1
|
package eu.dnetlib.data.transform;
|
2
|
|
3
|
import java.util.Collection;
|
4
|
import java.util.List;
|
5
|
import java.util.Map;
|
6
|
import java.util.Set;
|
7
|
import java.util.stream.Collectors;
|
8
|
import java.util.stream.Stream;
|
9
|
|
10
|
import com.google.common.base.Functions;
|
11
|
import com.google.common.base.Predicate;
|
12
|
import com.google.common.collect.Iterables;
|
13
|
import com.google.common.collect.Lists;
|
14
|
import com.google.common.collect.Maps;
|
15
|
import com.google.common.collect.Sets;
|
16
|
import com.google.protobuf.Descriptors.FieldDescriptor;
|
17
|
import com.google.protobuf.Message.Builder;
|
18
|
import eu.dnetlib.data.proto.FieldTypeProtos;
|
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.ResultProtos;
|
27
|
import eu.dnetlib.data.proto.ResultProtos.Result;
|
28
|
import eu.dnetlib.data.proto.ResultProtos.Result.Instance;
|
29
|
import eu.dnetlib.data.proto.SpecialTrustProtos.SpecialTrust;
|
30
|
import eu.dnetlib.data.proto.TypeProtos.Type;
|
31
|
import eu.dnetlib.pace.config.DedupConfig;
|
32
|
import org.apache.commons.lang.StringUtils;
|
33
|
|
34
|
public class OafEntityMerger {
|
35
|
|
36
|
private static final String DEDUP_CLASSID = "sysimport:dedup";
|
37
|
|
38
|
private static final String DNET_PROVENANCE_SCHEME = "dnet:provenanceActions";
|
39
|
|
40
|
private final Predicate<StringField> skipEmptyStringField = s -> (s != null) && (s.getValue() != null) && !s.getValue().isEmpty();
|
41
|
|
42
|
private final Predicate<String> skipEmptyString = s -> StringUtils.isNotBlank(s);
|
43
|
|
44
|
public static Oaf.Builder merge(final String id, final Iterable<Oaf> entities) {
|
45
|
return merge(null, id, entities);
|
46
|
}
|
47
|
|
48
|
public static Oaf.Builder merge(final DedupConfig dedupConf, final String id, final Iterable<Oaf> entities) {
|
49
|
return new OafEntityMerger().mergeEntities(dedupConf, id, entities);
|
50
|
}
|
51
|
|
52
|
public static Oaf.Builder merge(final Oaf.Builder builder) {
|
53
|
return new OafEntityMerger().doMergeEntities(builder);
|
54
|
}
|
55
|
|
56
|
public Oaf.Builder mergeEntities(final DedupConfig dedupConf, final String id, final Iterable<Oaf> entities) {
|
57
|
|
58
|
Oaf.Builder builder = Oaf.newBuilder();
|
59
|
String trust = "0.0";
|
60
|
|
61
|
for (final Oaf oaf : TrustOrdering.sort(entities)) {
|
62
|
// doublecheck we're dealing only with main entities
|
63
|
if (!oaf.getKind().equals(Kind.entity)) throw new IllegalArgumentException("expected OafEntity!");
|
64
|
|
65
|
final String currentTrust = oaf.getDataInfo().getTrust();
|
66
|
if (!currentTrust.equals(SpecialTrust.NEUTRAL.toString())) {
|
67
|
trust = currentTrust;
|
68
|
}
|
69
|
|
70
|
builder.mergeFrom(oaf);
|
71
|
|
72
|
if (oaf.getEntity().getType().equals(Type.result)) {
|
73
|
builder.getEntityBuilder().getResultBuilder().getMetadataBuilder().clearAuthor();
|
74
|
builder.getEntityBuilder().getResultBuilder().getMetadataBuilder().addAllAuthor(oaf.getEntity().getResult().getMetadata().getAuthorList());
|
75
|
}
|
76
|
}
|
77
|
|
78
|
builder = doMergeEntities(builder);
|
79
|
builder.getEntityBuilder().setId(id);
|
80
|
builder.getDataInfoBuilder()
|
81
|
