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.List;
|
8
|
import java.util.Map;
|
9
|
import java.util.Set;
|
10
|
import java.util.stream.Collectors;
|
11
|
import java.util.stream.Stream;
|
12
|
import java.util.stream.StreamSupport;
|
13
|
|
14
|
import org.apache.commons.lang3.StringUtils;
|
15
|
|
16
|
import com.fasterxml.jackson.databind.ObjectMapper;
|
17
|
|
18
|
import eu.dnetlib.data.proto.FieldTypeProtos.DataInfo;
|
19
|
import eu.dnetlib.data.proto.FieldTypeProtos.KeyValue;
|
20
|
import eu.dnetlib.data.proto.FieldTypeProtos.Qualifier;
|
21
|
import eu.dnetlib.data.proto.FieldTypeProtos.StringField;
|
22
|
import eu.dnetlib.data.proto.FieldTypeProtos.StructuredProperty;
|
23
|
import eu.dnetlib.data.proto.KindProtos.Kind;
|
24
|
import eu.dnetlib.data.proto.OafProtos.Oaf;
|
25
|
import eu.dnetlib.data.proto.OafProtos.OafEntity;
|
26
|
import eu.dnetlib.data.proto.OafProtos.OafRel;
|
27
|
import eu.dnetlib.data.proto.OrganizationProtos.Organization;
|
28
|
import eu.dnetlib.data.proto.TypeProtos.Type;
|
29
|
import eu.dnetlib.grid.process.model.GridOrganization;
|
30
|
import eu.dnetlib.grid.process.model.GridResponse;
|
31
|
import eu.dnetlib.miscutils.datetime.DateUtils;
|
32
|
import eu.dnetlib.miscutils.functional.hash.Hashing;
|
33
|
|
34
|
public class GridUtils {
|
35
|
|
36
|
public static Stream<GridOrganization> streamOrganizations(final String jsonFile) {
|
37
|
try {
|
38
|
return StreamSupport.stream(GridUtils.getOrganizations(new FileInputStream(jsonFile)).spliterator(), false);
|
39
|
} catch (final FileNotFoundException e) {
|
40
|
e.printStackTrace();
|
41
|
return Stream.empty();
|
42
|
}
|
43
|
}
|
44
|
|
45
|
protected static Iterable<GridOrganization> getOrganizations(final InputStream input) {
|
46
|
try {
|
47
|
return new ObjectMapper().readValue(input, GridResponse.class).getInstitutes();
|
48
|
} catch (final Throwable e) {
|
49
|
e.printStackTrace();
|
50
|
return new ArrayList<>();
|
51
|
}
|
52
|
}
|
53
|
|
54
|
public static List<Oaf> toProtos(final GridOrganization org, final Map<String, String> parents, final Datasource ds) {
|
55
|
|
56
|
final String gridId = org.getId();
|
57
|
final String shortName = findShortName(org);
|
58
|
final String parent = parents.get(gridId);
|
59
|
final String mainName = calculateName(parent, org.getName());
|
60
|
final Set<String> otherNames = calculateNames(parent, org.getAcronyms());
|
61
|
|
62
|
final Map<String, String> orgRels = otherNames.stream()
|
63
|
.collect(Collectors.toMap(s -> calculateOpenaireId(ds.getPrefix(), gridId, s), s -> s));
|
64
|
orgRels.put(calculateOpenaireId(ds.getPrefix(), gridId, mainName), mainName);
|
65
|
|
66
|
final Qualifier.Builder country = org.getAddresses().stream()
|
67
|
.map(addr -> Qualifier.newBuilder()
|
68
|
.setClassid(addr.getCountry_code())
|
69
|
.setClassname(addr.getCountry())
|
70
|
.setSchemeid("dnet:countries")
|
71
|
.setSchemename("dnet:countries"))
|
72
|
.filter(q -> StringUtils.isNotBlank(q.getClassid()))
|
73
|
.filter(q -> StringUtils.isNotBlank(q.getClassname()))
|
74
|
.findFirst()
|
75
|
.orElse(null);
|
76
|
|
77
|
final String url = org.getLinks().stream().findFirst().orElse("");
|
78
|
|
79
|
final List<Oaf> res = new ArrayList<>();
|
80
|
|
81
|
res.add(toProto(gridId, mainName, shortName, url, country, org.getAcronyms(), orgRels, ds));
|
82
|
|
83
|
res.addAll(otherNames.stream()
