| 1 |
52209
|
claudio.at
|
package eu.dnetlib.data.transform.xml2;
|
| 2 |
51942
|
claudio.at
|
|
| 3 |
|
|
import java.util.List;
|
| 4 |
|
|
import java.util.Map;
|
| 5 |
52041
|
claudio.at
|
import java.util.Map.Entry;
|
| 6 |
52048
|
claudio.at
|
import java.util.Objects;
|
| 7 |
52041
|
claudio.at
|
import java.util.function.Function;
|
| 8 |
51942
|
claudio.at
|
|
| 9 |
52041
|
claudio.at
|
import com.google.common.collect.Streams;
|
| 10 |
|
|
import com.google.protobuf.Descriptors.Descriptor;
|
| 11 |
51942
|
claudio.at
|
import com.ximpleware.AutoPilot;
|
| 12 |
52041
|
claudio.at
|
import com.ximpleware.VTDGen;
|
| 13 |
51942
|
claudio.at
|
import com.ximpleware.VTDNav;
|
| 14 |
52041
|
claudio.at
|
import eu.dnetlib.data.proto.FieldTypeProtos;
|
| 15 |
51953
|
claudio.at
|
import eu.dnetlib.data.proto.FieldTypeProtos.*;
|
| 16 |
51942
|
claudio.at
|
import eu.dnetlib.data.proto.FieldTypeProtos.OAIProvenance.OriginDescription;
|
| 17 |
52041
|
claudio.at
|
import eu.dnetlib.data.proto.KindProtos.Kind;
|
| 18 |
51953
|
claudio.at
|
import eu.dnetlib.data.proto.OafProtos.Oaf;
|
| 19 |
|
|
import eu.dnetlib.data.proto.OafProtos.OafEntity;
|
| 20 |
52041
|
claudio.at
|
import eu.dnetlib.data.proto.OafProtos.OafRel;
|
| 21 |
|
|
import eu.dnetlib.data.proto.RelTypeProtos.RelType;
|
| 22 |
|
|
import eu.dnetlib.data.proto.RelTypeProtos.SubRelType;
|
| 23 |
|
|
import eu.dnetlib.data.proto.ResultProtos.Result;
|
| 24 |
|
|
import eu.dnetlib.data.proto.ResultProtos.Result.*;
|
| 25 |
|
|
import eu.dnetlib.data.proto.TypeProtos.Type;
|
| 26 |
51944
|
sandro.lab
|
import eu.dnetlib.miscutils.collections.Pair;
|
| 27 |
52041
|
claudio.at
|
import eu.dnetlib.pace.model.Person;
|
| 28 |
51945
|
sandro.lab
|
import org.apache.commons.lang3.StringUtils;
|
| 29 |
52041
|
claudio.at
|
import org.apache.commons.lang3.exception.ExceptionUtils;
|
| 30 |
|
|
import org.apache.commons.logging.Log;
|
| 31 |
|
|
import org.apache.commons.logging.LogFactory;
|
| 32 |
51942
|
claudio.at
|
|
| 33 |
52041
|
claudio.at
|
import static eu.dnetlib.data.transform.xml.AbstractDNetXsltFunctions.oafSimpleId;
|
| 34 |
|
|
import static eu.dnetlib.data.transform.xml.AbstractDNetXsltFunctions.oafSplitId;
|
| 35 |
52209
|
claudio.at
|
import static eu.dnetlib.data.transform.xml2.Utils.*;
|
| 36 |
|
|
import static eu.dnetlib.data.transform.xml2.VtdUtilityParser.*;
|
| 37 |
52041
|
claudio.at
|
import static java.lang.String.format;
|
| 38 |
51942
|
claudio.at
|
|
| 39 |
52048
|
claudio.at
|
public abstract class AbstractResultVtdParser implements Function<String, Oaf> {
|
| 40 |
51942
|
claudio.at
|
|
| 41 |
52048
|
claudio.at
|
private static final Log log = LogFactory.getLog(AbstractResultVtdParser.class);
|
| 42 |
52041
|
claudio.at
|
|
| 43 |
51953
|
claudio.at
|
protected boolean invisible = false;
|
| 44 |
52041
|
claudio.at
|
protected String provenance = "";
|
| 45 |
51953
|
claudio.at
|
protected String trust = "0.9";
|
| 46 |
|
|
|
| 47 |
52209
|
claudio.at
|
protected SpecificationMap specs;
|
| 48 |
51974
|
claudio.at
|
|
| 49 |
52209
|
claudio.at
|
public AbstractResultVtdParser(final Map<String, String> fields) {
|
| 50 |
|
|
this.specs = buildSpecs(fields);
|
| 51 |
|
|
}
|
| 52 |
|
|
|
| 53 |
|
|
public AbstractResultVtdParser(final boolean invisible, final String provenance, final String trust, final Map<String, String> fields) {
