source: other-projects/trunk/greenstone3-extension/mat/src/org/greenstone/gsdl3_extension/mat/servlet/PrintHTML.java@ 18104

Last change on this file since 18104 was 18104, checked in by cc108, 15 years ago

new MatServlet source code
File size: 107.5 KB
Line 
1package org.greenstone.gsdl3_extension.mat.servlet;
2
3import java.io.*;
4import java.text.SimpleDateFormat;
5import java.util.ArrayList;
6import java.util.Calendar;
7import java.util.Collection;
8import java.util.HashMap;
9import java.util.Iterator;
10import java.util.Map;
11import java.util.Set;
12
13import javax.xml.parsers.DocumentBuilder;
14import javax.xml.parsers.DocumentBuilderFactory;
15
16import org.w3c.dom.Document;
17import org.w3c.dom.Element;
18import org.w3c.dom.Node;
19import org.w3c.dom.NodeList;
20import org.greenstone.gsdl3.*;
21import org.greenstone.gsdl3.util.GlobalProperties;
22
23public class PrintHTML {
24
25 private final String NORMAL = "normal";
26 private final String WORST = "worst";
27 private final String BEST = "best";
28 private final String title1 ="Hide Empty Metadata ,";
29 private final String title2 ="Hide Completed Metadata ,";
30 private final String title3 ="Hide Documents with Empty Metadata ,";
31 private final String title4 ="Hide Documents with Completed Metadata ,";
32 private final String title5 ="No Available Graph";
33 private final String metadataSet1 = "Dublin Core";
34 private final String metadataSet2 = "Extracted";
35 private final String spaceLeft = "<I>&laquo;";
36 private final String spaceRight = "&raquo;</I>";
37 private final String htmlSpace = "&nbsp;";
38 private final String oddChar ="odd Character";
39 private final String destination;
40 private final String space ="space";
41 private String hostName;
42 private String cssString;
43 private int port_number = 9090;
44 private GlobalProperties globalProperty = null;
45 private String fileSeparator = File.separator;
46
47 String linkIdentifier = "";
48
49
50
51 private HashMap url;
52
53 MetadataStats mds;
54 DataMaker dm;
55 Calendar cl;
56 SimpleDateFormat sdf;
57 PrintWriter printWriter;
58
59 public PrintHTML(MetadataStats ms, PrintWriter arg){
60
61 mds = ms;
62 printWriter = arg;
63 dm = new DataMaker(mds);
64 cl=Calendar.getInstance();
65 linkIdentifier = "dc.Identifier";
66 destination = mds.HTMLDirectory;
67 sdf = new SimpleDateFormat(" dd MMM yyyy 'at' HH:mm:ss z 'GMT'Z");
68 setupMetadataLink("metadataElementURL");
69
70
71 try {
72 java.net.InetAddress localMachine = java.net.InetAddress.getLocalHost();
73 hostName = localMachine.getHostName();
74 }catch (java.net.UnknownHostException uhe) {
75 uhe.printStackTrace();
76 }
77
78 cssString = "<link rel=\"stylesheet\" href=\"http://"+hostName+":"+port_number+"/greenstone3/mat/script/mat.css\" type=\"text/css\" >";
79 }
80
81 private ArrayList getHideEmptyMetadataDetail(ArrayList dataset, String[] id, String[] metadataName){
82
83 printWriter.write("<!-- generating all possible page -->");
84 printWriter.flush();
85
86 if(id.length==0 || metadataName.length==0){
87 return new ArrayList();
88 }
89
90 ArrayList detailList = new ArrayList();
91 ArrayList alist = new ArrayList();
92 ArrayList nameList = new ArrayList();
93
94 for(int i = 0; i< metadataName.length; i++){
95 if(!dm.IsElementEmpty(metadataName[i])){
96 alist.add(dm.getMetadataRows(metadataName[i]));
97 nameList.add(metadataName[i]);
98 }
99 }
100
101 String[] metaDataElementName = new String[nameList.size()];
102
103 for(int i = 0; i< nameList.size(); i++){
104 metaDataElementName[i] = nameList.get(i).toString();
105 }
106
107 if(id.length==0 || metaDataElementName.length==0){
108 return new ArrayList();
109 }
110
111 detailList.add(alist);
112 detailList.add(id);
113 detailList.add(metaDataElementName);
114 printWriter.write("<!-- generating all possible page-->");
115 printWriter.flush();
116
117 return detailList;
118 }
119
120 private ArrayList getHideFullMetadataDetail(ArrayList dataset, String[] id, String[] metadataName){
121
122 printWriter.write("<!-- generating all possible page-->");
123 printWriter.flush();
124
125 if(id.length==0 || metadataName.length==0){
126 return new ArrayList();
127 }
128
129 ArrayList detailList = new ArrayList();
130 ArrayList alist = new ArrayList();
131 ArrayList nameList = new ArrayList();
132
133 for(int i = 0; i< metadataName.length; i++){
134 if(!dm.IsElementFull(metadataName[i])){
135 alist.add(dm.getMetadataRows(metadataName[i]));
136 nameList.add(metadataName[i]);
137 }
138 }
139
140 String[] metaDataElementName = new String[nameList.size()];
141
142 for(int i = 0; i< nameList.size(); i++){
143 metaDataElementName[i] = nameList.get(i).toString();
144 }
145
146 if(id.length==0 || metaDataElementName.length==0){
147 return new ArrayList();
148 }
149
150 detailList.add(alist);
151 detailList.add(id);
152 detailList.add(metaDataElementName);
153
154 printWriter.write("<!-- generating all possible page-->");
155 printWriter.flush();
156
157 return detailList;
158 }
159
160 private ArrayList getHideEmptyDocumentDetail(ArrayList dataset, String[] id, String[] metadataName){
161
162 printWriter.write("<!-- generating all possible page-->");
163 printWriter.flush();
164
165 if(id.length==0 || metadataName.length==0){
166 return new ArrayList();
167 }
168
169 ArrayList detailList = new ArrayList();
170 ArrayList alist = dm.removeDocument(dataset, id, 0);
171 ArrayList idList = dm.getRemovedID();
172 ArrayList newIDList = new ArrayList();
173
174 for(int i = 0; i< id.length; i++){
175 if(!idList.contains(id[i])){
176 newIDList.add(id[i]);
177 }
178 }
179
180 id = new String[newIDList.size()];
181 for(int i = 0; i<newIDList.size(); i++){
182 id[i] = newIDList.get(i).toString();
183 }
184
185 if(id.length==0 || metadataName.length==0){
186 return new ArrayList();
187 }
188
189 detailList.add(alist);
190 detailList.add(id);
191 detailList.add(metadataName);
192
193 printWriter.write("<!-- generating all possible page-->");
194 printWriter.flush();
195
196 return detailList;
197 }
198
199 private ArrayList getHideFullDocumentDetail(ArrayList dataset, String[] id, String[] metadataName){
200
201 printWriter.write("<!-- generating all possible page-->");
202 printWriter.flush();
203
204 if(id.length==0 || metadataName.length==0){
205 return new ArrayList();
206 }
207
208 ArrayList detailList = new ArrayList();
209 ArrayList alist = dm.removeDocument(dataset, id ,1);
210 ArrayList idList = dm.getRemovedID();
211 ArrayList newIDList = new ArrayList();
212
213 for(int i = 0; i< id.length; i++){
214 if(!idList.contains(id[i])){
215 newIDList.add(id[i]);
216 }
217 }
218
219 id = new String[newIDList.size()];
220 for(int i = 0; i<newIDList.size(); i++){
221 id[i] = newIDList.get(i).toString();
222 }
223
224 if(id.length==0 || metadataName.length==0){
225 return new ArrayList();
226 }
227
228 detailList.add(alist);
229 detailList.add(id);
230 detailList.add(metadataName);
231
232 printWriter.write("<!-- generating all possible page-->");
233 printWriter.flush();
234
235 return detailList;
236 }
237
238 public void generateAllPossibleGraph(ArrayList Namelist,String[]ids,String[] names,String condition){
239
240 printWriter.write("<!-- generating all possible page-->");
241 printWriter.flush();
242
243 ArrayList dataList= Namelist;
244 String[] id = ids;
245 String[] metadataName = names;
246
247 ArrayList alist = new ArrayList();
248 ArrayList detailList = new ArrayList();
249 ArrayList new_dataList = new ArrayList();
250 String[] idList;
251 String[] metadataNameList;
252
253 //---------------------------------------------------------------------------------------------------------SS
254 //-----------------------------------------------------------------------SSSS
255 if(id.length==0 || metadataName.length==0){
256 generateEmptyGraph("SSSS",WORST,condition);
257 generateEmptyGraph("SSSS",BEST,condition);
258 }
259
260 else{
261 new_dataList = transformDataList(dataList,id);
262 WriteHTML("SSSS",new_dataList,id,metadataName,WORST,"Show completed graph",condition);
263 WriteHTML("SSSS",new_dataList,id,metadataName,BEST,"Show completed graph",condition);
264 }
265 //-----------------------------------------------------------------------SSSH
266 detailList = getHideFullDocumentDetail(dataList,id,metadataName);
267 if(detailList.size()==0){
268 generateEmptyGraph("SSSH",WORST,condition);
269 generateEmptyGraph("SSSH",BEST,condition);
270 }
271 else{
272 alist = (ArrayList)detailList.get(0);
273 idList = (String[])detailList.get(1);
274 metadataNameList = (String[])detailList.get(2);
275 new_dataList = transformDataList(alist,idList);
276 WriteHTML("SSSH",new_dataList,idList,metadataNameList,WORST,title4,condition);
277 WriteHTML("SSSH",new_dataList,idList,metadataNameList,BEST,title4,condition);
278 }
279 //-----------------------------------------------------------------------SSHS
280 detailList = getHideEmptyDocumentDetail(dataList,id,metadataName);
281 if(detailList.size()==0){
282 generateEmptyGraph("SSHS",WORST,condition);
283 generateEmptyGraph("SSHS",BEST,condition);
284 generateEmptyGraph("SSHH",WORST,condition);
285 generateEmptyGraph("SSHH",BEST,condition);
286 }
287 else{
288 alist = (ArrayList)detailList.get(0);
289 idList = (String[])detailList.get(1);
290 metadataNameList = (String[])detailList.get(2);
291 new_dataList = transformDataList(alist,idList);
292 WriteHTML("SSHS",new_dataList,idList,metadataNameList,WORST,title3,condition);
293 WriteHTML("SSHS",new_dataList,idList,metadataNameList,BEST,title3,condition);
294 //-----------------------------------------------------------------------SSHH
295 detailList = getHideFullDocumentDetail(alist,idList,metadataNameList);
296 if(detailList.size()==0){
297 generateEmptyGraph("SSHH",WORST,condition);
298 generateEmptyGraph("SSHH",BEST,condition);
299 }
300 else{
301 alist = (ArrayList)detailList.get(0);
302 idList = (String[])detailList.get(1);
303 metadataNameList = (String[])detailList.get(2);
304 new_dataList = transformDataList(alist,idList);
305 WriteHTML("SSHH",new_dataList,idList,metadataNameList,WORST,title3+title4,condition);
306 WriteHTML("SSHH",new_dataList,idList,metadataNameList,BEST,title3+title4,condition);
307 }
308 }
309
310 //---------------------------------------------------------------------------------------------------------SH
311 ArrayList xList = new ArrayList();
312 String[] idListCopy;
313 String[] metadataNameListCopy;
314
315 //-----------------------------------------------------------------------SHSS
316 detailList = getHideFullMetadataDetail(dataList,id,metadataName);
317
318 if(detailList.size()==0){
319 generateEmptyGraph("SHSS",WORST,condition);
320 generateEmptyGraph("SHSS",BEST,condition);
321
322 generateEmptyGraph("SHSH",WORST,condition);
323 generateEmptyGraph("SHSH",BEST,condition);
324
325 generateEmptyGraph("SHHS",WORST,condition);
326 generateEmptyGraph("SHHS",BEST,condition);
327
328 generateEmptyGraph("SHHH",WORST,condition);
329 generateEmptyGraph("SHHH",BEST,condition);
330 }
331 else{
332 alist = (ArrayList)detailList.get(0);
333 idList = (String[])detailList.get(1);
334 metadataNameList = (String[])detailList.get(2);
335
336 xList = (ArrayList)alist.clone();
337 idListCopy = (String[]) idList.clone();
338 metadataNameListCopy = (String[])metadataNameList.clone();
339 new_dataList = transformDataList(alist,idList);
340 WriteHTML("SHSS",new_dataList,idList,metadataNameList,WORST,title2,condition);
341 WriteHTML("SHSS",new_dataList,idList,metadataNameList,BEST,title2,condition);
342
343 //-----------------------------------------------------------------------SHHS
344 detailList = getHideEmptyDocumentDetail((ArrayList)xList.clone(),(String[])idListCopy.clone(),(String[])metadataNameListCopy.clone());
345 if(detailList.size()==0){
346 generateEmptyGraph("SHHS",WORST,condition);
347 generateEmptyGraph("SHHS",BEST,condition);
348 generateEmptyGraph("SHHH",WORST,condition);
349 generateEmptyGraph("SHHH",BEST,condition);
350 }
351 else{
352 alist = (ArrayList)detailList.get(0);
353 idList = (String[])detailList.get(1);
354 metadataNameList = (String[])detailList.get(2);
355 new_dataList = transformDataList(alist,idList);
356 WriteHTML("SHHS",new_dataList,idList,metadataNameList,WORST,title2+title3,condition);
357 WriteHTML("SHHS",new_dataList,idList,metadataNameList,BEST,title2+title3,condition);
358
359 //-----------------------------------------------------------------------SHHH
360 detailList = getHideFullDocumentDetail(alist,idList,metadataNameList);
361 if(detailList.size()==0){
362 generateEmptyGraph("SHHH",WORST,condition);
363 generateEmptyGraph("SHHH",BEST,condition);
364 }
365 else{
366 alist = (ArrayList)detailList.get(0);
367 idList = (String[])detailList.get(1);
368 metadataNameList = (String[])detailList.get(2);
369 new_dataList = transformDataList(alist,idList);
370 WriteHTML("SHHH",new_dataList,idList,metadataNameList,WORST,title2+title3+title4,condition);
