Context Navigation

AcquiredStructureAnalyzer.java @ 222

リビジョン 173, 15.5 KB (コミッタ: atsuko, 10 年前)
bug fix
属性 svn:mime-type の設定値 `text/plain`

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

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。

異なるフォーマットでダウンロード:

元の形式