|
| 54 |
|
|
this(fields);
|
| 55 |
51953
|
claudio.at
|
this.invisible = invisible;
|
| 56 |
|
|
this.provenance = provenance;
|
| 57 |
|
|
this.trust = trust;
|
| 58 |
|
|
}
|
| 59 |
|
|
|
| 60 |
52209
|
claudio.at
|
protected abstract String getResulttype(final String cobjcategory);
|
| 61 |
|
|
|
| 62 |
52041
|
claudio.at
|
@Override
|
| 63 |
|
|
public Oaf apply(final String xml) {
|
| 64 |
|
|
try {
|
| 65 |
|
|
final VTDGen vg = parseXml(xml);
|
| 66 |
|
|
final VTDNav vn = vg.getNav();
|
| 67 |
|
|
final AutoPilot ap = new AutoPilot(vn);
|
| 68 |
51942
|
claudio.at
|
|
| 69 |
52041
|
claudio.at
|
final boolean skiprecord = Boolean.valueOf(getFirstValue(ap, vn, xpath("record", "header", "skipRecord")));
|
| 70 |
|
|
int metadata = countNodes(ap, vn, format("count(%s)", xpath("record", "metadata")));
|
| 71 |
51942
|
claudio.at
|
|
| 72 |
52041
|
claudio.at
|
if (metadata == 0 || skiprecord) {
|
| 73 |
|
|
return null;
|
| 74 |
|
|
}
|
| 75 |
51942
|
claudio.at
|
|
| 76 |
52041
|
claudio.at
|
final String objIdentifier = oafSimpleId(Type.result.name(), getFirstValue(ap, vn, xpath("record", "header", "objIdentifier")));
|
| 77 |
|
|
if (StringUtils.isBlank(objIdentifier)) {
|
| 78 |
|
|
return null;
|
| 79 |
|
|
}
|
| 80 |
51942
|
claudio.at
|
|
| 81 |
52209
|
claudio.at
|
for(final Entry<Descriptor, SpecificationDescriptor> spec : specs.entrySet()) {
|
| 82 |
|
|
final Descriptor d = spec.getKey();
|
| 83 |
|
|
final SpecificationDescriptor md = spec.getValue();
|
| 84 |
|
|
|
| 85 |
|
|
for(Entry<String, Pair<String, Function<List<Node>, Object>>> entry : md.getFields().entrySet()) {
|
| 86 |
|
|
final String fieldName = entry.getKey();
|
| 87 |
|
|
final Pair<String, Function<List<Node>, Object>> pair = entry.getValue();
|
| 88 |
|
|
final String xpath = pair.getKey();
|
| 89 |
|
|
final Function<List<Node>, Object> function = pair.getValue();
|
| 90 |
|
|
try {
|
| 91 |
|
|
addField(md.getBuilder(), d.findFieldByName(fieldName), function.apply(getNodes(ap, vn, xpath)));
|
| 92 |
|
|
} catch (Throwable e) {
|
| 93 |
|
|
throw new VtdException(String.format("Error mapping field '%s' from xpath '%s' for record '%s'", fieldName, xpath, objIdentifier), e);
|
| 94 |
|
|
}
|
| 95 |
|
|
}
|
| 96 |
|
|
}
|
| 97 |
|
|
|
| 98 |
|
|
return Oaf.newBuilder()
|
| 99 |
|
|
.setKind(Kind.entity)
|
| 100 |
|
|
.setDataInfo(ensureDataInfo(ap, vn, DataInfo.newBuilder()))
|
| 101 |
|
|
.setEntity(((OafEntity.Builder) specs.get(OafEntity.getDescriptor())
|
| 102 |
|
|
.getBuilder()
|
| 103 |
|
|
.setField(
|
| 104 |
|
|
OafEntity.getDescriptor().findFieldByName(Type.result.name()),
|
| 105 |
|
|
((Result.Builder) specs.get(Result.getDescriptor()).getBuilder())
|
| 106 |
|
|
.setMetadata((Metadata) specs.get(Metadata.getDescriptor()).getBuilder().build())
|
| 107 |
|
|
.addInstance((Instance) specs.get(Instance.getDescriptor()).getBuilder().build())
|
| 108 |
|
|
.build()))
|
| 109 |
|
|
.setId(objIdentifier)
|
| 110 |
|
|
.setOaiprovenance(getOaiProvenance(ap, vn))
|
| 111 |
|
|
.build())
|
| 112 |
|
|
.build();
|
| 113 |
52041
|
claudio.at
|
} catch (Throwable e) {
|
| 114 |
52062
|
claudio.at
|
log.error(xml);
|
| 115 |
52041
|
claudio.at
|
log.error(ExceptionUtils.getStackTrace(e));
|
| 116 |
|
|
return null;
