| 1 | package org.biohackathon.SPARQLBuilder.OWL;
|
|---|
| 2 |
|
|---|
| 3 | import java.util.*;
|
|---|
| 4 |
|
|---|
| 5 | import org.biohackathon.SPARQLBuilder.endpointMetadata.MetadataManager;
|
|---|
| 6 |
|
|---|
| 7 | import jp.riken.accc.db.sparqlBuilderMetadata.crawler.dataStructure.sparql.crawler.ClassPartition;
|
|---|
| 8 | import jp.riken.accc.db.sparqlBuilderMetadata.crawler.dataStructure.sparql.crawler.ClassRelation;
|
|---|
| 9 | import jp.riken.accc.db.sparqlBuilderMetadata.crawler.dataStructure.sparql.crawler.CrawledMetadata;
|
|---|
| 10 | import jp.riken.accc.db.sparqlBuilderMetadata.crawler.dataStructure.sparql.crawler.Dataset;
|
|---|
| 11 | import jp.riken.accc.db.sparqlBuilderMetadata.crawler.dataStructure.sparql.crawler.Label;
|
|---|
| 12 | import jp.riken.accc.db.sparqlBuilderMetadata.crawler.dataStructure.sparql.crawler.PropertyPartition;
|
|---|
| 13 |
|
|---|
| 14 |
|
|---|
| 15 | //public class OWLQueryBuilderForCrawlerImpl implements OWLQueryBuilder {
|
|---|
| 16 | public class AcquiredStructureAnalyzer implements RDFSchemaAnalyzer {
|
|---|
| 17 |
|
|---|
| 18 | // key: endpointURI, value: crawled metadata
|
|---|
| 19 | private HashMap<String, CrawledMetadata> crawledMetadataTable= null;
|
|---|
| 20 |
|
|---|
| 21 | public String[] getEndpointURIs(){
|
|---|
| 22 | if( crawledMetadataTable == null ){
|
|---|
| 23 | return new String[0];
|
|---|
| 24 | }else{
|
|---|
| 25 | return crawledMetadataTable.keySet().toArray(new String[0]);
|
|---|
| 26 | }
|
|---|
| 27 | }
|
|---|
| 28 |
|
|---|
| 29 | public String[] getGraphURIs(String endpointURI){
|
|---|
| 30 | if( crawledMetadataTable == null ){
|
|---|
| 31 | return new String[0];
|
|---|
| 32 | }else{
|
|---|
| 33 | CrawledMetadata crawledMetadata = crawledMetadataTable.get(endpointURI);
|
|---|
| 34 | if( crawledMetadata == null ){
|
|---|
| 35 | return new String[0];
|
|---|
| 36 | }else{
|
|---|
| 37 | return crawledMetadata.getGraphURIs();
|
|---|
| 38 | }
|
|---|
| 39 | }
|
|---|
| 40 | }
|
|---|
| 41 |
|
|---|
| 42 |
|
|---|
| 43 | public AcquiredStructureAnalyzer(MetadataManager metadataManager){
|
|---|
| 44 | CrawledMetadata[] cmList = metadataManager.getCrawlerMetadataList();
|
|---|
| 45 | crawledMetadataTable = new HashMap<String, CrawledMetadata>();
|
|---|
| 46 |
|
|---|
| 47 | if( cmList != null ){
|
|---|
| 48 | for(CrawledMetadata cm: cmList){
|
|---|
| 49 | String endpointURI = cm.getEndpointURI();
|
|---|
| 50 | crawledMetadataTable.put(endpointURI, cm);
|
|---|
| 51 | }
|
|---|
| 52 | }
|
|---|
| 53 | }
|
|---|
| 54 |
|
|---|
| 55 |
|
|---|
| 56 |
|
|---|
| 57 | public SClass[] listClasses() throws Exception{
|
|---|
| 58 | return getOWLClasses(null, null);
|
|---|
| 59 | }
|
|---|
| 60 |
|
|---|
| 61 |
|
|---|
| 62 | public SClass[] getOWLClasses(String[] keywords, String language) throws Exception{
|
|---|
| 63 | return getOWLClassList(keywords, language).toArray(new SClass[0]);
|
|---|
| 64 | }
|
|---|
| 65 |
|
|---|
| 66 |
|
|---|
| 67 | public List<SClass> getOWLClassList(String[] keywords, String language) throws Exception{
|
|---|