.setInvisible(false)
|
82
|
.setInferred(true)
|
83
|
.setDeletedbyinference(false)
|
84
|
.setTrust(trust)
|
85
|
.setInferenceprovenance(dedupConf != null ? dedupConf.getWf().getConfigurationId() : "")
|
86
|
.setProvenanceaction(getProvenanceAction());
|
87
|
|
88
|
if ((dedupConf != null) && dedupConf.getWf().isIncludeChildren()) {
|
89
|
for (final Oaf oaf : Iterables.limit(entities, dedupConf.getWf().getMaxChildren())) {
|
90
|
builder.getEntityBuilder().addChildren(oaf.getEntity());
|
91
|
}
|
92
|
}
|
93
|
|
94
|
return builder;
|
95
|
}
|
96
|
|
97
|
private Qualifier.Builder getProvenanceAction() {
|
98
|
return Qualifier.newBuilder().setClassid(DEDUP_CLASSID).setClassname(DEDUP_CLASSID).setSchemeid(DNET_PROVENANCE_SCHEME)
|
99
|
.setSchemename(DNET_PROVENANCE_SCHEME);
|
100
|
}
|
101
|
|
102
|
public Oaf.Builder doMergeEntities(final Oaf.Builder builder) {
|
103
|
|
104
|
for (final String field : OafUtils.getFieldNames(OafEntity.getDescriptor(), OafEntity.COLLECTEDFROM_FIELD_NUMBER)) {
|
105
|
setKeyValues(builder.getEntityBuilder(), field);
|
106
|
}
|
107
|
for (final String field : OafUtils.getFieldNames(OafEntity.getDescriptor(), OafEntity.PID_FIELD_NUMBER)) {
|
108
|
setStructuredProperty(builder.getEntityBuilder(), field);
|
109
|
}
|
110
|
for (final String field : OafUtils.getFieldNames(OafEntity.getDescriptor(), OafEntity.ORIGINALID_FIELD_NUMBER)) {
|
111
|
setUniqueString(builder.getEntityBuilder(), field);
|
112
|
}
|
113
|
|
114
|
switch (builder.getEntity().getType()) {
|
115
|
case datasource:
|
116
|
break;
|
117
|
case organization:
|
118
|
break;
|
119
|
case project:
|
120
|
break;
|
121
|
case result:
|
122
|
final Result.Metadata.Builder result = builder.getEntityBuilder().getResultBuilder().getMetadataBuilder();
|
123
|
setTitle(result);
|
124
|
mergeInstances(builder.getEntityBuilder().getResultBuilder());
|
125
|
|
126
|
// for (String field : Lists.newArrayList("subject", "relevantdate")) {
|
127
|
for (final String field : OafUtils.getFieldNames(Result.Metadata.getDescriptor(), Result.Metadata.SUBJECT_FIELD_NUMBER,
|
128
|
Result.Metadata.RELEVANTDATE_FIELD_NUMBER)) {
|
129
|
setStructuredProperty(result, field);
|
130
|
}
|
131
|
for (final String field : OafUtils.getFieldNames(Result.Metadata.getDescriptor(), Result.Metadata.DESCRIPTION_FIELD_NUMBER)) {
|
132
|
setLongestStringField(result, field);
|
133
|
}
|
134
|
for (final String field : OafUtils.getFieldNames(Result.Metadata.getDescriptor(), Result.Metadata.SOURCE_FIELD_NUMBER)) {
|
135
|
setUniqueStringField(result, field);
|
136
|
}
|
137
|
|
138
|
mergeContexts(result);
|
139
|
mergeCountries(result);
|
140
|
|
141
|
break;
|
142
|
default:
|
143
|
break;
|
144
|
}
|
145
|
return builder;
|
146
|
}
|
147
|
|
148
|
private void mergeCountries(Result.Metadata.Builder result) {
|
149
|
final Map<String, Qualifier.Builder> cMap = Maps.newHashMap();
|
150
|
for(Qualifier country : result.getCountryList()) {
|
151
|
if (!cMap.containsKey(country.getClassid())) {
|
152
|
cMap.put(country.getClassid(), Qualifier.newBuilder(country));
|
153
|
}
|
154
|
}
|
155
|
if (!cMap.isEmpty()) {
|
156
|
result.clearCountry();
|
157
|
for(Qualifier.Builder country : cMap.values()) {