|
| 117 |
|
|
}
|
| 118 |
51942
|
claudio.at
|
}
|
| 119 |
|
|
|
| 120 |
52209
|
claudio.at
|
public SpecificationMap buildSpecs(final Map<String, String> fields) {
|
| 121 |
|
|
final SpecificationMap specs = new SpecificationMap();
|
| 122 |
52041
|
claudio.at
|
|
| 123 |
|
|
specs.put(Result.getDescriptor(), SpecificationDescriptor.newInstance())
|
| 124 |
|
|
.setBuilder(Result.newBuilder())
|
| 125 |
|
|
.put("externalReference", fields.get("externalReference"), nodes -> nodes.stream()
|
| 126 |
52209
|
claudio.at
|
.map(node -> {
|
| 127 |
|
|
final ExternalReference.Builder extref = ExternalReference.newBuilder();
|
| 128 |
|
|
if (StringUtils.isNotBlank(node.getTextValue())) {
|
| 129 |
|
|
extref.setUrl(node.getTextValue());
|
| 130 |
|
|
}
|
| 131 |
|
|
final Map<String, String> a = node.getAttributes();
|
| 132 |
|
|
final String source = a.get("source");
|
| 133 |
|
|
if (StringUtils.isNotBlank(source)) {
|
| 134 |
|
|
extref.setSitename(source);
|
| 135 |
|
|
}
|
| 136 |
|
|
final String identifier = a.get("identifier");
|
| 137 |
|
|
if (StringUtils.isNotBlank(identifier)) {
|
| 138 |
|
|
extref.setRefidentifier(identifier);
|
| 139 |
|
|
}
|
| 140 |
|
|
final String title = a.get("title");
|
| 141 |
|
|
if (StringUtils.isNotBlank(title)) {
|
| 142 |
|
|
extref.setLabel(title);
|
| 143 |
|
|
}
|
| 144 |
|
|
final String query = a.get("query");
|
| 145 |
|
|
if (StringUtils.isNotBlank(query)) {
|
| 146 |
|
|
extref.setQuery(query);
|
| 147 |
|
|
}
|
| 148 |
|
|
final String type = a.get("type");
|
| 149 |
|
|
if (StringUtils.isNotBlank(type)) {
|
| 150 |
|
|
extref.setQualifier(getSimpleQualifier(type, DNET_EXT_REF_TYPOLOGIES));
|
| 151 |
|
|
}
|
| 152 |
|
|
return extref.build();
|
| 153 |
|
|
}));
|
| 154 |
52041
|
claudio.at
|
|
| 155 |
|
|
specs.put(Instance.getDescriptor(), SpecificationDescriptor.newInstance())
|
| 156 |
|
|
.setBuilder(Instance.newBuilder())
|
| 157 |
|
|
.put("license", fields.get("license"), nodes -> nodes.stream()
|
| 158 |
52053
|
claudio.at
|
.filter(node -> {
|
| 159 |
|
|
final Map<String, String> a = node.getAttributes();
|
| 160 |
|
|
switch (node.getName()) {
|
| 161 |
|
|
case "rights":
|
| 162 |
|
|
return a.containsKey(RIGHTS_URI) && a.get(RIGHTS_URI).matches(URL_REGEX);
|
| 163 |
|
|
case "license":
|
| 164 |
|
|
return true;
|
| 165 |
|
|
default:
|
| 166 |
|
|
return false;
|
| 167 |
|
|
}
|
| 168 |
|
|
})
|
| 169 |
52041
|
claudio.at
|
.map(Node::getTextValue))
|
| 170 |
|
|
.put("accessright", fields.get("accessright"), nodes -> nodes.stream()
|
| 171 |
|
|
.map(Node::getTextValue)
|
| 172 |
|
|
.map(rights -> mappingAccess.containsKey(rights) ? mappingAccess.get(rights) : "UNKNOWN")
|
| 173 |
|
|
.map(code -> getQualifier(code, getClassName(code), DNET_ACCESS_MODES, DNET_ACCESS_MODES)))
|
| 174 |
|
|
.put("instancetype", fields.get("instancetype"), nodes -> nodes.stream()
|
| 175 |
|
|
.map(Node::getTextValue)
|
| 176 |
|
|
.map(code -> getQualifier(code, getClassName(code), DNET_PUBLICATION_RESOURCE, DNET_PUBLICATION_RESOURCE)))
|
| 177 |
|
|
.put("hostedby", fields.get("hostedby"), nodes -> nodes.stream()
|
| 178 |
|
|
.map(node -> getKV(oafSplitId("datasource", node.getAttributes().get("id")), node.getAttributes().get("name"))))
|
| 179 |
|
|
.put("url", fields.get("url"), nodes -> nodes.stream()
