Context Navigation

AcquiredStructureAnalyzer.java @ 133

リビジョン 133, 15.4 KB (コミッタ: atsuko, 10 年前)
最新版に追従
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	JenaModelGenerator jmGene = new JenaModelGenerator("c:\\temp\\reactomeF18s.ttl");
38	AcquiredStructureAnalyzer impl
39	= new AcquiredStructureAnalyzer(jmGene.getEndpointURI(), jmGene.getGraphURIs(), jmGene.getModel());
40	// System.out.println("--------------------------");
41	// SClass[] scs = impl.getOWLClasses(null, null, null, true);
42	// System.out.println("list classes:---------------");
43	// for(SClass sc: scs){
44	// System.out.println(sc.toString());
45	// }
46	// System.out.println("--------------------------");
47
48	ClassLink[] cls = impl.getNextClass(null,"http://www.biopax.org/release/biopax-level3.owl#Protein",100,true );
49	for(ClassLink cl: cls){
50	System.out.println(cl.toString());
51	}
52	System.out.println("--------------------------");
53
54	}
55
56
57	public AcquiredStructureAnalyzer(String endpointURI, String[] graphURIs, Model model){
58	this.model = model;
59	this.endpointURI = endpointURI;
60	this.graphURIs = graphURIs;
61	}
62
63	private String[] filterGraphURIs(String[] orgGraphURIs){
64	// TODO
65	return graphURIs;
66	}
67
68
69	public SClass[] listClasses(String[] graphURIs, boolean countInstances) throws Exception{
70	return getOWLClasses(graphURIs, null, null, countInstances);
71	}
72
73
74
75
76	public SClass[] getOWLClasses(String[] graphURIs, String[] keywords, String language, boolean countInstances) throws Exception{
77	String[] targetGraphURIs = filterGraphURIs(graphURIs);
78
79	StringBuffer queryStr = new StringBuffer();
80	queryStr.append("PREFIX owl: <http://www.w3.org/2002/07/owl#>\n");
81	queryStr.append("PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>\n");
82	queryStr.append("PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>\n");
83	queryStr.append("SELECT DISTINCT ?c ?pLabel ?entities\n");
84	// if (targetGraphURIs != null) {
85	// for (String graphURI : targetGraphURIs) {
86	// queryStr.append("FROM <");
87	// queryStr.append(graphURI);
88	// queryStr.append(">\n");
89	// }
90	// }
91	queryStr.append("WHERE{\n");
92	queryStr.append(" ?cp <").append(URICollection.PROPERTY_VOID_CLASS).append("> ?c. \n");
93	queryStr.append(" ?cp <");
94	queryStr.append(URICollection.PROPERTY_VOID_ENTITIES);
95	queryStr.append("> ?entities. \n");
96	queryStr.append(" OPTIONAL{ ?c <");
97	queryStr.append(URICollection.PROPERTY_RDFS_LABEL);
98	queryStr.append("> ?pLabel. }\n");
99
100	if (keywords != null && keywords.length != 0) {
101
102	queryStr.append(" ?c rdfs:label ");
103	queryStr.append("?keywords").append(".\n");
104	queryStr.append(" filter((LANG(?keywords) = \'").append(language);
105	queryStr.append("\') && \n (");
106
107	for (int i = 0; i < keywords.length; i++) {
108	if (i > 0)
109	queryStr.append(" \|\| \n ");
110
111	queryStr.append("regex(str(").append("?keywords")
112	.append("),\"");
113	queryStr.append(keywords[i]);
114	queryStr.append("\", \"i\" )");
115
116	}
117	queryStr.append("))\n");
118
119	}
120	queryStr.append("}");
121	System.out.println(queryStr.toString());
122
123	Query query = QueryFactory.create(queryStr.toString());
124
125	QueryExecution qexec = null;
126	ResultSet results = null;
127	try {
128	// long start = System.currentTimeMillis();
129	qexec = QueryExecutionFactory.create(query, model);
130	results = qexec.execSelect();
131	// long end = System.currentTimeMillis();
132	// System.out.println("EXEC TIME: " + (end - start));
133	} catch (Exception ex) {
134	ex.printStackTrace();
135	throw ex;
136	}
137
138	HashMap<String, SClass> classMap = new HashMap<String, SClass>();
139	for (; results.hasNext();) {
