[17] | 1 | package hozo.sparql.plugin.compare;
|
---|
[9] | 2 |
|
---|
| 3 | import java.io.File;
|
---|
| 4 | import java.io.FileInputStream;
|
---|
| 5 | import java.io.FileNotFoundException;
|
---|
| 6 | import java.io.FileOutputStream;
|
---|
| 7 | import java.io.IOException;
|
---|
| 8 | import java.io.OutputStreamWriter;
|
---|
| 9 | import java.io.PrintWriter;
|
---|
| 10 | import java.io.UnsupportedEncodingException;
|
---|
| 11 | import java.sql.Timestamp;
|
---|
| 12 | import java.util.ArrayList;
|
---|
| 13 | import java.util.Date;
|
---|
| 14 | import java.util.HashMap;
|
---|
| 15 | import java.util.List;
|
---|
| 16 | import java.util.Map;
|
---|
| 17 |
|
---|
| 18 | import com.hp.hpl.jena.rdf.model.RDFNode;
|
---|
[35] | 19 | //import com.ibm.icu.text.SimpleDateFormat;
|
---|
[9] | 20 |
|
---|
[17] | 21 | import hozo.sparql.EndpointSettings;
|
---|
| 22 | import hozo.sparql.EndpointSettingsManager;
|
---|
| 23 | import hozo.sparql.PlainSparqlAccessor;
|
---|
| 24 | import hozo.sparql.SparqlQueryListener;
|
---|
[9] | 25 |
|
---|
| 26 | public class CompareSubject {
|
---|
| 27 |
|
---|
| 28 | private File csvFile;
|
---|
| 29 |
|
---|
| 30 | private File propertyFile;
|
---|
| 31 | private HashMap<String, PlainSparqlAccessor> accessorMap;
|
---|
| 32 |
|
---|
| 33 | private static final String SEPARATOR = ",";
|
---|
| 34 |
|
---|
| 35 | private Thread thread;
|
---|
| 36 | private boolean finalizeThread = false;
|
---|
| 37 | private SparqlQueryListener listener;
|
---|
| 38 |
|
---|
| 39 | public CompareSubject(File csvFile, File endpointFile){
|
---|
| 40 | this(csvFile, endpointFile, null);
|
---|
| 41 | }
|
---|
| 42 |
|
---|
| 43 | public CompareSubject(File csvFile, File endpointFile, SparqlQueryListener listener){
|
---|
| 44 | this.csvFile = csvFile;
|
---|
| 45 | this.propertyFile = endpointFile;
|
---|
| 46 | this.listener = listener;
|
---|
| 47 |
|
---|
| 48 | this.accessorMap = new HashMap<String, PlainSparqlAccessor>();
|
---|
| 49 | }
|
---|
| 50 |
|
---|
| 51 |
|
---|
| 52 | public void outputResult(File out, CompareResultListener listener){
|
---|
| 53 | thread = new QueryThread(new Object[]{out}, listener){
|
---|
| 54 | public void run(){
|
---|
| 55 | try {
|
---|
| 56 | File file = (File)((Object[])getOption())[0];
|
---|
| 57 | getCompareResultListener().resultReceived(outputResult(file));
|
---|
| 58 | } catch(Exception e){
|
---|
| 59 | throw new RuntimeException(e);
|
---|
| 60 | }
|
---|
| 61 | }
|
---|
| 62 | };
|
---|
| 63 | thread.setUncaughtExceptionHandler(listener);
|
---|
| 64 | thread.start();
|
---|
| 65 | }
|
---|
| 66 |
|
---|
| 67 |
|
---|
| 68 | private boolean outputResult(File out) throws IOException {
|
---|
| 69 | if (out.getParentFile() != null && !out.getParentFile().exists()){
|
---|
| 70 | out.getParentFile().mkdirs();
|
---|
| 71 | }
|
---|