|
| 180 |
|
|
.map(Node::getTextValue)
|
| 181 |
|
|
.filter(s -> s.trim().matches(URL_REGEX)))
|
| 182 |
|
|
.put("dateofacceptance", fields.get("dateofacceptance"), nodes -> nodes.stream()
|
| 183 |
|
|
.map(Node::getTextValue));
|
| 184 |
|
|
|
| 185 |
|
|
specs.put(Metadata.getDescriptor(), SpecificationDescriptor.newInstance())
|
| 186 |
|
|
.setBuilder(Metadata.newBuilder())
|
| 187 |
|
|
.put("title", fields.get("title"), nodes -> nodes.stream()
|
| 188 |
|
|
.map(node -> {
|
| 189 |
|
|
final Qualifier.Builder q = Qualifier.newBuilder().setSchemeid(DNET_TITLE_TYPOLOGIES).setSchemename(DNET_TITLE_TYPOLOGIES);
|
| 190 |
|
|
switch (node.getAttributes().get(TITLE_TYPE) + "") {
|
| 191 |
|
|
case "AlternativeTitle":
|
| 192 |
|
|
q.setClassid("alternative title").setClassname("alternative title");
|
| 193 |
|
|
break;
|
| 194 |
|
|
case "Subtitle":
|
| 195 |
|
|
q.setClassid("subtitle").setClassname("subtitle");
|
| 196 |
|
|
break;
|
| 197 |
|
|
case "TranslatedTitle":
|
| 198 |
|
|
q.setClassid("translated title").setClassname("translated title");
|
| 199 |
|
|
break;
|
| 200 |
|
|
default:
|
| 201 |
|
|
q.setClassid("main title").setClassname("main title");
|
| 202 |
|
|
break;
|
| 203 |
|
|
}
|
| 204 |
|
|
return StructuredProperty.newBuilder().setValue(node.getTextValue()).setQualifier(q).build();
|
| 205 |
|
|
}))
|
| 206 |
|
|
.put("description", fields.get("description"), nodes -> nodes.stream()
|
| 207 |
|
|
.map(Node::getTextValue))
|
| 208 |
|
|
.put("storagedate", fields.get("storagedate"), nodes -> nodes.stream()
|
| 209 |
|
|
.map(Node::getTextValue))
|
| 210 |
|
|
.put("lastmetadataupdate", fields.get("lastmetadataupdate"), nodes -> nodes.stream()
|
| 211 |
|
|
.map(Node::getTextValue))
|
| 212 |
|
|
.put("embargoenddate", fields.get("embargoenddate"), nodes -> nodes.stream()
|
| 213 |
|
|
.map(Node::getTextValue))
|
| 214 |
|
|
.put("dateofacceptance", fields.get("dateofacceptance"), nodes -> nodes.stream()
|
| 215 |
|
|
.map(Node::getTextValue))
|
| 216 |
|
|
.put("author", fields.get("author"), nodes -> Streams.mapWithIndex(
|
| 217 |
|
|
nodes.stream()
|
| 218 |
|
|
.map(Node::getTextValue),
|
| 219 |
|
|
(creator, i) -> new Pair<>(i, creator))
|
| 220 |
|
|
.map(pair -> {
|
| 221 |
|
|
final Author.Builder author = Author.newBuilder();
|
| 222 |
|
|
author.setFullname(pair.getValue());
|
| 223 |
|
|
author.setRank(pair.getKey().intValue() + 1);
|
| 224 |
|
|
final Person p = new Person(pair.getValue(), false);
|
| 225 |
|
|
if (p.isAccurate()) {
|
| 226 |
|
|
author.setName(p.getNormalisedFirstName());
|
| 227 |
|
|
author.setSurname(p.getNormalisedSurname());
|
| 228 |
|
|
}
|
| 229 |
|
|
return author.build();
|
| 230 |
|
|
}))
|
| 231 |
|
|
.put("contributor", fields.get("contributor"), nodes -> nodes.stream()
|
| 232 |
|
|
.map(Node::getTextValue))
|
| 233 |
|
|
.put("subject", fields.get("subject"), nodes -> nodes.stream()
|
| 234 |
|
|
.map(node -> {
|
| 235 |
|
|
final Map<String, String> a = node.getAttributes();
|
| 236 |
|
|
final String classId = StringUtils.isNotBlank(a.get(CLASSID)) ? a.get(CLASSID) : KEYWORD;
|
| 237 |
|
|
final String className = StringUtils.isNotBlank(a.get(CLASSNAME)) ? a.get(CLASSNAME) : KEYWORD;
|
| 238 |
|
|
final String schemeId = StringUtils.isNotBlank(a.get(SCHEMEID)) ? a.get(SCHEMEID) : DNET_SUBJECT_TYPOLOGIES;