|
84
|
.map(name -> toProto(gridId, name, shortName, url, country, new ArrayList<>(), orgRels, ds))
|
85
|
.collect(Collectors.toList()));
|
86
|
|
87
|
return res;
|
88
|
}
|
89
|
|
90
|
private static Oaf toProto(final String gridId,
|
91
|
final String name,
|
92
|
final String shortName,
|
93
|
final String url,
|
94
|
final Qualifier.Builder country,
|
95
|
final List<String> acronyms,
|
96
|
final Map<String, String> orgRels,
|
97
|
final Datasource ds) {
|
98
|
|
99
|
// TODO ADD ALTERNATIVE NAMES AND RELS
|
100
|
|
101
|
final String openaireId = calculateOpenaireId(ds.getPrefix(), gridId, name);
|
102
|
|
103
|
final KeyValue collectedFrom = KeyValue.newBuilder()
|
104
|
.setKey(ds.getId())
|
105
|
.setValue(ds.getName()).build();
|
106
|
|
107
|
final OafEntity.Builder entity = OafEntity.newBuilder()
|
108
|
.setId(openaireId)
|
109
|
.addPid(StructuredProperty.newBuilder()
|
110
|
.setValue(gridId)
|
111
|
.setQualifier(Qualifier.newBuilder()
|
112
|
.setClassid("grid")
|
113
|
.setClassname("grid")
|
114
|
.setSchemeid("dnet:pid_types")
|
115
|
.setSchemename("dnet:pid_types")))
|
116
|
.addCollectedfrom(collectedFrom)
|
117
|
.setType(Type.organization)
|
118
|
.setOrganization(Organization.newBuilder()
|
119
|
.setMetadata(Organization.Metadata.newBuilder()
|
120
|
.setLegalname(StringField.newBuilder().setValue(name))
|
121
|
.setLegalshortname(StringField.newBuilder().setValue(shortName))
|
122
|
.setWebsiteurl(StringField.newBuilder().setValue(url))
|
123
|
.addAllAlternativeNames(acronyms.stream().map(a -> StringField.newBuilder().setValue(a).build()).collect(Collectors.toList()))
|
124
|
.setCountry(country)));
|
125
|
|
126
|
orgRels.entrySet().stream()
|
127
|
.filter(e -> !e.getKey().equals(openaireId))
|
128
|
.forEach(e -> entity.addCachedRel(
|
129
|
OafRel.newBuilder()
|
130
|
.setTarget(e.getKey())
|
131
|
// TODO ??? e.getValue()
|
132
|
.addCollectedfrom(collectedFrom)));
|
133
|
|
134
|
return Oaf.newBuilder()
|
135
|
.setKind(Kind.entity)
|
136
|
.setLastupdatetimestamp(DateUtils.now())
|
137
|
.setEntity(entity)
|
138
|
.setDataInfo(DataInfo.newBuilder()
|
139
|
.setTrust("0.90")
|
140
|
.setInferred(false)
|
141
|
.setProvenanceaction(Qualifier.newBuilder()
|
142
|
.setClassid("UNKNOWN")
|
143
|
.setClassname("UNKNOWN")
|
144
|
.setSchemeid("dnet:provenanceActions")
|
145
|
.setSchemename("dnet:provenanceActions")))
|
146
|
.build();
|
147
|
}
|
148
|
|
149
|
private static String calculateOpenaireId(final String prefix, final String gridId, final String name) {
|
150
|
return String.format("20|%s::%s", prefix, Hashing.md5(gridId + " " + name));
|
151
|
}
|
152
|
|
153
|
private static String calculateName(final String parent, final String simpleName) {
|
154
|
return StringUtils.isBlank(parent) ? simpleName : parent + " - " + simpleName;
|
155
|
}
|
156
|
|
157
|
private static Set<String> calculateNames(final String parent, final List<String> list) {
|
158
|
return list.stream()
|
159
|
.map(s -> calculateName(parent, s))
|
160
|
.collect(Collectors.toSet());
|
161
|
}
|
162
|
|
163
|
public static String findShortName(final GridOrganization org) {
|
164
|
return org.getAcronyms()
|
165
|
.stream()
|
166
|
.filter(StringUtils::isNotBlank)
|
167
|
.filter(s -> s.length() < 10)
|
168
|
.filter(s -> s.equals(s.toUpperCase()))
|
169
|
.findFirst()
|
170
|
.orElse(org.getName());
|
171
|
|
172
|
}
|
173
|
}
|