Context Navigation

AcquiredStructureAnalyzer.java @ 267

リビジョン 267, 15.7 KB (コミッタ: atsuko, 10 年前)
メタデータ仕様変更，フェデレートサーチ対応向け

行番号
1	package org.biohackathon.SPARQLBuilder.OWL;
2
3	import java.util.*;
4
5	import jp.riken.accc.db.rdf.crawler.dataStructure.sparql.JenaModelGenerator;
6	import jp.riken.accc.db.rdf.crawler.dataStructure.sparql.URICollection;
7
8	import com.hp.hpl.jena.query.Query;
9	import com.hp.hpl.jena.query.QueryExecution;
10	import com.hp.hpl.jena.query.QueryExecutionFactory;
11	import com.hp.hpl.jena.query.QueryFactory;
12	import com.hp.hpl.jena.query.QuerySolution;
13	import com.hp.hpl.jena.query.ResultSet;
14	import com.hp.hpl.jena.rdf.model.Literal;
15	import com.hp.hpl.jena.rdf.model.Model;
16	import com.hp.hpl.jena.rdf.model.Property;
17	import com.hp.hpl.jena.rdf.model.Resource;
18
19	//public class OWLQueryBuilderForCrawlerImpl implements OWLQueryBuilder {
20	public class AcquiredStructureAnalyzer implements RDFSchemaAnalyzer {
21
22	private Model model = null;
23	private String endpointURI = null;
24	private String[] graphURIs = null;
25
26	public String getEndpointURI(){
27	return endpointURI;
28	}
29
30	public String[] getGraphURIs(){
31	return graphURIs;
32	}
33
34	/*
35	public static void main(String[] args) throws Exception{
36
37	JenaModelGenerator jmGene = new JenaModelGenerator("c:\\temp\\biosamplesF11.ttl");
38	// JenaModelGenerator jmGene = new JenaModelGenerator("c:\\temp\\reactomeF18s.ttl");
39	AcquiredStructureAnalyzer impl
40	= new AcquiredStructureAnalyzer(jmGene.getEndpointURI(), jmGene.getGraphURIs(), jmGene.getModel());
41
42	System.out.println("--------------------------");
43	SClass[] scs = impl.getOWLClasses(null, null, null, true);
44	System.out.println("list classes:---------------");
45	for(SClass sc: scs){
46	System.out.println(sc.toString());
47	}
48	System.out.println("--------------------------");
49
50	// ClassLink[] cls = impl.getNextClass(null,"http://www.biopax.org/release/biopax-level3.owl#Protein",100,true );
51	// for(ClassLink cl: cls){
52	// System.out.println(cl.toString());
53	// }
54	// System.out.println("--------------------------");
55
56	}
57	*/
58
59	public AcquiredStructureAnalyzer(String endpointURI, String[] graphURIs, Model model){
60	this.model = model;
61	this.endpointURI = endpointURI;
62	this.graphURIs = graphURIs;
63	}
64
65	private String[] filterGraphURIs(String[] orgGraphURIs){
66	// TODO
67	return graphURIs;
68	}
69
70
71	public SClass[] listClasses(String[] graphURIs, boolean countInstances) throws Exception{
72	return getOWLClasses(graphURIs, null, null, countInstances);
73	}
74
75
76
77
78	public SClass[] getOWLClasses(String[] graphURIs, String[] keywords, String language, boolean countInstances) throws Exception{
79	return getOWLClassList(graphURIs, keywords, language, countInstances).toArray(new SClass[0]);
80	}
81
82	public List<SClass> getOWLClassList(String[] graphURIs, String[] keywords, String language, boolean countInstances) throws Exception{
83	String[] targetGraphURIs = filterGraphURIs(graphURIs);
84
85	StringBuffer queryStr = new StringBuffer();
86	queryStr.append("PREFIX owl: <http://www.w3.org/2002/07/owl#>\n");
87	queryStr.append("PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>\n");
88	queryStr.append("PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>\n");
