source: other-projects/trunk/greenstone3-extension/mat/src/org/greenstone/gsdl3_extension/mat/servlet/PrintHTML.java@ 18000

Last change on this file since 18000 was 18000, checked in by cc108, 15 years ago

new source code
File size: 107.1 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;
20
21public class PrintHTML {
22
23 private final String NORMAL = "normal";
24 private final String WORST = "worst";
25 private final String BEST = "best";
26 private final String title1 ="Hide Empty Metadata ,";
27 private final String title2 ="Hide Completed Metadata ,";
28 private final String title3 ="Hide Documents with Empty Metadata ,";
29 private final String title4 ="Hide Documents with Completed Metadata ,";
30 private final String title5 ="No Available Graph";
31 private final String metadataSet1 = "Dublin Core";
32 private final String metadataSet2 = "Extracted";
33 private final String spaceLeft = "<I>&laquo;";
34 private final String spaceRight = "&raquo;</I>";
35 private final String htmlSpace = "&nbsp;";
36 private final String oddChar ="odd Character";
37 private final String destination;
38 private final String space ="space";
39 private String hostName;
40 private String cssString;
41 private int port_number = 9090;
42
43 String linkIdentifier = "";
44
45
46
47 private HashMap url;
48
49 MetadataStats mds;
50 DataMaker dm;
51 Calendar cl;
52 SimpleDateFormat sdf;
53 PrintWriter printWriter;
54
55 public PrintHTML(MetadataStats ms, PrintWriter arg){
56
57 mds = ms;
58 printWriter = arg;
59 dm = new DataMaker(mds);
60 cl=Calendar.getInstance();
61 linkIdentifier = "dc.Identifier";
62 destination = mds.HTMLDirectory;
63 sdf = new SimpleDateFormat(" dd MMM yyyy 'at' HH:mm:ss z 'GMT'Z");
64 setupMetadataLink("metadataElementURL");
65
66
67 try {
68 java.net.InetAddress localMachine = java.net.InetAddress.getLocalHost();
69 hostName = localMachine.getHostName();
70 }catch (java.net.UnknownHostException uhe) {
71 uhe.printStackTrace();
72 }
73
74 cssString = "<link rel=\"stylesheet\" href=\"http://"+hostName+":"+port_number+"/greenstone3/mat/script/mat.css\" type=\"text/css\" >";
75 }
76
77 private ArrayList getHideEmptyMetadataDetail(ArrayList dataset, String[] id, String[] metadataName){
78
79 printWriter.write("<!-- generating all possible page -->");
80 printWriter.flush();
81
82 if(id.length==0 || metadataName.length==0){
83 return new ArrayList();
84 }
85
86 ArrayList detailList = new ArrayList();
87 ArrayList alist = new ArrayList();
88 ArrayList nameList = new ArrayList();
89
90 for(int i = 0; i< metadataName.length; i++){
91 if(!dm.IsElementEmpty(metadataName[i])){
92 alist.add(dm.getMetadataRows(metadataName[i]));
93 nameList.add(metadataName[i]);
94 }
95 }
96
97 String[] metaDataElementName = new String[nameList.size()];
98
99 for(int i = 0; i< nameList.size(); i++){
100 metaDataElementName[i] = nameList.get(i).toString();
101 }
102
103 if(id.length==0 || metaDataElementName.length==0){
104 return new ArrayList();
105 }
106
107 detailList.add(alist);
108 detailList.add(id);
109 detailList.add(metaDataElementName);
110 printWriter.write("<!-- generating all possible page-->");
111 printWriter.flush();
112
113 return detailList;
114 }
115
116 private ArrayList getHideFullMetadataDetail(ArrayList dataset, String[] id, String[] metadataName){
117
118 printWriter.write("<!-- generating all possible page-->");
119 printWriter.flush();
120
121 if(id.length==0 || metadataName.length==0){
122 return new ArrayList();
123 }
124
