1
|
package eu.dnetlib.data.transform;
|
2
|
|
3
|
import java.util.List;
|
4
|
import java.util.Map;
|
5
|
import java.util.Set;
|
6
|
|
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.Maps;
|
11
|
import com.google.common.collect.Sets;
|
12
|
import com.google.protobuf.Descriptors.FieldDescriptor;
|
13
|
import com.google.protobuf.Message.Builder;
|
14
|
import eu.dnetlib.data.proto.DNGFProtos.DNGF;
|
15
|
import eu.dnetlib.data.proto.DNGFProtos.DNGFEntity;
|
16
|
import eu.dnetlib.data.proto.DatasetProtos.Dataset;
|
17
|
import eu.dnetlib.data.proto.FieldTypeProtos.KeyValue;
|
18
|
import eu.dnetlib.data.proto.FieldTypeProtos.Qualifier;
|
19
|
import eu.dnetlib.data.proto.FieldTypeProtos.StringField;
|
20
|
import eu.dnetlib.data.proto.FieldTypeProtos.StructuredProperty;
|
21
|
import eu.dnetlib.data.proto.KindProtos.Kind;
|
22
|
import eu.dnetlib.data.proto.PersonProtos.Person;
|
23
|
import eu.dnetlib.data.proto.PersonProtos.Person.CoAuthor;
|
24
|
import eu.dnetlib.data.proto.PersonProtos.Person.MergedPerson;
|
25
|
import eu.dnetlib.data.proto.PersonProtos.Person.Metadata;
|
26
|
import eu.dnetlib.data.proto.PublicationProtos.Publication;
|
27
|
import eu.dnetlib.data.proto.SpecialTrustProtos.SpecialTrust;
|
28
|
import eu.dnetlib.pace.config.DedupConfig;
|
29
|
import org.apache.commons.lang3.StringUtils;
|
30
|
|
31
|
public class DNGFEntityMerger {
|
32
|
|
33
|
private static final String DEDUP_CLASSID = "sysimport:dedup";
|
34
|
|
35
|
private static final String DNET_PROVENANCE_SCHEME = "dnet:provenanceActions";
|
36
|
|
37
|
private final Predicate<StringField> skipEmptyStringField = s -> (s != null) && (s.getValue() != null) && !s.getValue().isEmpty();
|
38
|
|
39
|
private final Predicate<String> skipEmptyString = s -> StringUtils.isNotBlank(s);
|
40
|
|
41
|
public static DNGF.Builder merge(final String id, final Iterable<DNGF> entities) {
|
42
|
return merge(null, id, entities);
|
43
|
}
|
44
|
|
45
|
public static DNGF.Builder merge(final DedupConfig dedupConf, final String id, final Iterable<DNGF> entities) {
|
46
|
return new DNGFEntityMerger().mergeEntities(dedupConf, id, entities);
|
47
|
}
|
48
|
|
49
|
public static DNGF.Builder merge(final DNGF.Builder builder) {
|
50
|
return new DNGFEntityMerger().doMergeEntities(builder);
|
51
|
}
|
52
|
|
53
|
public DNGF.Builder mergeEntities(final DedupConfig dedupConf, final String id, final Iterable<DNGF> entities) {
|
54
|
|
55
|
DNGF.Builder builder = DNGF.newBuilder();
|
56
|
String trust = "0.0";
|
57
|
|
58
|
for (final DNGF dngf : TrustOrdering.sort(entities)) {
|
59
|
// doublecheck we're dealing only with main entities
|
60
|
if (!dngf.getKind().equals(Kind.entity)) throw new IllegalArgumentException("expected DNGFEntity!");
|
61
|
|
62
|
final String currentTrust = dngf.getDataInfo().getTrust();
|
63
|
if (!currentTrust.equals(SpecialTrust.NEUTRAL.toString())) {
|
64
|
trust = currentTrust;
|
65
|
}
|
66
|
builder.mergeFrom(dngf);
|
67
|
}
|
68
|
|
69
|
builder = doMergeEntities(builder);
|
70
|
builder.getEntityBuilder().setId(id);
|
71
|
builder.getDataInfoBuilder()
|
72
|
.setInferred(true)
|
73
|
.setDeletedbyinference(false)
|
74
|
.setTrust(trust)
|
75
|
.setInferenceprovenance(dedupConf != null ? dedupConf.getWf().getConfigurationId() : "")