371 WriteHTML("SHHH",new_dataList,idList,metadataNameList,BEST,title2+title3+title4,condition);
372 }
373 }
374
375 //-----------------------------------------------------------------------SHSH
376 detailList = getHideFullDocumentDetail((ArrayList)xList.clone(),(String[])idListCopy.clone(),(String[])metadataNameListCopy.clone());
377 if(detailList.size()==0){
378 generateEmptyGraph("SHSH",WORST,condition);
379 generateEmptyGraph("SHSH",BEST,condition);
380 }
381 else{
382 alist = (ArrayList)detailList.get(0);
383 idList = (String[])detailList.get(1);
384 metadataNameList = (String[])detailList.get(2);
385 new_dataList = transformDataList(alist,idList);
386 WriteHTML("SHSH",new_dataList,idList,metadataNameList,WORST,title2+title4,condition);
387 WriteHTML("SHSH",new_dataList,idList,metadataNameList,BEST,title2+title4,condition);
388 }
389 }
390
391 //---------------------------------------------------------------------------------------------------------HS
392 ArrayList xList1 = new ArrayList();
393 String[] idListCopy1;
394 String[] metadataNameListCopy1;
395
396 //-----------------------------------------------------------------------HSSS
397 detailList = getHideEmptyMetadataDetail(dataList,id,metadataName);
398
399 if(detailList.size()==0){
400 generateEmptyGraph("HSSS",WORST,condition);
401 generateEmptyGraph("HSSS",BEST,condition);
402
403 generateEmptyGraph("HSHS",WORST,condition);
404 generateEmptyGraph("HSHS",BEST,condition);
405
406 generateEmptyGraph("HSHH",WORST,condition);
407 generateEmptyGraph("HSHH",BEST,condition);
408
409 generateEmptyGraph("HSSH",WORST,condition);
410 generateEmptyGraph("HSSH",BEST,condition);
411 }
412 else{
413 alist = (ArrayList)detailList.get(0);
414 idList = (String[])detailList.get(1);
415 metadataNameList = (String[])detailList.get(2);
416
417 xList1 = (ArrayList)alist.clone();
418 idListCopy1 = (String[]) idList.clone();
419 metadataNameListCopy1 = (String[])metadataNameList.clone();
420
421 new_dataList = transformDataList(alist,idList);
422 WriteHTML("HSSS",new_dataList,idList,metadataNameList,WORST,title1,condition);
423 WriteHTML("HSSS",new_dataList,idList,metadataNameList,BEST,title1,condition);
424
425 //-----------------------------------------------------------------------HSHS
426 detailList = getHideEmptyDocumentDetail((ArrayList)xList1.clone(),(String[])idListCopy1.clone(),(String[])metadataNameListCopy1.clone());
427 if(detailList.size()==0){
428 generateEmptyGraph("HSHS",WORST,condition);
429 generateEmptyGraph("HSHS",BEST,condition);
430
431 generateEmptyGraph("HSHH",WORST,condition);
432 generateEmptyGraph("HSHH",BEST,condition);
433 }
434 else{
435 alist = (ArrayList)detailList.get(0);
436 idList = (String[])detailList.get(1);
437 metadataNameList = (String[])detailList.get(2);
438
439 new_dataList = transformDataList(alist,idList);
440 WriteHTML("HSHS",new_dataList,idList,metadataNameList,WORST,title1+title3,condition);
441 WriteHTML("HSHS",new_dataList,idList,metadataNameList,BEST,title1+title3,condition);
442
443 //-----------------------------------------------------------------------HSHH
444 detailList = getHideFullDocumentDetail(alist,idList,metadataNameList);
445 if(detailList.size()==0){
446 generateEmptyGraph("HSHH",WORST,condition);
447 generateEmptyGraph("HSHH",BEST,condition);
448 }
449 else{
450 alist = (ArrayList)detailList.get(0);
451 idList = (String[])detailList.get(1);
452 metadataNameList = (String[])detailList.get(2);
453
454 new_dataList = transformDataList(alist,idList);
455 WriteHTML("HSHH",new_dataList,idList,metadataNameList,WORST,title1+title3+title4,condition);
456 WriteHTML("HSHH",new_dataList,idList,metadataNameList,BEST,title1+title3+title4,condition);
457 }
458 }
459
460 //-----------------------------------------------------------------------HSSH
461 detailList = getHideFullDocumentDetail((ArrayList)xList1.clone(),(String[])idListCopy1.clone(),(String[])metadataNameListCopy1.clone());
462 if(detailList.size()==0){
463 generateEmptyGraph("HSSH",WORST,condition);
464 generateEmptyGraph("HSSH",BEST,condition);
465 }
466 else{
467 alist = (ArrayList)detailList.get(0);
468 idList = (String[])detailList.get(1);
469 metadataNameList = (String[])detailList.get(2);
470
471 new_dataList = transformDataList(alist,idList);
472 WriteHTML("HSSH",new_dataList,idList,metadataNameList,WORST,title1+title4,condition);
473 WriteHTML("HSSH",new_dataList,idList,metadataNameList,BEST,title1+title4,condition);
474 }
475 }
476
477 //---------------------------------------------------------------------------------------------------------HH
478 ArrayList xList2 = new ArrayList();
479 String[] idListCopy2;
480 String[] metadataNameListCopy2;
481
482 //-----------------------------------------------------------------------HHSS
483 detailList = getHideEmptyMetadataDetail(dataList,id,metadataName);
484 if(detailList.size()==0){
485 generateEmptyGraph("HHSS",WORST,condition);
486 generateEmptyGraph("HHSS",BEST,condition);
487
488 generateEmptyGraph("HHHS",WORST,condition);
489 generateEmptyGraph("HHHS",BEST,condition);
490
491 generateEmptyGraph("HHHH",WORST,condition);
492 generateEmptyGraph("HHHH",BEST,condition);
493
494 generateEmptyGraph("HHSH",WORST,condition);
495 generateEmptyGraph("HHSH",BEST,condition);
496 }
497 else{
498 alist = (ArrayList)detailList.get(0);
499 idList = (String[])detailList.get(1);
500 metadataNameList = (String[])detailList.get(2);
501
502 detailList = getHideFullMetadataDetail(alist,idList,metadataNameList);
503
504 if(detailList.size()==0){
505 generateEmptyGraph("HHSS",WORST,condition);
506 generateEmptyGraph("HHSS",BEST,condition);
507
508 generateEmptyGraph("HHHS",WORST,condition);
509 generateEmptyGraph("HHHS",BEST,condition);
510
511 generateEmptyGraph("HHHH",WORST,condition);
512 generateEmptyGraph("HHHH",BEST,condition);
513
514 generateEmptyGraph("HHSH",WORST,condition);
515 generateEmptyGraph("HHSH",BEST,condition);
516 }
517 else{
518 alist = (ArrayList)detailList.get(0);
519 idList = (String[])detailList.get(1);
520 metadataNameList = (String[])detailList.get(2);
521
522 xList2 = (ArrayList)alist.clone();
523 idListCopy2 = (String[]) idList.clone();
524 metadataNameListCopy2 = (String[])metadataNameList.clone();
525
526 new_dataList = transformDataList(alist,idList);
527 WriteHTML("HHSS",new_dataList,idList,metadataNameList,WORST,title1+title2,condition);
528 WriteHTML("HHSS",new_dataList,idList,metadataNameList,BEST,title1+title2,condition);
529
530 //---------------------------------------------------------------------- HHSH
531 detailList = getHideFullDocumentDetail((ArrayList)xList2.clone(),(String[])idListCopy2.clone(),(String[])metadataNameListCopy2.clone());
532 if(detailList.size()==0){
533 generateEmptyGraph("HHSH",WORST,condition);
534 generateEmptyGraph("HHSH",BEST,condition);
535 }
536 else{
537 alist = (ArrayList)detailList.get(0);
538 idList = (String[])detailList.get(1);
539 metadataNameList = (String[])detailList.get(2);
540
541 new_dataList = transformDataList(alist,idList);
542
543 WriteHTML("HHSH",new_dataList,idList,metadataNameList,WORST,title1+title2+title4,condition);
544 WriteHTML("HHSH",new_dataList,idList,metadataNameList,BEST,title1+title2+title4,condition);
545 }
546 //----------------------------------------------------------------------HHHS
547 detailList = getHideEmptyDocumentDetail((ArrayList)xList2.clone(),(String[])idListCopy2.clone(),(String[])metadataNameListCopy2.clone());
548 if(detailList.size()==0){
549 generateEmptyGraph("HHHS",WORST,condition);
550 generateEmptyGraph("HHHS",BEST,condition);
551
552 generateEmptyGraph("HHHH",WORST,condition);
553 generateEmptyGraph("HHHH",BEST,condition);
554 }
555 else{
556 alist = (ArrayList)detailList.get(0);
557 idList = (String[])detailList.get(1);
558 metadataNameList = (String[])detailList.get(2);
559
560 new_dataList = transformDataList(alist,idList);
561
562 WriteHTML("HHHS",new_dataList,idList,metadataNameList,WORST,title1+title2+title3,condition);
563 WriteHTML("HHHS",new_dataList,idList,metadataNameList,BEST,title1+title2+title3,condition);
564
565 //----------------------------------------------------------------------HHHH
566 detailList = getHideFullDocumentDetail((ArrayList)alist.clone(),(String[])idList.clone(),(String[])metadataNameList.clone());
567 if(detailList.size()==0){
568 generateEmptyGraph("HHHH",WORST,condition);
569 generateEmptyGraph("HHHH",BEST,condition);
570 }
571 else{
572 alist = (ArrayList)detailList.get(0);
573 idList = (String[])detailList.get(1);
574 metadataNameList = (String[])detailList.get(2);
575
576 new_dataList = transformDataList(alist,idList);
577
578 WriteHTML("HHHH",new_dataList,idList,metadataNameList,WORST,title1+title2+title3+title4,condition);
579 WriteHTML("HHHH",new_dataList,idList,metadataNameList,BEST,title1+title2+title3+title4,condition);
580 }
581 }
582 }
583 }
584 }
585
586 private ArrayList transformDataList(ArrayList list, String[] ids){
587
588 printWriter.write("<!-- generating all possible page-->");
589 printWriter.flush();
590
591 ArrayList wholeList = new ArrayList();
592
593 for(int i = 0; i< ids.length; i++){
594 ArrayList idList = new ArrayList();
595 for(int j = 0; j<list.size();j++){
596 int[] datarows = (int[]) list.get(j);
597 idList.add(new Integer (datarows[i]));
598 }
599 wholeList.add(idList);
600 }
601
602 for(int i = 0; i< wholeList.size(); i++){
603 ArrayList alist = (ArrayList)wholeList.get(i);
604 int[] rows = new int[alist.size()];
605 for(int j = 0; j< alist.size(); j++){
606 rows[j] = ((Integer)alist.get(j)).intValue();
607 }
608 wholeList.remove(i);
609 wholeList.add(i,rows);
610 }
611
612 printWriter.write("<!-- generating all possible page-->");
613 printWriter.flush();
614
615 return wholeList;
616
617 }
618
619 private void generateEmptyGraph(String fileName,String condition,String suffix){
620
621 printWriter.write("<!-- generating all possible page-->");
622 printWriter.flush();
623
624 String suf = suffix;
625 if(!suffix.equalsIgnoreCase("dublin")){
626 suf = "other";
627 }
628
629 String cases = condition;
630 String collectionFolder = mds.getCollectionName();
631
632 try{
633 FileWriter fstream = new FileWriter(destination+suf+"_"+fileName+"_"+cases+".html");
634 BufferedWriter out = new BufferedWriter(fstream);
635 out.write("<!-- This comment keeps IE6/7 in the reliable quirks mode -->\r\n");
636 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">\r\n");
637 out.write("<html>\r\n");
638 out.write("<head>\r\n<title> No Available Chart</title>\r\n");
639 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
640 out.write("<link rel=\"stylesheet\" href=\"../script/doby2.css\" type=\"text/css\"/>\r\n");
641 out.write("</head>\r\n");
642 out.write("<p><a href=\" Overall.html \">Summary</a></p>");
643 out.write("<body>\r\n");
644 out.write("<p>No data available to render chart.</p>\r\n");
645 out.write("<p>Reason: Document number is zero or Metadata element number is zero </p>\r\n");
646 out.write("</body></html>\r\n");
647 out.close();
648 fstream.close();
649
650 }catch(Exception e){}
651
652 printWriter.write("<!-- generating all possible page-->");
653 printWriter.flush();
654 }
655
656 private void WriteHTML(String fileName, ArrayList dataset, String[] ids, String[] metadataName, String condition, String title, String suffix){
657
658 int blueDot = 0;
659 String cases = condition;
660 String suf = suffix;
661 if(!suffix.equalsIgnoreCase("dublin")){
662 suf = "other";
663 }
664 try{
665 printWriter.write("<!-- generating all possible page-->");
666 printWriter.flush();
667
668 ArrayList tempList = mds.getMetadataNameList();
669 ArrayList urlIDList;
670 if(tempList.contains("dc.Identifier") && linkIdentifier.equals("dc.Identifier")){
671 urlIDList = dm.getURLMap("dc.Identifier");
672 }
673 else if(tempList.contains("nzir_internal.Link") && linkIdentifier.equals("nzir_internal.Link")){
674 urlIDList = dm.getURLMap("nzir_internal.Link");
675 }
676 else{
677 urlIDList = new ArrayList();
678 }
679
680 FileWriter fstream = new FileWriter(destination+suf+"_"+fileName+"_"+cases+".html");
681 BufferedWriter out = new BufferedWriter(fstream);
682
683 out.write("<!-- This comment keeps IE6/7 in the reliable quirks mode -->\r\n");
684 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">\r\n");
685 out.write("<html>\r\n");
686 out.write("<head>\r\n<title>"+title+"</title>\r\n");
687 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
688 out.write("<link rel=\"stylesheet\" href=\"../script/doby2.css\" type=\"text/css\"/>\r\n");
689 out.write("<script type=\"text/javascript\" src=\"../script/getInfomation.js\"></script>\r\n");
690 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/yahoo/yahoo-min.js\"></script>\r\n");
691 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/event/event-min.js\"></script>\r\n");
692 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/connection/connection-min.js\"></script>\r\n");
