Project

General

Profile

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 parent = parents.get(gridId);
58
		final String mainName = calculateName(parent, org.getName());
59
		final String shortName = findShortName(org);
60
		final String mainOpenaireId = calculateOpenaireId(ds.getPrefix(), gridId, mainName);
61

    
62
		final Map<String, String> orgNames = calculateNames(parent, org.getAcronyms())
63
				.stream()
64
				.collect(Collectors.toMap(
65
						s -> calculateOpenaireId(ds.getPrefix(), gridId, s),
66
						s -> s));
67
		orgNames.put(mainOpenaireId, mainName);
68

    
69
		final Qualifier country = org.getAddresses().stream()
70
				.map(addr -> Qualifier.newBuilder()
71
						.setClassid(addr.getCountry_code())
72
						.setClassname(addr.getCountry())
73
						.setSchemeid("dnet:countries")
74
						.setSchemename("dnet:countries"))
75
				.filter(q -> StringUtils.isNotBlank(q.getClassid()))
76
				.filter(q -> StringUtils.isNotBlank(q.getClassname()))
77
				.map(q -> q.build())
78
				.findFirst()
79
				.orElse(null);
80

    
81
		final KeyValue collectedFrom = KeyValue.newBuilder()
82
				.setKey(ds.getId())
83
				.setValue(ds.getName()).build();
84

    
85
		return orgNames.entrySet()
86
				.stream()
87
				.map(e -> toProto(e.getKey(),
88
						gridId,
89
						e.getValue(),
90
						shortName,
91
						org.getLinks().stream().findFirst().orElse(""),
92
						country,
93
						e.getKey().equals(mainOpenaireId) ? org.getAcronyms() : new ArrayList<>(),
94
						orgNames, collectedFrom))
95
				.collect(Collectors.toList());
96

    
97
	}
98

    
99
	private static Oaf toProto(final String openaireId,
100
			final String gridId,
101
			final String name,
102
			final String shortName,
103
			final String url,
104
			final Qualifier country,
105
			final List<String> acronyms,
106
			final Map<String, String> orgRels,
107
			final KeyValue collectedFrom) {
108

    
109
		// TODO ADD ALTERNATIVE NAMES AND RELS
110

    
111
		final OafEntity.Builder entity = OafEntity.newBuilder()
112
				.setId(openaireId)
113
				.addPid(StructuredProperty.newBuilder()
114
						.setValue(gridId)
115
						.setQualifier(Qualifier.newBuilder()
116
								.setClassid("grid")
117
								.setClassname("grid")
118
								.setSchemeid("dnet:pid_types")
119
								.setSchemename("dnet:pid_types")))
120
				.addCollectedfrom(collectedFrom)
121
				.setType(Type.organization)
122
				.setOrganization(Organization.newBuilder()
123
						.setMetadata(Organization.Metadata.newBuilder()
124
								.setLegalname(StringField.newBuilder().setValue(name))
125
								.setLegalshortname(StringField.newBuilder().setValue(shortName))
126
								.setWebsiteurl(StringField.newBuilder().setValue(url))
127
								.addAllAlternativeNames(acronyms.stream().map(a -> StringField.newBuilder().setValue(a).build()).collect(Collectors.toList()))
128
								.setCountry(country)));
129

    
130
		orgRels.entrySet().stream()
131
				.filter(e -> !e.getKey().equals(openaireId))
132
				.forEach(e -> entity.addCachedRel(
133
						OafRel.newBuilder()
134
								.setTarget(e.getKey())
135
								// TODO ??? e.getValue()
136
								.addCollectedfrom(collectedFrom)));
137

    
138
		return Oaf.newBuilder()
139
				.setKind(Kind.entity)
140
				.setLastupdatetimestamp(DateUtils.now())
141
				.setEntity(entity)
142
				.setDataInfo(DataInfo.newBuilder()
143
						.setTrust("0.90")
144
						.setInferred(false)
145
						.setProvenanceaction(Qualifier.newBuilder()
146
								.setClassid("UNKNOWN")
147
								.setClassname("UNKNOWN")
148
								.setSchemeid("dnet:provenanceActions")
149
								.setSchemename("dnet:provenanceActions")))
150
				.build();
151
	}
152

    
153
	private static String calculateOpenaireId(final String prefix, final String gridId, final String name) {
154
		return String.format("20|%s::%s", prefix, Hashing.md5(gridId + " " + name));
155
	}
156

    
157
	private static String calculateName(final String parent, final String simpleName) {
158
		return StringUtils.isBlank(parent) ? simpleName : parent + " - " + simpleName;
159
	}
160

    
161
	private static Set<String> calculateNames(final String parent, final List<String> list) {
162
		return list.stream()
163
				.map(s -> calculateName(parent, s))
164
				.collect(Collectors.toSet());
165
	}
166

    
167
	public static String findShortName(final GridOrganization org) {
168
		return org.getAcronyms()
169
				.stream()
170
				.filter(StringUtils::isNotBlank)
171
				.filter(s -> s.length() < 10)
172
				.filter(s -> s.equals(s.toUpperCase()))
173
				.findFirst()
174
				.orElse(org.getName());
175

    
176
	}
177
}
(3-3/3)