Revision b0c4a3aa
Added by Andreas Kohlbecker over 8 years ago
| src/main/java/org/bgbm/biovel/drf/checklist/EEA_BDC_Client.java | ||
|---|---|---|
| 12 | 12 |
import org.bgbm.biovel.drf.query.TinkerPopClient; |
| 13 | 13 |
import org.bgbm.biovel.drf.store.Neo4jStore; |
| 14 | 14 |
import org.bgbm.biovel.drf.store.Store; |
| 15 |
import org.bgbm.biovel.drf.tnr.msg.Classification; |
|
| 15 | 16 |
import org.bgbm.biovel.drf.tnr.msg.NameType; |
| 16 | 17 |
import org.bgbm.biovel.drf.tnr.msg.Query; |
| 17 | 18 |
import org.bgbm.biovel.drf.tnr.msg.Query.Request; |
| ... | ... | |
| 27 | 28 |
import org.bgbm.biovel.drf.utils.TnrMsgUtils; |
| 28 | 29 |
import org.neo4j.graphdb.Relationship; |
| 29 | 30 |
|
| 30 |
import com.tinkerpop.blueprints.Direction; |
|
| 31 | 31 |
import com.tinkerpop.blueprints.Graph; |
| 32 | 32 |
import com.tinkerpop.blueprints.Vertex; |
| 33 | 33 |
import com.tinkerpop.blueprints.impls.neo4j2.Neo4j2Vertex; |
| ... | ... | |
| 47 | 47 |
public static final String COPYRIGHT_URL = "http://www.eea.europa.eu/legal/eea-data-policy"; |
| 48 | 48 |
|
| 49 | 49 |
private static final String SPECIES_RDF_FILE_URL = "http://localhost/download/species.rdf.gz"; // http://eunis.eea.europa.eu/rdf/species.rdf.gz |
| 50 |
private static final String TAXONOMY_RDF_FILE_URL = "http://localhost/download/taxonomy.rdf.gz"; // http://eunis.eea.europa.eu/rdf/taxonomy.rdf.gz |
|
| 50 | 51 |
private static final String LEGALREFS_RDF_FILE_URL = "http://localhost/download/legalrefs.rdf.gz"; // http://eunis.eea.europa.eu/rdf/legalrefs.rdf.gz |
| 51 | 52 |
private static final String REFERENCES_RDF_FILE_URL = "http://localhost/download/references.rdf.gz"; // http://eunis.eea.europa.eu/rdf/references.rdf.gz |
| 53 |
|
|
| 52 | 54 |
private static final boolean REFRESH_TDB = false; |
| 53 | 55 |
|
| 54 | 56 |
public static final EnumSet<SearchMode> SEARCH_MODES = EnumSet.of( |
| ... | ... | |
| 101 | 103 |
return abbreviation; |
| 102 | 104 |
} |
| 103 | 105 |
|
| 104 |
public String propertyURI(String name) {
|
|
| 106 |
public String property(String name) {
|
|
| 105 | 107 |
return schemaUri + name; |
| 106 | 108 |
} |
| 107 | 109 |
|
| ... | ... | |
| 148 | 150 |
private void updateStore(Store neo4jStore) {
|
| 149 | 151 |
try {
|
| 150 | 152 |
neo4jStore.loadIntoStore( |
| 151 |
SPECIES_RDF_FILE_URL, |
|
| 152 |
LEGALREFS_RDF_FILE_URL, |
|
| 153 |
REFERENCES_RDF_FILE_URL |
|
| 153 |
// SPECIES_RDF_FILE_URL, |
|
| 154 |
TAXONOMY_RDF_FILE_URL |
|
| 155 |
// LEGALREFS_RDF_FILE_URL, |
|
| 156 |
// REFERENCES_RDF_FILE_URL |
|
| 154 | 157 |
); |
| 155 | 158 |
} catch (Exception e) {
|
| 156 | 159 |
throw new RuntimeException("Loading "
|
| 157 | 160 |
+ SPECIES_RDF_FILE_URL + ", " |
| 161 |
+ TAXONOMY_RDF_FILE_URL + ", " |
|
| 158 | 162 |
+ LEGALREFS_RDF_FILE_URL + ", " |
| 159 | 163 |
+ REFERENCES_RDF_FILE_URL + |
| 160 | 164 |
" into Neo4jStore failed", e); |
| ... | ... | |
| 205 | 209 |
taxon.setTaxonName(taxonName); |
| 206 | 210 |
taxon.setIdentifier(v.getId().toString()); |
| 207 | 211 |
taxon.setAccordingTo(queryClient.relatedVertexValue(v, RdfSchema.DWC, "nameAccordingToID")); |