693 out.write("</head>\r\n");
694 out.write ("<p><a href=\" Overall.html \">Summary</a></p>");
695 out.write("<body id=\""+mds.getCollectionName()+"\" onLoad=\"reconfig()\">\r\n");
696 out.write("<div id=\"container\">\r\n");
697 out.write("<div class=\"tableContainer\">\r\n");
698 out.write ("<table cellspacing=\"0\">\r\n");
699 out.write ("<thead>\r\n");
700 out.write ("<tr>\r\n");
701 out.write("<td class=\"qh\">Info\r\n");
702 out.write("<td class=\"qh\">URL\r\n");
703
704 for(int a = 0; a< metadataName.length; a++){
705 out.write("<td>"+ metadataName[a]+"\r\n");
706 }
707
708 out.write ("</thead>\r\n<tfoot>\r\n<tr>\r\n");
709 out.write("<td class=\"qh\">&nbsp;\r\n");
710 out.write("<td class=\"qh\">&nbsp;\r\n");
711
712 for(int a = 0; a< metadataName.length; a++){
713 out.write("<td>"+ dm.Mean(metadataName[a])+"%\r\n");
714 }
715
716 out.write ("</tfoot>\r\n<tbody>\r\n");
717
718 if(cases.equals("normal")){
719 for(int i = ids.length; i>0; i--){
720 out.write ("<tr>");
721 int[] datarows = (int[])dataset.get(i-1);
722
723 for(int j = 0; j<datarows.length; j++){
724
725 if(datarows[j]==1){
726 out.write ("<td class=\"b\">");
727 blueDot++;
728 }
729 else {
730 out.write ("<td>&nbsp;");
731 }
732 }
733 }
734 }
735 else{
736 ArrayList idList = new ArrayList();
737 for(int i = 0; i<ids.length;i++){
738 idList.add(ids[i]);
739 }
740
741
742 HashMap hp = new HashMap();
743
744 for(int i = ids.length; i>0; i--){
745 String idValue = ids[i-1];
746 int dots = 0;
747 int[] datarows = (int[])dataset.get(i-1);
748 for(int j = 0; j<datarows.length; j++){
749 if(datarows[j]==1){
750 dots++;
751 }
752 }
753 hp.put(idValue,new Integer(dots));
754
755 }
756 ArrayList alist = dm.sortMap(hp);
757 String[] idIntValue = new String[alist.size()];
758 for(int i = 0; i< alist.size(); i++){
759 Map.Entry entry = (Map.Entry) alist.get(i);
760 String idElement = ((String) entry.getKey());
761 idIntValue[i] = idElement;
762 }
763
764 if(cases.equals("worst")){
765
766 for(int i = 0; i<idIntValue.length; i++){
767
768 printWriter.write("<!-- generating all possible page -->");
769 printWriter.flush();
770
771 int value = idList.indexOf(idIntValue[i]);
772 int[] datarows = (int[])dataset.get(value);
773
774 out.write ("<tr id=\""+idIntValue[i].substring(4)+"\">\r\n");
775 out.write("<td class=\"E\" onclick=\"GD(this)\">&nbsp;\r\n");
776
777 if(urlIDList.contains(idIntValue[i])){
778 out.write("<td class=\"qh\" onclick=\"GX(this)\"><span title=\"open URL in new window\">open</span>\r\n");
779 }
780 else{
781 out.write("<td>\r\n");
782 }
783
784 for(int j = 0; j<datarows.length; j++){
785
786 if(datarows[j]==1){
787 out.write ("<td class=\"b\">\r\n");
788 blueDot++;
789 }
790 else {
791 out.write ("<td class=\"w\">\r\n");
792 }
793 }
794 }
795 }
796
797 if(cases.equals("best")){
798
799 for(int i = idIntValue.length; i>0; i--){
800 printWriter.write("<!-- generating all possible page -->");
801 printWriter.flush();
802
803 int value = idList.indexOf(idIntValue[i-1]);
804 int[] datarows = (int[])dataset.get(value);
805
806 out.write ("<tr id=\""+idIntValue[i-1].substring(4)+"\">\r\n");
807 out.write("<td class=\"E\" onclick=\"GD(this)\">&nbsp;\r\n");
808
809 if(urlIDList.contains(idIntValue[i-1])){
810 out.write("<td class=\"qh\" onclick=\"GX(this)\"><span title=\"open URL in new window\">open</span>\r\n");
811 }
812 else{
813 out.write("<td>\r\n");
814 }
815
816 for(int j = 0; j<datarows.length; j++){
817
818 if(datarows[j]==1){
819 out.write ("<td class=\"b\">\r\n");
820 blueDot++;
821 }
822 else {
823 out.write ("<td class=\"w\">\r\n");
824 }
825 }
826 }
827 }
828 }
829
830 out.write ("</tbody>\r\n");
831 out.write ("</table>\r\n</div></div>\r\n");
832
833 int t1 = ids.length;
834 int t3 = metadataName.length;
835 int t4 = t1*t3;
836
837 out.write("<table>\r\n<tbody class=\"table1\">");
838 out.write("<tr>\r\n");
839
840 HashMap tempMap = mds.getMetadataSetMap();
841 MetadataSet ms = (MetadataSet)tempMap.get(suffix);
842
843 out.write("<td class=\"bfont\">This subset shows "+t1+" out of "+mds.getDocNum()+" documents");
844 out.write("<td class=\"bfont\">"+blueDot+" out of "+(mds.getDocNum()*t3)+" metadata items are defined");
845 out.write("<tr>\r\n");
846 out.write("<td class=\"bfont\">This subset shows "+t3+" out of "+(ms.getIndexsList().size())+" metadata elements");
847 out.write("<td class=\"bfont\">Subset completeness: "+dm.round((double)(blueDot*100/t4),5)+"%");
848 out.write("</table>\r\n");
849 out.write ("<p> <a href=\"Overall.html \">Summary</a></p>");
850 out.write ("</body></html>");
851 //Close the output stream
852
853 out.close();
854 fstream.close();
855 }catch (Exception e){
856 e.printStackTrace(printWriter);
857 }
858
859 printWriter.write("<!-- generating all possible page-->");
860 printWriter.flush();
861 }
862
863
864 public void generateOverallStatisticsPage(HashMap MetadataSetMap){
865
866 String fileName = "Overall";
867 ArrayList wholeList = new ArrayList();
868 HashMap hp = mds.getMetadataSetMap();
869 Collection c = hp.values();
870 Iterator i = c.iterator();
871 int counter = 0;
872 String[][] MetadataData = new String[c.size()][2];
873
874 while(i.hasNext()){
875 MetadataSet mds = (MetadataSet)i.next();
876 wholeList.add(mds);
877 ArrayList newMDS = new ArrayList();
878 newMDS.add(mds);
879
880 MetadataData[counter][0] = mds.getName();
881 MetadataData[counter][1] = dm.getSingleMetadataSetCompleteness(newMDS)+"%";
882
883 counter++;
884 }
885
886 try{
887 printWriter.write("<!-- generating all possible page-->");
888 printWriter.flush();
889
890 String str = sdf.format(cl.getTime());
891 String collectionFolder = mds.getCollectionName();
892
893 FileWriter fstream = new FileWriter(destination+fileName+".html");
894 BufferedWriter out = new BufferedWriter(fstream);
895
896 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
897 out.write("<html>\r\n");
898 out.write("<head><title>Summary</title>\r\n");
899 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
900 out.write("<style type=\"text/css\">\r\n");
901 out.write(".tam{height:20; text-align:right}\r\n");
902 out.write("th{height:20; text-align:left}\r\n");
903 out.write("body{font-family: Arial;}\r\n");
904 out.write("</style>\r\n");
905 out.write("<script type=\"text/javascript\" src=\"../script/status3.js\"></script>\r\n");
906 out.write("</head>\r\n");
907 out.write("<body>\r\n");
908 out.write ("<td> <div style=\"float:left;\"> <a href=\"http://"+hostName+":"+port_number+"/greenstone3/mat\">Mat Home</a></div>");
909 out.write("<div style=\"float:right;\"><a href=\"http://chnm.gmu.edu/tools/surveys/4386/\">Please send feedback about the Mat tool</a></div></td><br>");
910 out.write("<h3 align=\"center\">Summary</h3>\r\n");
911 out.write("<table border=1 align=\"center\">\r\n");
912 out.write("<colgroup width=\"300\" span=\"2\">\r\n");
913 out.write("<tr>\r\n");
914 out.write("<th>OAI URL: </th>\r\n");
915
916 if(mds.getOAIURL().length()>50){
917 out.write("<td class=\"tam\"><font size=\"2px\"><a href=\""+mds.getOAIURL()+"\">"+mds.getOAIURL().substring(0,47)+"....</a></font>\r\n");
918 }
919 else{
920 out.write("<td class=\"tam\"><font size=\"2px\"><a href=\""+mds.getOAIURL()+"\">"+mds.getOAIURL()+"</a></font>\r\n");
921 }
922
923 out.write("</tr>\r\n");
924 out.write("<tr>\r\n");
925 out.write("<th>Number of Records:\r\n");
926 out.write("<td class=\"tam\">"+mds.getDocNum()+"\r\n");
927 out.write("</tr>\r\n");
928 out.write("<table border=1 align=\"center\">\r\n");
929 out.write("<colgroup width=\"300\" span=\"2\">\r\n");
930 out.write("<tr>\r\n");
931 out.write("<th>Metadata:\r\n");
932 out.write("<td class=\"tam\"><b>Completeness</b>\r\n");
933 out.write("</tr><br>\r\n");
934
935 for(int a = 0; a<MetadataData.length; a++){
936 out.write ("<tr>\r\n");
937 if(MetadataData[a][0].equalsIgnoreCase("dublin")){
938 out.write ("<th><a href =\""+MetadataData[a][0]+".html\">"+metadataSet1+"</a>");
939 }
940 else if (MetadataData[a][0].equalsIgnoreCase("extracted")){
941 out.write ("<th><a href =\""+MetadataData[a][0]+".html\">"+metadataSet2+"</a>");
942 }
943 else{
944 out.write ("<th><a href =\""+MetadataData[a][0]+".html\">"+MetadataData[a][0]+"</a>");
945 }
946 out.write ("<td class=\"tam\"> "+MetadataData[a][1]);
947 out.write ("</tr>\r\n");
948 }
949
950 out.write ("</table>");
951 out.write("<br>\r\n");
952 out.write("<FORM name=\"test\" onsubmit=\"checkStatus()\" action=\"\">");
953 out.write("<table border=1 align=\"center\">\r\n");
954 out.write ("<colgroup width=\"606\" span=\"1\">\r\n");
955 out.write("<tr>\r\n<th>Customize Visualization");
956 out.write("<tr>\r\n<td><LABEL FOR=\"H1\"><INPUT align=\"left\" TYPE=\"checkbox\" ID=\"H1\">Hide Empty Metadata Elements</LABEL>");
957 out.write("<tr>\r\n<td><LABEL FOR=\"H2\"><INPUT align=\"left\" TYPE=\"checkbox\" ID=\"H2\">Hide Completed Metadata Elements</LABEL>");
958 out.write("<tr>\r\n<td><LABEL FOR=\"H3\"><INPUT align=\"left\" TYPE=\"checkbox\" ID=\"H3\">Hide Documents with Empty Metadata Elements</LABEL>");
959 out.write("<tr>\r\n<td><LABEL FOR=\"H4\"><INPUT align=\"left\" TYPE=\"checkbox\" ID=\"H4\">Hide Documents with Completed Metadata Elements</LABEL>");
960 out.write("<tr>\r\n<th>Metadata: ");
961
962 if(MetadataData.length==1){
963 if(MetadataData[0][0].equalsIgnoreCase("dublin")){
964 out.write("<tr>\r\n<td><LABEL FOR=\"C1\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C1\" checked> Dublin Core </LABEL>");
965 }
966 else {
967 out.write("<tr>\r\n<td><LABEL FOR=\"C4\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C4\" checked>"+MetadataData[0][0]+"</LABEL>");
968 }
969 }
970 else{
971 for(int a = 0; a<MetadataData.length; a++){
972 if(a==0){
973 if(MetadataData[a][0].equalsIgnoreCase("dublin")){
974 out.write("<tr>\r\n<td><LABEL FOR=\"C1\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C1\" checked>Dublin Core</LABEL>");
975 }
976 else {
977 out.write("<tr>\r\n<td><LABEL FOR=\"C4\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C4\" checked>"+MetadataData[a][0]+"</LABEL>");
978 }
979 }
980 else{
981 if(MetadataData[a][0].equalsIgnoreCase("dublin")){
982 out.write("<tr>\r\n<td><LABEL FOR=\"C1\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C1\">"+MetadataData[a][0]+"</LABEL>");
983 }
984 else{
985 out.write("<tr>\r\n<td><LABEL FOR=\"C4\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C4\">"+MetadataData[a][0]+"</LABEL>");
986 }
987 }
988 }
989 }
990
991 out.write("<tr>\r\n<th>Order By Completeness : ");
992 out.write("<tr>\r\n<td><LABEL FOR=\"R1\"><INPUT align=\"left\" TYPE=\"radio\" name = \"order\" ID=\"R1\">Best Case to Worst Case</LABEL>");
993 out.write("<tr>\r\n<td><LABEL FOR=\"R2\"><INPUT align=\"left\" TYPE=\"radio\" name = \"order\" ID=\"R2\" checked >Worst Case to Best Case</LABEL>");
994 out.write("</table><p align=\"center\"><INPUT TYPE=\"button\" VALUE=\"Show Visualization\" onClick=\"checkStatus()\"> </FORM>");
995 out.write("<p align=\"center\">"+str+"</p>");
996 out.write("</body></html>");
997
998 out.close();
999 fstream.close();
1000 }catch (Exception e){
1001 System.err.println("Error: " + e.getMessage());
1002 }
1003 printWriter.write("<!-- generating all possible page-->");
1004 printWriter.flush();
1005 }
1006
1007 public void WriteMetadataSetDetailHTML(MetadataSet mdset){
1008
1009 String fileName = mdset.getName();
1010
1011 try{
1012 printWriter.write("<!-- generating all possible page-->");
1013 printWriter.flush();
1014
1015 String str = sdf.format(cl.getTime());
1016 String collectionFolder = mds.getCollectionName();
1017
1018 FileWriter fstream = new FileWriter(destination+fileName+".html");
1019 BufferedWriter out = new BufferedWriter(fstream);
1020
1021 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
1022 out.write("<html>\r\n");
1023 out.write("<head>\r\n<title>Metadata Detail</title>\r\n");
1024 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
1025 out.write("<style type=\"text/css\">\r\n");
1026 out.write("td{height:20;text-align:right;}\r\n");
1027 out.write("th{height:20;text-align:left;}\r\n");
1028 out.write("body{font-family: Arial;}\r\n");
1029 out.write("</style>\r\n");
1030 out.write( "</head>\r\n");
1031 out.write ("<body>\r\n");
1032 out.write ("<td> <div style=\"float:left;\"> <a href=\" Overall.html \">Summary</a></div>");
1033 out.write("<div style=\"float:right;\"><a href=\"http://chnm.gmu.edu/tools/surveys/4386/\">Please send feedback about the Mat tool</a></div></td><br>");
1034 if(fileName.equalsIgnoreCase("dublin")){
1035 out.write ("<h3 align=\"center\">Metadata Detail: "+metadataSet1+"</h3>\r\n");
1036 }
1037 else if (fileName.equalsIgnoreCase("extracted")){
1038 out.write ("<h3 align=\"center\">Metadata Detail: "+metadataSet2+"</h3>\r\n");
1039 }
1040 else{
1041 out.write ("<h3 align=\"center\">Metadata Detail: "+fileName+"</h3>\r\n");
1042 }
1043
1044 out.write ("<table border=\"1\" align=\"center\">\r\n");