|
| 239 |
|
|
final String schemeName = StringUtils.isNotBlank(a.get(SCHEMENAME)) ? a.get(SCHEMENAME) : DNET_SUBJECT_TYPOLOGIES;
|
| 240 |
|
|
return getStructuredProperty(node.getTextValue(), classId, className, schemeId, schemeName);
|
| 241 |
|
|
}))
|
| 242 |
|
|
.put("format", fields.get("format"), nodes -> nodes.stream()
|
| 243 |
|
|
.map(Node::getTextValue))
|
| 244 |
|
|
.put("source", fields.get("source"), nodes -> nodes.stream()
|
| 245 |
|
|
.map(Node::getTextValue))
|
| 246 |
|
|
.put("size", fields.get("size"), nodes -> nodes.stream()
|
| 247 |
|
|
.map(Node::getTextValue))
|
| 248 |
|
|
.put("version", fields.get("version"), nodes -> nodes.stream()
|
| 249 |
|
|
.map(Node::getTextValue))
|
| 250 |
|
|
.put("publisher", fields.get("publisher"), nodes -> nodes.stream()
|
| 251 |
|
|
.map(Node::getTextValue))
|
| 252 |
|
|
.put("language", fields.get("language"), nodes -> nodes.stream()
|
| 253 |
|
|
.map(Node::getTextValue)
|
| 254 |
|
|
.map(code -> getQualifier(code, getClassName(code), DNET_LANGUAGES, DNET_LANGUAGES)))
|
| 255 |
|
|
.put("resourcetype", fields.get("resourcetype"), nodes -> nodes.stream()
|
| 256 |
|
|
.map(node -> node.getAttributes().get("resourceTypeGeneral"))
|
| 257 |
|
|
.map(resourceType -> getSimpleQualifier(resourceType, DNET_DATA_CITE_RESOURCE)))
|
| 258 |
|
|
.put("resulttype", fields.get("resulttype"), nodes -> nodes.stream()
|
| 259 |
|
|
.map(Node::getTextValue)
|
| 260 |
|
|
.map(cobjcategory -> getSimpleQualifier(getResulttype(cobjcategory), DNET_RESULT_TYPOLOGIES)))
|
| 261 |
|
|
.put("concept", fields.get("concept"), nodes -> nodes.stream()
|
| 262 |
52209
|
claudio.at
|
.filter(node -> node.getAttributes() != null && StringUtils.isNotBlank(node.getAttributes().get("id")))
|
| 263 |
|
|
.map(node -> Context.newBuilder().setId(node.getAttributes().get("id"))))
|
| 264 |
52041
|
claudio.at
|
.put("journal", fields.get("journal"), nodes -> nodes.stream()
|
| 265 |
52209
|
claudio.at
|
.map(node -> {
|
| 266 |
|
|
final Journal.Builder journal = Journal.newBuilder();
|
| 267 |
|
|
if (StringUtils.isNotBlank(node.getTextValue())) {
|
| 268 |
|
|
journal.setName(node.getTextValue());
|
| 269 |
52041
|
claudio.at
|
}
|
| 270 |
52209
|
claudio.at
|
if (node.getAttributes() != null) {
|
| 271 |
|
|
final Map<String, String> a = node.getAttributes();
|
| 272 |
|
|
if (StringUtils.isNotBlank(a.get("issn"))) {
|
| 273 |
|
|
journal.setIssnPrinted(a.get("issn"));
|
| 274 |
|
|
}
|
| 275 |
|
|
if (StringUtils.isNotBlank(a.get("eissn"))) {
|
| 276 |
|
|
journal.setIssnOnline(a.get("eissn"));
|
| 277 |
|
|
}
|
| 278 |
|
|
if (StringUtils.isNotBlank(a.get("lissn"))) {
|
| 279 |
|
|
journal.setIssnLinking(a.get("lissn"));
|
| 280 |
|
|
}
|
| 281 |
|
|
if (StringUtils.isNotBlank(a.get("sp"))) {
|
| 282 |
|
|
journal.setSp(a.get("sp"));
|
| 283 |
|
|
}
|
| 284 |
|
|
if (StringUtils.isNotBlank(a.get("ep"))) {
|
| 285 |
|
|
journal.setEp(a.get("ep"));
|
| 286 |
|
|
}
|
| 287 |
|
|
if (StringUtils.isNotBlank(a.get("iss"))) {
|
| 288 |
|
|
journal.setIss(a.get("iss"));
|
| 289 |
|
|
}
|
| 290 |
|
|
if (StringUtils.isNotBlank(a.get("vol"))) {
|
| 291 |
|
|
journal.setVol(a.get("vol"));
|
| 292 |
|
|
}
|
| 293 |
52041
|
claudio.at
|
}
|
| 294 |
52209
|
claudio.at
|
return journal;
|
| 295 |
|
|
}));
|
| 296 |
