|
1
|
package eu.dnetlib.data.transform.xml;
|
|
2
|
|
|
3
|
import java.util.List;
|
|
4
|
import java.util.Map;
|
|
5
|
|
|
6
|
import com.google.common.collect.Lists;
|
|
7
|
import com.google.common.collect.Maps;
|
|
8
|
import eu.dnetlib.data.mapreduce.util.OafRowKeyDecoder;
|
|
9
|
import eu.dnetlib.data.proto.FieldTypeProtos.StructuredProperty;
|
|
10
|
import eu.dnetlib.data.proto.OafProtos.Oaf;
|
|
11
|
import eu.dnetlib.data.proto.OafProtos.OafEntity;
|
|
12
|
import eu.dnetlib.data.proto.ResultProtos.Result;
|
|
13
|
import eu.dnetlib.data.proto.ResultProtos.Result.Instance;
|
|
14
|
import eu.dnetlib.data.proto.TypeProtos.Type;
|
|
15
|
import org.w3c.dom.NamedNodeMap;
|
|
16
|
import org.w3c.dom.Node;
|
|
17
|
import org.w3c.dom.NodeList;
|
|
18
|
|
|
19
|
public class OdfToHbaseXsltFunctions extends CommonDNetXsltFunctions {
|
|
20
|
|
|
21
|
private static Map<String, String> mappingAccess = Maps.newHashMap();
|
|
22
|
|
|
23
|
static {
|
|
24
|
|
|
25
|
mappingAccess.put("info:eu-repo/semantics/openAccess", "OPEN");
|
|
26
|
mappingAccess.put("info:eu-repo/semantics/closedAccess", "CLOSED");
|
|
27
|
mappingAccess.put("info:eu-repo/semantics/restrictedAccess", "RESTRICTED");
|
|
28
|
mappingAccess.put("info:eu-repo/semantics/embargoedAccess", "EMBARGO");
|
|
29
|
|
|
30
|
// Transformator now maps the access rights into proper values, not sure if it does for all datasets.
|
|
31
|
mappingAccess.put("OPEN", "OPEN");
|
|
32
|
mappingAccess.put("CLOSED", "CLOSED");
|
|
33
|
mappingAccess.put("RESTRICTED", "RESTRICTED");
|
|
34
|
mappingAccess.put("EMBARGO", "EMBARGO");
|
|
35
|
|
|
36
|
}
|
|
37
|
|
|
38
|
public static String odfResult(
|
|
39
|
final String resultId,
|
|
40
|
final NodeList about,
|
|
41
|
final NodeList metadata,
|
|
42
|
final NodeList titles,
|
|
43
|
final NodeList subjects,
|
|
44
|
final NodeList publisher,
|
|
45
|
final NodeList descriptions,
|
|
46
|
final NodeList dates,
|
|
47
|
final NodeList dateaccepted,
|
|
48
|
final NodeList resourceTypes,
|
|
49
|
final NodeList formats,
|
|
50
|
final NodeList sizes,
|
|
51
|
final NodeList languages,
|
|
52
|
final NodeList cobjcategory,
|
|
53
|
final NodeList contributors,
|
|
54
|
final NodeList rights,
|
|
55
|
final NodeList version,
|
|
56
|
final NodeList pidList,
|
|
57
|
final String provenance,
|
|
58
|
final String trust,
|
|
59
|
final String hostedbyId,
|
|
60
|
final String hostedbyName,
|
|
61
|
final String collectedfromId,
|
|
62
|
final String collectedfromName,
|
|
63
|
final NodeList originalIds,
|
|
64
|
final String instanceUri,
|
|
65
|
final String dateOfCollection,
|
|
66
|
final String dateOfTransformation) {
|
|
67
|
|
|
68
|
try {
|
|
69
|
final String entityId = OafRowKeyDecoder.decode(resultId).getKey();
|
|
70
|
|
|
71
|
final Result.Builder result = Result.newBuilder();
|
|
72
|
Result.Metadata.Builder metadataProto = Result.Metadata.newBuilder();
|
|
73
|
|
|
74
|
// subject
|
|
75
|
for (int i = 0; i < subjects.getLength(); i++) {
|
|
76
|
Node currentNode = subjects.item(i);
|
|
77
|
NodeList childNodes = currentNode.getChildNodes();
|
|
78
|
if (childNodes.getLength() > 0) {
|
|
79
|
String subjectValue = childNodes.item(0).getNodeValue();
|
|
80
|
addField(metadataProto, Result.Metadata.getDescriptor().findFieldByName("subject"),
|
|
81
|
getStructuredProperty(subjectValue, "keyword", "keyword", "dnet:result_subject", "dnet:result_subject"));
|
|
82
|
}
|
|
83
|
}
|
|
84
|
|
|
85
|
// title
|
|
86
|
for (int i = 0; i < titles.getLength(); i++) {
|
|
87
|
Node currentNode = titles.item(i);
|
|
88
|
NodeList childNodes = currentNode.getChildNodes();
|
|
89
|
if (childNodes.getLength() > 0) {
|
|
90
|