89	queryStr.append("SELECT DISTINCT ?c ?pLabel ?entities\n");
90	// if (targetGraphURIs != null) {
91	// for (String graphURI : targetGraphURIs) {
92	// queryStr.append("FROM <");
93	// queryStr.append(graphURI);
94	// queryStr.append(">\n");
95	// }
96	// }
97	queryStr.append("WHERE{\n");
98	queryStr.append(" ?cp <").append(URICollection.PROPERTY_VOID_CLASS).append("> ?c. \n");
99	queryStr.append(" ?cp <");
100	queryStr.append(URICollection.PROPERTY_VOID_ENTITIES);
101	queryStr.append("> ?entities. \n");
102	queryStr.append(" OPTIONAL{ ?c <");
103	queryStr.append(URICollection.PROPERTY_RDFS_LABEL);
104	queryStr.append("> ?pLabel. }\n");
105
106	if (keywords != null && keywords.length != 0) {
107
108	queryStr.append(" ?c rdfs:label ");
109	queryStr.append("?keywords").append(".\n");
110	queryStr.append(" filter((LANG(?keywords) = \'").append(language);
111	queryStr.append("\') && \n (");
112
113	for (int i = 0; i < keywords.length; i++) {
114	if (i > 0)
115	queryStr.append(" \|\| \n ");
116
117	queryStr.append("regex(str(").append("?keywords")
118	.append("),\"");
119	queryStr.append(keywords[i]);
120	queryStr.append("\", \"i\" )");
121
122	}
123	queryStr.append("))\n");
124
125	}
126	queryStr.append("}");
127	System.out.println(queryStr.toString());
128
129	Query query = QueryFactory.create(queryStr.toString());
130
131	QueryExecution qexec = null;
132	ResultSet results = null;
133	try {
134	// long start = System.currentTimeMillis();
135	qexec = QueryExecutionFactory.create(query, model);
136	results = qexec.execSelect();
137	// long end = System.currentTimeMillis();
138	// System.out.println("EXEC TIME: " + (end - start));
139	} catch (Exception ex) {
140	ex.printStackTrace();
141	throw ex;
142	}
143
144	HashMap<String, SClass> classMap = new HashMap<String, SClass>();
145	for (; results.hasNext();) {
146	QuerySolution sol = results.next();
147	Resource res = sol.getResource("c");
148	if (res != null && res.getURI() != null) {
149	String uri = res.getURI();
150	int numOfInstances = 0;
151	if (countInstances) {
152	numOfInstances = sol.getLiteral("entities").getInt();
153	} //
154	Literal labelLiteral = sol.getLiteral("pLabel");
155	SClass sClass = null;
156	if (classMap.containsKey(uri)) {
157	sClass = classMap.get(uri);
158	} else {
159	sClass = new SClass(uri, null, numOfInstances);
160	classMap.put(uri, sClass);
161	}
162	if (labelLiteral != null) {
163	String label = labelLiteral.getString();
164	String lang = labelLiteral.getLanguage();
165	sClass.addLabel(new Label(label, lang));
166	}
167	}
168	}
169	qexec.close();
170	return new LinkedList<SClass>(classMap.values());
171	}
172
173	/*
174
175	public Instance[] getInstances(String[] graphURIs, String keyword) throws Exception;
176	*/
177
178
179	public ClassLink[] getNextClass(String[] graphURIs, String originClass, int limit, boolean countLinks) throws Exception{
180	String[] targetGraphURIs = filterGraphURIs(graphURIs);
181
182	StringBuffer queryStr = new StringBuffer();
183	queryStr.append("PREFIX owl: <http://www.w3.org/2002/07/owl#>\n");
184	queryStr.append("PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>\n");
185	queryStr.append("PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>\n");
186
187	// SELECT