| 72 | out.createNewFile();
|
---|
| 73 | if (!out.canWrite()){
|
---|
| 74 | return false;
|
---|
| 75 | }
|
---|
| 76 |
|
---|
| 77 | return outputResult(new PrintWriter(new OutputStreamWriter(new FileOutputStream(out), "SJIS")));
|
---|
| 78 | }
|
---|
| 79 |
|
---|
| 80 | public void stop(){
|
---|
| 81 | println(getCurrentTime()+"Stop Request");
|
---|
| 82 | finalizeThread = true;
|
---|
| 83 | }
|
---|
| 84 |
|
---|
| 85 |
|
---|
| 86 | private boolean outputResult(PrintWriter out){
|
---|
| 87 | HashMap<String, List<String>> propertyHash = readPropertyFile();
|
---|
| 88 |
|
---|
| 89 | getResources(propertyHash, out);
|
---|
| 90 |
|
---|
| 91 | out.flush();
|
---|
| 92 | out.close();
|
---|
| 93 |
|
---|
| 94 | println(getCurrentTime()+"query end.");
|
---|
| 95 |
|
---|
| 96 | return true; // TODO
|
---|
| 97 | }
|
---|
| 98 |
|
---|
| 99 | private HashMap<String, List<String>> readPropertyFile(){
|
---|
| 100 | HashMap<String, List<String>> propertyHash = new HashMap<String, List<String>>();
|
---|
| 101 | List<String> contents = FileUtil.readFileText(propertyFile, "UTF-8");
|
---|
| 102 |
|
---|
| 103 | for (String line : contents){
|
---|
| 104 | List<String> property = FileUtil.splitLine(line, SEPARATOR);
|
---|
| 105 | if (property != null && property.size() > 1){
|
---|
| 106 | // 荳繧ォ繝ゥ繝逶ョ縺ッendpoint
|
---|
| 107 | List<String> properties = new ArrayList<String>();
|
---|
| 108 | propertyHash.put(property.get(0), properties);
|
---|
| 109 | for (int i=1; i<property.size(); i++){
|
---|
| 110 | properties.add(property.get(i));
|
---|
| 111 | }
|
---|
| 112 | }
|
---|
| 113 | }
|
---|
| 114 |
|
---|
| 115 | return propertyHash;
|
---|
| 116 | }
|
---|
| 117 |
|
---|
| 118 |
|
---|
| 119 | private void println(){
|
---|
| 120 | println("");
|
---|
| 121 | }
|
---|
| 122 |
|
---|
| 123 | private void println(String str){
|
---|
| 124 | print(str + "\n");
|
---|
| 125 | }
|
---|
| 126 |
|
---|
| 127 | private void print(String str){
|
---|
| 128 | if (listener != null){
|
---|
| 129 | listener.sparqlExecuted(str);
|
---|
| 130 | }
|
---|
| 131 | }
|
---|
| 132 |
|
---|
| 133 |
|
---|
| 134 | private void getResources(HashMap<String, List<String>> propertyHash, PrintWriter out){
|
---|
| 135 | List<String> endpoints = null;
|
---|
| 136 | List<String> files = FileUtil.readFileText(csvFile, "SJIS");
|
---|
| 137 |
|
---|
| 138 | println();
|
---|
| 139 | println(getCurrentTime()+"query start.");
|
---|
| 140 |
|
---|
| 141 | if (files != null && files.size() > 0){
|
---|
| 142 | // 荳陦檎岼縺ッ繝倥ャ繝
|
---|
| 143 | endpoints = getEndpoints(files.get(0));
|
---|
| 144 |
|
---|
| 145 | setEndpoint(endpoints);
|
---|
| 146 | }
|
---|
| 147 |
|
---|
| 148 | // header蜃コ蜉� |
---|
| 149 | StringBuilder line = new StringBuilder();
|
---|
| 150 | line.append("");
|
---|
| 151 | line.append(SEPARATOR);
|
---|