1045 out.write ("<colgroup width=\"200\" span=\"2\">\r\n");
1046 out.write ("<tr>\r\n");
1047 out.write ("<th>Elements:\r\n");
1048 out.write ("<td><b>Completeness</b>\r\n");
1049 out.write ("</tr>\r\n");
1050
1051 ArrayList elementList = mdset.getIndexsList();
1052 HashMap hp = new HashMap();
1053 int num = elementList.size();
1054
1055 for(int i = 0; i<num; i++){
1056 String elementName = (String)elementList.get(i);
1057 Double elementValue = new Double(dm.Mean(elementName));
1058 hp.put(elementName, elementValue);
1059 }
1060
1061 elementList = new ArrayList();
1062 elementList = dm.sortMap(hp);
1063 num = elementList.size();
1064
1065 for(int i = 0; i<num; i++){
1066 Map.Entry entry = (Map.Entry) elementList.get(i);
1067 String elementName = (String) entry.getKey();
1068 out.write ("<tr>\r\n");
1069 out.write ("<th><a href=\""+ elementName +".html\">"+elementName+"</a>\r\n");
1070 out.write ("<td>"+ dm.Mean(elementName)+"%\r\n");
1071 out.write ("</tr>\r\n");
1072 }
1073
1074 out.write("</table>\r\n");
1075 out.write ("<p align=\"center\"> <a href=\" Overall.html \">Summary</a></p>");
1076 out.write("<p align=\"center\">"+str+"</p>");
1077 out.write ("</body></html>\r\n");
1078
1079 out.close();
1080 fstream.close();
1081 }catch (Exception e){
1082 e.printStackTrace(printWriter);
1083 }
1084
1085 printWriter.write("<!-- generating all possible page-->");
1086 printWriter.flush();
1087
1088 }
1089
1090 public void generateMetadataElementDetailPage(MetadataSet mds,PrintWriter out){
1091
1092 ArrayList nameList = mds.getIndexsList();
1093
1094 for(int i = 0; i<nameList.size();i++){
1095 WriteMetadataElementDetailHTML((String)nameList.get(i),mds.getName(),out);
1096 }
1097 }
1098
1099 public void WriteMetadataElementDetailHTML(String name, String linkName,PrintWriter outx){
1100
1101 String fileName = name;
1102
1103 try{
1104 printWriter.write("<!-- generating all possible page-->");
1105 printWriter.flush();
1106
1107 String collectionFolder = mds.getCollectionName();
1108
1109 FileWriter fstream = new FileWriter(destination+fileName+".html");
1110 BufferedWriter out = new BufferedWriter(fstream);
1111
1112 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
1113 out.write ("<html>\r\n");
1114 out.write("<head>\r\n<title> "+ name +" </title>\r\n");
1115 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
1116 out.write("<style type=\"text/css\">\r\n");
1117 out.write("td{height:20; text-align:left;}\r\n");
1118 out.write(".tam{height:20; text-align:center}\r\n");
1119 out.write("th{height:20; text-align:left;}\r\n");
1120 out.write("body{font-family: Arial;}\r\n");
1121 out.write("</style>\r\n");
1122 out.write("<script type=\"text/javascript\" src=\"http://"+hostName+":"+port_number+"/greenstone3/mat/script/status3.js\"></script>");
1123 out.write("</head>\r\n");
1124 out.write("<body>\r\n");
1125 out.write ("<td> <div style=\"float:left;\"> <a href=\" Overall.html \">Summary</a></div>");
1126 out.write("<div style=\"float:right;\"><a href=\"http://chnm.gmu.edu/tools/surveys/4386/\">Please send feedback about the Mat tool</a></div></td>");
1127 out.write("&raquo;");
1128
1129 if(linkName.equalsIgnoreCase("dublin")){
1130 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+metadataSet1+")</a>");
1131 }
1132 else if (linkName.equalsIgnoreCase("extracted")){
1133 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+metadataSet2+")</a>");
1134 }
1135 else{
1136 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+linkName+")</a>");
1137 }
1138
1139 if(name.startsWith("dc.")){
1140 int dot = name.lastIndexOf('.');
1141 dot++;
1142 String nameReplace = name.substring(dot);
1143
1144 if(url.containsKey(nameReplace)){
1145 out.write ("<h3 align=\"center\">Metadata Element Detail:<a href=\""+ url.get(nameReplace) +"\">"+ name +"</a></h3>\r\n");
1146 }
1147 else{
1148 out.write ("<h3 align=\"center\">Metadata Element Detail: "+ name +"</a></h3>\r\n");
1149 }
1150 }
1151 else{
1152 out.write ("<h3 align=\"center\">Metadata Element Detail: "+ name +"</h3>\r\n");
1153 }
1154
1155 out.write ("<table border=\"1\" align=\"center\" width=1000>\r\n");
1156 out.write ("<colgroup width=\"500\" span=\"2\">\r\n");
1157 out.write ("<tr>\r\n");
1158 out.write ("<td><b>Total Number of Records</b>\r\n");
1159 out.write ("<td class=\"tam\">"+mds.getDocNum()+"\r\n");
1160 out.write ("</tr>\r\n");
1161 out.write ("<tr>\r\n");
1162 out.write ("<td><b>Unique Values</b>\r\n");
1163 out.write ("<td class=\"tam\">"+dm.getDistinctNumber(name)+"\r\n");
1164 out.write ("</tr>\r\n");
1165 out.write ("<tr>\r\n");
1166 out.write ("<td><b>Total times element used</b>\r\n");
1167 out.write ("<td class=\"tam\">"+dm.getFrequency(name) +"\r\n");
1168 out.write ("</tr>\r\n");
1169 out.write ("<tr>\r\n");
1170 out.write ("<td><b>No. of records containing element</b>\r\n");
1171 out.write ("<td class=\"tam\"> "+dm.getDocumentUsedElement(name)+"\r\n");
1172 out.write ("</tr>\r\n");
1173 out.write ("<tr>\r\n");
1174
1175 double percentage = dm.Mean(name);
1176
1177 out.write ("<th title = \"The completeness means the arithmetic average \"> <div style=\"float:left;\"> Completeness</div>");
1178 out.write ("<td class=\"tam\"> "+dm.Mean(name) +"%\r\n");
1179 out.write ("</tr>\r\n");
1180 out.write ("<tr>\r\n");
1181 out.write ("<td><b><div style=\"float:left;\">Minimum "+name +" usage in any record</div></b><div style=\"float:right;\"> <a href=\"#\" onClick=\"helpWindow('Minimum','"+name+"')\" style=\"font-size:small;\">What's this?</a></div>\r\n");
1182 out.write ("<td class=\"tam\"> "+dm.getMinRange(name) +"\r\n");
1183 out.write ("</tr>\r\n");
1184
1185 out.write ("<tr>\r\n");
1186 out.write ("<td><b><div style=\"float:left;\">Maximum "+name +" usage in any record</div></b><div style=\"float:right;\"> <a href=\"#\" onClick=\"helpWindow('Maximum','"+name+"')\" style=\"font-size:small;\">What's this?</a></div>\r\n");
1187 out.write ("<td class=\"tam\"> "+dm.getMaxRange(name) +"\r\n");
1188 out.write ("</tr>\r\n");
1189
1190 out.write ("<tr>\r\n");
1191 out.write ("<td><b><div style=\"float:left;\">Average "+name +" usage/record</div></b><div style=\"float:right;\"> <a href=\"#\" onClick=\"helpWindow('Average','"+name+"')\" style=\"font-size:small;\">What's this?</a></div>\r\n");
1192 out.write ("<td class=\"tam\"> "+dm.Average(name) +"\r\n");
1193 out.write ("</tr>\r\n");
1194
1195 out.write ("<tr>\r\n");
1196 out.write ("<td><b><div style=\"float:left;\">Mode of "+name +" usage/record</div></b><div style=\"float:right;\"><a href=\"#\" onClick=\"helpWindow('Mode','"+name+"')\" style=\"font-size:small;\">What's this?</a></div>\r\n");
1197 out.write ("<td class=\"tam\"> "+dm.getMode(name) +"\r\n");
1198 out.write ("</tr>\r\n");
1199
1200 out.write ("<tr>\r\n");
1201 out.write ("<td><b><div style=\"float:left;\">Coverage of the mode of "+name +" usage/record</div></b><div style=\"float:right;\"><a href=\"#\" onClick=\"helpWindow('Mode Frequency','"+name+"')\" style=\"font-size:small;\">What's this?</a></div>\r\n");
1202 out.write ("<td class=\"tam\">"+dm.ModeFrequency(name) +"%\r\n");
1203 out.write ("</tr>\r\n");
1204
1205 if(mds.getOaiPrefix().equalsIgnoreCase("oai_dc")){
1206
1207 HashMap suggestionMap = generateMetadataElementSortList(fileName,"ASCII",linkName,outx);
1208
1209 if(suggestionMap.size()>0){
1210 boolean status = compareElement(fileName,collectionFolder,suggestionMap,linkName);
1211 if(status){
1212 out.write("<tr><td class=\"tam\"><a href =\""+fileName+"_Suggestion.html\">View Potential Duplicate List</a>");
1213 if(percentage<100){
1214 out.write ("<td class=\"tam\"><a href =\""+fileName+"_IncompletedList.html\">Records missing "+fileName+"</a></td>");
1215 createIncompletedList(fileName,linkName,collectionFolder,outx);
1216 }
1217 else{
1218 out.write("<td class=\"tam\">No Records Missing "+ fileName);
1219 }
1220 }
1221 else{
1222 out.write("<tr><td class=\"tam\">No Potential Duplicates");
1223 if(percentage<100){
1224 out.write ("<td class=\"tam\"><a href =\""+fileName+"_IncompletedList.html\">Records missing "+fileName+"</a></td>");
1225 createIncompletedList(fileName,linkName,collectionFolder,outx);
1226 }
1227 else{
1228 out.write("<td class=\"tam\">No Records Missing "+ fileName);
1229 }
1230 }
1231 }
1232 else{
1233 out.write("<tr><td class=\"tam\">No Potential Duplicates");
1234 if(percentage<100){
1235 out.write ("<td class=\"tam\"><a href =\""+fileName+"_IncompletedList.html\">Records missing "+fileName+"</a></td>");
1236 createIncompletedList(fileName,linkName,collectionFolder,outx);
1237 }
1238 else{
1239 out.write("<td class=\"tam\">No Records Missing "+ fileName);
1240 }
1241 }
1242 generateMetadataElementSortList(fileName,"Frequency-based",linkName,outx);
1243 }
1244 else{
1245 HashMap suggestionMap = generateMetadataElementSortListNoIdentifier(fileName,"ASCII",linkName,outx);
1246 generateMetadataElementSortListNoIdentifier(fileName,"Frequency-based",linkName,outx);
1247 }
1248
1249 out.write ("<tr>\r\n");
1250 out.write ("<td class=\"tam\"><a href =\""+fileName+"_Frequency-based.html\">"+"View Full Frequency Sorted list</a>" +
1251 " <td class=\"tam\"><a href =\""+fileName+"_ASCII.html\">"+"View Full ASCII Sorted list</a></td>");
1252 out.write ("</tr>\r\n");
1253 out.write ("</table><br>\r\n");
1254 out.write ("<table border=\"1\" align=\"center\" width=1000>\r\n");
1255 out.write ("<colgroup width=\"500\" span=\"2\">\r\n");
1256 out.write ("<tr>\r\n");
1257 out.write ("<th>ASCII-Based\r\n");
1258 out.write ("<th>First Five\r\n");
1259 out.write ("</tr>\r\n");
1260
1261 String[] temp = dm.getSortList(name,"ASCII");
1262 String[] temp2 = {"&nbsp;","&nbsp;","&nbsp;","&nbsp;","&nbsp;"};
1263 int length = 0;
1264 if(temp.length>=5){
1265 length=5;
1266 }
1267 else if(temp.length<5){
1268 length = temp.length;
1269 }
1270
1271 for(int i =0; i<length; i++){
1272 temp2[i] = temp[i];
1273 }
1274
1275 int x = temp2.length;
1276 for(int a = 0; a<x; a++){
1277 out.write ("<tr>\r\n");
1278 if(!temp2[a].equals("&nbsp;")){
1279 out.write ("<th>"+(a+1)+"\r\n");
1280 }
1281 else{
1282 out.write ("<th>&nbsp;\r\n");
1283 }
1284 if(temp2[a].startsWith("http://")){
1285 if(temp2[a].length()>60){
1286 out.write ("<td><a href = \""+temp2[a]+"\">"+temp2[a].substring(0, 60)+"...</a>");
1287 }
1288 else {
1289 out.write ("<td><a href = \""+temp2[a]+"\">"+temp2[a]+"</a>");
1290 }
1291 }
1292 else{
1293 char singleChar = temp2[a].charAt(0);
1294 if(temp2[a].length()>61 ){
1295 if(temp2[a].startsWith(" ") && ((int)singleChar!=65279)){
1296 temp2[a] = temp2[a].substring(1);
1297 out.write ("<td>"+spaceLeft+space+spaceRight+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#topFive\"> ... </a>\r\n");
1298 }
1299 else if (((int)singleChar==65279)){
1300 temp2[a] = temp2[a].substring(1);
1301 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#topFive\"> ... </a>\r\n");
1302 }
1303 else{
1304 out.write ("<td>"+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#topFive\"> ... </a>\r\n");
1305 }
1306 }
1307 else {
1308 if(temp2[a].startsWith(" ")&& ((int)singleChar!=65279)){
1309 temp2[a] = temp2[a].substring(1);
1310 out.write ("<td>"+spaceLeft+space+spaceRight+temp2[a]+"\r\n");
1311 }
1312 else if (((int)singleChar==65279)){
1313 temp2[a] = temp2[a].substring(1);
1314 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp2[a]+"\r\n");
1315 }
1316 else{
1317 out.write ("<td>"+temp2[a]+"\r\n");
1318 }
1319 }
1320 }
1321 out.write ("</tr>\r\n");
1322 }
1323
1324 String[] temp3 ={"&nbsp;","&nbsp;","&nbsp;","&nbsp;","&nbsp;"};
1325 length = 0;
1326 int start = temp.length;
1327
1328 if(temp.length>=5){
1329 length= 5;
1330 }
1331 else if(temp.length<5){
1332 length = temp.length;
1333 }
1334
1335 for(int i = length; i>0; i--){
1336 temp3[i-1] = temp[start-1];
1337 start--;
1338 }
1339
1340 out.write ("<tr>\r\n");
1341 out.write ("<th>......\r\n");
1342 out.write ("<th>Last Five\r\n");
1343 out.write ("</tr>\r\n");
1344
1345 int counter = temp.length;
1346 start = temp.length;
1347 x = temp3.length;
1348
1349 for(int a = 0; a<x; a++){
1350 out.write ("<tr>\r\n");
1351
1352 if(!temp3[a].equals("&nbsp;")){
1353 out.write ("<th>"+(start-length+1+a));
1354 }
1355 else{
1356 out.write ("<th>&nbsp;");
1357 }
1358 if(temp3[a].startsWith("http://")){
1359 if(temp3[a].length()>60){
1360 out.write ("<td><a href = \""+temp3[a]+"\">"+temp3[a].substring(0, 60)+"...</a>");
1361 }
1362 else {
1363 out.write ("<td><a href = \""+temp3[a]+"\">"+temp3[a]+"</a>");
1364 }
1365 }
1366 else{
1367 char singleChar = temp3[a].charAt(0);
1368 if(temp3[a].length()>61){
1369 if(temp3[a].startsWith(" ") && (int)singleChar!=65279){
1370 temp3[a] = temp3[a].substring(1);
1371 out.write ("<td>"+spaceLeft+space+spaceRight+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#lastFive\"> ... </a>\r\n");
1372 }
1373 else if((int)singleChar==65279){
1374 temp3[a] = temp3[a].substring(1);
1375 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#lastFive\"> ... </a>\r\n");