|
158
|
result.addCountry(country.build());
|
159
|
}
|
160
|
}
|
161
|
}
|
162
|
|
163
|
private void mergeContexts(Result.Metadata.Builder result) {
|
164
|
final Map<String, Result.Context.Builder> cMap = Maps.newHashMap();
|
165
|
for(Result.Context c : result.getContextList()) {
|
166
|
if (!cMap.containsKey(c.getId())) {
|
167
|
//TODO merge DataInfo
|
168
|
cMap.put(c.getId(), Result.Context.newBuilder(c));
|
169
|
}
|
170
|
}
|
171
|
if (!cMap.isEmpty()) {
|
172
|
result.clearContext();
|
173
|
for (Result.Context.Builder b : cMap.values()) {
|
174
|
result.addContext(b.build());
|
175
|
}
|
176
|
}
|
177
|
}
|
178
|
|
179
|
private void mergeInstances(final Result.Builder builder) {
|
180
|
final Map<String, Instance.Builder> map = Maps.newHashMap();
|
181
|
for(Instance i : builder.getInstanceList()) {
|
182
|
|
183
|
final String key = i.getHostedby().getKey() + i.getAccessright().getClassid() + i.getInstancetype().getClassid();
|
184
|
if (!map.containsKey(key)) {
|
185
|
map.put(key, Instance.newBuilder().mergeFrom(i));
|
186
|
continue;
|
187
|
}
|
188
|
|
189
|
map.get(key).addAllUrl(i.getUrlList());
|
190
|
}
|
191
|
|
192
|
for(Instance.Builder i : map.values()) {
|
193
|
final Set<String> urls = Sets.newHashSet();
|
194
|
urls.addAll(i.getUrlList());
|
195
|
i.clearUrl().addAllUrl(urls);
|
196
|
}
|
197
|
builder.clearInstance();
|
198
|
builder.addAllInstance(Iterables.transform(map.values(), b -> b.build()));
|
199
|
}
|
200
|
|
201
|
/**
|
202
|
* Helper method, avoid duplicated StructuredProperties in the given builder for the given fieldName
|
203
|
*
|
204
|
* @param builder
|
205
|
* @param fieldName
|
206
|
*/
|
207
|
@SuppressWarnings("unchecked")
|
208
|
private void setStructuredProperty(final Builder builder, final String fieldName) {
|
209
|
final Map<String, StructuredProperty> map = Maps.newHashMap();
|
210
|
final FieldDescriptor fd = builder.getDescriptorForType().findFieldByName(fieldName);
|
211
|
final List<StructuredProperty> sps = (List<StructuredProperty>) builder.getField(fd);
|
212
|
|
213
|
if ((sps != null) && !sps.isEmpty()) {
|
214
|
for (final StructuredProperty sp : sps) {
|
215
|
if (StringUtils.isNotBlank(sp.getValue())) {
|
216
|
map.put(sp.getValue().toLowerCase(), sp);
|
217
|
}
|
218
|
}
|
219
|
|
220
|
if (!map.isEmpty()) {
|
221
|
builder.clearField(fd).setField(fd, Lists.newArrayList(map.values()));
|
222
|
}
|
223
|
}
|
224
|
}
|
225
|
|
226
|
/**
|
227
|
* Helper method, avoid duplicated KeyValues in the given builder for the given fieldName
|
228
|
*
|
229
|
* @param builder
|
230
|
* @param fieldName
|
231
|
*/
|
232
|
@SuppressWarnings("unchecked")
|
233
|
private void setKeyValues(final Builder builder, final String fieldName) {
|
234
|
final Map<String, KeyValue> map = Maps.newHashMap();
|
235
|
final FieldDescriptor fd = builder.getDescriptorForType().findFieldByName(fieldName);
|
236
|
final List<KeyValue> kvs = (List<KeyValue>) builder.getField(fd);
|
237
|
|
238
|
if ((kvs != null) && !kvs.isEmpty()) {
|
239
|
for (final KeyValue sp : kvs) {
|
240
|
map.put(sp.getKey(), sp);
|
241
|
}
|
242
|
|
243
|
if (!map.isEmpty()) {
|
244
|
