
|
return Sets.newHashSet(Iterables.transform(Lists.newArrayList(Type.values()), typeName));
|
82 |
|
|
}
|
83 |
|
|
|
84 |
|
|
public static Set<String> listRelationships() {
|
85 |
|
|
final Set<String> cfs = Sets.newHashSet();
|
86 |
|
|
cfs.add(RelType.datasourceOrganization + _ + SubRelType.provision + _ + Provision.RelName.isProvidedBy);
|
87 |
|
|
cfs.add(RelType.datasourceOrganization + _ + SubRelType.provision + _ + Provision.RelName.provides);
|
88 |
|
|
|
89 |
|
|
cfs.add(RelType.personPerson + _ + SubRelType.coauthorship + _ + CoAuthorship.RelName.isCoauthorOf);
|
90 |
|
|
|
91 |
|
|
cfs.add(RelType.personResult + _ + SubRelType.authorship + _ + Authorship.RelName.isAuthorOf);
|
92 |
|
|
cfs.add(RelType.personResult + _ + SubRelType.authorship + _ + Authorship.RelName.hasAuthor);
|
93 |
|
|
|
94 |
|
|
cfs.add(RelType.projectOrganization + _ + SubRelType.participation + _ + Participation.RelName.hasParticipant);
|
95 |
|
|
cfs.add(RelType.projectOrganization + _ + SubRelType.participation + _ + Participation.RelName.isParticipant);
|
96 |
|
|
|
97 |
|
|
cfs.add(RelType.projectPerson + _ + SubRelType.contactPerson + _ + ContactPerson.RelName.isContact);
|
98 |
|
|
cfs.add(RelType.projectPerson + _ + SubRelType.contactPerson + _ + ContactPerson.RelName.hasContact);
|
99 |
|
|
|
100 |
|
|
cfs.add(RelType.resultProject + _ + SubRelType.outcome + _ + Outcome.RelName.isProducedBy);
|
101 |
|
|
cfs.add(RelType.resultProject + _ + SubRelType.outcome + _ + Outcome.RelName.produces);
|
102 |
|
|
|
103 |
|
|
cfs.add(RelType.resultResult + _ + SubRelType.similarity + _ + Similarity.RelName.hasAmongTopNSimilarDocuments);
|
104 |
|
|
cfs.add(RelType.resultResult + _ + SubRelType.similarity + _ + Similarity.RelName.isAmongTopNSimilarDocuments);
|
105 |
|
|
|
106 |
|
|
cfs.add(RelType.resultResult + _ + SubRelType.publicationDataset + _ + PublicationDataset.RelName.isRelatedTo);
|
107 |
|
|
|
108 |
|
|
return cfs;
|
109 |
|
|
}
|
110 |
|
|
|
111 |
|
|
}
|