|
76
|
.setProvenanceaction(getProvenanceAction());
|
77
|
|
78
|
if ((dedupConf != null) && dedupConf.getWf().isIncludeChildren()) {
|
79
|
for (final DNGF dngf : Iterables.limit(entities, dedupConf.getWf().getMaxChildren())) {
|
80
|
builder.getEntityBuilder().addChildren(dngf.getEntity());
|
81
|
}
|
82
|
}
|
83
|
|
84
|
return builder;
|
85
|
}
|
86
|
|
87
|
private Qualifier.Builder getProvenanceAction() {
|
88
|
return Qualifier.newBuilder().setClassid(DEDUP_CLASSID).setClassname(DEDUP_CLASSID).setSchemeid(DNET_PROVENANCE_SCHEME)
|
89
|
.setSchemename(DNET_PROVENANCE_SCHEME);
|
90
|
}
|
91
|
|
92
|
public DNGF.Builder doMergeEntities(final DNGF.Builder builder) {
|
93
|
|
94
|
for (final String field : DNGFUtils.getFieldNames(DNGFEntity.getDescriptor(), DNGFEntity.COLLECTEDFROM_FIELD_NUMBER)) {
|
95
|
setKeyValues(builder.getEntityBuilder(), field);
|
96
|
}
|
97
|
for (final String field : DNGFUtils.getFieldNames(DNGFEntity.getDescriptor(), DNGFEntity.PID_FIELD_NUMBER)) {
|
98
|
setStructuredProperty(builder.getEntityBuilder(), field);
|
99
|
}
|
100
|
for (final String field : DNGFUtils.getFieldNames(DNGFEntity.getDescriptor(), DNGFEntity.ORIGINALID_FIELD_NUMBER)) {
|
101
|
setUniqueString(builder.getEntityBuilder(), field);
|
102
|
}
|
103
|
|
104
|
switch (builder.getEntity().getType()) {
|
105
|
case datasource:
|
106
|
break;
|
107
|
case organization:
|
108
|
break;
|
109
|
case person:
|
110
|
final Person.Builder person = builder.getEntityBuilder().getPersonBuilder().setAnchor(true);
|
111
|
|
112
|
for (final String field : DNGFUtils.getFieldNames(Person.Metadata.getDescriptor(), Metadata.SECONDNAMES_FIELD_NUMBER)) {
|
113
|
setSingleString(person.getMetadataBuilder(), field);
|
114
|
}
|
115
|
|
116
|
final Map<String, MergedPerson> mergedMap = Maps.newHashMap();
|
117
|
for(MergedPerson merged : person.getMergedpersonList()) {
|
118
|
mergedMap.put(merged.getId(), merged);
|
119
|
}
|
120
|
person.clearMergedperson().addAllMergedperson(mergedMap.values());
|
121
|
|
122
|
final Map<String, CoAuthor> coAuthorMap = Maps.newHashMap();
|
123
|
for(CoAuthor coAuthor : person.getCoauthorList()) {
|
124
|
coAuthorMap.put(coAuthor.getId(), coAuthor);
|
125
|
}
|
126
|
person.clearCoauthor().addAllCoauthor(coAuthorMap.values());
|
127
|
|
128
|
break;
|
129
|
case project:
|
130
|
break;
|
131
|
case publication:
|
132
|
final Publication.Metadata.Builder pub = builder.getEntityBuilder().getPublicationBuilder().getMetadataBuilder();
|
133
|
setTitle(pub);
|
134
|
|
135
|
// for (String field : Lists.newArrayList("subject", "relevantdate")) {
|
136
|
for (final String field : DNGFUtils.getFieldNames(Publication.Metadata.getDescriptor(), Publication.Metadata.SUBJECT_FIELD_NUMBER,
|
137
|
Publication.Metadata.RELEVANTDATE_FIELD_NUMBER)) {
|
138
|
setStructuredProperty(pub, field);
|
139
|
}
|
140
|
for (final String field : DNGFUtils.getFieldNames(Publication.Metadata.getDescriptor(), Publication.Metadata.DESCRIPTION_FIELD_NUMBER)) {
|
141
|
setLongestStringField(pub, field);
|
142
|
}
|
143
|
for (final String field : DNGFUtils.getFieldNames(Publication.Metadata.getDescriptor(), Publication.Metadata.SOURCE_FIELD_NUMBER)) {
|
144
|
setUniqueStringField(pub, field);