1376
1377 }
1378 else{
1379 out.write ("<td>"+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#lastFive\"> ... </a>\r\n");
1380 }
1381 }
1382 else{
1383 if(temp3[a].startsWith(" ")&& (int)singleChar!=65279){
1384 temp3[a] = temp3[a].substring(1);
1385 out.write ("<td>"+spaceLeft+space+spaceRight+temp3[a]+"\r\n");
1386 }
1387 else if((int)singleChar==65279){
1388 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp3[a]+"\r\n");
1389 }
1390 else{
1391 out.write ("<td>"+temp3[a]+"\r\n");
1392 }
1393 }
1394 out.write ("</tr>\r\n");
1395 }
1396 }
1397 out.write ("</table><br>\r\n");
1398 out.write ("<table border=\"1\" align=\"center\" width=1000>\r\n");
1399 out.write ("<colgroup width=\"500\" span=\"2\">\r\n");
1400 out.write ("<tr>\r\n");
1401 out.write ("<th>Frequency-Based:\r\n");
1402 out.write ("<th>First Five\r\n");
1403 out.write ("</tr>\r\n");
1404
1405 HashMap xMap = dm.getDistinctValueMap(name);
1406 temp = dm.getSortList(name,"Frequency-based");
1407 temp2 = new String[] {"&nbsp;","&nbsp;","&nbsp;","&nbsp;","&nbsp;"};
1408 length = 0;
1409
1410 if(temp.length>=5){
1411 length=5;
1412 }
1413 else if(temp.length<5){length = temp.length;}
1414
1415 for(int i =0; i<length; i++){
1416 temp2[i] = temp[i];
1417 }
1418
1419 ArrayList tempList = new ArrayList();
1420
1421 for(int i =0; i<length; i++){
1422 if(((Integer)xMap.get(temp2[i])).intValue()==1){
1423 tempList.add(temp2[i]);
1424 }
1425 }
1426
1427 x = temp2.length;
1428 for(int a = 0; a<x; a++){
1429 out.write ("<tr>\r\n");
1430
1431 if(!temp2[a].equals("&nbsp;")){
1432 char singleChar = temp2[a].charAt(0);
1433 out.write ("<th>"+(a+1)+". (No. of occurrences: "+((Integer)xMap.get(temp2[a])).toString()+")\r\n");
1434 if(temp2[a].startsWith("http://") && (int)singleChar != 65279){
1435 if(temp2[a].length()>61){
1436 out.write ("<td><a href = \""+temp2[a]+"\">"+temp2[a].substring(0, 60)+"...</a>");
1437 }
1438 else {
1439 out.write ("<td><a href = \""+temp2[a]+"\">"+temp2[a]+"</a>");
1440 }
1441 }
1442
1443 else{
1444 if(temp2[a].length()>61){
1445
1446 if(temp2[a].startsWith(" ")){
1447 temp2[a] = temp2[a].substring(1);
1448 out.write ("<td>"+spaceLeft+space+spaceRight+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#topFive\"> ... </a>\r\n");
1449 }
1450
1451 else if((int)singleChar == 65279){
1452 temp2[a] = temp2[a].substring(1);
1453 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#topFive\"> ... </a>\r\n");
1454 }
1455 else{
1456 out.write ("<td>"+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#topFive\"> ... </a>\r\n");
1457 }
1458
1459 }
1460 else{
1461 if(temp2[a].startsWith(" ")){
1462 temp2[a] = temp2[a].substring(1);
1463 out.write ("<td>"+spaceLeft+space+spaceRight+temp2[a]+"\r\n");
1464 }
1465 else if((int)singleChar == 65279){
1466 temp2[a] = temp2[a].substring(1);
1467 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp2[a]+"\r\n");
1468 }
1469 else{
1470 out.write ("<td>"+temp2[a]+"\r\n");
1471 }
1472 }
1473 }
1474 }
1475 else{
1476 out.write ("<th>&nbsp;\r\n");
1477 out.write ("<td>\r\n");
1478 }
1479 out.write ("</tr>\r\n");
1480 }
1481
1482 temp3 = new String[]{"&nbsp;","&nbsp;","&nbsp;","&nbsp;","&nbsp;"};
1483 length = 0;
1484 start = temp.length;
1485 if(temp.length>=5){length= 5;}
1486 else if(temp.length<5){length = temp.length;}
1487
1488 for(int i = length; i>0; i--){
1489 temp3[i-1] = temp[start-1];
1490 start--;
1491 }
1492
1493 out.write ("<tr>\r\n");
1494 out.write ("<th>......\r\n");
1495 out.write ("<th>Last Five\r\n");
1496 out.write ("</tr>\r\n");
1497
1498 x = temp3.length;
1499 start = temp.length;
1500 for(int a = 0; a<x; a++){
1501 out.write ("<tr>\r\n");
1502 if(!temp3[a].equals("&nbsp;")){
1503
1504
1505 out.write ("<th>"+(start-length+1+a)+". (No. of occurrences: "+((Integer)xMap.get(temp3[a])).toString()+")\r\n");
1506
1507 char singleChar = temp3[a].charAt(0);
1508 if(temp3[a].startsWith("http://")){
1509 if(temp3[a].length()>60){
1510 out.write ("<td><a href = \""+temp3[a]+"\">"+temp3[a].substring(0, 60)+"...</a>");
1511 }
1512 else {
1513 out.write ("<td><a href = \""+temp3[a]+"\">"+temp3[a]+"</a>");
1514 }
1515 }
1516 else{
1517 if(temp3[a].length()>61){
1518
1519 if(temp3[a].startsWith(" ")){
1520 temp3[a] = temp3[a].substring(1);
1521 out.write ("<td>"+spaceLeft+space+spaceRight+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#lastFive\"> ... </a>\r\n");
1522 }
1523 else if((int)singleChar == 65279){
1524 temp3[a] = temp3[a].substring(1);
1525 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#lastFive\"> ... </a>\r\n");
1526
1527 }
1528 else{
1529 out.write ("<td>"+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#lastFive\"> ... </a>\r\n");
1530 }
1531
1532 }
1533 else{
1534 if(temp3[a].startsWith(" ")){
1535 out.write ("<td>"+spaceLeft+space+spaceRight+temp3[a]+"\r\n");
1536 }
1537
1538 else if((int)singleChar == 65279){
1539 temp3[a] = temp3[a].substring(1);
1540 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp3[a]+"\r\n");
1541
1542 }
1543
1544 else{
1545 out.write ("<td>"+temp3[a]+"\r\n");
1546 }
1547 }
1548 }
1549
1550 }
1551 else{
1552 //out.write ("<th>"+(a+1)+"\r\n");
1553 out.write ("<th>&nbsp;\r\n");
1554 out.write ("<th>\r\n");
1555 }
1556 out.write ("</tr>\r\n");
1557 }
1558
1559
1560 out.write ("</table>\r\n");
1561
1562 ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
1563 out.write ("<p> <a href=\" Overall.html \">Summary</a>");
1564 out.write ("&raquo;");
1565
1566 if(linkName.equalsIgnoreCase("dublin")){
1567 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+metadataSet1+")</a>");
1568
1569 }
1570 else if (linkName.equalsIgnoreCase("extracted")){
1571 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+metadataSet2+")</a>");
1572
1573 }
1574 else{
1575 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+linkName+")</a>");
1576 }
1577 //out.write(str+"</p>");
1578 out.write ("</body></html>\r\n");
1579
1580 //Close the output stream
1581 out.close();
1582 fstream.close();
1583 }catch (Exception e){
1584
1585 e.printStackTrace(printWriter);
1586 }
1587
1588 printWriter.write("<!-- generating all possible page-->");
1589 printWriter.flush();
1590 //System.out.println("detail ends");
1591
1592 }
1593
1594
1595 private HashMap generateMetadataElementSortList(String title, String sort, String metadataSetName,PrintWriter outx){
1596
1597 String fileName = title+"_"+sort;
1598 String collectionFolder = mds.getCollectionName();
1599
1600 String IDENTIFIER = linkIdentifier;
1601 SearchLink sl = new SearchLink(mds.StatsDirectory);
1602 HashMap suggestionMap = new HashMap();
1603
1604 HashMap valueMap = sl.createValueMap(title,collectionFolder);
1605
1606 HashMap linkMap = sl.createLinkMap(IDENTIFIER,collectionFolder);
1607 HashMap internalIDMap = dm.getInternalIdentifier(title);
1608 boolean status = false;
1609
1610 try{
1611 printWriter.write("<!-- generating all possible page-->");
1612 printWriter.flush();
1613
1614 FileWriter fstream = new FileWriter(destination+fileName+".html");
1615 BufferedWriter out = new BufferedWriter(fstream);
1616
1617 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
1618 out.write ("<html>\r\n");
1619 out.write("<head>\r\n<title>Metadata Element Sort List</title>\r\n");
1620 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
1621 out.write("<script type=\"text/javascript\" src=\"../script/getInfomation.js\"></script>\r\n");
1622 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/yahoo/yahoo-min.js\"></script>\r\n");
1623 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/event/event-min.js\"></script>\r\n");
1624 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/connection/connection-min.js\"></script>\r\n");
1625
1626 out.write("<style type=\"text/css\">\r\n");
1627 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
1628 out.write(".tam{height:20; text-align:center}\r\n");
1629 out.write("th{height:20; text-align:center;}\r\n");
1630 out.write("body{font-family: Arial;}\r\n");
1631 out.write("</style>\r\n");
1632 out.write("</head><body id=\""+collectionFolder+"\">\r\n");
1633
1634 out.write("<td><div style=\"float:left;\"><a href=\" Overall.html \">Summary</a>");
1635 out.write("&raquo;");
1636 ////////////////////////////////////////////////////////////////////////////////////
1637 if(metadataSetName.equalsIgnoreCase("dublin")){
1638 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
1639 }
1640 else if (metadataSetName.equalsIgnoreCase("extracted")){
1641 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
1642 }
1643 else{
1644 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
1645 }
1646 out.write("&raquo;");
1647 out.write("<a href=\""+title+".html\">"+title+"</a></div>");
1648 out.write("<div style=\"float:right;\"><a href=\"http://chnm.gmu.edu/tools/surveys/4386/\">Please send feedback about the Mat tool</a></div></td><br>");
1649
1650
1651 String[] list = dm.getSortList(title, sort);
1652
1653
1654 if(list.length>=1){
1655
1656
1657 if(sort.equals("ASCII")){
1658
1659 out.write("<table border=\"1\" align=\"center\" width=\"800\">\r\n");
1660 out.write ("<h2 align=\"center\">"+ title+"</h2>\r\n");
1661 out.write ("<th class=\"tam\">ASCII Sort\r\n");
1662 out.write ("<th class=\"tam\">Element Values\r\n");
1663 out.write ("<th class=\"tam\">Source Documents\r\n");
1664 out.write ("<th class=\"tam\">Internal Link\r\n");
1665
1666 out.write ("<a name='topFive'>\r\n");
1667
1668 int counter = 0;
1669
1670 for(int i = 0; i<list.length; i++){
1671 if(list.length<=5 && i==0){
1672 out.write ("<a name='lastFive'>\r\n");
1673 }
1674 else if((list.length>5) && (list.length-5==i)){
1675 out.write ("<a name='lastFive'>\r\n");
1676 }
1677
1678
1679 InternalLink il= (InternalLink)internalIDMap.get(list[i]);
1680 ArrayList alist2 = il.retrieveList();
1681 String id = (String)alist2.get(0);
1682 id = id.substring(4);
1683 out.write("<tr id=\""+id+"\" >\r\n");
1684
1685
1686 if(list[i].length()>=201){
1687 if(list[i].startsWith("http://")){
1688
1689 if(title.equals(IDENTIFIER)){
1690 if(list[i].length()>=100){
1691 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1692
1693 }
1694 else{
1695 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1696 }
1697 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
1698 }
1699 else{
1700 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1701
1702 if(list[i].length()>=100){
1703 String url = (String)alist.get(0);
1704 if(alist.size()==1){
1705 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1706 if(url.startsWith("http://")){
1707 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1708 }else{
1709 out.write("<td>Source\r\n");
1710 }
1711 }
1712 else if(alist.size()>1){
1713 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1714 if(url.startsWith("http://")){
1715 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1716 }else{
1717 out.write("<td>Source\r\n");
1718 }
1719
1720 }
1721 else{
1722 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1723 out.write("<td>No Source Available\r\n");
1724 }
1725 suggestionMap.put(list[i], url);
1726 }
1727 else{
1728
1729 String url = (String)alist.get(0);
1730 if(alist.size()==1){
1731 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1732 //out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1733 if(url.startsWith("http://")){
1734 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1735 }else{
1736 out.write("<td>Source\r\n");
1737 }
1738 }
1739 else if(alist.size()>1){
1740 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1741 //out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1742 if(url.startsWith("http://")){
1743 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1744 }else{
1745 out.write("<td>Source\r\n");
1746 }
1747 }
1748 else{
1749 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1750 out.write("<td>No Source Available\r\n");
1751 }
1752 suggestionMap.put(list[i], url);
1753 }
1754 }
1755
1756
1757
1758 }
1759 else{
1760
1761 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1762
1763 //////////////////////////////////////
1764
1765 if(list[i].startsWith(" ")){
1766 String elements = list[i];
1767 list[i] = list[i].substring(1);
1768
1769 if(alist.size()==1){
1770 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i].substring(0, 200)+"...<td>"+
1771 "<a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1772 suggestionMap.put(elements, (String)alist.get(0));
1773 }
1774 else if(alist.size()>1){
1775 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i].substring(0, 200)+"...<td>"+
1776 "<a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1777 suggestionMap.put(elements, (String)alist.get(0));
1778 }
1779 else{
1780 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i].substring(0, 200)+"...<td>No Source Available \r\n");
1781 suggestionMap.put(elements, "No Source Available");
1782 }
1783 }
1784 else{
1785
1786 if(alist.size()==1){