String titleValue = childNodes.item(0).getNodeValue();
|
|
91
|
String classname = "main title";
|
|
92
|
String classid = "main title";
|
|
93
|
if (currentNode.hasAttributes()) {
|
|
94
|
NamedNodeMap attributes = currentNode.getAttributes();
|
|
95
|
Node titleType = attributes.getNamedItem("titleType");
|
|
96
|
|
|
97
|
if (titleType != null && titleType.getNodeValue().equals("AlternativeTitle")) {
|
|
98
|
classname = "alternative title";
|
|
99
|
classid = "alternative title";
|
|
100
|
}
|
|
101
|
if (titleType != null && titleType.getNodeValue().equals("Subtitle")) {
|
|
102
|
classname = "subtitle";
|
|
103
|
classid = "subtitle";
|
|
104
|
}
|
|
105
|
if (titleType != null && titleType.getNodeValue().equals("TranslatedTitle")) {
|
|
106
|
classname = "translated title";
|
|
107
|
classid = "translated title";
|
|
108
|
}
|
|
109
|
}
|
|
110
|
addField(metadataProto, Result.Metadata.getDescriptor().findFieldByName("title"),
|
|
111
|
getStructuredProperty(titleValue, classname, classid, "dnet:dataCite_title", "dnet:dataCite_title"));
|
|
112
|
}
|
|
113
|
}
|
|
114
|
|
|
115
|
// description
|
|
116
|
for (int i = 0; i < descriptions.getLength(); i++) {
|
|
117
|
Node currentNode = descriptions.item(i);
|
|
118
|
if (currentNode != null && currentNode.hasChildNodes()) {
|
|
119
|
String descriptionValue = currentNode.getChildNodes().item(0).getNodeValue();
|
|
120
|
addField(metadataProto, Result.Metadata.getDescriptor().findFieldByName("description"), descriptionValue);
|
|
121
|
}
|
|
122
|
}
|
|
123
|
|
|
124
|
// contributors
|
|
125
|
for (int i = 0; i < contributors.getLength(); i++) {
|
|
126
|
Node currentNode = contributors.item(i);
|
|
127
|
if (currentNode != null && currentNode.hasChildNodes()) {
|
|
128
|
String contributorValue = currentNode.getChildNodes().item(0).getNodeValue();
|
|
129
|
addField(metadataProto, Result.Metadata.getDescriptor().findFieldByName("contributor"), contributorValue);
|
|
130
|
}
|
|
131
|
}
|
|
132
|
|
|
133
|
// publisher
|
|
134
|
addField(metadataProto, Result.Metadata.getDescriptor().findFieldByName("publisher"), getFirstItem(publisher));
|
|
135
|
|
|
136
|
// dates
|
|
137
|
for (int i = 0; i < dates.getLength(); i++) {
|
|
138
|
Node currentNode = dates.item(i);
|
|
139
|
if (currentNode != null && currentNode.hasAttributes() && currentNode.hasChildNodes()) {
|
|
140
|
String dateAttribute = currentNode.getAttributes().getNamedItem("dateType").getNodeValue();
|
|
141
|
String dateValue = currentNode.getChildNodes().item(0).getNodeValue();
|
|
142
|
String protoAttribute = "relevantdate";
|
|
143
|
if ("Accepted".equals(dateAttribute)) {
|
|
144
|
protoAttribute = "dateofacceptance";
|
|
145
|
} else if ("Issued".equals(dateAttribute)) {
|
|
146
|
protoAttribute = "storagedate";
|
|
147
|
} else if ("Updated".equals(dateAttribute)) {
|
|
148
|
protoAttribute = "lastmetadataupdate";
|
|
149
|
} else if ("Available".equals(dateAttribute)) {
|
|
150
|
protoAttribute = "embargoenddate";
|
|
151
|
}
|
|
152
|
if (protoAttribute.equals("relevantdate") == false) {
|
|
153
|
addField(metadataProto, Result.Metadata.getDescriptor().findFieldByName(protoAttribute), dateValue);
|
|
154
|
} else {
|
|
155
|
addField(metadataProto, Result.Metadata.getDescriptor().findFieldByName(protoAttribute),
|
|
156
|
getStructuredProperty(dateValue, "UNKNOWN", "UNKNOWN", "dnet:dataCite_date", "dnet:dataCite_date"));
|
|
157
|
}
|
|
158
|
}
|
|
159
|
}
|
|
160
|
|
|
161
|
// dateofacceptance
|
|
162
|
addField(metadataProto, Result.Metadata.getDescriptor().findFieldByName("dateofacceptance"), getFirstItem(dateaccepted));
|
|
163
|
|
|
164
|
// size
|
|
165
|
addField(metadataProto, Result.Metadata.getDescriptor().findFieldByName("size"), getFirstItem(sizes));