| 152 | for (String endpoint : endpoints){
|
---|
| 153 | line.append(endpoint);
|
---|
| 154 | line.append(SEPARATOR);
|
---|
| 155 | line.append("");
|
---|
| 156 | line.append(SEPARATOR);
|
---|
| 157 | line.append("");
|
---|
| 158 | line.append(SEPARATOR);
|
---|
| 159 | }
|
---|
| 160 | out.append(line.toString());
|
---|
| 161 | out.append("\n");
|
---|
| 162 |
|
---|
| 163 | for (int i=1; i<files.size(); i++){
|
---|
| 164 | if (!finalizeThread){
|
---|
| 165 | getResource(endpoints, FileUtil.splitLine(files.get(i), SEPARATOR), propertyHash, out);
|
---|
| 166 | } else {
|
---|
| 167 | break;
|
---|
| 168 | }
|
---|
| 169 | }
|
---|
| 170 |
|
---|
| 171 | }
|
---|
| 172 |
|
---|
| 173 | /**
|
---|
| 174 | * 荳陦後#縺ィ縺ォ
|
---|
| 175 | * @param endpoints
|
---|
| 176 | * @param contents
|
---|
| 177 | * @param propertyHash
|
---|
| 178 | * @param out
|
---|
| 179 | */
|
---|
| 180 | private void getResource(List<String> endpoints, List<String> contents, HashMap<String, List<String>> propertyHash, PrintWriter out){
|
---|
| 181 | int index=1;
|
---|
| 182 | int max = 0;
|
---|
| 183 |
|
---|
| 184 | String label = contents.get(0);
|
---|
| 185 | List<HashMap<String, String>> ret = new ArrayList<HashMap<String,String>>();
|
---|
| 186 |
|
---|
| 187 | print(getCurrentTime()+"searching ["+label+"]");
|
---|
| 188 |
|
---|
| 189 | for (String endpoint : endpoints){
|
---|
| 190 | print("*");
|
---|
| 191 | HashMap<String, String> properToResultHash = new HashMap<String, String>();
|
---|
| 192 | ret.add(properToResultHash);
|
---|
| 193 | // 繧ィ繝ウ繝峨�繧、繝ウ繝医#縺ィ縺ョ繝ェ繧ス繝シ繧ケ蜿門セ� |
---|
| 194 | List<String> properties = propertyHash.get(endpoint);
|
---|
| 195 | if (properties != null){
|
---|
| 196 | HashMap<String, List<String>> results = getResource(endpoint, contents.get(index), properties);
|
---|
| 197 | if (max < results.size()){
|
---|
| 198 | max = results.size();
|
---|
| 199 | }
|
---|
| 200 | for (String property : results.keySet()){
|
---|
| 201 | // property縺斐→縺ョ邨先棡
|
---|
| 202 | String result = getResult(results.get(property));
|
---|
| 203 | properToResultHash.put(property, result);
|
---|
| 204 | }
|
---|
| 205 | }
|
---|
| 206 | index++;
|
---|
| 207 | }
|
---|
| 208 | println();
|
---|
| 209 |
|
---|
| 210 | // TODO properToResultHash縺ョ譛螟ァ蛟、繧貞叙蠕励☆繧� |
---|
| 211 | // 縺昴�謨ー縺縺題。後r霑ス蜉縺吶k縲� |
---|
| 212 |
|
---|
| 213 | /* 1, endpoint1, , endpoint2, ,.....
|
---|
| 214 | * label1, property1-1, object1-1, property2-1, object2-1 ,....o縲縲竊� |
---|
| 215 | * label1, property1-2, object1-2, property2-2, object2-2 ,....o 縺薙�遽�峇繧貞昇繧� |
---|
| 216 | * label1, property1-3, object1-3, , ,....o縲縲竊� |
---|
| 217 | * label2, property1-4, object1-1, property2-3, object2-3 ,.....
|
---|
| 218 | * label2, property1-5, object1-2, property2-4, object2-4 ,.....
|
---|
| 219 | * label2, , , property2-5, object2-5 ,.....