| 68 | ArrayList<SClass> classList = new ArrayList<SClass>();
|
|---|
| 69 |
|
|---|
| 70 | Set<String> endpointURISet = crawledMetadataTable.keySet();
|
|---|
| 71 | for(String endpointURI: endpointURISet){
|
|---|
| 72 | CrawledMetadata cm = crawledMetadataTable.get(endpointURI);
|
|---|
| 73 | // default
|
|---|
| 74 | Dataset dataset = cm.getDefaultDataset();
|
|---|
| 75 | List<SClass> tempClassList = getOWLClassList(endpointURI, null, dataset, keywords, language);
|
|---|
| 76 | for(SClass sClass: tempClassList){
|
|---|
| 77 | classList.add(sClass);
|
|---|
| 78 | }
|
|---|
| 79 |
|
|---|
| 80 | // graphs
|
|---|
| 81 | String[] graphURIs = cm.getGraphURIs();
|
|---|
| 82 | if( graphURIs != null ){
|
|---|
| 83 | for(String graphURI: graphURIs){
|
|---|
| 84 | dataset = cm.getDataset(graphURI);
|
|---|
| 85 | tempClassList = getOWLClassList(endpointURI, graphURI, dataset, keywords, language);
|
|---|
| 86 | for(SClass sClass: tempClassList){
|
|---|
| 87 | classList.add(sClass);
|
|---|
| 88 | }
|
|---|
| 89 | }
|
|---|
| 90 | }
|
|---|
| 91 | }
|
|---|
| 92 | return classList;
|
|---|
| 93 | }
|
|---|
| 94 |
|
|---|
| 95 | private List<SClass> getOWLClassList(String endpointURI, String graphURI, Dataset dataset, String[] keywords, String language) throws Exception{
|
|---|
| 96 | ArrayList<SClass> results = new ArrayList<SClass>();
|
|---|
| 97 |
|
|---|
| 98 | ClassPartition[] classPartitionList = dataset.getClassPartitions();
|
|---|
| 99 | if( classPartitionList == null || classPartitionList.length == 0 ){
|
|---|
| 100 | return new ArrayList<SClass>();
|
|---|
| 101 | }
|
|---|
| 102 | for( ClassPartition cp: classPartitionList){
|
|---|
| 103 | String classURI = cp.classDef.classURI;
|
|---|
| 104 | Label[] rLabels = cp.classDef.labels;
|
|---|
| 105 | org.biohackathon.SPARQLBuilder.OWL.Label[] labels = null;
|
|---|
| 106 | if( rLabels == null ){
|
|---|
| 107 | labels = new org.biohackathon.SPARQLBuilder.OWL.Label[0];
|
|---|
| 108 | }else{
|
|---|
| 109 | labels = new org.biohackathon.SPARQLBuilder.OWL.Label[rLabels.length];
|
|---|
| 110 | for(int i = 0; i < rLabels.length; i++ ) {
|
|---|
| 111 | labels[i] = new org.biohackathon.SPARQLBuilder.OWL.Label(rLabels[i].value, rLabels[i].language);
|
|---|
| 112 | }
|
|---|
| 113 | }
|
|---|
| 114 | int entities = cp.entities;
|
|---|
| 115 |
|
|---|
| 116 | if( keywords == null || keywords.length == 0 ){
|
|---|
| 117 | SClass sClass = new SClass(classURI, labels, entities, endpointURI, graphURI);
|
|---|
| 118 | results.add(sClass);
|
|---|
| 119 | }else{
|
|---|
| 120 | boolean hit = false;
|
|---|
| 121 | for(org.biohackathon.SPARQLBuilder.OWL.Label label: labels){
|
|---|
| 122 | if( language == null || label.getLanguage().equals(language)){
|
|---|
| 123 | String value = label.getLabel();
|
|---|
| 124 | if( value != null ){
|
|---|
| 125 | value = value.toLowerCase().trim();
|
|---|
| 126 | for(String keyword: keywords){
|
|---|
| 127 | if( value.contains(keyword.toLowerCase().trim())){
|
|---|
| 128 | hit = true;
|
|---|
| 129 | break;
|
|---|
| 130 | }
|
|---|
| 131 | }
|
|---|
| 132 | }
|
|---|
| 133 | }
|
|---|
| 134 | if( hit ){
|