1787 out.write("<td>"+(counter+1)+"<td>"+list[i].substring(0, 200)+"...<td>" +
1788 "<a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1789 suggestionMap.put(list[i], (String)alist.get(0));
1790 }
1791 else if(alist.size()>1){
1792 out.write("<td>"+(counter+1)+"<td>"+list[i].substring(0, 200)+"...<td>" +
1793 "<a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1794 suggestionMap.put(list[i], (String)alist.get(0));
1795 }
1796 else{
1797 out.write("<td>"+(counter+1)+"<td>"+list[i].substring(0, 200)+"... <td>No Source Available\r\n");
1798 suggestionMap.put(list[i],"No Source Available");
1799 }
1800 }
1801 }
1802
1803 }
1804 else{
1805 if(list[i].startsWith("http://")){
1806
1807 if(title.equals(IDENTIFIER)){
1808 if(list[i].length()>=100){
1809 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1810
1811 }
1812 else{
1813 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1814 }
1815 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
1816 }
1817 else{
1818 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1819
1820 if(list[i].length()>=100){
1821 if(alist.size()==1){
1822 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1823 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1824 }
1825 else if(alist.size()>1){
1826 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1827 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1828 }
1829 else{
1830 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1831 out.write("<td>No Source Available\r\n");
1832 }
1833 }
1834 else{
1835 if(alist.size()==1){
1836 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1837 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1838 suggestionMap.put(list[i], (String)alist.get(0));
1839 }
1840 else if(alist.size()>1){
1841 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1842 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1843 suggestionMap.put(list[i], (String)alist.get(0));
1844 }
1845 else{
1846 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1847 out.write("<td>No Source Available\r\n");
1848 suggestionMap.put(list[i],"No Source Available");
1849 }
1850 }
1851 }
1852
1853 }
1854
1855
1856 else{
1857
1858 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1859
1860 if(list[i].startsWith(" ")){
1861 String elements = list[i];
1862 list[i] = list[i].substring(1);
1863
1864 if(alist.size()==1){
1865 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i]+
1866 "<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1867 suggestionMap.put(list[i], (String)alist.get(0));
1868 }
1869 else if(alist.size()>1){
1870 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i]+
1871 "<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1872 suggestionMap.put(list[i], (String)alist.get(0));
1873 }
1874 else{
1875 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i]+"<td>No Source Available\r\n");
1876 suggestionMap.put(elements,"No Source Available");
1877 }
1878
1879 }
1880 else{
1881 if(alist.size()==1){
1882 out.write("<td>"+(counter+1)+"<td>"+list[i]+"<td><a href=\""+alist.get(0)+"\">Source</a>\r\n");
1883 suggestionMap.put(list[i],(String)alist.get(0));
1884 }
1885 else if(alist.size()>1){
1886 out.write("<td>"+(counter+1)+"<td>"+list[i]+"<td><a href=\""+alist.get(0)+"\">Source</a>...\r\n");
1887 suggestionMap.put(list[i],(String)alist.get(0));
1888 }
1889 else{
1890 out.write("<td>"+(counter+1)+"<td>"+list[i]+"<td>No Source Available\r\n");
1891 suggestionMap.put(list[i],"No Source Available");
1892 }
1893 //out.write("<td>"+(counter+1)+"<td>"+list[i]+"\r\n");
1894 }
1895 }
1896 }
1897 counter++;
1898 out.write("<td onclick=\"GD(this)\">View");
1899 }
1900 if(suggestionMap.size()>1 && !title.equals(IDENTIFIER)){
1901 status = true;
1902 }
1903 }
1904 else{
1905
1906
1907 out.write("<table border=\"1\" align=\"center\" width=\"800\">\r\n");
1908 out.write ("<h2 align=\"center\">"+ title+"</h2>\r\n");
1909 out.write ("<th class=\"tam\">&nbsp;\r\n");
1910 out.write ("<th class=\"tam\">Frequency\r\n");
1911 out.write ("<th class=\"tam\">Element Values\r\n");
1912 out.write ("<th class=\"tam\">Source Documents\r\n");
1913 out.write ("<th class=\"tam\">Internal Link\r\n");
1914 out.write ("<a name='topFive'>\r\n");
1915
1916 HashMap xMap = dm.getDistinctValueMap(title);
1917 int counter = 0;
1918 for(int i = 0; i<list.length; i++){
1919 if(list.length<=5 && i==0){
1920 out.write ("<a name='lastFive'>\r\n");
1921 }
1922 else if((list.length>5) && (list.length-5==i)){
1923 out.write ("<a name='lastFive'>\r\n");
1924 }
1925
1926 InternalLink il= (InternalLink)internalIDMap.get(list[i]);
1927 ArrayList alist2 = il.retrieveList();
1928 String id = (String)alist2.get(0);
1929 id = id.substring(4);
1930 out.write("<tr id=\""+id+"\" >\r\n");
1931
1932 out.write("<td>"+(counter+1)+"<th> "+((Integer)xMap.get(list[i])).toString()+"\r\n");
1933
1934 if(list[i].length()>=201){
1935
1936 if(list[i].startsWith("http://")){
1937 if(title.equals(IDENTIFIER)){
1938 if(list[i].length()>=100){
1939 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1940 }
1941 else{
1942 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1943 }
1944
1945 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
1946 }
1947 else{
1948 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1949 if(list[i].length()>=100){
1950 if(alist.size()==1){
1951 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1952 out.write("<td><a href=\""+alist.get(0)+"\">Source</a>\r\n");
1953 }
1954 else if (alist.size()>1){
1955 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1956 out.write("<td><a href=\""+alist.get(0)+"\">Source</a>...\r\n");
1957 }
1958 else{
1959 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1960 out.write("<td>No Source Available\r\n");
1961 }
1962
1963 }
1964 else{
1965 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1966 }
1967 }
1968 }
1969
1970
1971
1972
1973 else{
1974 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1975
1976 if(list[i].startsWith(" ")){
1977 list[i]=list[i].substring(1);
1978 if(alist.size()==1){
1979 out.write("<td>"+spaceLeft+space+spaceRight+list[i].substring(0,200)+"...<td>"+
1980 "<a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1981 }
1982 else if(alist.size()>1){
1983 out.write("<td>"+spaceLeft+space+spaceRight+list[i].substring(0,200)+"...<td>"+
1984 "<a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1985 }
1986 else{
1987 out.write("<td>"+spaceLeft+space+spaceRight+list[i].substring(0,200)+"...<td>No Source Available\r\n");
1988 }
1989 }
1990 else{
1991
1992 if(alist.size()==1){
1993 out.write("<td>"+list[i].substring(0,200)+"...<td>" +
1994 "<a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1995 }
1996 else if(alist.size()>1){
1997 out.write("<td>"+list[i].substring(0,200)+"...<td>"+
1998 "<a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1999 }
2000 else{
2001 out.write("<td>"+list[i].substring(0,200)+"...<td>No Source Available\r\n");
2002 }
2003 }
2004 }
2005 }
2006 else{
2007 if(list[i].startsWith("http://")){
2008 if(title.equals(IDENTIFIER)){
2009 if(list[i].length()>=100){
2010 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2011 }
2012 else{
2013 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2014 }
2015 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2016 }
2017 else{
2018 //ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],"dc.Identifier", collectionFolder,valueMap,linkMap);
2019 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
2020
2021 if(list[i].length()>=100){
2022 if(alist.size()==1){
2023 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2024 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
2025 }
2026 else if(alist.size()>1){
2027 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2028 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
2029 }
2030 else{
2031 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2032 out.write("<td>No Source Available\r\n");
2033 }
2034 }
2035 else{
2036 if(alist.size()==1){
2037 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2038 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
2039 }
2040 else if(alist.size()>1){
2041 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2042 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
2043 }
2044 else{
2045 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2046 out.write("<td>No Source Available\r\n");
2047 }
2048 }
2049
2050 }
2051
2052 }
2053 else{
2054
2055 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
2056
2057 if(list[i].startsWith(" ")){
2058 list[i]=list[i].substring(1);
2059 if(alist.size()==1){
2060 out.write("<td>"+spaceLeft+space+spaceRight+list[i]+
2061 "<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
2062 }
2063 else if(alist.size()>1){
2064 out.write("<td>"+spaceLeft+space+spaceRight+list[i]+
2065 "<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
2066 }
2067 else{
2068 out.write("<td>"+spaceLeft+space+spaceRight+list[i]+"<td>No Source Available\r\n");
2069 }
2070 }
2071 else{
2072 if(alist.size()==1){
2073 out.write("<td>"+list[i]+"<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
2074 }
2075 else if(alist.size()>1){
2076 out.write("<td>"+list[i]+"<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
2077 }
2078 else{
2079 out.write("<td>"+list[i]+"<td>No Source Available\r\n");
2080 }
2081 }
2082 }
2083 }
2084 counter++;
2085 out.write("<td onclick=\"GD(this)\">View");
2086 }
2087 }
2088 out.write("</table>\r\n");
2089 }
2090 else{
2091 out.write("<p>Sorry! The list is empty</p>");
2092 }
2093 out.write("<tr>\r\n");
2094
2095
2096 out.write("<p><a href=\" Overall.html \">Summary</a>");
2097 out.write("&raquo;");
2098 ////////////////////////////////////////////////////////////////////////////////////
2099 if(metadataSetName.equalsIgnoreCase("dublin")){
2100 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2101 }
2102 else if (metadataSetName.equalsIgnoreCase("extracted")){
2103 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2104 }
2105 else{
2106 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2107 }
2108 out.write("&raquo;");
2109 out.write("<a href=\""+title+".html\">"+title+"</a></p>");
2110 out.write("</body></html>\r\n");
2111 out.close();
2112 fstream.close();
2113 }catch(Exception e){
2114
2115 e.printStackTrace(outx);
2116 }
2117
2118 if(status){
2119 return suggestionMap;
2120 }
2121 else{
2122 return new HashMap();
2123 }
2124 }
2125
2126
2127 private void setupMetadataLink(String fileName){
2128
2129 String destination = globalProperty.getGSDL3Home()+fileSeparator+"mat"+fileSeparator+"script"+fileSeparator+fileName+".xml";
2130
2131 try{
2132 DocumentBuilderFactory docBuilderFactory = DocumentBuilderFactory.newInstance();
2133 DocumentBuilder docBuilder = docBuilderFactory.newDocumentBuilder();
2134 Document doc = docBuilder.newDocument();
2135 //doc = docBuilder.parse (new File("/research/cc108/greenstone3/web/mat/script/"+fileName+".xml"));
2136 doc = docBuilder.parse (destination);
2137 Element rootNode = doc.getDocumentElement();
2138
2139 NodeList listOfName = rootNode.getElementsByTagName("metadataElement");
2140 url = new HashMap();
2141 for(int i = 0; i<listOfName.getLength(); i++){
2142 Node NameNode = listOfName.item(i);
2143 Element docElement = (Element)NameNode;
2144 NodeList valueList = docElement.getElementsByTagName("URL");
2145 Node urlParentNode = valueList.item(0);
2146 String urlText = urlParentNode.getChildNodes().item(0).getNodeValue();
2147
2148 valueList = docElement.getElementsByTagName("name");
2149 Node urlNameParentNode = valueList.item(0);
2150 String urlNameText = urlNameParentNode.getChildNodes().item(0).getNodeValue();
2151
2152 url.put(urlNameText, urlText);
2153
2154 }
2155
2156
2157 }catch(Exception e){
2158 e.printStackTrace(printWriter);
2159 }
2160 }
2161
2162 private void createIncompletedList(String fileName, String metadataSetName, String collectionFolder,PrintWriter outx){
2163
2164 HashMap hp = dm.getIdentifierLink(linkIdentifier);
2165 String[] ids = dm.getDocumentIDList(fileName);
2166
2167 for(int i = 0; i<ids.length; i++){
2168 hp.remove(ids[i]);
2169 }
2170
2171 try{
2172 FileWriter fstream = new FileWriter(destination+fileName+"_IncompletedList.html");
2173 BufferedWriter out = new BufferedWriter(fstream);
2174
2175 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
2176 out.write ("<html>\r\n");
2177 out.write("<head>\r\n<title>Incompleted Document List</title>\r\n");
2178 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
2179 out.write("<style type=\"text/css\">\r\n");
2180 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
2181 out.write(".tam{height:20; text-align:center}\r\n");
2182 out.write("th{height:20; text-align:center;}\r\n");
2183 out.write("body{font-family: Arial;}\r\n");
2184 out.write("</style>\r\n");
2185 out.write("</head><body>\r\n");
2186
2187 out.write("<p><a href=\" Overall.html \">Summary</a>");
2188 out.write("&raquo;");
