Context Navigation

AcquiredStructureAnalyzer.java @ 93

リビジョン 93, 11.5 KB (コミッタ: nori, 10 年前)
クローラ出力ファイルを指定する機能を追加
属性 svn:mime-type の設定値 `text/plain`

Rev	行番号
[80]	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
	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.Resource;
	17
	18	//public class OWLQueryBuilderForCrawlerImpl implements OWLQueryBuilder {
	19	public class AcquiredStructureAnalyzer implements RDFSchemaAnalyzer {
	20
	21	private Model model = null;
	22	private String endpointURI = null;
	23	private String[] graphURIs = null;
	24
	25	public String getEndpointURI(){
	26	return endpointURI;
	27	}
	28
	29	public String[] getGraphURIs(){
	30	return graphURIs;
	31	}
	32
	33
	34	public static void main(String[] args) throws Exception{
	35	JenaModelGenerator jmGene = new JenaModelGenerator("c:\\temp\\allie.ttl");
	36	AcquiredStructureAnalyzer impl
	37	= new AcquiredStructureAnalyzer(jmGene.getEndpointURI(), jmGene.getGraphURIs(), jmGene.getModel());
	38	SClass[] scs = impl.getOWLClasses(null, null, null, true);
	39	for(SClass sc: scs){
	40	System.out.println(sc.toString());
	41	}
	42	ClassLink[] cls = impl.getNextClass(null,"http://purl.org/goodrelations/v1#Offering",100,true );
	43	for(ClassLink cl: cls){
	44	System.out.println(cl.toString());
	45	}
	46
	47	}
	48
	49
	50	public AcquiredStructureAnalyzer(String endpointURI, String[] graphURIs, Model model){
	51	this.model = model;
	52	this.endpointURI = endpointURI;
	53	this.graphURIs = graphURIs;
	54	}
	55
	56	private String[] filterGraphURIs(String[] orgGraphURIs){
	57	// TODO
	58	return graphURIs;
	59	}
	60
	61
	62	public SClass[] listClasses(String[] graphURIs, boolean countInstances) throws Exception{
	63	return getOWLClasses(graphURIs, null, null, countInstances);
	64	}
	65
	66
[93]	67
	68
[80]	69	public SClass[] getOWLClasses(String[] graphURIs, String[] keywords, String language, boolean countInstances) throws Exception{
	70	String[] targetGraphURIs = filterGraphURIs(graphURIs);
	71
	72	StringBuffer queryStr = new StringBuffer();
	73	queryStr.append("PREFIX owl: <http://www.w3.org/2002/07/owl#>\n");
	74	queryStr.append("PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>\n");
	75	queryStr.append("PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>\n");
	76	queryStr.append("SELECT DISTINCT ?c ?pLabel ?numOfInstances\n");
	77	if (targetGraphURIs != null) {
	78	for (String graphURI : targetGraphURIs) {
	79	queryStr.append("FROM <");
	80	queryStr.append(graphURI);
	81	queryStr.append(">\n");
	82	}
	83	}
	84	queryStr.append("WHERE{\n");
	85
	86	//
	87	queryStr.append(" ?c rdf:type rdfs:Class. \n");
	88	queryStr.append(" ?c <http://sparqlbuilder.org/numberOfInstances> ?numOfInstances. \n");
	89	queryStr.append(" OPTIONAL{ ?c rdfs:label ?pLabel. }\n");
	90
	91	if (keywords != null && keywords.length != 0) {
	92
	93	queryStr.append(" ?c rdfs:label ");
	94	queryStr.append("?keywords").append(".\n");
	95	queryStr.append(" filter((LANG(?keywords) = \'").append(language);
	96	queryStr.append("\') && \n (");
	97
	98	// (LANG(?keywords) = 'en') &&
	99
	100	for (int i = 0; i < keywords.length; i++) {
	101	if (i > 0)
	102	queryStr.append(" \|\| \n ");
	103
	104	queryStr.append("regex(str(").append("?keywords")
	105	.append("),\"");
	106	queryStr.append(keywords[i]);
	107	queryStr.append("\", \"i\" )");
	108
	109	}
	110	queryStr.append("))\n");
	111
	112	}
	113	queryStr.append("}");
	114	System.out.println(queryStr.toString());
	115
	116	Query query = QueryFactory.create(queryStr.toString());
	117
	118	QueryExecution qexec = null;
	119	ResultSet results = null;
	120	try {
	121	long start = System.currentTimeMillis();
	122	qexec = QueryExecutionFactory.create(query, model);
	123	results = qexec.execSelect();
	124	long end = System.currentTimeMillis();
	125	System.out.println("EXEC TIME: " + (end - start));
	126	} catch (Exception ex) {
	127	ex.printStackTrace();
	128	throw ex;
	129	}
	130
	131	HashMap<String, SClass> classMap = new HashMap<String, SClass>();