52041
|
claudio.at
|
|
| 297 |
|
|
specs.put(OafEntity.getDescriptor(), SpecificationDescriptor.newInstance())
|
| 298 |
52209
|
claudio.at
|
.setBuilder(OafEntity.newBuilder().setType(Type.result))
|
| 299 |
52041
|
claudio.at
|
.put("originalId", fields.get("originalId"), nodes -> nodes.stream()
|
| 300 |
|
|
.map(Node::getTextValue)
|
| 301 |
|
|
.map(s -> StringUtils.contains(s, ID_SEPARATOR) ? StringUtils.substringAfter(s, ID_SEPARATOR) : s)
|
| 302 |
|
|
.filter(s -> !s.trim().matches(URL_REGEX)))
|
| 303 |
|
|
.put("collectedfrom", fields.get("collectedfrom"), nodes -> nodes.stream()
|
| 304 |
|
|
.map(node -> getKV(
|
| 305 |
|
|
oafSplitId(Type.datasource.name(), node.getAttributes().get("id")),
|
| 306 |
|
|
node.getAttributes().get("name"))))
|
| 307 |
|
|
.put("pid", fields.get("pid"), nodes -> nodes.stream()
|
| 308 |
|
|
.filter(pid -> {
|
| 309 |
|
|
final Map<String, String> a = pid.getAttributes();
|
| 310 |
|
|
return a.containsKey(IDENTIFIER_TYPE) || a.containsKey(ALTERNATE_IDENTIFIER_TYPE);
|
| 311 |
|
|
})
|
| 312 |
|
|
.filter(pid -> {
|
| 313 |
|
|
final Map<String, String> a = pid.getAttributes();
|
| 314 |
|
|
return !"url".equalsIgnoreCase(a.get(IDENTIFIER_TYPE)) && !"url".equalsIgnoreCase(a.get(ALTERNATE_IDENTIFIER_TYPE));
|
| 315 |
|
|
})
|
| 316 |
|
|
.map(pid -> {
|
| 317 |
|
|
final Map<String, String> a = pid.getAttributes();
|
| 318 |
|
|
final String identifierType = a.get(IDENTIFIER_TYPE);
|
| 319 |
|
|
final String altIdentifierType = a.get(ALTERNATE_IDENTIFIER_TYPE);
|
| 320 |
|
|
return StructuredProperty.newBuilder()
|
| 321 |
|
|
.setValue(pid.getTextValue())
|
| 322 |
|
|
.setQualifier(getSimpleQualifier(
|
| 323 |
|
|
StringUtils.isNotBlank(identifierType) ?
|
| 324 |
|
|
identifierType : altIdentifierType, DNET_PID_TYPES))
|
| 325 |
|
|
.build();
|
| 326 |
|
|
}))
|
| 327 |
|
|
.put("dateofcollection", fields.get("dateofcollection"), nodes -> nodes.stream()
|
| 328 |
|
|
.map(Node::getTextValue))
|
| 329 |
|
|
.put("dateoftransformation", fields.get("dateoftransformation"), nodes -> nodes.stream()
|
| 330 |
|
|
.map(Node::getTextValue))
|
| 331 |
|
|
.put("cachedRel", fields.get("cachedRel"), nodes -> nodes.stream()
|
| 332 |
52209
|
claudio.at
|
.map(node -> getOafRel(node,
|
| 333 |
52048
|
claudio.at
|
OafRel.newBuilder()
|
| 334 |
52209
|
claudio.at
|
.setSource("")
|
| 335 |
|
|
.setChild(false)))
|
| 336 |
52048
|
claudio.at
|
.filter(Objects::nonNull)
|
| 337 |
|
|
.map(oafRel -> oafRel.build()));
|
| 338 |
52209
|
claudio.at
|
return specs;
|
| 339 |
52041
|
claudio.at
|
}
|
| 340 |
|
|
|
| 341 |
52209
|
claudio.at
|
private static OafRel.Builder getOafRel(final Node node, final OafRel.Builder oafRel) {
|
| 342 |
52048
|
claudio.at
|
final Map<String, String> a = node.getAttributes();
|
| 343 |
|
|
|
| 344 |
52062
|
claudio.at
|
switch (node.getName()) {
|
| 345 |
52048
|
claudio.at
|
case PROJECTID:
|
| 346 |
52062
|
claudio.at
|
if (StringUtils.isBlank(node.getTextValue())) {
|
| 347 |
|
|
return null;
|
| 348 |
|
|
}
|
| 349 |
52048
|
claudio.at
|
return oafRel
|
| 350 |
|
|
.setTarget(oafSplitId(Type.project.name(), StringUtils.trim(node.getTextValue())))
|
| 351 |
|
|
.setRelType(RelType.resultProject)
|
| 352 |
|
|
.setSubRelType(SubRelType.outcome)
|
| 353 |
|
|
.setRelClass("isProducedBy");
|
| 354 |