188	queryStr.append("SELECT DISTINCT ?indPropCat ?c ?dat ?d ?p ?numLnkInsStart ?numLnkInsEnd ?numInsDom ?numInsRan ?numTriples\n");
189
190	// if (targetGraphURIs != null) {
191	// for (String graphURI : targetGraphURIs) {
192	// queryStr.append("FROM <");
193	// queryStr.append(graphURI);
194	// queryStr.append(">\n");
195	// }
196	// }
197
198	queryStr.append("WHERE{\n");
199	queryStr.append(" ?cr <").append(URICollection.PROPERTY_RDF_TYPE).append("> <").append(URICollection.RESOURCE_SB_CLASS_RELATION).append(">. \n");
200	queryStr.append(" ?propPart <").append(URICollection.PROPERTY_SB_CLASS_RELATION).append("> ?cr. \n");
201	queryStr.append(" ?propPart <").append(URICollection.PROPERTY_VOID_PROPERTY).append("> ?p. \n");
202
203	queryStr.append(" {");
204	queryStr.append(" ?cr <").append(URICollection.PROPERTY_SB_SUBJECT_CLASS).append("> <" + originClass + ">. \n");
205	queryStr.append(" ?cr <").append(URICollection.PROPERTY_SB_OBJECT_CLASS).append("> ?d. \n");
206	queryStr.append(" ?cr <").append(URICollection.PROPERTY_VOID_DISTINCT_SUBJECTS).append("> ?numLnkInsStart. \n");
207	queryStr.append(" ?cr <").append(URICollection.PROPERTY_VOID_DISTINCT_OBJECTS).append("> ?numLnkInsEnd. \n");
208	queryStr.append(" ?cr <").append(URICollection.PROPERTY_VOID_TRIPLES).append("> ?numTriples. \n");
209	// queryStr.append(" ?cr <http://sparqlbuilder.org/startClassLimitedQ> ?isStartClsLim. \n");
210	// queryStr.append(" ?cr <http://sparqlbuilder.org/endClassLimitedQ> ?isEndClsLim. \n");
211	queryStr.append("}\n");
212	queryStr.append(" UNION\n");
213	queryStr.append(" {");
214	queryStr.append(" ?cr <").append(URICollection.PROPERTY_SB_SUBJECT_CLASS).append("> <" + originClass + ">. \n");
215	queryStr.append(" ?cr <").append(URICollection.PROPERTY_SB_OBJECT_DATATYPE).append("> ?dat. \n");
216	queryStr.append(" ?cr <").append(URICollection.PROPERTY_VOID_DISTINCT_SUBJECTS).append("> ?numLnkInsStart. \n");
217	queryStr.append(" ?cr <").append(URICollection.PROPERTY_VOID_DISTINCT_OBJECTS).append("> ?numLnkInsEnd. \n");
218	queryStr.append(" ?cr <").append(URICollection.PROPERTY_VOID_TRIPLES).append("> ?numTriples. \n");
219	// queryStr.append(" ?cr <http://sparqlbuilder.org/startClassLimitedQ> ?isStartClsLim. \n");
220	// queryStr.append(" ?cr <http://sparqlbuilder.org/endClassLimitedQ> ?isEndClsLim. \n");
221	queryStr.append("}\n");
222	queryStr.append(" UNION\n");
223	queryStr.append(" {");
224	queryStr.append(" ?cr <").append(URICollection.PROPERTY_SB_OBJECT_CLASS).append("> <" + originClass + ">. \n");
225	queryStr.append(" ?cr <").append(URICollection.PROPERTY_SB_SUBJECT_CLASS).append("> ?c. \n");
226	queryStr.append(" ?cr <").append(URICollection.PROPERTY_VOID_DISTINCT_SUBJECTS).append("> ?numLnkInsEnd. \n");
227	queryStr.append(" ?cr <").append(URICollection.PROPERTY_VOID_DISTINCT_OBJECTS).append("> ?numLnkInsStart. \n");
228	queryStr.append(" ?cr <").append(URICollection.PROPERTY_VOID_TRIPLES).append("> ?numTriples. \n");
229	// queryStr.append(" ?cr <http://sparqlbuilder.org/startClassLimitedQ> ?isEndClsLim. \n");
230	// queryStr.append(" ?cr <http://sparqlbuilder.org/endClassLimitedQ> ?isStartClsLim. \n");
231	queryStr.append("}\n");
232
233	// queryStr.append(" ?propPart <").append(URICollection.PROPERTY_RDF_TYPE).append("> <").append(URICollection.PROPERTY_VOID_PROPERTY_PARTITION).append(">. \n");