|---|
| 135 | SClass sClass = new SClass(classURI, labels, entities, endpointURI, graphURI);
|
|---|
| 136 | results.add(sClass);
|
|---|
| 137 | }
|
|---|
| 138 | }
|
|---|
| 139 | }
|
|---|
| 140 | }
|
|---|
| 141 | return results;
|
|---|
| 142 | }
|
|---|
| 143 |
|
|---|
| 144 |
|
|---|
| 145 |
|
|---|
| 146 |
|
|---|
| 147 |
|
|---|
| 148 | public ClassLink[] getNextClass(String originClass, int limit) throws Exception{
|
|---|
| 149 |
|
|---|
| 150 | ArrayList<ClassLink> classLinkList = new ArrayList<ClassLink>();
|
|---|
| 151 |
|
|---|
| 152 | Set<String> endpointURISet = crawledMetadataTable.keySet();
|
|---|
| 153 | for(String endpointURI: endpointURISet){
|
|---|
| 154 | CrawledMetadata cm = crawledMetadataTable.get(endpointURI);
|
|---|
| 155 | // default
|
|---|
| 156 | Dataset dataset = cm.getDefaultDataset();
|
|---|
| 157 | List<ClassLink> tempClassLinkList = getNextClass(endpointURI, null, dataset, originClass, limit);
|
|---|
| 158 | for(ClassLink classLink: tempClassLinkList){
|
|---|
| 159 | classLinkList.add(classLink);
|
|---|
| 160 | }
|
|---|
| 161 |
|
|---|
| 162 | // graphs
|
|---|
| 163 | String[] graphURIs = cm.getGraphURIs();
|
|---|
| 164 | if( graphURIs != null ){
|
|---|
| 165 | for(String graphURI: graphURIs){
|
|---|
| 166 | dataset = cm.getDataset(graphURI);
|
|---|
| 167 | tempClassLinkList = getNextClass(endpointURI, graphURI, dataset, originClass, limit);
|
|---|
| 168 | for(ClassLink classLink: tempClassLinkList){
|
|---|
| 169 | classLinkList.add(classLink);
|
|---|
| 170 | }
|
|---|
| 171 | }
|
|---|
| 172 | }
|
|---|
| 173 | }
|
|---|
| 174 | return classLinkList.toArray(new ClassLink[0]);
|
|---|
| 175 | }
|
|---|
| 176 |
|
|---|
| 177 |
|
|---|
| 178 |
|
|---|
| 179 | private List<ClassLink> getNextClass(String endpointURI, String graphURI, Dataset dataset, String originClass, int limit) throws Exception{
|
|---|
| 180 | ArrayList<ClassLink> classLinkList = new ArrayList<ClassLink>();
|
|---|
| 181 |
|
|---|
| 182 | PropertyPartition[] pps = dataset.getPropertyPartitions();
|
|---|
| 183 | if( pps == null ){
|
|---|
| 184 | return classLinkList;
|
|---|
| 185 | }
|
|---|
| 186 |
|
|---|
| 187 | for(PropertyPartition pp: pps){
|
|---|
| 188 | ClassRelation[] classRelations = pp.classRelations;
|
|---|
| 189 | if( classRelations != null ){
|
|---|
| 190 | for(ClassRelation classRelation: classRelations){
|
|---|
| 191 | String subjClassURI = classRelation.subjectClassURI;
|
|---|
| 192 | String objClassURI = classRelation.objectClassURI;
|
|---|
| 193 | boolean forward = false;
|
|---|
| 194 | boolean reverse = false;
|
|---|
| 195 | if( objClassURI != null && objClassURI.equals(originClass) ){
|
|---|
| 196 | if( subjClassURI != null ){
|
|---|
| 197 | reverse = true;
|
|---|
| 198 | }
|
|---|
| 199 | }
|
|---|
| 200 | if(subjClassURI != null && subjClassURI.equals(originClass)){
|
|---|
| 201 | if( objClassURI != null || classRelation.objectDatatypeURI != null ){
|
|---|
| 202 | forward = true;
|
|---|
| 203 | }
|
|---|
| 204 | }
|
|---|
| 205 | ClassLink classLink = null;
|
|---|
| 206 | if( forward && !reverse ){
|
|---|