| 208 |
URI typeUri = queryClient.relatedVertexURI(v, RdfSchema.RDF, "type");
|
|
| 212 |
URI typeUri = queryClient.vertexURI(v, RdfSchema.RDF, "type");
|
|
| 209 | 213 |
taxon.setTaxonomicStatus(typeUri.getFragment()); |
| 210 | 214 |
|
| 211 | 215 |
createSources(v, taxon); |
| 212 | 216 |
|
| 213 |
/* |
|
| 214 |
|
|
| 215 | 217 |
// classification |
| 216 | 218 |
Classification c = null; |
| 217 |
Resource parentR = queryClient.objectAsResource(taxonR, RdfSchema.EUNIS_SPECIES, "taxonomy"); |
|
| 218 |
while (parentR != null) {
|
|
| 219 |
Vertex parentV= null; |
|
| 220 |
try {
|
|
| 221 |
parentV = queryClient.relatedVertex(v, RdfSchema.EUNIS_SPECIES, "taxonomy"); |
|
| 222 |
} catch (Exception e) {
|
|
| 223 |
logger.error("No taxonomy information for " + v.toString());
|
|
| 224 |
} |
|
| 219 | 225 |
|
| 220 |
String level = queryClient.objectAsString(parentR, RdfSchema.EUNIS_TAXONOMY, "level"); |
|
| 221 |
String parentTaxonName = queryClient.objectAsString(parentR, RdfSchema.EUNIS_TAXONOMY, "name"); |
|
| 226 |
while (parentV != null) {
|
|
| 227 |
logger.debug("parent taxon: " + parentV.toString());
|
|
| 228 |
String level = queryClient.relatedVertexValue(parentV, RdfSchema.EUNIS_TAXONOMY, "level"); |
|
| 229 |
String parentTaxonName = queryClient.relatedVertexValue(parentV, RdfSchema.EUNIS_TAXONOMY, "name"); |
|
| 222 | 230 |
|
| 223 | 231 |
RankLevel rankLevel = null; |
| 224 | 232 |
try {
|
| ... | ... | |
| 253 | 261 |
break; |
| 254 | 262 |
} |
| 255 | 263 |
} |
| 256 |
Resource lastParentR = parentR;
|
|
| 257 |
parentR = queryClient.objectAsResource(parentR, RdfSchema.EUNIS_TAXONOMY, "parent");
|
|
| 258 |
if(lastParentR.equals(parentR)) {
|
|
| 264 |
Vertex lastParentV = parentV;
|
|
| 265 |
parentV = queryClient.relatedVertex(parentV, RdfSchema.EUNIS_TAXONOMY, "parent");
|
|
| 266 |
if(lastParentV.equals(parentV)) {
|
|
| 259 | 267 |
// avoid endless looping when data is not correct |
| 260 | 268 |
break; |
| 261 | 269 |
} |
| ... | ... | |
| 263 | 271 |
if(c != null) {
|
| 264 | 272 |
taxon.setClassification(c); |
| 265 | 273 |
} |
| 266 |
*/ |
|
| 267 | 274 |
return taxon; |
| 268 | 275 |
} |
| 269 | 276 |
|
| ... | ... | |
| 280 | 287 |
|
| 281 | 288 |
try {
|
| 282 | 289 |
List<Vertex> titleVs = taxonPipe |
| 283 |
.outE(RdfSchema.EUNIS_SPECIES.propertyURI("hasLegalReference")).inV()
|
|
| 284 |
.outE(RdfSchema.DCTERMS.propertyURI("source")).inV().dedup()
|
|
| 285 |
.outE(RdfSchema.DCTERMS.propertyURI("title")).inV()
|
|
| 290 |
.outE(RdfSchema.EUNIS_SPECIES.property("hasLegalReference")).inV()
|
|
| 291 |
.outE(RdfSchema.DCTERMS.property("source")).inV().dedup()
|
|
| 292 |
.outE(RdfSchema.DCTERMS.property("title")).inV()
|
|
| 286 | 293 |
.toList(); |
| 287 | 294 |
for(Vertex tv : titleVs) {
|
| 288 | 295 |
Source source = new Source(); |
| ... | ... | |
| 317 | 324 |
|
| 318 | 325 |
try {
|
| 319 | 326 |
List<Vertex> synonymVs = taxonPipe |
| 320 |
.inE(RdfSchema.EUNIS_SPECIES.propertyURI("eunisPrimaryName")).outV().dedup()
|
|
| 327 |
.inE(RdfSchema.EUNIS_SPECIES.property("eunisPrimaryName")).outV().dedup()
|
|
| 321 | 328 |
.toList(); |
| 322 | 329 |