|
|
|
| 355 |
|
|
case RELATED_PUBLICATION:
|
| 356 |
|
|
case RELATED_DATASET:
|
| 357 |
52062
|
claudio.at
|
if (StringUtils.isBlank(a.get("id"))) {
|
| 358 |
|
|
return null;
|
| 359 |
|
|
}
|
| 360 |
52048
|
claudio.at
|
return oafRel
|
| 361 |
|
|
.setTarget(oafSimpleId(Type.result.name(), StringUtils.trim(a.get("id"))))
|
| 362 |
|
|
.setRelType(RelType.resultResult)
|
| 363 |
|
|
.setSubRelType(SubRelType.publicationDataset)
|
| 364 |
|
|
.setRelClass("isRelatedTo");
|
| 365 |
|
|
|
| 366 |
|
|
case RELATED_IDENTIFIER:
|
| 367 |
52062
|
claudio.at
|
if (StringUtils.isBlank(node.getTextValue())) {
|
| 368 |
|
|
return null;
|
| 369 |
|
|
}
|
| 370 |
52048
|
claudio.at
|
return oafRel
|
| 371 |
|
|
.setTarget(node.getTextValue())
|
| 372 |
|
|
.setRelType(RelType.resultResult)
|
| 373 |
|
|
.setSubRelType(SubRelType.relationship)
|
| 374 |
|
|
.setRelClass(a.get(RELATION_TYPE))
|
| 375 |
|
|
.setCachedTarget(
|
| 376 |
|
|
OafEntity.newBuilder()
|
| 377 |
|
|
.setType(Type.result)
|
| 378 |
52209
|
claudio.at
|
.setId("") //TODO
|
| 379 |
52048
|
claudio.at
|
.addPid(
|
| 380 |
|
|
StructuredProperty.newBuilder()
|
| 381 |
|
|
.setValue(node.getTextValue())
|
| 382 |
|
|
.setQualifier(getSimpleQualifier(a.get(RELATED_IDENTIFIER_TYPE), DNET_PID_TYPES))
|
| 383 |
|
|
.build()));
|
| 384 |
|
|
default:
|
| 385 |
|
|
return null;
|
| 386 |
|
|
}
|
| 387 |
|
|
}
|
| 388 |
|
|
|
| 389 |
52041
|
claudio.at
|
private OriginDescription getOriginDescription(final AutoPilot ap, final VTDNav vn, final String basePath) throws VtdException {
|
| 390 |
51942
|
claudio.at
|
final OriginDescription.Builder od = OriginDescription.newBuilder();
|
| 391 |
|
|
if (getNodes(ap, vn, basePath).isEmpty()) {
|
| 392 |
|
|
return od.build();
|
| 393 |
|
|
}
|
| 394 |
51974
|
claudio.at
|
final Map<String, String> odAttr = getNode(ap, vn, basePath).getAttributes();
|
| 395 |
|
|
|
| 396 |
|
|
final String harvestDate = odAttr.get("harvestDate");
|
| 397 |
|
|
if (StringUtils.isNotBlank(harvestDate)) {
|
| 398 |
|
|
od.setHarvestDate(harvestDate);
|
| 399 |
|
|
}
|
| 400 |
|
|
final String altered = odAttr.get("altered");
|
| 401 |
|
|
if (StringUtils.isNotBlank(altered)) {
|
| 402 |
|
|
od.setAltered(Boolean.valueOf(altered));
|
| 403 |
|
|
}
|
| 404 |
|
|
final String baseUrl = getFirstValue(ap, vn, basePath + xpath("baseURL"));
|
| 405 |
|
|
if (StringUtils.isNotBlank(basePath)) {
|
| 406 |
|
|
od.setBaseURL(baseUrl);
|
| 407 |
|
|
}
|
| 408 |
|
|
final String identifier = getFirstValue(ap, vn, basePath + xpath("identifier"));
|
| 409 |
|
|
if (StringUtils.isNotBlank(identifier)) {
|
| 410 |
|
|
od.setIdentifier(identifier);
|
| 411 |
|
|
}
|
| 412 |
|
|
final String datestamp = getFirstValue(ap, vn, basePath + xpath("datestamp"));
|
| 413 |
|
|
if (StringUtils.isNotBlank(datestamp)) {
|
| 414 |
|
|
od.setDatestamp(datestamp);
|
| 415 |
|
|
}
|
| 416 |
|
|
final String metadataNamespace = getFirstValue(ap, vn, basePath + xpath("metadataNamespace"));
|
| 417 |
|
|
if (StringUtils.isNotBlank(metadataNamespace)) {
|
| 418 |
|
|
od.setMetadataNamespace(metadataNamespace);
|
| 419 |
|
|
}
|
| 420 |
|
|
final OriginDescription originDescription = getOriginDescription(ap, vn, basePath + xpath("originDescription"));
|
| 421 |
|
|
if (originDescription.hasHarvestDate()) {
|
| 422 |