|
---|
| 220 | */
|
---|
| 221 |
|
---|
| 222 |
|
---|
| 223 | for (int i=0; i<max; i++){
|
---|
| 224 | StringBuilder line = new StringBuilder();
|
---|
| 225 | line.append(label);
|
---|
| 226 | line.append(SEPARATOR);
|
---|
| 227 | int lp = 1;
|
---|
| 228 | for (HashMap<String, String> values : ret){
|
---|
| 229 | String[] hoge = values.keySet().toArray(new String[]{});
|
---|
| 230 | if (hoge.length > i && !contents.get(lp).trim().isEmpty()){
|
---|
| 231 | String resultProperty = hoge[i];
|
---|
| 232 | String resultObject = values.get(resultProperty);
|
---|
| 233 | line.append(contents.get(lp));
|
---|
| 234 | line.append(SEPARATOR);
|
---|
| 235 | line.append(resultProperty);
|
---|
| 236 | line.append(SEPARATOR);
|
---|
| 237 | line.append(resultObject);
|
---|
| 238 | line.append(SEPARATOR);
|
---|
| 239 | } else {
|
---|
| 240 | line.append(SEPARATOR);
|
---|
| 241 | line.append(SEPARATOR);
|
---|
| 242 | line.append(SEPARATOR);
|
---|
| 243 | }
|
---|
| 244 | lp++;
|
---|
| 245 | }
|
---|
| 246 | out.append(line.toString());
|
---|
| 247 | out.append("\n");
|
---|
| 248 | }
|
---|
| 249 |
|
---|
| 250 | }
|
---|
| 251 |
|
---|
| 252 | private String getResult(List<String> result){
|
---|
| 253 | if (result.size() == 0){
|
---|
| 254 | return "";
|
---|
| 255 | }
|
---|
| 256 | if (result.size() == 1){
|
---|
| 257 | return result.get(0);
|
---|
| 258 | }
|
---|
| 259 |
|
---|
| 260 | StringBuilder sb = new StringBuilder();
|
---|
| 261 | for (int i=0; i<result.size()-1; i++){
|
---|
| 262 | sb.append(result.get(i));
|
---|
| 263 | sb.append("<>"); // TODO 繧サ繝代Ξ繝シ繧ソ
|
---|
| 264 | }
|
---|
| 265 | sb.append(result.get(result.size()-1));
|
---|
| 266 |
|
---|
| 267 | return sb.toString();
|
---|
| 268 | }
|
---|
| 269 |
|
---|
| 270 | /**
|
---|
| 271 | *
|
---|
| 272 | * @param endpoint
|
---|
| 273 | * @param content
|
---|
| 274 | * @param properties
|
---|
| 275 | * @return property縺ィ縺昴�result縺ョmap
|
---|
| 276 | */
|
---|
| 277 | private HashMap<String, List<String>> getResource(String endpoint, String content, List<String> properties){
|
---|
| 278 | HashMap<String, List<String>> ret = new HashMap<String, List<String>>();
|
---|
| 279 | PlainSparqlAccessor sa = accessorMap.get(endpoint);
|
---|
| 280 | for (String property : properties){
|
---|
| 281 | List<String> res = new ArrayList<String>();
|
---|
| 282 | ret.put(property, res);
|
---|
| 283 |
|
---|
| 284 | if (!content.trim().isEmpty()){
|
---|
| 285 | String query = makeQuery(content, property);
|
---|
| 286 | try {
|
---|
| 287 | List<Map<String, RDFNode>> results = sa.executeQuery(query);
|
---|
| 288 | if (results != null){
|
---|
| 289 | for (Map<String, RDFNode> result : results){
|
---|
| 290 | RDFNode s = result.get("o");
|
---|
| 291 | res.add(s.toString());
|
---|
| 292 | }
|
---|
| 293 | }
|
---|
| 294 | } catch (Exception e) {
|
---|
| 295 | e.printStackTrace();
|
---|
| 296 | }
|
---|
| 297 | }
|
---|
| 298 | }
|
---|
| 299 |
|
---|
| 300 | return ret;
|
---|
| 301 | }
|
---|
| 302 |
|
---|
| 303 | private String makeQuery(String s, String p){
|
---|
| 304 | return "select distinct ?o where {\n" +
|
---|
| 305 | "<" + s.trim() + "> <" + p.trim() + "> ?o \n" +
|
---|
| 306 | "}";
|
---|
| 307 |
|
---|
| 308 | }
|
---|
| 309 |
|
---|
| 310 | public void addEndpoint(String endpoint){
|
---|
| 311 | EndpointSettings settings = EndpointSettingsManager.instance.getSetting(endpoint);
|
---|
| 312 |
|
---|
| 313 | PlainSparqlAccessor sa = new PlainSparqlAccessor(settings);
|
---|
| 314 | accessorMap.put(endpoint, sa);
|
---|