2189 if(metadataSetName.equalsIgnoreCase("dublin")){
2190 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2191 }
2192 else if (metadataSetName.equalsIgnoreCase("extracted")){
2193 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2194 }
2195 else{
2196 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2197 }
2198 out.write("&raquo;");
2199 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
2200
2201 out.write("<h3 align=center>"+fileName+" does not appear in the following documents</h3>");
2202 int counter = 0;
2203 Set s = hp.keySet();
2204 Iterator i = s.iterator();
2205 out.write("<table border=1 align=center><tr><th>Document ID<th>Source Link\r\n");
2206 while(i.hasNext()){
2207 counter++;
2208 String keys = (String)i.next();
2209 InternalLink il = (InternalLink) hp.get(keys);
2210 ArrayList alist = il.retrieveList();
2211 String url = (String)alist.get(0);
2212 out.write("<tr><td>"+counter+"<td><a href=\""+url+"\">"+url+"</a>\r\n");
2213 }
2214 out.write("</table></body></html>\r\n");
2215
2216
2217 out.write("<p><a href=\" Overall.html \">Summary</a>");
2218 out.write("&raquo;");
2219 if(metadataSetName.equalsIgnoreCase("dublin")){
2220 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2221 }
2222 else if (metadataSetName.equalsIgnoreCase("extracted")){
2223 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2224 }
2225 else{
2226 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2227 }
2228 out.write("&raquo;");
2229 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
2230
2231 out.close();
2232 fstream.close();
2233 }catch(IOException ex){ex.printStackTrace();}
2234
2235
2236 }
2237
2238 public boolean compareElement(String fileName,String collectionFolder,HashMap suggestionMap, String metadataSetName){
2239
2240 boolean status = false;
2241
2242 Set kset = suggestionMap.keySet();
2243 ArrayList arrayList = new ArrayList();
2244 Iterator is = kset.iterator();
2245 while(is.hasNext()){
2246 arrayList.add((String)is.next());
2247 }
2248 HashMap distanceMap = new HashMap();
2249
2250 int totalLength = 0;
2251
2252 int arrayListLength = arrayList.size();
2253 int counter = 0;
2254
2255 printWriter.write("<!-- generating all possible page-->");
2256 printWriter.flush();
2257 double distance ;
2258 try{
2259 for(int i = 0; i<arrayListLength; i++){
2260 String keyword = (String)arrayList.get(i);
2261
2262 int spaceCounterFront1 = 0;
2263 int spaceCounterEnd1 = 0;
2264 String keywordClone = keyword;
2265 String _keywordClone = keyword;
2266 String bkKeyWord = keyword;
2267
2268 keywordClone = removeUnusedCharacter(keywordClone);
2269
2270 while(true){
2271
2272 if(_keywordClone.length()>1){
2273 if(_keywordClone.charAt(0)==' '){
2274 _keywordClone = _keywordClone.substring(1);
2275 spaceCounterFront1++;
2276 }
2277 else{
2278 if(_keywordClone.charAt(_keywordClone.length()-1)==' '){
2279 _keywordClone = _keywordClone.substring(0,_keywordClone.length()-1);
2280 spaceCounterEnd1++;
2281 }
2282 else{
2283 break;
2284 }
2285 }
2286 }
2287 else{
2288 break;
2289 }
2290 }
2291
2292 int keywordLength = keywordClone.length();
2293 totalLength = totalLength + keywordLength;
2294 for(int j = i+1; j<arrayListLength; j++){
2295 String keyword2 = (String)arrayList.get(j);
2296 counter ++;
2297 double pre_cost = 0;
2298 String keywordClone2 = keyword2;
2299 String _keywordClone2 = keyword2;
2300 String bkKeyWord2 = keyword2;
2301 costModel cm = new costModel();
2302 cm = removeUnusedCharacter(keywordClone2,pre_cost);
2303 keywordClone2 = cm.getString();
2304 pre_cost = cm.getCost();
2305
2306 int spaceCounterFront2 = 0;
2307 int spaceCounterEnd2 = 0;
2308
2309 while(true){
2310 if(_keywordClone2.length()>1){
2311 if(_keywordClone2.charAt(0)==' '){
2312 _keywordClone2 = _keywordClone2.substring(1);
2313 spaceCounterFront2++;
2314 }
2315 else{
2316 if(_keywordClone2.charAt(_keywordClone2.length()-1)==' '){
2317 _keywordClone2 = _keywordClone2.substring(0,_keywordClone2.length()-1);
2318 spaceCounterEnd2++;
2319 }
2320 else{
2321 break;
2322 }
2323 }}
2324 else{
2325 break;
2326 }
2327 }
2328
2329 if(counter == 50000){
2330 printWriter.write("<!-- generating all possible page-->");
2331 printWriter.flush();
2332 counter = 0;
2333 }
2334 int keyword2Length = keywordClone2.length();
2335
2336 if(keywordLength>(keyword2Length+2) || (keywordLength+2)<keyword2Length ){}
2337 else{
2338
2339 distance = calculateEditDistance(keywordClone.toLowerCase().toCharArray(),keywordClone2.toLowerCase().toCharArray());
2340
2341 distance = distance + pre_cost;
2342
2343 if(distance<=2){
2344 if(distanceMap.containsKey(_keywordClone)){
2345 InternalLink il = (InternalLink)distanceMap.get(_keywordClone);
2346
2347 String keywordHolder = _keywordClone2;
2348
2349 while(true){
2350 if(keywordHolder.contains(" ")){
2351 keywordHolder = keywordHolder.replaceFirst(" ",spaceLeft+space+spaceRight+" ");
2352 }
2353 else{
2354 break;
2355 }
2356 }
2357
2358 for(int a = 0; a<spaceCounterFront2; a++){
2359 keywordHolder = spaceLeft+space+spaceRight+keywordHolder;
2360 }
2361 for(int a = 0; a<spaceCounterEnd2; a++){
2362 keywordHolder = keywordHolder+spaceLeft+space+spaceRight;
2363 }
2364
2365 InternalLink il2 = new InternalLink();
2366 il2.setValue(keywordHolder);
2367 if(suggestionMap.containsKey(bkKeyWord2)){
2368 il2.increaseElement((String)suggestionMap.get(bkKeyWord2));
2369 //System.out.println("input this url(3):"+(String)alistkw2.get(0));
2370 }
2371 else{
2372 il2.increaseElement("&nbsp;");
2373 }
2374
2375 il.increaseNode(il2);
2376
2377 distanceMap.put(_keywordClone, il);
2378
2379 }
2380 else{
2381 InternalLink il = new InternalLink();
2382 String keywordHolder = _keywordClone;
2383
2384 while(true){
2385 if(keywordHolder.contains(" ")){
2386 keywordHolder = keywordHolder.replaceFirst(" ",spaceLeft+space+spaceRight+" ");
2387 }
2388 else{
2389 break;
2390 }
2391 }
2392
2393 for(int a = 0; a<spaceCounterFront1; a++){
2394 keywordHolder = spaceLeft+space+spaceRight+keywordHolder;
2395 }
2396 for(int a = 0; a<spaceCounterEnd1; a++){
2397 keywordHolder = keywordHolder+spaceLeft+space+spaceRight;
2398 }
2399
2400 InternalLink il2 = new InternalLink();
2401 il2.setValue(keywordHolder);
2402 if(suggestionMap.containsKey(bkKeyWord)){
2403 il2.increaseElement((String)suggestionMap.get(bkKeyWord));
2404
2405 }
2406 else{
2407 il2.increaseElement("&nbsp;");
2408 }
2409
2410 il.increaseNode(il2);
2411
2412
2413 keywordHolder = _keywordClone2;
2414
2415 while(true){
2416 if(keywordHolder.contains(" ")){
2417 keywordHolder = keywordHolder.replaceFirst(" ",spaceLeft+space+spaceRight+" ");
2418 }
2419 else{
2420 break;
2421 }
2422 }
2423 for(int a = 0; a<spaceCounterFront2; a++){
2424 keywordHolder = spaceLeft+space+spaceRight+keywordHolder;
2425 }
2426 for(int a = 0; a<spaceCounterEnd2; a++){
2427 keywordHolder = keywordHolder+spaceLeft+space+spaceRight;
2428 }
2429
2430 InternalLink il3 = new InternalLink();
2431 il3.setValue(keywordHolder);
2432 if(suggestionMap.containsKey(bkKeyWord2)){
2433 il3.increaseElement((String)suggestionMap.get(bkKeyWord2));
2434
2435 }
2436 else{
2437 il3.increaseElement("&nbsp;");
2438 }
2439
2440 il.increaseNode(il3);
2441
2442 distanceMap.put(_keywordClone, il);
2443 }
2444 }
2445 }
2446
2447 }
2448
2449 }
2450
2451
2452 printWriter.write("<!-- generating all possible page-->");
2453 printWriter.flush();
2454
2455 if(distanceMap.size()!=0){
2456 generateHTML(distanceMap,fileName,metadataSetName);
2457 status = true;
2458 }
2459 }catch(Exception ex){ex.printStackTrace();}
2460 return status;
2461
2462 }
2463
2464
2465 private void generateHTML(HashMap distanceMap, String fileName,String metadataSetName){
2466 printWriter.write("<!-- generating all possible page-->");
2467 printWriter.flush();
2468
2469 String collectionFolder = mds.getCollectionName();
2470
2471 try{
2472 FileWriter fstream = new FileWriter(destination+fileName+"_Suggestion.html");
2473 BufferedWriter out = new BufferedWriter(fstream);
2474 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
2475 out.write ("<html>\r\n");
2476 out.write("<head>\r\n<title>Potential Duplicate List</title>\r\n");
2477 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
2478 out.write("<style type=\"text/css\">\r\n");
2479 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
2480 out.write(".tam{height:20; text-align:center}\r\n");
2481 out.write("th{height:20; text-align:left;}\r\n");
2482 out.write("body{font-family: Arial;}\r\n");
2483 out.write("</style>\r\n");
2484 out.write("</head><body>\r\n");
2485 out.write("<p><a href=\" Overall.html \">Summary</a>");
2486 out.write("&raquo;");
2487 if(metadataSetName.equalsIgnoreCase("dublin")){
2488 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2489 }
2490 else if (metadataSetName.equalsIgnoreCase("extracted")){
2491 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2492 }
2493 else{
2494 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2495 }
2496 out.write("&raquo;");
2497 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
2498 out.write("<h3 align=center>Potential Duplicate List</h3>");
2499 out.write("<table border=1 align=center>\r\n");
2500
2501 Set s = distanceMap.keySet();
2502 Iterator i = s.iterator();
2503 while(i.hasNext()){
2504
2505 String keyword = (String)i.next();
2506 InternalLink il = (InternalLink)distanceMap.get(keyword);
2507
2508 ArrayList alist = il.retrieveNodeList();
2509
2510 out.write("<tr><th>Original Text<th>Source Link\r\n");
2511 for(int a = 0; a<alist.size(); a++){
2512 InternalLink il2 = new InternalLink();
2513 il2 = (InternalLink)alist.get(a);
2514 String url = (String)il2.retrieveList().get(0);
2515 out.write("<tr><td>"+il2.getValue()+"<td><a href=\""+url+"\">"+url+"</a>");
2516
2517 }
2518 out.write("</tr>");
2519
2520 }
2521 out.write("</table>\r\n");
2522
2523 out.write("<p><a href=\" Overall.html \">Summary</a>");
2524 out.write("&raquo;");
2525 if(metadataSetName.equalsIgnoreCase("dublin")){
2526 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2527 }
2528 else if (metadataSetName.equalsIgnoreCase("extracted")){
2529 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2530 }
2531 else{
2532 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2533 }
2534 out.write("&raquo;");
2535 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
2536 out.write("</body></html>\r\n");
2537 out.close();
2538 fstream.close();
2539 }catch(IOException ex){ex.printStackTrace();}
2540
2541 }
2542
2543 public int calculateEditDistance(char[] args1, char[] args2){
2544
2545 int n = args1.length;
2546 int m = args2.length;
2547 if (n == 0) {
2548 return m;
2549 }
2550 else if (m == 0) {
2551 return n;
2552 }
2553
2554 int[] p = new int[n + 1];
2555 int[] d = new int[n + 1];
2556 int[] _d;
2557
2558
2559 int i;
2560 int j;
2561
2562 int cost; // cost
2563
2564 for (i = 0; i <= n; i++) {
2565 p[i] = i;
2566 }
2567 for (j = 1; j <= m; j++) {
2568
2569 d[0] = j;
2570 for (i = 1; i <= n; i++) {
2571 cost = (args1[i-1] == args2[j-1]) ? 0 : 1;
2572 d[i] = Math.min(Math.min(d[i - 1] + 1, p[i] + 1),
2573 p[i - 1] + cost);
2574 }
2575
2576 // copy current distance counts to 'previous row' distance counts
2577 _d = p;
2578 p = d;
2579 d = _d;
2580 }
2581 // our last action in the above loop was to switch d and p, so p now
2582 // actually has the most recent cost counts
2583 return p[n];
2584 }
2585
2586 private String removeUnusedCharacter(String target){
2587 //printWriter.write("<!-- generating all possible page-->");
2588 //printWriter.flush();
2589
2590 // remove \n
2591 while(true){
2592 if(target.contains("\\n")){
2593 target = target.replaceFirst("\\\\n","");
2594 }
2595 else{
2596 break;
2597 }
2598 }
2599
2600 //remove spaces at the end of string
2601 if(target.length()>1){
2602 while(true){
2603 if(target.length()>1){
2604 if(target.charAt(target.length()-1)==' '){
2605 target = target.substring(0,target.length()-1);
2606 }
2607 else{
2608 break;
2609 }
2610 }
2611 else{
2612 break;
2613 }
2614 }
2615 }
2616 //remove leading spaces
2617 if(target.length()>1){
2618 while(true){
2619
2620 if(target.length()>1){
2621 if(target.charAt(0)==' '){
2622 target = target.substring(1,target.length());
2623 }
2624 else{
2625 break;
2626 }}
2627 else{
2628 break;
2629 }
2630 }
2631 }
2632 //remove multiple spaces between words
2633 while(true){
2634 if(target.contains(" ")){
2635 target = target.replaceFirst(" "," ");
2636 }
2637 else{
2638 break;
2639 }
2640 }
2641
2642 return target;
2643 }
2644
2645
2646 private costModel removeUnusedCharacter(String target, double cost){
2647
2648
2649 costModel cm = new costModel();
2650
2651 // remove \n plus cost
2652 while(true){
2653
2654 if(target.contains("\\n")){
2655 target = target.replaceFirst("\\\\n","");
2656 cost = cost + 0.2;
2657 }
2658 else{
2659 break;
2660 }
2661 }
2662
2663 //remove spaces at the end of string
2664 if(target.length()>1){
2665 while(true){
2666 if(target.length()>1){