234	queryStr.append(" ?propPart <").append(URICollection.PROPERTY_SB_PROPERTY_CATEGORY).append("> ?indPropCat. \n");
235
236	queryStr.append(" ?propPart <").append(URICollection.PROPERTY_VOID_DISTINCT_SUBJECTS).append("> ?numInsDom. \n");
237	queryStr.append(" ?propPart <").append(URICollection.PROPERTY_VOID_DISTINCT_OBJECTS).append("> ?numInsRan. \n");
238
239	queryStr.append("}\n");
240
241
242	if (limit > 0) {
243	queryStr.append("limit ");
244	queryStr.append(limit);
245	queryStr.append("\n");
246	}
247
248	// System.out.println("getNextClasses SPARQL Query: ");
249	// System.out.println(queryStr.toString());
250
251	Query query = QueryFactory.create(queryStr.toString());
252	QueryExecution qexec = null;
253	ResultSet results = null;
254	try {
255	long start = System.currentTimeMillis();
256	qexec = QueryExecutionFactory.create(query, model);
257	results = qexec.execSelect();
258	long end = System.currentTimeMillis();
259	System.out.println("EXEC TIME: " + (end - start));
260	} catch (Exception ex) {
261	ex.printStackTrace();
262	throw ex;
263	}
264
265	ArrayList<ClassLink> solCLs = new ArrayList<ClassLink>();
266	for (; results.hasNext();) {
267	QuerySolution sol = results.next();
268	Resource pro = sol.getResource("p");
269	String clsURI = null;
270	String datURI = null;
271	if (pro != null) {
272	int indPropCat = 4;
273	Literal indPropCatLit = sol.getLiteral("indPropCat");
274	if( indPropCatLit != null ){
275	indPropCat = indPropCatLit.getInt();
276	}
277	if( indPropCat < 4 ) {
278	String proURI = pro.getURI();
279	Resource ccls = sol.getResource("c");
280	Resource dcls = sol.getResource("d");
281	Resource dat = sol.getResource("dat");
282	Direction direction = null;
283	if(ccls != null && dcls == null ){
284	// direction forward
285	direction = Direction.reverse;
286	clsURI = ccls.getURI();
287	}else{
288	if( ccls == null && dcls != null ){
289	direction = Direction.forward;
290	clsURI = dcls.getURI();
291	}else{
292	if( ccls == null && dat != null && dcls == null ){
293	clsURI = null;
294	direction = Direction.forward;
295	datURI = dat.getURI();
296	}
297	}
298	}
299
300
301	int numTriples = 0;
302	Literal numTriplesLit = sol.getLiteral("numTriples");
303	if( numTriplesLit != null ){
304	numTriples = numTriplesLit.getInt();
305	}
306
307	int numLnkInsStart = 0;
308	Literal numInsStartLit = sol.getLiteral("numLnkInsStart");
309	if( numInsStartLit != null ){
310	numLnkInsStart = numInsStartLit.getInt();
311	}
312	int numLnkInsEnd = 0;
313	Literal numInsEndLit = sol.getLiteral("numLnkInsEnd");
314	if( numInsEndLit != null ){
315	numLnkInsEnd = numInsEndLit.getInt();
316	}
317
318	int numInsDom = 0;
319	Literal numInsDomLit = sol.getLiteral("numInsDom");
320	if( numInsDomLit != null ){
321	numInsDom = numInsDomLit.getInt();
322	}
323	int numInsRan = 0;
324	Literal numInsRanLit = sol.getLiteral("numInsRan");
325	if( numInsRanLit != null ){
326	numInsRan = numInsRanLit.getInt();
327	}
328
329	boolean isStartClsLim = false;
330	Literal isStartClsLimLit = sol.getLiteral("isStartClsLim");
331	if( isStartClsLimLit != null ){
332	isStartClsLim = isStartClsLimLit.getBoolean();
333	}
334	boolean isEndClsLim = false;
335	Literal isEndClsLimLit = sol.getLiteral("isEndClsLim");
336	if( isEndClsLimLit != null ){
