| 1 |
26600
|
sandro.lab
|
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 |
28092
|
claudio.at
|
import eu.dnetlib.data.proto.FieldTypeProtos.StructuredProperty;
|
| 10 |
26600
|
sandro.lab
|
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 |
40014
|
claudio.at
|
import org.w3c.dom.NamedNodeMap;
|
| 16 |
|
|
import org.w3c.dom.Node;
|
| 17 |
|
|
import org.w3c.dom.NodeList;
|
| 18 |
26600
|
sandro.lab
|
|
| 19 |
40198
|
claudio.at
|
public class OdfToHbaseXsltFunctions extends CommonDNetXsltFunctions {
|
| 20 |
26600
|
sandro.lab
|
|
| 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 |
33325
|
claudio.at
|
// 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 |
26600
|
sandro.lab
|
}
|
| 37 |
|
|
|
| 38 |
40198
|
claudio.at
|
public static String odfResult(
|
| 39 |
|
|
final String resultId,
|
| 40 |
|
|
final NodeList about,
|
| 41 |
26600
|
sandro.lab
|
final NodeList metadata,
|
| 42 |
|
|
final NodeList titles,
|
| 43 |
|
|
final NodeList subjects,
|
| 44 |
|
|
final NodeList publisher,
|
| 45 |
|
|
final NodeList descriptions,
|
| 46 |
|
|
final NodeList dates,
|
| 47 |
30969
|
claudio.at
|
final NodeList dateaccepted,
|
| 48 |
26600
|
sandro.lab
|
final NodeList resourceTypes,
|
| 49 |
|
|
final NodeList formats,
|
| 50 |
|
|
final NodeList sizes,
|
| 51 |
|
|
final NodeList languages,
|
| 52 |
31200
|
claudio.at
|
final NodeList cobjcategory,
|
| 53 |
41526
|
claudio.at
|
final NodeList contributors,
|
| 54 |
26600
|
sandro.lab
|
final NodeList rights,
|
| 55 |
|
|
final NodeList version,
|
| 56 |
40014
|
claudio.at
|
final NodeList pidList,
|
| 57 |
26600
|
sandro.lab
|
final String provenance,
|
| 58 |
|
|
final String trust,
|
| 59 |
|
|
final String hostedbyId,
|
| 60 |
|
|
final String hostedbyName,
|
| 61 |
|
|
final String collectedfromId,
|
| 62 |
|
|
final String collectedfromName,
|
| 63 |
40014
|
claudio.at
|
final NodeList originalIds,
|
| 64 |
26600
|
sandro.lab
|
final String instanceUri,
|
| 65 |
40313
|
claudio.at
|
final String dateOfCollection,
|
| 66 |
|
|
final String dateOfTransformation) {
|
| 67 |
26600
|
sandro.lab
|
|
| 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 |
44552
|
claudio.at
|
getStructuredProperty(subjectValue, "keyword", "keyword", "dnet:subject_classification_typologies", "dnet:subject_classification_typologies"));
|
| 82 |
26600
|
sandro.lab
|
}
|
| 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 |
28092
|
claudio.at
|
|
| 97 |
33325
|
claudio.at
|
if (titleType != null && titleType.getNodeValue().equals("AlternativeTitle")) {
|
| 98 |
26600
|
sandro.lab
|
classname = "alternative title";
|
| 99 |
|
|
classid = "alternative title";
|
| 100 |
|
|
}
|
| 101 |
33325
|
claudio.at
|
if (titleType != null && titleType.getNodeValue().equals("Subtitle")) {
|
| 102 |
26600
|
sandro.lab
|
classname = "subtitle";
|
| 103 |
|
|
classid = "subtitle";
|
| 104 |
|
|
}
|
| 105 |
33325
|
claudio.at
|
if (titleType != null && titleType.getNodeValue().equals("TranslatedTitle")) {
|
| 106 |
26600
|
sandro.lab
|
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 |
33325
|
claudio.at
|
if (currentNode != null && currentNode.hasChildNodes()) {
|
| 119 |
26600
|
sandro.lab
|
String descriptionValue = currentNode.getChildNodes().item(0).getNodeValue();
|
| 120 |
|
|
addField(metadataProto, Result.Metadata.getDescriptor().findFieldByName("description"), descriptionValue);
|
| 121 |
|
|
}
|
| 122 |
|
|
}
|
| 123 |
|
|
|
| 124 |
41526
|
claudio.at
|
// 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 |
26600
|
sandro.lab
|
// 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 |
33325
|
claudio.at
|
if (currentNode != null && currentNode.hasAttributes() && currentNode.hasChildNodes()) {
|
| 140 |
26600
|
sandro.lab
|
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 |
30969
|
claudio.at
|
// dateofacceptance
|
| 162 |
|
|
addField(metadataProto, Result.Metadata.getDescriptor().findFieldByName("dateofacceptance"), getFirstItem(dateaccepted));
|
| 163 |
|
|
|
| 164 |
26600
|
sandro.lab
|
// 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 |
42500
|
alessia.ba
|
//resource type
|
| 178 |
|
|
