1
|
package eu.dnetlib.grid.process.utils;
|
2
|
|
3
|
import java.io.FileInputStream;
|
4
|
import java.io.FileNotFoundException;
|
5
|
import java.io.InputStream;
|
6
|
import java.util.ArrayList;
|
7
|
import java.util.HashSet;
|
8
|
import java.util.List;
|
9
|
import java.util.Map;
|
10
|
import java.util.Set;
|
11
|
import java.util.stream.Collectors;
|
12
|
import java.util.stream.Stream;
|
13
|
import java.util.stream.StreamSupport;
|
14
|
|
15
|
import org.apache.commons.lang3.StringUtils;
|
16
|
|
17
|
import com.fasterxml.jackson.databind.ObjectMapper;
|
18
|
|
19
|
import eu.dnetlib.data.proto.FieldTypeProtos.DataInfo;
|
20
|
import eu.dnetlib.data.proto.FieldTypeProtos.KeyValue;
|
21
|
import eu.dnetlib.data.proto.FieldTypeProtos.Qualifier;
|
22
|
import eu.dnetlib.data.proto.FieldTypeProtos.StringField;
|
23
|
import eu.dnetlib.data.proto.FieldTypeProtos.StructuredProperty;
|
24
|
import eu.dnetlib.data.proto.KindProtos.Kind;
|
25
|
import eu.dnetlib.data.proto.OafProtos.Oaf;
|
26
|
import eu.dnetlib.data.proto.OafProtos.OafEntity;
|
27
|
import eu.dnetlib.data.proto.OafProtos.OafRel;
|
28
|
import eu.dnetlib.data.proto.OrganizationProtos.Organization;
|
29
|
import eu.dnetlib.data.proto.RelTypeProtos.RelType;
|
30
|
import eu.dnetlib.data.proto.RelTypeProtos.SubRelType;
|
31
|
import eu.dnetlib.data.proto.TypeProtos.Type;
|
32
|
import eu.dnetlib.grid.process.model.GridLabel;
|
33
|
import eu.dnetlib.grid.process.model.GridOrganization;
|
34
|
import eu.dnetlib.grid.process.model.GridRel;
|
35
|
import eu.dnetlib.grid.process.model.GridResponse;
|
36
|
import eu.dnetlib.miscutils.datetime.DateUtils;
|
37
|
import eu.dnetlib.miscutils.functional.hash.Hashing;
|
38
|
|
39
|
public class GridUtils {
|
40
|
|
41
|
public static Stream<GridOrganization> streamOrganizations(final String jsonFile) {
|
42
|
try {
|
43
|
return StreamSupport.stream(GridUtils.getOrganizations(new FileInputStream(jsonFile)).spliterator(), false);
|
44
|
} catch (final FileNotFoundException e) {
|
45
|
e.printStackTrace();
|
46
|
return Stream.empty();
|
47
|
}
|
48
|
}
|
49
|
|
50
|
protected static Iterable<GridOrganization> getOrganizations(final InputStream input) {
|
51
|
try {
|
52
|
return new ObjectMapper().readValue(input, GridResponse.class).getInstitutes();
|
53
|
} catch (final Throwable e) {
|
54
|
e.printStackTrace();
|
55
|
return new ArrayList<>();
|
56
|
}
|
57
|
}
|
58
|
|
59
|
public static List<Oaf> toProtos(final GridOrganization org, final Map<String, String> parents, final Datasource ds) {
|
60
|
|
61
|
final String gridId = org.getId();
|
62
|
final String parent = findParentName(org, parents);
|
63
|
final String mainName = applyPrefix(parent, org.getName());
|
64
|
final String shortName = findShortName(org);
|
65
|
final String mainOpenaireId = calculateOpenaireId(ds.getPrefix(), gridId);
|
66
|
final Set<String> alternativeNames = calculateAlternativeNames(org);
|
67
|
|
68
|
final Map<String, String> orgNames = alternativeNames.stream()
|
69
|
.map(s -> applyPrefix(parent, s))
|
70
|
.collect(Collectors.toMap(
|
71
|
s -> calculateOpenaireId(ds.getPrefix(), gridId, s),
|
72
|
s -> s));
|
73
|
orgNames.put(mainOpenaireId, mainName);
|
74
|
|
75
|
final Qualifier country = org.getAddresses().stream()
|
76
|
.map(addr -> Qualifier.newBuilder()
|
77
|
.setClassid(addr.getCountry_code())
|
78
|
.setClassname(addr.getCountry())
|
79
|
.setSchemeid("dnet:countries")
|
80
|
.setSchemename("dnet:countries"))
|
81
|
.filter(q -> StringUtils.isNotBlank(q.getClassid()))
|
82
|
.filter(q -> StringUtils.isNotBlank(q.getClassname()))
|
83
|
.map(q -> q.build())
|
84
|
.findFirst()
|
85
|
.orElse(null);
|
86
|
|
87
|
final KeyValue collectedFrom = KeyValue.newBuilder()
|
88
|
.setKey(ds.getId())
|
89
|
.setValue(ds.getName()).build();
|
90
|
|
91
|
return orgNames.entrySet()
|
92
|
.stream()
|
93
|
.map(e -> toProtos(e.getKey(),
|
94
|
gridId,
|
95
|
e.getValue(),
|
96
|
shortName,
|
97
|
org.getLinks().stream().findFirst().orElse(""),
|
98
|
country,
|
99
|