337	isEndClsLim = isEndClsLimLit.getBoolean();
338	}
339
340	ClassLink cl = new ClassLink(proURI, clsURI, datURI, direction,
341	numTriples, numInsDom, numInsRan, numLnkInsStart, numLnkInsEnd, isStartClsLim, isEndClsLim);
342	solCLs.add(cl);
343
344	}
345	}
346	}
347	qexec.close();
348	return solCLs.toArray(new ClassLink[0]);
349	}
350
351
352
353	/*
354
355	public ClassLink[] getNextClassViaInstanceLink(String[] graphURIs, String originClass, int limit) throws Exception;
356
357	public Path[] getPaths(String startClass, String endClass, int mode, boolean countLinks) throws Exception;
358
359	public String createSPARQL(Path path) throws Exception;
360
361	InstanceLink[] getNextInstancesViaInstanceLink(String[] graphURIs, String originInstance,
362	int limit) throws Exception;
363	*/
364
365	public LabelMap[] getLabels(String[] graphURIs, String[] resourceURIs,
366	String language) throws Exception {
367	if (resourceURIs == null \|\| resourceURIs.length == 0) {
368	return new LabelMap[0];
369	}
370	StringBuffer queryStr = new StringBuffer();
371	queryStr.append("PREFIX owl: <http://www.w3.org/2002/07/owl#>\n");
372	queryStr.append("PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>\n");
373	queryStr.append("PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>\n");
374	queryStr.append("SELECT DISTINCT ?res ?label \n");
375	if (graphURIs != null) {
376	for (String graphURI : graphURIs) {
377	queryStr.append("FROM <");
378	queryStr.append(graphURI);
379	queryStr.append(">\n");
380	}
381	}
382	queryStr.append("WHERE{\n");
383	queryStr.append(" ?res rdfs:label ?label.\n");
384	queryStr.append(" FILTER(?res IN (");
385	boolean f = false;
386	for (String resourceURI : resourceURIs) {
387	if (f) {
388	queryStr.append(", ");
389	}
390	f = true;
391	queryStr.append("<");
392	queryStr.append(resourceURI);
393	queryStr.append(">");
394	}
395	queryStr.append("))\n");
396	queryStr.append("}");
397
398	// System.out.println(queryStr.toString());
399
400	Query query = QueryFactory.create(queryStr.toString());
401	QueryExecution qexec = QueryExecutionFactory.create(query, model);
402
403	ResultSet results = qexec.execSelect();
404	HashMap<String, LabelMap> lMap = new HashMap<String, LabelMap>();
405	for (; results.hasNext();) {
406	QuerySolution sol = results.next();
407	String uri = sol.getResource("res").getURI();
408	Literal literal = sol.getLiteral("label");
409	if (literal != null) {
410	String label = literal.getString();
411	String lang = literal.getLanguage();
412	if (language != null && language.equals(lang)) {
413	Label lbl = new Label(label, lang);
414	if (lMap.containsKey(uri)) {
415	LabelMap lm = lMap.get(uri);
416	lm.addLabel(lbl);
417	} else {
418	LabelMap lm = new LabelMap(uri, new Label[] { lbl });
419	lMap.put(uri, lm);
420	}
421	}
422	}
423	}
424	return lMap.values().toArray(new LabelMap[0]);
425	}
426
427	/*
428	public ClassLink[] countLinks(String[] graphURIs, String startClassURI, ClassLink[] classLinks) throws Exception;
429	public SClass[] countInstances(String[] graphURIs, SClass[] classes) throws Exception;
430
431	*/
432
433	}

Note: リポジトリブラウザについてのヘルプは TracBrowser を参照してください。

Context Navigation

root/SPARQLBuilderWWW2016/src/java/org/biohackathon/SPARQLBuilder/OWL/AcquiredStructureAnalyzer.java @ 267

異なるフォーマットでダウンロード: