Project

General

Profile

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.data.proto.TypeProtos.Type;
25
import eu.dnetlib.pace.config.DedupConfig;
26
import org.apache.commons.lang.StringUtils;
27

    
28
public class OafEntityMerger {
29

    
30
	private static final String DEDUP_CLASSID = "sysimport:dedup";
31

    
32
	private static final String DNET_PROVENANCE_SCHEME = "dnet:provenanceActions";
33

    
34
	private final Predicate<StringField> skipEmptyStringField = s -> (s != null) && (s.getValue() != null) && !s.getValue().isEmpty();
35

    
36
	private final Predicate<String> skipEmptyString = s -> StringUtils.isNotBlank(s);
37

    
38
	public static Oaf.Builder merge(final String id, final Iterable<Oaf> entities) {
39
		return merge(null, id, entities);
40
	}
41

    
42
	public static Oaf.Builder merge(final DedupConfig dedupConf, final String id, final Iterable<Oaf> entities) {
43
		return new OafEntityMerger().mergeEntities(dedupConf, id, entities);
44
	}
45

    
46
	public static Oaf.Builder merge(final Oaf.Builder builder) {
47
		return new OafEntityMerger().doMergeEntities(builder);
48
	}
49

    
50
	public Oaf.Builder mergeEntities(final DedupConfig dedupConf, final String id, final Iterable<Oaf> entities) {
51

    
52
		Oaf.Builder builder = Oaf.newBuilder();
53
		String trust = "0.0";
54

    
55
		for (final Oaf oaf : TrustOrdering.sort(entities)) {
56
			// doublecheck we're dealing only with main entities
57
			if (!oaf.getKind().equals(Kind.entity)) throw new IllegalArgumentException("expected OafEntity!");
58

    
59
			final String currentTrust = oaf.getDataInfo().getTrust();
60
			if (!currentTrust.equals(SpecialTrust.NEUTRAL.toString())) {
61
				trust = currentTrust;
62
			}
63

    
64
			builder.mergeFrom(oaf);
65

    
66
			if (oaf.getEntity().getType().equals(Type.result)) {
67
				builder.getEntityBuilder().getResultBuilder().getMetadataBuilder().clearAuthor();
68
				builder.getEntityBuilder().getResultBuilder().getMetadataBuilder().addAllAuthor(oaf.getEntity().getResult().getMetadata().getAuthorList());
69
			}
70
		}
71

    
72
		builder = doMergeEntities(builder);
73
		builder.getEntityBuilder().setId(id);
74
		builder.getDataInfoBuilder()
75
				.setInvisible(false)
76
				.setInferred(true)
77
				.setDeletedbyinference(false)
78
				.setTrust(trust)
79
				.setInferenceprovenance(dedupConf != null ? dedupConf.getWf().getConfigurationId() : "")
80
				.setProvenanceaction(getProvenanceAction());
81

    
82
		if ((dedupConf != null) && dedupConf.getWf().isIncludeChildren()) {
83
			for (final Oaf oaf : Iterables.limit(entities, dedupConf.getWf().getMaxChildren())) {
84
				builder.getEntityBuilder().addChildren(oaf.getEntity());
85
			}
86
		}
87

    
88
		return builder;
89
	}
90

    
91
	private Qualifier.Builder getProvenanceAction() {
92
		return Qualifier.newBuilder().setClassid(DEDUP_CLASSID).setClassname(DEDUP_CLASSID).setSchemeid(DNET_PROVENANCE_SCHEME)
93
				.setSchemename(DNET_PROVENANCE_SCHEME);
94
	}
95

    
96
	public Oaf.Builder doMergeEntities(final Oaf.Builder builder) {
97

    
98
		for (final String field : OafUtils.getFieldNames(OafEntity.getDescriptor(), OafEntity.COLLECTEDFROM_FIELD_NUMBER)) {
99
			setKeyValues(builder.getEntityBuilder(), field);
100
		}
101
		for (final String field : OafUtils.getFieldNames(OafEntity.getDescriptor(), OafEntity.PID_FIELD_NUMBER)) {
102
			setStructuredProperty(builder.getEntityBuilder(), field);
103
		}
104
		for (final String field : OafUtils.getFieldNames(OafEntity.getDescriptor(), OafEntity.ORIGINALID_FIELD_NUMBER)) {
105
			setUniqueString(builder.getEntityBuilder(), field);
106
		}
107