| 315 | }
|
---|
| 316 |
|
---|
| 317 | private void setEndpoint(List<String> endpoints){
|
---|
| 318 | for (String ep : endpoints){
|
---|
| 319 | addEndpoint(ep);
|
---|
| 320 | }
|
---|
| 321 | }
|
---|
| 322 |
|
---|
| 323 | private List<String> getEndpoints(String line){
|
---|
| 324 | List<String> headers = FileUtil.splitLine(line, SEPARATOR);
|
---|
| 325 |
|
---|
| 326 | if (headers.size() > 0){
|
---|
| 327 | headers.remove(0);
|
---|
| 328 | }
|
---|
| 329 |
|
---|
| 330 | return headers;
|
---|
| 331 | }
|
---|
| 332 |
|
---|
| 333 | private String getCurrentTime(){
|
---|
| 334 | Timestamp time = new Timestamp(new Date().getTime());
|
---|
[35] | 335 | return null;
|
---|
| 336 | // return new SimpleDateFormat("yyyy/MM/dd HH:mm:ss").format(time) + ":";
|
---|
[9] | 337 | }
|
---|
| 338 |
|
---|
| 339 |
|
---|
| 340 | private class QueryThread extends Thread {
|
---|
| 341 |
|
---|
| 342 | private Object option;
|
---|
| 343 | private CompareResultListener listener;
|
---|
| 344 |
|
---|
| 345 |
|
---|
| 346 | public QueryThread(Object option, CompareResultListener listener){
|
---|
| 347 | this.option = option;
|
---|
| 348 | this.listener = listener;
|
---|
| 349 | }
|
---|
| 350 |
|
---|
| 351 | protected Object getOption(){
|
---|
| 352 | return this.option;
|
---|
| 353 | }
|
---|
| 354 |
|
---|
| 355 | protected CompareResultListener getCompareResultListener(){
|
---|
| 356 | return this.listener;
|
---|
| 357 | }
|
---|
| 358 |
|
---|
| 359 | }
|
---|
| 360 |
|
---|
| 361 | public static void main(String[] args){
|
---|
| 362 | String epFile = "C:\\Users\\kato\\Desktop\\out\\endp_prop.txt";
|
---|
| 363 | String inFile = "C:\\works\\daily\\20130926\\input\\蜍戊ゥ樒エ「蠑�csv";
|
---|
| 364 | String outFile = "C:\\works\\daily\\20130926\\input\\蜍戊ゥ樒エ「蠑廟out.csv";
|
---|
| 365 |
|
---|
| 366 | try {
|
---|
| 367 | /*
|
---|
| 368 | for (int i=0; i<args.length; i++){
|
---|
| 369 | if (args[i].startsWith("-") && i < (args.length-1)){
|
---|
| 370 |
|
---|
| 371 | if (args[i].equals("-o")){
|
---|
| 372 | outFile = args[i+1];
|
---|
| 373 | }
|
---|
| 374 | if (args[i].equals("-i")){
|
---|
| 375 | inFile = args[i+1];
|
---|
| 376 | }
|
---|
| 377 | if (args[i].equals("-e")){
|
---|
| 378 | epFile = args[i+1];
|
---|
| 379 | }
|
---|
| 380 | i++;
|
---|
| 381 | }
|
---|
| 382 |
|
---|
| 383 | }
|
---|
| 384 | */
|
---|
| 385 |
|
---|
| 386 | EndpointSettings[] settings = EndpointSettings.inputXML(new FileInputStream("settings.xml"));
|
---|
| 387 | if (settings != null){
|
---|
| 388 | EndpointSettingsManager.instance.setSettings(settings);
|
---|
| 389 |
|
---|
| 390 | }
|
---|
| 391 |
|
---|
| 392 |
|
---|
| 393 | if (epFile == null || inFile == null || outFile == null){
|
---|
| 394 | System.out.println("usage:java -jar SparqlTestTool.jar Compare -e endpointFile -i inputFile -o outputFile -t type(s:subject/l:label object/o:all object)");
|
---|
| 395 | } else {
|
---|
| 396 | CompareSubject ct = new CompareSubject(new File(inFile), new File(epFile));
|
---|
| 397 | ct.outputResult(new File(outFile));
|
---|
| 398 | }
|
---|
| 399 | } catch (UnsupportedEncodingException e) {
|
---|
| 400 | // TODO 閾ェ蜍慕函謌舌&繧後◆ catch 繝悶Ο繝�け
|
---|
| 401 | e.printStackTrace();
|
---|
| 402 | } catch (FileNotFoundException e) {
|
---|
| 403 | // TODO 閾ェ蜍慕函謌舌&繧後◆ catch 繝悶Ο繝�け
|
---|
| 404 | e.printStackTrace();
|
---|
| 405 | } catch (IOException e) {
|
---|
| 406 | // TODO 閾ェ蜍慕函謌舌&繧後◆ catch 繝悶Ο繝�け
|
---|
| 407 | e.printStackTrace();
|
---|
| 408 | }
|
---|
| 409 |
|
---|
| 410 | }
|
---|
| 411 |
|
---|
| 412 | }
|
---|