| 207 | classLink = new ClassLink();
|
|---|
| 208 | classLink.setDirection(Direction.forward);
|
|---|
| 209 | classLink.setNumOfOriginClassInstances(classRelation.distinctSubjects);
|
|---|
| 210 | if( objClassURI != null ){
|
|---|
| 211 | classLink.setLinkedClassURI(objClassURI);
|
|---|
| 212 | classLink.setNumOfLinkedClassInstances(classRelation.distinctObjects);
|
|---|
| 213 | }else{
|
|---|
| 214 | classLink.setLinkedLiteralDatatypeURI(classRelation.objectDatatypeURI);
|
|---|
| 215 | classLink.setNumOfLinkedInstances(classRelation.triples);
|
|---|
| 216 | }
|
|---|
| 217 | classLink.setNumOfOriginInstances(pp.distinctSubjects);
|
|---|
| 218 | classLink.setNumOfLinkedInstances(pp.distinctObjects);
|
|---|
| 219 |
|
|---|
| 220 | }
|
|---|
| 221 | if( !forward && reverse ){
|
|---|
| 222 | classLink = new ClassLink();
|
|---|
| 223 | classLink.setDirection(Direction.reverse);
|
|---|
| 224 | classLink.setLinkedClassURI(objClassURI);
|
|---|
| 225 | classLink.setNumOfOriginClassInstances(classRelation.distinctObjects);
|
|---|
| 226 | classLink.setNumOfOriginInstances(pp.distinctObjects);
|
|---|
| 227 | classLink.setNumOfLinkedInstances(pp.distinctSubjects);
|
|---|
| 228 | classLink.setNumOfLinkedClassInstances(classRelation.distinctSubjects);
|
|---|
| 229 | }
|
|---|
| 230 | if( forward && reverse){
|
|---|
| 231 | classLink = new ClassLink();
|
|---|
| 232 | classLink.setDirection(Direction.both);
|
|---|
| 233 | classLink.setLinkedClassURI(objClassURI);
|
|---|
| 234 | classLink.setNumOfOriginClassInstances(classRelation.distinctSubjects);
|
|---|
| 235 | classLink.setNumOfOriginInstances(pp.distinctSubjects);
|
|---|
| 236 | classLink.setNumOfLinkedInstances(pp.distinctObjects);
|
|---|
| 237 | classLink.setNumOfLinkedClassInstances(classRelation.distinctObjects);
|
|---|
| 238 | }
|
|---|
| 239 | // hit
|
|---|
| 240 | if( classLink != null ){
|
|---|
| 241 | classLink.setEndpointURI(endpointURI);
|
|---|
| 242 | classLink.setGraphURI(graphURI);
|
|---|
| 243 | classLink.setPropertyURI(pp.propertyDef.propertyURI);
|
|---|
| 244 | classLink.setNumOfLinks(classRelation.triples);
|
|---|
| 245 | classLinkList.add(classLink);
|
|---|
| 246 | }
|
|---|
| 247 | }
|
|---|
| 248 | }
|
|---|
| 249 | }
|
|---|
| 250 | return classLinkList;
|
|---|
| 251 | }
|
|---|
| 252 |
|
|---|
| 253 |
|
|---|
| 254 |
|
|---|
| 255 |
|
|---|
| 256 |
|
|---|
| 257 |
|
|---|
| 258 |
|
|---|
| 259 | public LabelMap[] getLabels(String[] resourceURIs, String language) throws Exception {
|
|---|
| 260 | if( resourceURIs == null || resourceURIs.length == 0 ){
|
|---|
| 261 | return new LabelMap[0];
|
|---|
| 262 | }
|
|---|
| 263 |
|
|---|
| 264 | HashSet<String> resourceURIset = new HashSet<String>();
|
|---|
| 265 | for(String resourceURI: resourceURIs){
|
|---|
| 266 | resourceURIset.add(resourceURI);
|
|---|
| 267 | }
|
|---|
| 268 |
|
|---|
| 269 | HashMap<String, LabelMap> labelMapTable = new HashMap<String, LabelMap>();
|
|---|
| 270 |
|
|---|
| 271 | Set<String> endpointURISet = crawledMetadataTable.keySet();
|
|---|
| 272 | for(String endpointURI: endpointURISet){
|
|---|
