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