for(Vertex synonymV : synonymVs) {
|
| 323 | 330 |
String typeUri = queryClient.relatedVertexValue(synonymV, RdfSchema.RDF, "type"); |
| ... | ... | |
| 379 | 386 |
pipe = new GremlinPipeline<Graph, Vertex>(hitVs); |
| 380 | 387 |
|
| 381 | 388 |
List<Vertex> vertices = new ArrayList<Vertex>(); |
| 382 |
pipe.in(RdfSchema.EUNIS_SPECIES.propertyURI("binomialName")).fill(vertices);
|
|
| 389 |
pipe.in(RdfSchema.EUNIS_SPECIES.property("binomialName")).fill(vertices);
|
|
| 383 | 390 |
|
| 384 | 391 |
updateQueriesWithResponse(vertices, null, null, checklistInfo, query); |
| 385 | 392 |
profiler.end(System.err); |
| ... | ... | |
| 434 | 441 |
List<Vertex> vertices = new ArrayList<Vertex>(); |
| 435 | 442 |
pipe = new GremlinPipeline<Graph, Vertex>(hitVs); |
| 436 | 443 |
Table table = new Table(); |
| 437 |
pipe.as("match").in(RdfSchema.DWC.propertyURI("vernacularName")).as("taxon").table(table).iterate();
|
|
| 444 |
pipe.as("match").in(RdfSchema.DWC.property("vernacularName")).as("taxon").table(table).iterate();
|
|
| 438 | 445 |
|
| 439 | 446 |
updateQueriesWithResponse( |
| 440 | 447 |
table.getColumn("taxon"), table.getColumn("match"),
|
| ... | ... | |
| 540 | 547 |
Vertex synonymV = taxonV; |
| 541 | 548 |
taxonV = null; |
| 542 | 549 |
try {
|
| 543 |
taxonV = synonymV.getEdges(Direction.OUT, RdfSchema.EUNIS_SPECIES.propertyURI("eunisPrimaryName")).iterator().next().getVertex(Direction.IN);
|
|
| 550 |
taxonV = queryClient.relatedVertex(synonymV, RdfSchema.EUNIS_SPECIES, "eunisPrimaryName");
|
|
| 544 | 551 |
} catch(Exception e) {
|
| 545 | 552 |
logger.error("No accepted taxon found for " + synonymV.toString() + " (" + synonymV.getProperty(GraphSail.VALUE) + ")");
|
| 546 | 553 |
} |
| src/main/java/org/bgbm/biovel/drf/query/TinkerPopClient.java | ||
|---|---|---|
| 86 | 86 |
} |
| 87 | 87 |
|
| 88 | 88 |
/** |
| 89 |
* @param subject |
|
| 89 |
* Returns the first Vertex of the edge with the property specified by |
|
| 90 |
* <code>nameSpace</code> and <code>localName</code>. Both, directions |
|
| 91 |
* of edges are taken into account whereas the OUT edge is tested first. |
|
| 92 |
* |
|
| 93 |
* @param v |
|
| 94 |
* @param nameSpace |
|
| 95 |
* @param localName |
|
| 96 |
* @return |
|
| 97 |
*/ |
|
| 98 |
public Vertex relatedVertex(Vertex v, RdfSchema nameSpace, String localName) {
|
|
| 99 |
Vertex relatedV = null; |
|
| 100 |
try {
|
|
| 101 |
relatedV = v.getEdges(Direction.OUT, nameSpace.property(localName)).iterator().next().getVertex(Direction.IN); |
|
| 102 |
} catch (NoSuchElementException e) {
|
|
| 103 |
try {
|
|
| 104 |
relatedV = v.getEdges(Direction.IN, nameSpace.property(localName)).iterator().next().getVertex(Direction.OUT); |
|
| 105 |
} catch (NoSuchElementException e2) {
|
|
| 106 |
logger.error("No taxonomy information for " + v.toString());
|
|
| 107 |
} |
|
| 108 |
} |
|
| 109 |
return relatedV; |
|
| 110 |
} |
|
| 111 |
|
|
| 112 |
/** |
|
| 113 |
* Returns the <code>value</code> property of the first Vertex of the edge with the property specified by |
|
| 114 |
* <code>nameSpace</code> and <code>localName</code>. Both, directions |
|
| 115 |