2667 if(target.charAt(target.length()-1)==' '){
2668 target = target.substring(0,target.length()-1);
2669 cost = cost + 0.2;
2670 }
2671 else{
2672 break;
2673 }}
2674 else{
2675 break;
2676 }
2677 }
2678 }
2679 //remove leading spaces
2680 if(target.length()>1){
2681 while(true){
2682
2683 if(target.length()>1){
2684 if(target.charAt(0)==' '){
2685 target = target.substring(1,target.length());
2686 cost = cost + 0.2;
2687 }
2688 else{
2689 break;
2690 }}
2691 else{
2692 break;
2693 }
2694 }
2695 }
2696 //remove multiple spaces between words
2697 while(true){
2698
2699 if(target.contains(" ")){
2700 target = target.replaceFirst(" "," ");
2701 cost = cost + 0.2;
2702 }
2703 else{
2704 break;
2705 }
2706 }
2707
2708 cm.setCost(cost);
2709 cm.setString(target);
2710
2711 return cm;
2712 }
2713
2714
2715
2716 private HashMap generateMetadataElementSortListNoIdentifier(String title, String sort, String metadataSetName,PrintWriter outx){
2717
2718 String fileName = title+"_"+sort;
2719 String collectionFolder = mds.getCollectionName();
2720 System.out.println("level 3");
2721 SearchLink sl = new SearchLink(mds.StatsDirectory);
2722 System.out.println("level 4");
2723 HashMap suggestionMap = new HashMap();
2724
2725 HashMap internalIDMap = dm.getInternalIdentifier(title);
2726 boolean status = false;
2727
2728 try{
2729 printWriter.write("<!-- generating all possible page-->");
2730 printWriter.flush();
2731
2732 FileWriter fstream = new FileWriter(destination+fileName+".html");
2733 BufferedWriter out = new BufferedWriter(fstream);
2734
2735 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
2736 out.write ("<html>\r\n");
2737 out.write("<head>\r\n<title>Metadata Element Sort List</title>\r\n");
2738 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
2739 out.write("<script type=\"text/javascript\" src=\"../script/getInfomation.js\"></script>\r\n");
2740 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/yahoo/yahoo-min.js\"></script>\r\n");
2741 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/event/event-min.js\"></script>\r\n");
2742 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/connection/connection-min.js\"></script>\r\n");
2743
2744 out.write("<style type=\"text/css\">\r\n");
2745 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
2746 out.write(".tam{height:20; text-align:center}\r\n");
2747 out.write("th{height:20; text-align:center;}\r\n");
2748 out.write("body{font-family: Arial;}\r\n");
2749 out.write("</style>\r\n");
2750 out.write("</head><body id=\""+collectionFolder+"\">\r\n");
2751
2752 out.write("<td><div style=\"float:left;\"><a href=\" Overall.html \">Summary</a>");
2753 out.write("&raquo;");
2754
2755 if(metadataSetName.equalsIgnoreCase("dublin")){
2756 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2757 }
2758 else if (metadataSetName.equalsIgnoreCase("extracted")){
2759 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2760 }
2761 else{
2762 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2763 }
2764 out.write("&raquo;");
2765 out.write("<a href=\""+title+".html\">"+title+"</a></div>");
2766
2767 out.write("<div style=\"float:right;\"><a href=\"http://chnm.gmu.edu/tools/surveys/4386/\">Please send feedback about the Mat tool</a></div></td><br>");
2768
2769
2770 String[] list = dm.getSortList(title, sort);
2771
2772
2773
2774 if(list.length>=1){
2775
2776
2777 if(sort.equals("ASCII")){
2778
2779 out.write("<table border=\"1\" align=\"center\" width=\"800\">\r\n");
2780 out.write ("<h2 align=\"center\">"+ title+"</h2>\r\n");
2781 out.write ("<th class=\"tam\">ASCII Sort\r\n");
2782 out.write ("<th class=\"tam\">Element Values\r\n");
2783 out.write ("<th class=\"tam\">Source Documents\r\n");
2784 out.write ("<th class=\"tam\">Internal Link\r\n");
2785
2786 out.write ("<a name='topFive'>\r\n");
2787
2788 int counter = 0;
2789
2790 for(int i = 0; i<list.length; i++){
2791
2792 if(list.length<=5 && i==0){
2793 out.write ("<a name='lastFive'>\r\n");
2794 }
2795 else if((list.length>5) && (list.length-5==i)){
2796 out.write ("<a name='lastFive'>\r\n");
2797 }
2798
2799
2800 InternalLink il= (InternalLink)internalIDMap.get(list[i]);
2801 ArrayList alist2 = il.retrieveList();
2802 String id = (String)alist2.get(0);
2803 id = id.substring(4);
2804 out.write("<tr id=\""+id+"\" >\r\n");
2805
2806 if(list[i].length()>=201){
2807 if(list[i].startsWith("http://")){
2808
2809 if(list[i].length()>=100){
2810 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2811 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2812 }
2813 else{
2814 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2815 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2816 }
2817
2818 }
2819 else{
2820
2821 if(list[i].startsWith(" ")){
2822 String elements = list[i];
2823 list[i] = list[i].substring(1);
2824 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i].substring(0, 200)+"...<td>No Source Available \r\n");
2825 }
2826 else{
2827 out.write("<td>"+(counter+1)+"<td>"+list[i].substring(0, 200)+"... <td>No Source Available\r\n");
2828 }
2829 }
2830
2831 }
2832 else{
2833
2834
2835 if(list[i].startsWith("http://")){
2836
2837 if(list[i].length()>=100){
2838 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
2839 }
2840 else{
2841 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2842 }
2843 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2844
2845 }
2846
2847
2848
2849 else{
2850
2851 if(list[i].startsWith(" ")){
2852 String elements = list[i];
2853 list[i] = list[i].substring(1);
2854
2855 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i]+"<td>No Source Available\r\n");
2856 suggestionMap.put(elements,"No Source Available");
2857
2858 }
2859 else{
2860 out.write("<td>"+(counter+1)+"<td>"+list[i]+"<td>No Source Available\r\n");
2861 }
2862 }
2863 }
2864 counter++;
2865 out.write("<td onclick=\"GD(this)\">View");
2866 }
2867
2868 }
2869 else{
2870
2871
2872 out.write("<table border=\"1\" align=\"center\" width=\"800\">\r\n");
2873 out.write ("<h2 align=\"center\">"+ title+"</h2>\r\n");
2874 out.write ("<th class=\"tam\">&nbsp;\r\n");
2875 out.write ("<th class=\"tam\">Frequency\r\n");
2876 out.write ("<th class=\"tam\">Element Values\r\n");
2877 out.write ("<th class=\"tam\">Source Documents\r\n");
2878 out.write ("<th class=\"tam\">Internal Link\r\n");
2879 out.write ("<a name='topFive'>\r\n");
2880
2881 HashMap xMap = dm.getDistinctValueMap(title);
2882 int counter = 0;
2883 for(int i = 0; i<list.length; i++){
2884 if(list.length<=5 && i==0){
2885 out.write ("<a name='lastFive'>\r\n");
2886 }
2887 else if((list.length>5) && (list.length-5==i)){
2888 out.write ("<a name='lastFive'>\r\n");
2889 }
2890
2891 InternalLink il= (InternalLink)internalIDMap.get(list[i]);
2892 ArrayList alist2 = il.retrieveList();
2893 String id = (String)alist2.get(0);
2894 id = id.substring(4);
2895 out.write("<tr id=\""+id+"\" >\r\n");
2896
2897 out.write("<td>"+(counter+1)+"<th> "+((Integer)xMap.get(list[i])).toString()+"\r\n");
2898
2899 if(list[i].length()>=201){
2900
2901 if(list[i].startsWith("http://")){
2902 if(list[i].length()>=100){
2903 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2904 }
2905 else{
2906 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2907 }
2908
2909 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2910 }
2911
2912
2913
2914
2915 else{
2916 if(list[i].startsWith(" ")){
2917 list[i]=list[i].substring(1);
2918
2919 out.write("<td>"+spaceLeft+space+spaceRight+list[i].substring(0,200)+"...<td>No Source Available\r\n");
2920
2921 }
2922 else{
2923
2924 out.write("<td>"+list[i].substring(0,200)+"...<td>No Source Available\r\n");
2925
2926 }
2927 }
2928 }
2929 else{
2930 if(list[i].startsWith("http://")){
2931 if(list[i].length()>=100){
2932 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2933 }
2934 else{
2935 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2936 }
2937 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2938
2939 }
2940 else{
2941
2942 if(list[i].startsWith(" ")){
2943 list[i]=list[i].substring(1);
2944
2945
2946 out.write("<td>"+spaceLeft+space+spaceRight+list[i]+"<td>No Source Available\r\n");
2947
2948 }
2949 else{
2950
2951 out.write("<td>"+list[i]+"<td>No Source Available\r\n");
2952
2953 }
2954 }
2955 }
2956 counter++;
2957 out.write("<td onclick=\"GD(this)\">View");
2958 }
2959 }
2960 out.write("</table>\r\n");
2961 }
2962 else{
2963 out.write("<p>Sorry! The list is empty</p>");
2964 }
2965 out.write("<tr>\r\n");
2966
2967
2968 out.write("<p><a href=\" Overall.html \">Summary</a>");
2969 out.write("&raquo;");
2970
2971 if(metadataSetName.equalsIgnoreCase("dublin")){
2972 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2973 }
2974 else if (metadataSetName.equalsIgnoreCase("extracted")){
2975 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2976 }
2977 else{
2978 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2979 }
2980 out.write("&raquo;");
2981 out.write("<a href=\""+title+".html\">"+title+"</a></p>");
2982 out.write("</body></html>\r\n");
2983 out.close();
2984 fstream.close();
2985 }catch(Exception e){
2986
2987 e.printStackTrace(outx);
2988 }
2989
2990 return new HashMap();
2991 }
2992
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024 private void createIncompletedListNoIdentifier(String fileName, String metadataSetName, String collectionFolder,PrintWriter outx){
3025 //HashMap hp = dm.getIdentifierLink("dc.Identifier");
3026 HashMap hp = dm.getIdentifierLinkNoIdentifier();
3027 String[] ids = dm.getDocumentIDList(fileName);
3028 System.out.println("Length: "+ids.length+" "+fileName+ " "+hp.size());
3029
3030 Set st = hp.keySet();
3031 Iterator it = st.iterator();
3032 while(it.hasNext()){
3033 System.out.println(it.next());
3034 }
3035 for(int i = 0; i<ids.length; i++){
3036 System.out.println("IDS:" +ids[i]);
3037 hp.remove(ids[i]);
3038 }
3039 System.out.println("Length: "+ids.length+" "+fileName+ " "+hp.size());
3040 //String collectionFolder = mds.getCollectionName();
3041 //System.out.println(hp.size());
3042 try{
3043 FileWriter fstream = new FileWriter(destination+fileName+"_IncompletedList.html");
3044 BufferedWriter out = new BufferedWriter(fstream);
3045
3046 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
3047 out.write ("<html>\r\n");
3048 out.write("<head>\r\n<title>Incompleted Document List</title>\r\n");
3049 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
3050 out.write("<style type=\"text/css\">\r\n");
3051 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
3052 out.write(".tam{height:20; text-align:center}\r\n");
3053 out.write("th{height:20; text-align:center;}\r\n");
3054 out.write("body{font-family: Arial;}\r\n");
3055 out.write("</style>\r\n");
3056 out.write("</head><body>\r\n");
3057
3058 out.write("<p><a href=\" Overall.html \">Summary</a>");
3059 out.write("&raquo;");
3060 if(metadataSetName.equalsIgnoreCase("dublin")){
3061 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
3062 }
3063 else if (metadataSetName.equalsIgnoreCase("extracted")){
3064 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
3065 }
3066 else{
3067 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
3068 }
3069 out.write("&raquo;");
3070 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
3071
3072 out.write("<h3 align=center>"+fileName+" does not appear in the following documents</h3>");
3073 int counter = 0;
3074 System.out.println("link.....");
3075 Set s = hp.keySet();
3076 Iterator i = s.iterator();
3077 System.out.println("link2.....");
3078 out.write("<table border=1 align=center><tr><th>Document ID<th>Source Link\r\n");
3079 while(i.hasNext()){
3080 //printWriter.write("<!-- generating all possible page-->");
3081 System.out.println("link.....3");
3082 counter++;
3083 String keys = (String)i.next();
3084 System.out.println("link.....4"+keys);
3085 InternalLink il = (InternalLink) hp.get(keys);
3086
3087 ArrayList alist = il.retrieveList();
3088 System.out.println("link.....5"+alist.size());
3089 String url = (String)alist.get(0);
3090 out.write("<tr><td>"+counter+"<td><a href=\""+url+"\">"+url+"</a>\r\n");
3091 }
3092 out.write("</table></body></html>\r\n");
3093
3094
3095 out.write("<p><a href=\" Overall.html \">Summary</a>");
3096 out.write("&raquo;");
3097 if(metadataSetName.equalsIgnoreCase("dublin")){
3098 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
3099 }
3100 else if (metadataSetName.equalsIgnoreCase("extracted")){
3101 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
3102 }
3103 else{
3104 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
3105 }
3106 out.write("&raquo;");
3107 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
3108
3109 out.close();
3110 fstream.close();
3111 }catch(IOException ex){ex.printStackTrace();}
3112
3113
3114 }
3115 ///////////////////////////////////////////////////////////////////////////////////////
3116}
3117
Note: See TracBrowser for help on using the repository browser.