|
145
|
}
|
146
|
|
147
|
// remove the inner authors, rely on the children
|
148
|
builder.getEntityBuilder().getPublicationBuilder().clearAuthor();
|
149
|
break;
|
150
|
case dataset:
|
151
|
final Dataset.Metadata.Builder dataset = builder.getEntityBuilder().getDatasetBuilder().getMetadataBuilder();
|
152
|
setTitle(dataset);
|
153
|
|
154
|
// for (String field : Lists.newArrayList("subject", "relevantdate")) {
|
155
|
for (final String field : DNGFUtils.getFieldNames(Publication.Metadata.getDescriptor(), Publication.Metadata.SUBJECT_FIELD_NUMBER,
|
156
|
Publication.Metadata.RELEVANTDATE_FIELD_NUMBER)) {
|
157
|
setStructuredProperty(dataset, field);
|
158
|
}
|
159
|
for (final String field : DNGFUtils.getFieldNames(Publication.Metadata.getDescriptor(), Publication.Metadata.DESCRIPTION_FIELD_NUMBER)) {
|
160
|
setLongestStringField(dataset, field);
|
161
|
}
|
162
|
for (final String field : DNGFUtils.getFieldNames(Publication.Metadata.getDescriptor(), Publication.Metadata.SOURCE_FIELD_NUMBER)) {
|
163
|
setUniqueStringField(dataset, field);
|
164
|
}
|
165
|
|
166
|
// remove the inner authors, rely on the children
|
167
|
builder.getEntityBuilder().getDatasetBuilder().clearAuthor();
|
168
|
break;
|
169
|
default:
|
170
|
break;
|
171
|
}
|
172
|
return builder;
|
173
|
}
|
174
|
|
175
|
/**
|
176
|
* Helper method, avoid duplicated StructuredProperties in the given builder for the given fieldName
|
177
|
*
|
178
|
* @param builder
|
179
|
* @param fieldName
|
180
|
*/
|
181
|
@SuppressWarnings("unchecked")
|
182
|
private void setStructuredProperty(final Builder builder, final String fieldName) {
|
183
|
final Map<String, StructuredProperty> map = Maps.newHashMap();
|
184
|
final FieldDescriptor fd = builder.getDescriptorForType().findFieldByName(fieldName);
|
185
|
final List<StructuredProperty> sps = (List<StructuredProperty>) builder.getField(fd);
|
186
|
|
187
|
if ((sps != null) && !sps.isEmpty()) {
|
188
|
for (final StructuredProperty sp : sps) {
|
189
|
map.put(sp.getValue(), sp);
|
190
|
}
|
191
|
|
192
|
if (!map.isEmpty()) {
|
193
|
builder.clearField(fd).setField(fd, Lists.newArrayList(map.values()));
|
194
|
}
|
195
|
}
|
196
|
}
|
197
|
|
198
|
/**
|
199
|
* Helper method, avoid duplicated KeyValues in the given builder for the given fieldName
|
200
|
*
|
201
|
* @param builder
|
202
|
* @param fieldName
|
203
|
*/
|
204
|
@SuppressWarnings("unchecked")
|
205
|
private void setKeyValues(final Builder builder, final String fieldName) {
|
206
|
final Map<String, KeyValue> map = Maps.newHashMap();
|
207
|
final FieldDescriptor fd = builder.getDescriptorForType().findFieldByName(fieldName);
|
208
|
final List<KeyValue> kvs = (List<KeyValue>) builder.getField(fd);
|
209
|
|
210
|
if ((kvs != null) && !kvs.isEmpty()) {
|
211
|
for (final KeyValue sp : kvs) {
|
212
|
map.put(sp.getKey(), sp);
|
213
|
}
|
214
|
|
215
|
if (!map.isEmpty()) {
|
216
|
builder.clearField(fd).setField(fd, Lists.newArrayList(map.values()));
|
217
|
}
|
218
|
}
|
219
|
}
|
220
|
|
221
|
@SuppressWarnings("unchecked")
|
222
|
private void setSingleString(final Builder builder, final String fieldName) {
|
223
|
|
224
|
final FieldDescriptor fd = builder.getDescriptorForType().findFieldByName(fieldName);