	132	for (; results.hasNext();) {
	133	QuerySolution sol = results.next();
	134	Resource res = sol.getResource("c");
	135	if (res != null) {
	136	String uri = res.getURI();
	137	int numOfInstances = 0;
	138	if (countInstances) {
	139	numOfInstances = sol.getLiteral("numOfInstances").getInt();
	140	} //
	141	Literal labelLiteral = sol.getLiteral("pLabel");
	142	SClass sClass = null;
	143	if (classMap.containsKey(uri)) {
	144	sClass = classMap.get(uri);
	145	} else {
	146	sClass = new SClass(uri, null, numOfInstances);
	147	classMap.put(uri, sClass);
	148	}
	149	if (labelLiteral != null) {
	150	String label = labelLiteral.getString();
	151	String lang = labelLiteral.getLanguage();
	152	sClass.addLabel(new Label(label, lang));
	153	}
	154	}
	155	}
	156	qexec.close();
	157	return classMap.values().toArray(new SClass[0]);
	158
	159	}
	160
	161	/*
	162
	163	public Instance[] getInstances(String[] graphURIs, String keyword) throws Exception;
	164	*/
	165
	166
	167	public ClassLink[] getNextClass(String[] graphURIs, String originClass, int limit, boolean countLinks) throws Exception{
	168	String[] targetGraphURIs = filterGraphURIs(graphURIs);
	169
	170	StringBuffer queryStr = new StringBuffer();
	171	queryStr.append("PREFIX owl: <http://www.w3.org/2002/07/owl#>\n");
	172	queryStr.append("PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>\n");
	173	queryStr.append("PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>\n");
	174
	175	// SELECT
	176	queryStr.append("SELECT DISTINCT ?c ?d ?p ?numInsStart ?numInsEnd ?numTriples \n");
	177
	178	if (targetGraphURIs != null) {
	179	for (String graphURI : targetGraphURIs) {
	180	queryStr.append("FROM <");
	181	queryStr.append(graphURI);
	182	queryStr.append(">\n");
	183	}
	184	}
	185
	186	queryStr.append("WHERE{\n");
	187	queryStr.append(" ?cr rdf:type <http://sparqlbuilder.org/ClassRelation>. \n");
[88]	188	queryStr.append(" <" + originClass + "> <http://sparqlbuilder.org/numberOfInstances> ?numInsStart. \n");
[80]	189	queryStr.append(" {");
	190	queryStr.append(" ?cr <http://sparqlbuilder.org/startClass> <" + originClass + ">. \n");
	191	queryStr.append(" ?cr <http://sparqlbuilder.org/endClass> ?c. \n");
	192	queryStr.append(" ?cr <http://sparqlbuilder.org/property> ?p. \n");
	193	queryStr.append(" ?cr <http://sparqlbuilder.org/numberOfTriples> ?numTriples. \n");
	194	queryStr.append(" ?c <http://sparqlbuilder.org/numberOfInstances> ?numInsEnd. \n");
	195	queryStr.append("}\n");
	196	queryStr.append(" UNION\n");
	197	queryStr.append(" {");
	198	queryStr.append(" ?cr <http://sparqlbuilder.org/endClass> <" + originClass + ">. \n");
	199	queryStr.append(" ?cr <http://sparqlbuilder.org/startClass> ?d. \n");
	200	queryStr.append(" ?cr <http://sparqlbuilder.org/property> ?p. \n");
	201	queryStr.append(" ?cr <http://sparqlbuilder.org/numberOfTriples> ?numTriples.\n");
	202	queryStr.append(" ?d <http://sparqlbuilder.org/numberOfInstances> ?numInsEnd. \n");
	203	queryStr.append("}\n");
	204	queryStr.append("}\n");
	205
	206
	207	if (limit > 0) {
	208	queryStr.append("limit ");
	209	queryStr.append(limit);
	210	queryStr.append("\n");
	211	}
	212
	213	// System.out.println("getNextClasses SPARQL Query: ");
	214	// System.out.println(queryStr.toString());
	215
	216	Query query = QueryFactory.create(queryStr.toString());
	217	QueryExecution qexec = null;
	218	ResultSet results = null;
	219	try {
	220	long start = System.currentTimeMillis();
	221	qexec = QueryExecutionFactory.create(query, model);
	222	results = qexec.execSelect();
	223	long end = System.currentTimeMillis();
	224	System.out.println("EXEC TIME: " + (end - start));
	225	} catch (Exception ex) {
	226	ex.printStackTrace();
	227	throw ex;
	228	}
	229
	230	ArrayList<ClassLink> solCLs = new ArrayList<ClassLink>();
	231	for (; results.hasNext();) {
	232	QuerySolution sol = results.next();
	233	Resource pro = sol.getResource("p");
	234	String clsURI = null;
	235	if (pro != null) {
	236	String proURI = pro.getURI();
	237	Resource ccls = sol.getResource("c");