    
108
		switch (builder.getEntity().getType()) {
109
		case datasource:
110
			break;
111
		case organization:
112
			break;
113
		case project:
114
			break;
115
		case result:
116
			final Result.Metadata.Builder result = builder.getEntityBuilder().getResultBuilder().getMetadataBuilder();
117
			setTitle(result);
118
			mergeInstances(builder.getEntityBuilder().getResultBuilder());
119

    
120
			// for (String field : Lists.newArrayList("subject", "relevantdate")) {
121
			for (final String field : OafUtils.getFieldNames(Result.Metadata.getDescriptor(), Result.Metadata.SUBJECT_FIELD_NUMBER,
122
					Result.Metadata.RELEVANTDATE_FIELD_NUMBER)) {
123
				setStructuredProperty(result, field);
124
			}
125
			for (final String field : OafUtils.getFieldNames(Result.Metadata.getDescriptor(), Result.Metadata.DESCRIPTION_FIELD_NUMBER)) {
126
				setLongestStringField(result, field);
127
			}
128
			for (final String field : OafUtils.getFieldNames(Result.Metadata.getDescriptor(), Result.Metadata.SOURCE_FIELD_NUMBER)) {
129
				setUniqueStringField(result, field);
130
			}
131

    
132
			break;
133
		default:
134
			break;
135
		}
136
		return builder;
137
	}
138

    
139
	private void mergeInstances(final Result.Builder builder) {
140
		final Map<String, Instance.Builder> map = Maps.newHashMap();
141
		for(Instance i : builder.getInstanceList()) {
142

    
143
			final String key = i.getHostedby().getKey() + i.getAccessright().getClassid() + i.getInstancetype().getClassid();
144
			if (!map.containsKey(key)) {
145
				map.put(key, Instance.newBuilder().mergeFrom(i));
146
				continue;
147
			}
148

    
149
			map.get(key).addAllUrl(i.getUrlList());
150
		}
151

    
152
		for(Instance.Builder i : map.values()) {
153
			final Set<String> urls = Sets.newHashSet();
154
			urls.addAll(i.getUrlList());
155
			i.clearUrl().addAllUrl(urls);
156
		}
157
		builder.clearInstance();
158
		builder.addAllInstance(Iterables.transform(map.values(), b -> b.build()));
159
	}
160

    
161
	/**
162
	 * Helper method, avoid duplicated StructuredProperties in the given builder for the given fieldName
163
	 *
164
	 * @param builder
165
	 * @param fieldName
166
	 */
167
	@SuppressWarnings("unchecked")
168
	private void setStructuredProperty(final Builder builder, final String fieldName) {
169
		final Map<String, StructuredProperty> map = Maps.newHashMap();
170
		final FieldDescriptor fd = builder.getDescriptorForType().findFieldByName(fieldName);
171
		final List<StructuredProperty> sps = (List<StructuredProperty>) builder.getField(fd);
172

    
173
		if ((sps != null) && !sps.isEmpty()) {
174
			for (final StructuredProperty sp : sps) {
175
				if (StringUtils.isNotBlank(sp.getValue())) {
176
					map.put(sp.getValue().toLowerCase(), sp);
177
				}
178
			}
179

    
180
			if (!map.isEmpty()) {
181
				builder.clearField(fd).setField(fd, Lists.newArrayList(map.values()));
182
			}
183
		}
184
	}
185