* of edges are taken into account whereas the OUT edge is tested first. |
|
| 116 |
* |
|
| 117 |
* @param pipe |
|
| 90 | 118 |
* @param nameSpace |
| 91 | 119 |
* @param localName |
| 92 | 120 |
* @return |
| 93 | 121 |
*/ |
| 94 | 122 |
public String relatedVertexValue(GremlinPipeline<Graph, Vertex> pipe, RdfSchema nameSpace, String localName) {
|
| 95 | 123 |
String txt = null; |
| 96 |
String edgeLabel = nameSpace.propertyURI(localName);
|
|
| 124 |
String edgeLabel = nameSpace.property(localName); |
|
| 97 | 125 |
try {
|
| 98 | 126 |
txt = pipe.outE(1, edgeLabel).inV().next().getProperty("value");
|
| 99 | 127 |
} catch (FastNoSuchElementException e) {
|
| ... | ... | |
| 114 | 142 |
return txt; |
| 115 | 143 |
} |
| 116 | 144 |
|
| 145 |
/** |
|
| 146 |
* Returns the <code>value</code> property of the first Vertex of the edge with the property specified by |
|
| 147 |
* <code>nameSpace</code> and <code>localName</code>. Both, directions |
|
| 148 |
* of edges are taken into account whereas the OUT edge is tested first. |
|
| 149 |
* |
|
| 150 |
* @param v |
|
| 151 |
* @param nameSpace |
|
| 152 |
* @param localName |
|
| 153 |
* @return |
|
| 154 |
*/ |
|
| 117 | 155 |
public String relatedVertexValue(Vertex v, RdfSchema nameSpace, String localName) {
|
| 118 | 156 |
String txt = null; |
| 119 |
String edgeLabel = nameSpace.propertyURI(localName);
|
|
| 157 |
String edgeLabel = nameSpace.property(localName); |
|
| 120 | 158 |
try {
|
| 121 | 159 |
txt = v.getEdges(Direction.OUT, edgeLabel).iterator().next().getVertex(Direction.IN) |
| 122 | 160 |
.getProperty(GraphSail.VALUE); |
| ... | ... | |
| 139 | 177 |
* @param localName |
| 140 | 178 |
* @return |
| 141 | 179 |
*/ |
| 142 |
public URI relatedVertexURI(Vertex v, RdfSchema nameSpace, String localName) {
|
|
| 180 |
public URI vertexURI(Vertex v, RdfSchema nameSpace, String localName) {
|
|
| 143 | 181 |
URI uri = null; |
| 144 | 182 |
try {
|
| 145 | 183 |
if (v.getProperty(GraphSail.KIND).equals(GraphSail.URI)) {
|
| src/test/java/org/bgbm/biovel/drf/checklist/EEA_BDC_ClientTest.java | ||
|---|---|---|
| 3 | 3 |
import static org.junit.Assert.assertEquals; |
| 4 | 4 |
import static org.junit.Assert.assertTrue; |
| 5 | 5 |
|
| 6 |
import org.bgbm.biovel.drf.tnr.msg.Classification; |
|
| 6 | 7 |
import org.bgbm.biovel.drf.tnr.msg.NameType; |
| 7 | 8 |
import org.bgbm.biovel.drf.tnr.msg.Response; |
| 8 | 9 |
import org.bgbm.biovel.drf.tnr.msg.TnrMsg; |
| ... | ... | |
| 48 | 49 |
assertEquals("Canis aureus", response.getMatchingNameString());
|
| 49 | 50 |
assertEquals("Canis aureus", response.getTaxon().getTaxonName().getCanonicalName());
|
| 50 | 51 |
assertTrue(response.getSynonym().size() > 0); |
| 52 |
Classification c = response.getTaxon().getClassification(); |
|
| 53 |
assertEquals("Canidae", c.getFamily());
|
|
| 54 |
assertEquals("Carnivora", c.getOrder());
|
|
| 55 |
assertEquals("Chordata", c.getPhylum());
|
|
| 56 |
assertEquals("Animalia", c.getKingdom());
|
|
| 51 | 57 |
} |
| 52 | 58 |
|
| 53 | 59 |
@Test |
Also available in: Unified diff
gathering classification information implemented with test