builder.clearField(fd).setField(fd, Lists.newArrayList(map.values()));
|
245
|
}
|
246
|
}
|
247
|
}
|
248
|
|
249
|
@SuppressWarnings("unchecked")
|
250
|
private void setSingleString(final Builder builder, final String fieldName) {
|
251
|
|
252
|
final FieldDescriptor fd = builder.getDescriptorForType().findFieldByName(fieldName);
|
253
|
final List<StringField> field = (List<StringField>) builder.getField(fd);
|
254
|
if ((field != null) && !field.isEmpty()) {
|
255
|
final StringField s = (StringField) Iterables.getLast(Iterables.filter(field, skipEmptyStringField), "");
|
256
|
|
257
|
if ((s != null) && (s.getValue() != null) && !s.getValue().isEmpty()) {
|
258
|
builder.clearField(fd).setField(fd, Lists.newArrayList(s));
|
259
|
}
|
260
|
}
|
261
|
}
|
262
|
|
263
|
@SuppressWarnings("unchecked")
|
264
|
private void setLongestStringField(final Builder builder, final String fieldName) {
|
265
|
|
266
|
final FieldDescriptor fd = builder.getDescriptorForType().findFieldByName(fieldName);
|
267
|
final List<StringField> field = (List<StringField>) builder.getField(fd);
|
268
|
|
269
|
if ((field != null) && !field.isEmpty()) {
|
270
|
final StringField.Builder max = StringField.newBuilder().setValue("");
|
271
|
int maxLength = 0;
|
272
|
for (final StringField sf : field) {
|
273
|
if (sf.getValue().length() > maxLength) {
|
274
|
maxLength = sf.getValue().length();
|
275
|
max.clear();
|
276
|
max.mergeFrom(sf);
|
277
|
}
|
278
|
}
|
279
|
|
280
|
builder.clearField(fd).setField(fd, Lists.newArrayList(max.build()));
|
281
|
}
|
282
|
}
|
283
|
|
284
|
@SuppressWarnings("unchecked")
|
285
|
private void setUniqueStringField(final Builder builder, final String fieldName) {
|
286
|
|
287
|
final FieldDescriptor fd = builder.getDescriptorForType().findFieldByName(fieldName);
|
288
|
final List<StringField> field = (List<StringField>) builder.getField(fd);
|
289
|
final Map<String, StringField> map = Maps.newHashMap();
|
290
|
if ((field != null) && !field.isEmpty()) {
|
291
|
for (final StringField s : Iterables.filter(field, skipEmptyStringField)) {
|
292
|
map.put(s.getValue(), s);
|
293
|
}
|
294
|
|
295
|
builder.clearField(fd).setField(fd, Lists.newArrayList(map.values()));
|
296
|
}
|
297
|
}
|
298
|
|
299
|
@SuppressWarnings("unchecked")
|
300
|
private void setUniqueString(final Builder builder, final String fieldName) {
|
301
|
|
302
|
final FieldDescriptor fd = builder.getDescriptorForType().findFieldByName(fieldName);
|
303
|
final List<String> field = (List<String>) builder.getField(fd);
|
304
|
final Set<String> set = Sets.newHashSet();
|
305
|
if ((field != null) && !field.isEmpty()) {
|
306
|
for (final String s : Iterables.filter(field, skipEmptyString)) {
|
307
|
set.add(s);
|
308
|
}
|
309
|
|
310
|
builder.clearField(fd).setField(fd, Lists.newArrayList(set));
|
311
|
}
|
312
|
}
|
313
|
|
314
|
private void setTitle(final Result.Metadata.Builder metadata) {
|
315
|
final Iterable<StructuredProperty> filtered = Iterables.filter(metadata.getTitleList(), OafUtils.mainTitleFilter());
|
316
|
|
317
|
if (!Iterables.isEmpty(filtered)) {
|
318
|
metadata.clearTitle().addTitle(Iterables.getLast(filtered));
|
319
|
}
|
320
|
}
|
321
|
|
322
|
}
|