|
|
166
|
|
|
167
|
// format
|
|
168
|
addField(metadataProto, Result.Metadata.getDescriptor().findFieldByName("format"), getFirstItem(formats));
|
|
169
|
|
|
170
|
// version
|
|
171
|
addField(metadataProto, Result.Metadata.getDescriptor().findFieldByName("version"), getFirstItem(version));
|
|
172
|
|
|
173
|
// language
|
|
174
|
addField(metadataProto, Result.Metadata.getDescriptor().findFieldByName("language"),
|
|
175
|
setQualifier(getDefaultQualifier("dnet:languages"), Lists.newArrayList(getFirstItem(languages))));
|
|
176
|
|
|
177
|
//resource type
|
|
178
|
addField(metadataProto, Result.Metadata.getDescriptor().findFieldByName("resourcetype"),
|
|
179
|
setQualifier(getDefaultQualifier("dnet:dataCite_resource"), Lists.newArrayList(getFirstItem(resourceTypes))));
|
|
180
|
|
|
181
|
// resultType
|
|
182
|
addField(metadataProto, Result.Metadata.getDescriptor().findFieldByName("resulttype"), getSimpleQualifier("dataset", "dnet:result_typologies"));
|
|
183
|
|
|
184
|
String tmpID;
|
|
185
|
String TmpName;
|
|
186
|
|
|
187
|
if (hostedbyId == null && hostedbyName == null) {
|
|
188
|
tmpID = collectedfromId;
|
|
189
|
TmpName = collectedfromName;
|
|
190
|
} else {
|
|
191
|
tmpID = hostedbyId;
|
|
192
|
TmpName = hostedbyName;
|
|
193
|
}
|
|
194
|
|
|
195
|
final Instance.Builder instance = Instance.newBuilder().setHostedby(getKV(tmpID, TmpName));
|
|
196
|
|
|
197
|
String tmpRigths = "UNKNOWN";
|
|
198
|
final String firstRight = getFirstItem(rights);
|
|
199
|
if (mappingAccess.containsKey(firstRight)) {
|
|
200
|
tmpRigths = mappingAccess.get(firstRight);
|
|
201
|
}
|
|
202
|
|
|
203
|
addField(instance, Instance.getDescriptor().findFieldByName("licence"),
|
|
204
|
setQualifier(getDefaultQualifier("dnet:access_modes"), Lists.newArrayList(tmpRigths)));
|
|
205
|
|
|
206
|
addField(instance, Instance.getDescriptor().findFieldByName("instancetype"),
|
|
207
|
setQualifier(getDefaultQualifier("dnet:dataCite_resource"), Lists.newArrayList(getFirstItem(cobjcategory))));
|
|
208
|
|
|
209
|
addField(instance, Instance.getDescriptor().findFieldByName("url"), instanceUri);
|
|
210
|
|
|
211
|
result.addInstance(instance);
|
|
212
|
|
|
213
|
List<StructuredProperty> pids = parsePids(pidList);
|
|
214
|
|
|
215
|
// original ids
|
|
216
|
final List<String> originalIdList = Lists.newArrayList();
|
|
217
|
for (int i = 0; i < originalIds.getLength(); i++) {
|
|
218
|
Node currentNode = originalIds.item(i);
|
|
219
|
if (currentNode != null && currentNode.hasChildNodes()) {
|
|
220
|
originalIdList.add(currentNode.getChildNodes().item(0).getNodeValue());
|
|
221
|
}
|
|
222
|
}
|
|
223
|
|
|
224
|
OafEntity.Builder entity =
|
|
225
|
getEntity(Type.result, entityId, getKV(collectedfromId, collectedfromName), originalIdList, dateOfCollection, dateOfTransformation, pids).setResult(
|
|
226
|
result.setMetadata(metadataProto));
|
|
227
|
|
|
228
|
entity.setOaiprovenance(getOAIProvenance(about));
|
|
229
|
|
|
230
|
Oaf oaf = getOaf(entity, getDataInfo(about, provenance, trust, false, false));
|
|
231
|
return base64(oaf.toByteArray());
|
|
232
|
} catch (Exception e) {
|
|
233
|
e.printStackTrace(System.err);
|
|
234
|
throw new RuntimeException(e);
|
|
235
|
}
|
|
236
|
|
|
237
|
}
|
|
238
|
|
|
239
|
public static String getFirstItem(final NodeList list) {
|
|
240
|
String out = "";
|
|
241
|
if (list != null) {
|
|
242
|
|
|
243
|
if (list.getLength() > 0 && list.item(0).getChildNodes() != null && list.item(0).getChildNodes().getLength() > 0) {
|
|
244
|
out = list.item(0).getChildNodes().item(0).getNodeValue();
|
|
245
|
}
|
|
246
|
}
|
|
247
|
return out;
|
|
248
|
}
|
|
249
|
|
|
250
|
}
|