|
225
|
final List<StringField> field = (List<StringField>) builder.getField(fd);
|
226
|
if ((field != null) && !field.isEmpty()) {
|
227
|
final StringField s = (StringField) Iterables.getLast(Iterables.filter(field, skipEmptyStringField), "");
|
228
|
|
229
|
if ((s != null) && (s.getValue() != null) && !s.getValue().isEmpty()) {
|
230
|
builder.clearField(fd).setField(fd, Lists.newArrayList(s));
|
231
|
}
|
232
|
}
|
233
|
}
|
234
|
|
235
|
@SuppressWarnings("unchecked")
|
236
|
private void setLongestStringField(final Builder builder, final String fieldName) {
|
237
|
|
238
|
final FieldDescriptor fd = builder.getDescriptorForType().findFieldByName(fieldName);
|
239
|
final List<StringField> field = (List<StringField>) builder.getField(fd);
|
240
|
|
241
|
if ((field != null) && !field.isEmpty()) {
|
242
|
final StringField.Builder max = StringField.newBuilder().setValue("");
|
243
|
int maxLength = 0;
|
244
|
for (final StringField sf : field) {
|
245
|
if (sf.getValue().length() > maxLength) {
|
246
|
maxLength = sf.getValue().length();
|
247
|
max.clear();
|
248
|
max.mergeFrom(sf);
|
249
|
}
|
250
|
}
|
251
|
|
252
|
builder.clearField(fd).setField(fd, Lists.newArrayList(max.build()));
|
253
|
}
|
254
|
}
|
255
|
|
256
|
@SuppressWarnings("unchecked")
|
257
|
private void setUniqueStringField(final Builder builder, final String fieldName) {
|
258
|
|
259
|
final FieldDescriptor fd = builder.getDescriptorForType().findFieldByName(fieldName);
|
260
|
final List<StringField> field = (List<StringField>) builder.getField(fd);
|
261
|
final Map<String, StringField> map = Maps.newHashMap();
|
262
|
if ((field != null) && !field.isEmpty()) {
|
263
|
for (final StringField s : Iterables.filter(field, skipEmptyStringField)) {
|
264
|
map.put(s.getValue(), s);
|
265
|
}
|
266
|
|
267
|
builder.clearField(fd).setField(fd, Lists.newArrayList(map.values()));
|
268
|
}
|
269
|
}
|
270
|
|
271
|
@SuppressWarnings("unchecked")
|
272
|
private void setUniqueString(final Builder builder, final String fieldName) {
|
273
|
|
274
|
final FieldDescriptor fd = builder.getDescriptorForType().findFieldByName(fieldName);
|
275
|
final List<String> field = (List<String>) builder.getField(fd);
|
276
|
final Set<String> set = Sets.newHashSet();
|
277
|
if ((field != null) && !field.isEmpty()) {
|
278
|
for (final String s : Iterables.filter(field, skipEmptyString)) {
|
279
|
set.add(s);
|
280
|
}
|
281
|
|
282
|
builder.clearField(fd).setField(fd, Lists.newArrayList(set));
|
283
|
}
|
284
|
}
|
285
|
|
286
|
private void setTitle(final Publication.Metadata.Builder metadata) {
|
287
|
final Iterable<StructuredProperty> filtered = Iterables.filter(metadata.getTitleList(), DNGFUtils.mainTitleFilter());
|
288
|
|
289
|
if (!Iterables.isEmpty(filtered)) {
|
290
|
metadata.clearTitle().addTitle(Iterables.getLast(filtered));
|
291
|
}
|
292
|
}
|
293
|
|
294
|
private void setTitle(final Dataset.Metadata.Builder metadata) {
|
295
|
final Iterable<StructuredProperty> filtered = Iterables.filter(metadata.getTitleList(), DNGFUtils.mainTitleFilter());
|
296
|
|
297
|
if (!Iterables.isEmpty(filtered)) {
|
298
|
metadata.clearTitle().addTitle(Iterables.getLast(filtered));
|
299
|
}
|
300
|
}
|
301
|
|
302
|
}
|