    
186
	/**
187
	 * Helper method, avoid duplicated KeyValues in the given builder for the given fieldName
188
	 *
189
	 * @param builder
190
	 * @param fieldName
191
	 */
192
	@SuppressWarnings("unchecked")
193
	private void setKeyValues(final Builder builder, final String fieldName) {
194
		final Map<String, KeyValue> map = Maps.newHashMap();
195
		final FieldDescriptor fd = builder.getDescriptorForType().findFieldByName(fieldName);
196
		final List<KeyValue> kvs = (List<KeyValue>) builder.getField(fd);
197

    
198
		if ((kvs != null) && !kvs.isEmpty()) {
199
			for (final KeyValue sp : kvs) {
200
				map.put(sp.getKey(), sp);
201
			}
202

    
203
			if (!map.isEmpty()) {
204
				builder.clearField(fd).setField(fd, Lists.newArrayList(map.values()));
205
			}
206
		}
207
	}
208

    
209
	@SuppressWarnings("unchecked")
210
	private void setSingleString(final Builder builder, final String fieldName) {
211

    
212
		final FieldDescriptor fd = builder.getDescriptorForType().findFieldByName(fieldName);
213
		final List<StringField> field = (List<StringField>) builder.getField(fd);
214
		if ((field != null) && !field.isEmpty()) {
215
			final StringField s = (StringField) Iterables.getLast(Iterables.filter(field, skipEmptyStringField), "");
216

    
217
			if ((s != null) && (s.getValue() != null) && !s.getValue().isEmpty()) {
218
				builder.clearField(fd).setField(fd, Lists.newArrayList(s));
219
			}
220
		}
221
	}
222

    
223
	@SuppressWarnings("unchecked")
224
	private void setLongestStringField(final Builder builder, final String fieldName) {
225

    
226
		final FieldDescriptor fd = builder.getDescriptorForType().findFieldByName(fieldName);
227
		final List<StringField> field = (List<StringField>) builder.getField(fd);
228

    
229
		if ((field != null) && !field.isEmpty()) {
230
			final StringField.Builder max = StringField.newBuilder().setValue("");
231
			int maxLength = 0;
232
			for (final StringField sf : field) {
233
				if (sf.getValue().length() > maxLength) {
234
					maxLength = sf.getValue().length();
235
					max.clear();
236
					max.mergeFrom(sf);
237
				}
238
			}
239

    
240
			builder.clearField(fd).setField(fd, Lists.newArrayList(max.build()));
241
		}
242
	}
243

    
244
	@SuppressWarnings("unchecked")
245
	private void setUniqueStringField(final Builder builder, final String fieldName) {
246

    
247
		final FieldDescriptor fd = builder.getDescriptorForType().findFieldByName(fieldName);
248
		final List<StringField> field = (List<StringField>) builder.getField(fd);
249
		final Map<String, StringField> map = Maps.newHashMap();
250
		if ((field != null) && !field.isEmpty()) {
251
			for (final StringField s : Iterables.filter(field, skipEmptyStringField)) {
252
				map.put(s.getValue(), s);
253
			}
254

    
255
			builder.clearField(fd).setField(fd, Lists.newArrayList(map.values()));
256
		}
257
	}
258

    
259
	@SuppressWarnings("unchecked")
260
	private void setUniqueString(final Builder builder, final String fieldName) {
261

    
262
		final FieldDescriptor fd = builder.getDescriptorForType().findFieldByName(fieldName);
263
		final List<String> field = (List<String>) builder.getField(fd);
264
		final Set<String> set = Sets.newHashSet();
265
		if ((field != null) && !field.isEmpty()) {
266
			for (final String s : Iterables.filter(field, skipEmptyString)) {
267
				set.add(s);
268
			}
269

    
270
			builder.clearField(fd).setField(fd, Lists.newArrayList(set));
271
		}
272
	}
273

    
274
	private void setTitle(final Result.Metadata.Builder metadata) {
275
		final Iterable<StructuredProperty> filtered = Iterables.filter(metadata.getTitleList(), OafUtils.mainTitleFilter());
276

    
277
		if (!Iterables.isEmpty(filtered)) {
278
			metadata.clearTitle().addTitle(Iterables.getLast(filtered));
279
		}
280
	}
281

    
282
}
(2-2/6)