140	QuerySolution sol = results.next();
141	Resource res = sol.getResource("c");
142	if (res != null && res.getURI() != null) {
143	String uri = res.getURI();
144	int numOfInstances = 0;
145	if (countInstances) {
146	numOfInstances = sol.getLiteral("entities").getInt();
147	} //
148	Literal labelLiteral = sol.getLiteral("pLabel");
149	SClass sClass = null;
150	if (classMap.containsKey(uri)) {
151	sClass = classMap.get(uri);
152	} else {
153	sClass = new SClass(uri, null, numOfInstances);
154	classMap.put(uri, sClass);
155	}
156	if (labelLiteral != null) {
157	String label = labelLiteral.getString();
158	String lang = labelLiteral.getLanguage();
159	sClass.addLabel(new Label(label, lang));
160	}
161	}
162	}
163	qexec.close();
164	return classMap.values().toArray(new SClass[0]);
165	}
166
167	/*
168
169	public Instance[] getInstances(String[] graphURIs, String keyword) throws Exception;
170	*/
171
172
173	public ClassLink[] getNextClass(String[] graphURIs, String originClass, int limit, boolean countLinks) throws Exception{
174	String[] targetGraphURIs = filterGraphURIs(graphURIs);
175
176	StringBuffer queryStr = new StringBuffer();
177	queryStr.append("PREFIX owl: <http://www.w3.org/2002/07/owl#>\n");
178	queryStr.append("PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>\n");
179	queryStr.append("PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>\n");
180
181	// SELECT
182	queryStr.append("SELECT DISTINCT ?indPropCat ?c ?dat ?d ?p ?numLnkInsStart ?numLnkInsEnd ?numInsDom ?numInsRan ?numTriples\n");
183
184	// if (targetGraphURIs != null) {
185	// for (String graphURI : targetGraphURIs) {
186	// queryStr.append("FROM <");
187	// queryStr.append(graphURI);
188	// queryStr.append(">\n");
189	// }
190	// }
191
192	queryStr.append("WHERE{\n");
193	queryStr.append(" ?cr <").append(URICollection.PROPERTY_RDF_TYPE).append("> <").append(URICollection.RESOURCE_SB_CLASS_RELATION).append(">. \n");
194	queryStr.append(" ?propPart <").append(URICollection.PROPERTY_SB_CLASS_RELATION).append("> ?cr. \n");
195	queryStr.append(" ?propPart <").append(URICollection.PROPERTY_VOID_PROPERTY).append("> ?p. \n");
196
197	queryStr.append(" {");
198	queryStr.append(" ?cr <").append(URICollection.PROPERTY_SB_SUBJECT_CLASS).append("> <" + originClass + ">. \n");
199	queryStr.append(" ?cr <").append(URICollection.PROPERTY_SB_OBJECT_CLASS).append("> ?d. \n");
200	queryStr.append(" ?cr <").append(URICollection.PROPERTY_VOID_DISTINCT_SUBJECTS).append("> ?numLnkInsStart. \n");
201	queryStr.append(" ?cr <").append(URICollection.PROPERTY_VOID_DISTINCT_OBJECTS).append("> ?numLnkInsEnd. \n");
202	queryStr.append(" ?cr <").append(URICollection.PROPERTY_VOID_TRIPLES).append("> ?numTriples. \n");
203	// queryStr.append(" ?cr <http://sparqlbuilder.org/startClassLimitedQ> ?isStartClsLim. \n");
204	// queryStr.append(" ?cr <http://sparqlbuilder.org/endClassLimitedQ> ?isEndClsLim. \n");
205	queryStr.append("}\n");
206	queryStr.append(" UNION\n");
207	queryStr.append(" {");
208	queryStr.append(" ?cr <").append(URICollection.PROPERTY_SB_SUBJECT_CLASS).append("> <" + originClass + ">. \n");
209	queryStr.append(" ?cr <").append(URICollection.PROPERTY_SB_OBJECT_DATATYPE).append("> ?dat. \n");
210	queryStr.append(" ?cr <").append(URICollection.PROPERTY_VOID_DISTINCT_SUBJECTS).append("> ?numLnkInsStart. \n");
211	queryStr.append(" ?cr <").append(URICollection.PROPERTY_VOID_DISTINCT_OBJECTS).append("> ?numLnkInsEnd. \n");
212	queryStr.append(" ?cr <").append(URICollection.PROPERTY_VOID_TRIPLES).append("> ?numTriples. \n");
213	// queryStr.append(" ?cr <http://sparqlbuilder.org/startClassLimitedQ> ?isStartClsLim. \n");
214	// queryStr.append(" ?cr <http://sparqlbuilder.org/endClassLimitedQ> ?isEndClsLim. \n");