| 273 | CrawledMetadata cm = crawledMetadataTable.get(endpointURI);
|
|---|
| 274 | // default
|
|---|
| 275 | Dataset dataset = cm.getDefaultDataset();
|
|---|
| 276 | labelMapTable = getLabels(dataset, resourceURIset, language, labelMapTable);
|
|---|
| 277 |
|
|---|
| 278 | // graphs
|
|---|
| 279 | String[] graphURIs = cm.getGraphURIs();
|
|---|
| 280 | if( graphURIs != null ){
|
|---|
| 281 | for(String graphURI: graphURIs){
|
|---|
| 282 | dataset = cm.getDataset(graphURI);
|
|---|
| 283 | labelMapTable = getLabels(dataset, resourceURIset, language, labelMapTable);
|
|---|
| 284 | }
|
|---|
| 285 | }
|
|---|
| 286 | }
|
|---|
| 287 | return labelMapTable.values().toArray(new LabelMap[0]);
|
|---|
| 288 | }
|
|---|
| 289 |
|
|---|
| 290 |
|
|---|
| 291 | private HashMap<String, LabelMap> getLabels(Dataset dataset, HashSet<String> resourceURISet , String language, HashMap<String, LabelMap> labelMapTable) throws Exception {
|
|---|
| 292 | ClassPartition[] cps = dataset.getClassPartitions();
|
|---|
| 293 | if( cps != null ){
|
|---|
| 294 | for(ClassPartition cp: cps){
|
|---|
| 295 | String uri = cp.classDef.classURI;
|
|---|
| 296 | if( resourceURISet.contains(uri)){
|
|---|
| 297 | Label[] rLabels = cp.classDef.labels;
|
|---|
| 298 | if( rLabels != null ){
|
|---|
| 299 | for(Label rLabel: rLabels){
|
|---|
| 300 | if( language == null || ( rLabel.language == null || rLabel.language.equals(language))){
|
|---|
| 301 | LabelMap labelMap = null;
|
|---|
| 302 | if( labelMapTable.containsKey(uri)){
|
|---|
| 303 | labelMap = labelMapTable.get(uri);
|
|---|
| 304 | }else{
|
|---|
| 305 | labelMap = new LabelMap();
|
|---|
| 306 | labelMapTable.put(uri, labelMap);
|
|---|
| 307 | labelMap.setResourceURI(uri);
|
|---|
| 308 | }
|
|---|
| 309 | labelMap.addLabel(new org.biohackathon.SPARQLBuilder.OWL.Label(rLabel.value, rLabel.language));
|
|---|
| 310 | }
|
|---|
| 311 | }
|
|---|
| 312 | }
|
|---|
| 313 | }
|
|---|
| 314 | }
|
|---|
| 315 | }
|
|---|
| 316 |
|
|---|
| 317 | PropertyPartition[] pps = dataset.getPropertyPartitions();
|
|---|
| 318 | if( pps != null ){
|
|---|
| 319 | for(PropertyPartition pp: pps){
|
|---|
| 320 | String uri = pp.propertyDef.propertyURI;
|
|---|
| 321 | if( resourceURISet.contains(uri)){
|
|---|
| 322 | Label[] rLabels = pp.propertyDef.labels;
|
|---|
| 323 | if( rLabels != null ){
|
|---|
| 324 | for(Label rLabel: rLabels){
|
|---|
| 325 | if( language == null || ( rLabel.language == null || rLabel.language.equals(language))){
|
|---|
| 326 | LabelMap labelMap = null;
|
|---|
| 327 | if( labelMapTable.containsKey(uri)){
|
|---|
| 328 | labelMap = labelMapTable.get(uri);
|
|---|
| 329 | }else{
|
|---|
| 330 | labelMap = new LabelMap();
|
|---|
| 331 | labelMapTable.put(uri, labelMap);
|
|---|
| 332 | labelMap.setResourceURI(uri);
|
|---|
| 333 | }
|
|---|
| 334 | labelMap.addLabel(new org.biohackathon.SPARQLBuilder.OWL.Label(rLabel.value, rLabel.language));
|
|---|
| 335 | }
|
|---|
| 336 | }
|
|---|
| 337 | }
|
|---|
| 338 | }
|
|---|
| 339 | }
|
|---|
| 340 | }
|
|---|
| 341 | return labelMapTable;
|
|---|
| 342 | }
|
|---|
| 343 |
|
|---|
| 344 | } |
|---|