addField(metadataProto, Result.Metadata.getDescriptor().findFieldByName("resourcetype"),
|
| 179 |
|
|
setQualifier(getDefaultQualifier("dnet:dataCite_resource"), Lists.newArrayList(getFirstItem(resourceTypes))));
|
| 180 |
|
|
|
| 181 |
26600
|
sandro.lab
|
// resultType
|
| 182 |
47480
|
claudio.at
|
final String cobjcategoryCode = getFirstItem(cobjcategory);
|
| 183 |
|
|
String resulttype = "";
|
| 184 |
|
|
switch (cobjcategoryCode) {
|
| 185 |
|
|
// add here the code to be excluded from the default mapping as 'dataset'
|
| 186 |
26600
|
sandro.lab
|
|
| 187 |
47480
|
claudio.at
|
/*
|
| 188 |
|
|
case "0029":
|
| 189 |
|
|
resulttype = "software";
|
| 190 |
|
|
break;
|
| 191 |
|
|
*/
|
| 192 |
|
|
default:
|
| 193 |
|
|
resulttype = "dataset";
|
| 194 |
|
|
}
|
| 195 |
|
|
|
| 196 |
|
|
addField(metadataProto, Result.Metadata.getDescriptor().findFieldByName("resulttype"), getSimpleQualifier(resulttype, "dnet:result_typologies"));
|
| 197 |
|
|
|
| 198 |
26600
|
sandro.lab
|
String tmpID;
|
| 199 |
|
|
String TmpName;
|
| 200 |
|
|
|
| 201 |
33325
|
claudio.at
|
if (hostedbyId == null && hostedbyName == null) {
|
| 202 |
26600
|
sandro.lab
|
tmpID = collectedfromId;
|
| 203 |
|
|
TmpName = collectedfromName;
|
| 204 |
|
|
} else {
|
| 205 |
|
|
tmpID = hostedbyId;
|
| 206 |
|
|
TmpName = hostedbyName;
|
| 207 |
|
|
}
|
| 208 |
|
|
|
| 209 |
|
|
final Instance.Builder instance = Instance.newBuilder().setHostedby(getKV(tmpID, TmpName));
|
| 210 |
|
|
|
| 211 |
|
|
String tmpRigths = "UNKNOWN";
|
| 212 |
|
|
final String firstRight = getFirstItem(rights);
|
| 213 |
|
|
if (mappingAccess.containsKey(firstRight)) {
|
| 214 |
|
|
tmpRigths = mappingAccess.get(firstRight);
|
| 215 |
|
|
}
|
| 216 |
|
|
|
| 217 |
|
|
addField(instance, Instance.getDescriptor().findFieldByName("licence"),
|
| 218 |
|
|
setQualifier(getDefaultQualifier("dnet:access_modes"), Lists.newArrayList(tmpRigths)));
|
| 219 |
|
|
|
| 220 |
|
|
addField(instance, Instance.getDescriptor().findFieldByName("instancetype"),
|
| 221 |
47480
|
claudio.at
|
setQualifier(getDefaultQualifier("dnet:dataCite_resource"), Lists.newArrayList(cobjcategoryCode)));
|
| 222 |
31200
|
claudio.at
|
|
| 223 |
26600
|
sandro.lab
|
addField(instance, Instance.getDescriptor().findFieldByName("url"), instanceUri);
|
| 224 |
|
|
|
| 225 |
|
|
result.addInstance(instance);
|
| 226 |
|
|
|
| 227 |
40014
|
claudio.at
|
List<StructuredProperty> pids = parsePids(pidList);
|
| 228 |
|
|
|
| 229 |
|
|
// original ids
|
| 230 |
|
|
final List<String> originalIdList = Lists.newArrayList();
|
| 231 |
|
|
for (int i = 0; i < originalIds.getLength(); i++) {
|
| 232 |
|
|
Node currentNode = originalIds.item(i);
|
| 233 |
|
|
if (currentNode != null && currentNode.hasChildNodes()) {
|
| 234 |
|
|
originalIdList.add(currentNode.getChildNodes().item(0).getNodeValue());
|
| 235 |
|
|
}
|
| 236 |
|
|
}
|
| 237 |
|
|
|
| 238 |
30969
|
claudio.at
|
OafEntity.Builder entity =
|
| 239 |
40313
|
claudio.at
|
getEntity(Type.result, entityId, getKV(collectedfromId, collectedfromName), originalIdList, dateOfCollection, dateOfTransformation, pids).setResult(
|
| 240 |
30969
|
claudio.at
|
result.setMetadata(metadataProto));
|
| 241 |
40198
|
claudio.at
|
|
| 242 |
|
|
entity.setOaiprovenance(getOAIProvenance(about));
|
| 243 |
|
|
|
| 244 |
|
|
Oaf oaf = getOaf(entity, getDataInfo(about, provenance, trust, false, false));
|
| 245 |
26600
|
sandro.lab
|
return base64(oaf.toByteArray());
|
| 246 |
|
|
} catch (Exception e) {
|
| 247 |
|
|
e.printStackTrace(System.err);
|
| 248 |
|
|
throw new RuntimeException(e);
|
| 249 |
|
|
}
|
| 250 |
|
|
|
| 251 |
|
|
}
|
| 252 |
|
|
|
| 253 |
|
|
public static String getFirstItem(final NodeList list) {
|
| 254 |
|
|
String out = "";
|
| 255 |
|
|
if (list != null) {
|
| 256 |
|
|
|
| 257 |
33325
|
claudio.at
|
if (list.getLength() > 0 && list.item(0).getChildNodes() != null && list.item(0).getChildNodes().getLength() > 0) {
|
| 258 |
26600
|
sandro.lab
|
out = list.item(0).getChildNodes().item(0).getNodeValue();
|
| 259 |
|
|
}
|
| 260 |
|
|
}
|
| 261 |
|
|
return out;
|
| 262 |
|
|
}
|
| 263 |
|
|
|
| 264 |
|
|
}
|