215	queryStr.append("}\n");
216	queryStr.append(" UNION\n");
217	queryStr.append(" {");
218	queryStr.append(" ?cr <").append(URICollection.PROPERTY_SB_OBJECT_CLASS).append("> <" + originClass + ">. \n");
219	queryStr.append(" ?cr <").append(URICollection.PROPERTY_SB_SUBJECT_CLASS).append("> ?d. \n");
220	queryStr.append(" ?cr <").append(URICollection.PROPERTY_VOID_DISTINCT_SUBJECTS).append("> ?numLnkInsEnd. \n");
221	queryStr.append(" ?cr <").append(URICollection.PROPERTY_VOID_DISTINCT_OBJECTS).append("> ?numLnkInsStart. \n");
222	queryStr.append(" ?cr <").append(URICollection.PROPERTY_VOID_TRIPLES).append("> ?numTriples. \n");
223	// queryStr.append(" ?cr <http://sparqlbuilder.org/startClassLimitedQ> ?isEndClsLim. \n");
224	// queryStr.append(" ?cr <http://sparqlbuilder.org/endClassLimitedQ> ?isStartClsLim. \n");
225	queryStr.append("}\n");
226
227	// queryStr.append(" ?propPart <").append(URICollection.PROPERTY_RDF_TYPE).append("> <").append(URICollection.PROPERTY_VOID_PROPERTY_PARTITION).append(">. \n");
228	queryStr.append(" ?propPart <").append(URICollection.PROPERTY_SB_PROPERTY_CATEGORY).append("> ?indPropCat. \n");
229
230	queryStr.append(" ?propPart <").append(URICollection.PROPERTY_VOID_DISTINCT_SUBJECTS).append("> ?numInsDom. \n");
231	queryStr.append(" ?propPart <").append(URICollection.PROPERTY_VOID_DISTINCT_OBJECTS).append("> ?numInsRan. \n");
232
233	queryStr.append("}\n");
234
235
236	if (limit > 0) {
237	queryStr.append("limit ");
238	queryStr.append(limit);
239	queryStr.append("\n");
240	}
241
242	// System.out.println("getNextClasses SPARQL Query: ");
243	// System.out.println(queryStr.toString());
244
245	Query query = QueryFactory.create(queryStr.toString());
246	QueryExecution qexec = null;
247	ResultSet results = null;
248	try {
249	long start = System.currentTimeMillis();
250	qexec = QueryExecutionFactory.create(query, model);
251	results = qexec.execSelect();
252	long end = System.currentTimeMillis();
253	System.out.println("EXEC TIME: " + (end - start));
254	} catch (Exception ex) {
255	ex.printStackTrace();
256	throw ex;
257	}
258
259	ArrayList<ClassLink> solCLs = new ArrayList<ClassLink>();
260	for (; results.hasNext();) {
261	QuerySolution sol = results.next();
262	Resource pro = sol.getResource("p");
263	String clsURI = null;
264	String datURI = null;
265	if (pro != null) {
266	int indPropCat = 4;
267	Literal indPropCatLit = sol.getLiteral("indPropCat");
268	if( indPropCatLit != null ){
269	indPropCat = indPropCatLit.getInt();
270	}
271	if( indPropCat < 4 ) {
272	String proURI = pro.getURI();
273	Resource ccls = sol.getResource("c");
274	Resource dcls = sol.getResource("d");
275	Resource dat = sol.getResource("dat");
276	Direction direction = null;
277	if(ccls != null && dcls == null ){
278	// direction forward
279	direction = Direction.forward;
280	clsURI = ccls.getURI();
281	}else{
282	if( ccls == null && dcls != null ){
283	direction = Direction.reverse;
284	clsURI = dcls.getURI();
285	}else{
286	if( ccls == null && dat != null && dcls == null ){
287	clsURI = null;
288	direction = Direction.forward;
289	datURI = dat.getURI();
290	}
291	}
292	}
293
294
295	int numTriples = 0;
296	Literal numTriplesLit = sol.getLiteral("numTriples");
297	if( numTriplesLit != null ){
298	numTriples = numTriplesLit.getInt();
299	}
300
301	int numLnkInsStart = 0;
302	Literal numInsStartLit = sol.getLiteral("numLnkInsStart");
303	if( numInsStartLit != null ){
304	numLnkInsStart = numInsStartLit.getInt();
305	}
306	int numLnkInsEnd = 0;
307	Literal numInsEndLit = sol.getLiteral("numLnkInsEnd");
308	if( numInsEndLit != null ){
309	numLnkInsEnd = numInsEndLit.getInt();
310	}
311
312	int numInsDom = 0;
313	Literal numInsDomLit = sol.getLiteral("numInsDom");
314	if( numInsDomLit != null ){
315	numInsDom = numInsDomLit.getInt();
316	}
317	int numInsRan = 0;
318	Literal numInsRanLit = sol.getLiteral("numInsRan");
319	if( numInsRanLit != null ){
320	numInsRan = numInsRanLit.getInt();
321	}
322
323	boolean isStartClsLim = false;
324	Literal isStartClsLimLit = sol.getLiteral("isStartClsLim");
325	if( isStartClsLimLit != null ){
326	isStartClsLim = isStartClsLimLit.getBoolean();
327	}
328	boolean isEndClsLim = false;
329	Literal isEndClsLimLit = sol.getLiteral("isEndClsLim");
330	if( isEndClsLimLit != null ){
331	isEndClsLim = isEndClsLimLit.getBoolean();
332	}
333
334	ClassLink cl = new ClassLink(proURI, clsURI, datURI, direction,
335	numTriples, numInsDom, numInsRan, numLnkInsStart, numLnkInsEnd, isStartClsLim, isEndClsLim);
336	solCLs.add(cl);
337
338	}
339	}
340	}
341	qexec.close();
342	return solCLs.toArray(new ClassLink[0]);
343	}
344
345
346
347	/*
348
349	public ClassLink[] getNextClassViaInstanceLink(String[] graphURIs, String originClass, int limit) throws Exception;
350
351	public Path[] getPaths(String startClass, String endClass, int mode, boolean countLinks) throws Exception;
352
353	public String createSPARQL(Path path) throws Exception;
354
355	InstanceLink[] getNextInstancesViaInstanceLink(String[] graphURIs, String originInstance,
356	int limit) throws Exception;
357	*/
358
359	public LabelMap[] getLabels(String[] graphURIs, String[] resourceURIs,
360	String language) throws Exception {
361	if (resourceURIs == null \|\| resourceURIs.length == 0) {
362	return new LabelMap[0];
363	}
364	StringBuffer queryStr = new StringBuffer();
365	queryStr.append("PREFIX owl: <http://www.w3.org/2002/07/owl#>\n");
366	queryStr.append("PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>\n");
367	queryStr.append("PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>\n");
368	queryStr.append("SELECT DISTINCT ?res ?label \n");
369	if (graphURIs != null) {
370	for (String graphURI : graphURIs) {
371	queryStr.append("FROM <");
372	queryStr.append(graphURI);
373	queryStr.append(">\n");
374	}
375	}
376	queryStr.append("WHERE{\n");
377	queryStr.append(" ?res rdfs:label ?label.\n");
378	queryStr.append(" FILTER(?res IN (");
379	boolean f = false;
380	for (String resourceURI : resourceURIs) {
381	if (f) {
382	queryStr.append(", ");
383	}
384	f = true;
385	queryStr.append("<");
386	queryStr.append(resourceURI);
387	queryStr.append(">");
388	}
389	queryStr.append("))\n");
390	queryStr.append("}");
391
392	// System.out.println(queryStr.toString());
393
394	Query query = QueryFactory.create(queryStr.toString());
395	QueryExecution qexec = QueryExecutionFactory.create(query, model);
396
397	ResultSet results = qexec.execSelect();
398	HashMap<String, LabelMap> lMap = new HashMap<String, LabelMap>();
399	for (; results.hasNext();) {
400	QuerySolution sol = results.next();
401	String uri = sol.getResource("res").getURI();
402	Literal literal = sol.getLiteral("label");
403	if (literal != null) {
404	String label = literal.getString();
405	String lang = literal.getLanguage();
406	if (language != null && language.equals(lang)) {
407	Label lbl = new Label(label, lang);
408	if (lMap.containsKey(uri)) {
409	LabelMap lm = lMap.get(uri);
410	lm.addLabel(lbl);
411	} else {
412	LabelMap lm = new LabelMap(uri, new Label[] { lbl });
413	lMap.put(uri, lm);
414	}
415	}
416	}
417	}
418	return lMap.values().toArray(new LabelMap[0]);
419	}
420
421	/*
422	public ClassLink[] countLinks(String[] graphURIs, String startClassURI, ClassLink[] classLinks) throws Exception;
423	public SClass[] countInstances(String[] graphURIs, SClass[] classes) throws Exception;
424
425	*/
426
427	}

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