e.getKey().equals(mainOpenaireId) ? alternativeNames : new HashSet<>(),
|
100
|
orgNames.keySet(),
|
101
|
collectedFrom))
|
102
|
.flatMap(l -> l.stream())
|
103
|
.collect(Collectors.toList());
|
104
|
}
|
105
|
|
106
|
private static List<Oaf> toProtos(final String openaireId,
|
107
|
final String gridId,
|
108
|
final String name,
|
109
|
final String shortName,
|
110
|
final String url,
|
111
|
final Qualifier country,
|
112
|
final Set<String> alternativeNames,
|
113
|
final Set<String> orgRels,
|
114
|
final KeyValue collectedFrom) {
|
115
|
|
116
|
final OafEntity.Builder entity = OafEntity.newBuilder()
|
117
|
.setId(openaireId)
|
118
|
.addPid(StructuredProperty.newBuilder()
|
119
|
.setValue(gridId)
|
120
|
.setQualifier(Qualifier.newBuilder()
|
121
|
.setClassid("grid")
|
122
|
.setClassname("grid")
|
123
|
.setSchemeid("dnet:pid_types")
|
124
|
.setSchemename("dnet:pid_types")))
|
125
|
.addCollectedfrom(collectedFrom)
|
126
|
.setType(Type.organization)
|
127
|
.setOrganization(Organization.newBuilder()
|
128
|
.setMetadata(Organization.Metadata.newBuilder()
|
129
|
.setLegalname(StringField.newBuilder().setValue(name))
|
130
|
.setLegalshortname(StringField.newBuilder().setValue(shortName))
|
131
|
.setWebsiteurl(StringField.newBuilder().setValue(url))
|
132
|
.addAllAlternativeNames(
|
133
|
alternativeNames.stream().map(a -> StringField.newBuilder().setValue(a).build()).collect(Collectors.toList()))
|
134
|
.setCountry(country)));
|
135
|
|
136
|
// Relations
|
137
|
final List<Oaf> oafs = orgRels.stream()
|
138
|
.filter(id -> !id.equals(openaireId))
|
139
|
.map(id -> Oaf.newBuilder()
|
140
|
.setKind(Kind.relation)
|
141
|
.setLastupdatetimestamp(DateUtils.now())
|
142
|
.setRel(OafRel.newBuilder()
|
143
|
.setSource(openaireId)
|
144
|
.setTarget(id)
|
145
|
.setRelType(RelType.organizationOrganization)
|
146
|
.setSubRelType(SubRelType.dedupSimilarity)
|
147
|
.setRelClass("isSimilarTo")
|
148
|
.setChild(false))
|
149
|
.build())
|
150
|
.collect(Collectors.toList());
|
151
|
|
152
|
// Entity
|
153
|
oafs.add(Oaf.newBuilder()
|
154
|
.setKind(Kind.entity)
|
155
|
.setLastupdatetimestamp(DateUtils.now())
|
156
|
.setEntity(entity)
|
157
|
.setDataInfo(DataInfo.newBuilder()
|
158
|
.setTrust("0.90")
|
159
|
.setInferred(false)
|
160
|
.setProvenanceaction(Qualifier.newBuilder()
|
161
|
.setClassid("UNKNOWN")
|
162
|
.setClassname("UNKNOWN")
|
163
|
.setSchemeid("dnet:provenanceActions")
|
164
|
.setSchemename("dnet:provenanceActions")))
|
165
|
.build());
|
166
|
|
167
|
return oafs;
|
168
|
}
|
169
|
|
170
|
private static String calculateOpenaireId(final String prefix, final String gridId) {
|
171
|
return String.format("20|%s::%s", prefix, Hashing.md5(gridId));
|
172
|
}
|
173
|
|
174
|
private static String calculateOpenaireId(final String prefix, final String gridId, final String name) {
|
175
|
return String.format("20|%s::%s", prefix, Hashing.md5(gridId + " " + name));
|
176
|
}
|
177
|
|
178
|
private static String applyPrefix(final String parent, final String simpleName) {
|
179
|
return StringUtils.isBlank(parent) ? simpleName : parent + " - " + simpleName;
|
180
|
}
|
181
|
|
182
|
private static Set<String> calculateAlternativeNames(final GridOrganization org) {
|
183
|
final Set<String> res = org.getLabels().stream()
|
184
|
.map(GridLabel::getLabel)
|
185
|
.collect(Collectors.toSet());
|
186
|
res.addAll(org.getAcronyms());
|
187
|
return res;
|
188
|
}
|
189
|
|
190
|
public static String findShortName(final GridOrganization org) {
|
191
|
return org.getAcronyms()
|
192
|
.stream()
|
193
|
.filter(StringUtils::isNotBlank)
|
194
|
.findFirst()
|
195
|
.orElse(org.getName());
|
196
|
}
|
197
|
|
198
|
private static String findParentName(final GridOrganization org, final Map<String, String> parents) {
|
199
|
return org.getRelationships()
|
200
|
.stream()
|
201
|
.filter(r -> r.getType().equalsIgnoreCase("Parent"))
|
202
|
.map(GridRel::getId)
|
203
|
.map(parents::get)
|
204
|
.filter(StringUtils::isNotBlank)
|
205
|
.findFirst()
|
206
|
.orElse(null);
|
207
|
}
|
208
|
}
|