	238	Resource dcls = sol.getResource("d");
	239	Direction direction = null;
	240	if(ccls != null && dcls == null ){
	241	// direction forward
	242	direction = Direction.forward;
	243	clsURI = ccls.getURI();
	244	}else{
	245	if( ccls == null && dcls != null ){
	246	direction = Direction.reverse;
	247	clsURI = dcls.getURI();
	248	}
	249	}
	250	int numTriples = 0;
	251	Literal numTriplesLit = sol.getLiteral("numTriples");
	252	if( numTriplesLit != null ){
	253	numTriples = numTriplesLit.getInt();
	254	}
[88]	255	int numInsStart = 0;
	256	Literal numInsStartLit = sol.getLiteral("numInsStart");
	257	if( numInsStartLit != null ){
	258	numInsStart = numInsStartLit.getInt();
	259	}
	260	int numInsEnd = 0;
	261	Literal numInsEndLit = sol.getLiteral("numInsEnd");
	262	if( numInsEndLit != null ){
	263	numInsEnd = numInsEndLit.getInt();
	264	}
[80]	265	ClassLink cl = new ClassLink(proURI, clsURI, direction,
[88]	266	numTriples, 0, 0, numInsEnd, numInsStart);
[80]	267	solCLs.add(cl);
	268	}
	269	}
	270	qexec.close();
	271	return solCLs.toArray(new ClassLink[0]);
	272	}
	273
	274
	275
	276	/*
	277
	278	public ClassLink[] getNextClassViaInstanceLink(String[] graphURIs, String originClass, int limit) throws Exception;
	279
	280	public Path[] getPaths(String startClass, String endClass, int mode, boolean countLinks) throws Exception;
	281
	282	public String createSPARQL(Path path) throws Exception;
	283
	284	InstanceLink[] getNextInstancesViaInstanceLink(String[] graphURIs, String originInstance,
	285	int limit) throws Exception;
	286	*/
	287
	288	public LabelMap[] getLabels(String[] graphURIs, String[] resourceURIs,
	289	String language) throws Exception {
	290	if (resourceURIs == null \|\| resourceURIs.length == 0) {
	291	return new LabelMap[0];
	292	}
	293	StringBuffer queryStr = new StringBuffer();
	294	queryStr.append("PREFIX owl: <http://www.w3.org/2002/07/owl#>\n");
	295	queryStr.append("PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>\n");
	296	queryStr.append("PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>\n");
	297	queryStr.append("SELECT DISTINCT ?res ?label \n");
	298	if (graphURIs != null) {
	299	for (String graphURI : graphURIs) {
	300	queryStr.append("FROM <");
	301	queryStr.append(graphURI);
	302	queryStr.append(">\n");
	303	}
	304	}
	305	queryStr.append("WHERE{\n");
	306	queryStr.append(" ?res rdfs:label ?label.\n");
	307	queryStr.append(" FILTER(?res IN (");
	308	boolean f = false;
	309	for (String resourceURI : resourceURIs) {
	310	if (f) {
	311	queryStr.append(", ");
	312	}
	313	f = true;
	314	queryStr.append("<");
	315	queryStr.append(resourceURI);
	316	queryStr.append(">");
	317	}
	318	queryStr.append("))\n");
	319	queryStr.append("}");
	320
	321	System.out.println(queryStr.toString());
	322
	323	Query query = QueryFactory.create(queryStr.toString());
	324	QueryExecution qexec = QueryExecutionFactory.create(query, model);
	325
	326	ResultSet results = qexec.execSelect();
	327	HashMap<String, LabelMap> lMap = new HashMap<String, LabelMap>();
	328	for (; results.hasNext();) {
	329	QuerySolution sol = results.next();
	330	String uri = sol.getResource("res").getURI();
	331	Literal literal = sol.getLiteral("label");
	332	if (literal != null) {
	333	String label = literal.getString();
	334	String lang = literal.getLanguage();
	335	if (language != null && language.equals(lang)) {
	336	Label lbl = new Label(label, lang);
	337	if (lMap.containsKey(uri)) {
	338	LabelMap lm = lMap.get(uri);
	339	lm.addLabel(lbl);
	340	} else {
	341	LabelMap lm = new LabelMap(uri, new Label[] { lbl });
	342	lMap.put(uri, lm);
	343	}
	344	}
	345	}
	346	}
	347	return lMap.values().toArray(new LabelMap[0]);
	348	}
	349	/*
	350	public ClassLink[] countLinks(String[] graphURIs, String startClassURI,
	351	ClassLink[] classLinks) throws Exception;
	352
	353	public SClass[] countInstances(String[] graphURIs, SClass[] classes) throws Exception;
	354
	355
	356	*/
	357	}

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