|
|
od.setOriginDescription(originDescription);
|
| 423 |
|
|
}
|
| 424 |
|
|
|
| 425 |
|
|
return od.build();
|
| 426 |
51942
|
claudio.at
|
}
|
| 427 |
|
|
|
| 428 |
52041
|
claudio.at
|
private OAIProvenance getOaiProvenance(final AutoPilot ap, final VTDNav vn) throws VtdException {
|
| 429 |
|
|
return OAIProvenance.newBuilder()
|
| 430 |
|
|
.setOriginDescription(getOriginDescription(ap, vn, xpath("record", "about", "provenance", "originDescription")))
|
| 431 |
|
|
.build();
|
| 432 |
|
|
}
|
| 433 |
51944
|
sandro.lab
|
|
| 434 |
52041
|
claudio.at
|
private FieldTypeProtos.DataInfo.Builder ensureDataInfo(
|
| 435 |
51953
|
claudio.at
|
final AutoPilot ap, final VTDNav vn,
|
| 436 |
52041
|
claudio.at
|
final DataInfo.Builder info) throws VtdException {
|
| 437 |
51944
|
sandro.lab
|
|
| 438 |
52041
|
claudio.at
|
if (info.isInitialized()) return info;
|
| 439 |
|
|
return buildDataInfo( ap, vn, invisible, provenance, trust, false, false);
|
| 440 |
51944
|
sandro.lab
|
}
|
| 441 |
|
|
|
| 442 |
52041
|
claudio.at
|
private FieldTypeProtos.DataInfo.Builder buildDataInfo(
|
| 443 |
|
|
final AutoPilot ap,
|
| 444 |
|
|
final VTDNav vn,
|
| 445 |
|
|
final boolean invisible,
|
| 446 |
|
|
final String defaultProvenanceaction,
|
| 447 |
|
|
final String defaultTrust,
|
| 448 |
|
|
final boolean defaultDeletedbyinference,
|
| 449 |
|
|
final boolean defaultInferred) throws VtdException {
|
| 450 |
51953
|
claudio.at
|
|
| 451 |
52041
|
claudio.at
|
final DataInfo.Builder dataInfoBuilder = FieldTypeProtos.DataInfo.newBuilder()
|
| 452 |
|
|
.setInvisible(invisible)
|
| 453 |
|
|
.setInferred(defaultInferred)
|
| 454 |
|
|
.setDeletedbyinference(defaultDeletedbyinference)
|
| 455 |
|
|
.setTrust(defaultTrust)
|
| 456 |
|
|
.setProvenanceaction(getSimpleQualifier(defaultProvenanceaction, DNET_PROVENANCE_ACTIONS));
|
| 457 |
51944
|
sandro.lab
|
|
| 458 |
52041
|
claudio.at
|
// checking instanceof because when receiving an empty <oaf:datainfo> we don't want to parse it.
|
| 459 |
51944
|
sandro.lab
|
|
| 460 |
52041
|
claudio.at
|
final String xpath = xpath("record", "about", "datainfo");
|
| 461 |
|
|
if (getNodes(ap, vn, xpath).size() > 0) {
|
| 462 |
|
|
final Map<String, String> provAction = getNode(ap, vn, xpath + xpath("provenanceaction")).getAttributes();
|
| 463 |
|
|
dataInfoBuilder
|
| 464 |
|
|
.setInvisible(Boolean.valueOf(getValue(getNode(ap, vn, xpath + xpath("invisible")), String.valueOf(invisible))))
|
| 465 |
|
|
.setInferred(Boolean.valueOf(getValue(getNode(ap, vn, xpath + xpath("inferred")), String.valueOf(defaultInferred))))
|
| 466 |
|
|
.setDeletedbyinference(Boolean.valueOf(
|
| 467 |
|
|
getValue(getNode(ap, vn, xpath + xpath("deletedbyinference")), String.valueOf(defaultDeletedbyinference))))
|
| 468 |
|
|
.setTrust(getValue(getNode(ap, vn, xpath + xpath("trust")), defaultTrust))
|
| 469 |
|
|
.setInferenceprovenance(getValue(getNode(ap, vn, xpath + xpath("inferenceprovenance")), ""))
|
| 470 |
|
|
.setProvenanceaction(getSimpleQualifier(
|
| 471 |
|
|
getValue(provAction.get(CLASSID), defaultProvenanceaction),
|
| 472 |
|
|
DNET_PROVENANCE_ACTIONS));
|
| 473 |
|
|
}
|
| 474 |
51944
|
sandro.lab
|
|
| 475 |
52041
|
claudio.at
|
return dataInfoBuilder;
|
| 476 |
51944
|
sandro.lab
|
}
|
| 477 |
|
|
|
| 478 |
51942
|
claudio.at
|
}
|