125 ArrayList detailList = new ArrayList();
126 ArrayList alist = new ArrayList();
127 ArrayList nameList = new ArrayList();
128
129 for(int i = 0; i< metadataName.length; i++){
130 if(!dm.IsElementFull(metadataName[i])){
131 alist.add(dm.getMetadataRows(metadataName[i]));
132 nameList.add(metadataName[i]);
133 }
134 }
135
136 String[] metaDataElementName = new String[nameList.size()];
137
138 for(int i = 0; i< nameList.size(); i++){
139 metaDataElementName[i] = nameList.get(i).toString();
140 }
141
142 if(id.length==0 || metaDataElementName.length==0){
143 return new ArrayList();
144 }
145
146 detailList.add(alist);
147 detailList.add(id);
148 detailList.add(metaDataElementName);
149
150 printWriter.write("<!-- generating all possible page-->");
151 printWriter.flush();
152
153 return detailList;
154 }
155
156 private ArrayList getHideEmptyDocumentDetail(ArrayList dataset, String[] id, String[] metadataName){
157
158 printWriter.write("<!-- generating all possible page-->");
159 printWriter.flush();
160
161 if(id.length==0 || metadataName.length==0){
162 return new ArrayList();
163 }
164
165 ArrayList detailList = new ArrayList();
166 ArrayList alist = dm.removeDocument(dataset, id, 0);
167 ArrayList idList = dm.getRemovedID();
168 ArrayList newIDList = new ArrayList();
169
170 for(int i = 0; i< id.length; i++){
171 if(!idList.contains(id[i])){
172 newIDList.add(id[i]);
173 }
174 }
175
176 id = new String[newIDList.size()];
177 for(int i = 0; i<newIDList.size(); i++){
178 id[i] = newIDList.get(i).toString();
179 }
180
181 if(id.length==0 || metadataName.length==0){
182 return new ArrayList();
183 }
184
185 detailList.add(alist);
186 detailList.add(id);
187 detailList.add(metadataName);
188
189 printWriter.write("<!-- generating all possible page-->");
190 printWriter.flush();
191
192 return detailList;
193 }
194
195 private ArrayList getHideFullDocumentDetail(ArrayList dataset, String[] id, String[] metadataName){
196
197 printWriter.write("<!-- generating all possible page-->");
198 printWriter.flush();
199
200 if(id.length==0 || metadataName.length==0){
201 return new ArrayList();
202 }
203
204 ArrayList detailList = new ArrayList();
205 ArrayList alist = dm.removeDocument(dataset, id ,1);
206 ArrayList idList = dm.getRemovedID();
207 ArrayList newIDList = new ArrayList();
208
209 for(int i = 0; i< id.length; i++){
210 if(!idList.contains(id[i])){
211 newIDList.add(id[i]);
212 }
213 }
214
215 id = new String[newIDList.size()];
216 for(int i = 0; i<newIDList.size(); i++){
217 id[i] = newIDList.get(i).toString();
218 }
219
220 if(id.length==0 || metadataName.length==0){
221 return new ArrayList();
222 }
223
224 detailList.add(alist);
225 detailList.add(id);
226 detailList.add(metadataName);
227
228 printWriter.write("<!-- generating all possible page-->");
229 printWriter.flush();
230
231 return detailList;
232 }
233
234 public void generateAllPossibleGraph(ArrayList Namelist,String[]ids,String[] names,String condition){
235
236 printWriter.write("<!-- generating all possible page-->");
237 printWriter.flush();
238
239 ArrayList dataList= Namelist;
240 String[] id = ids;
241 String[] metadataName = names;
242
243 ArrayList alist = new ArrayList();
244 ArrayList detailList = new ArrayList();
245 ArrayList new_dataList = new ArrayList();
246 String[] idList;
247 String[] metadataNameList;
248
249 //---------------------------------------------------------------------------------------------------------SS
250 //-----------------------------------------------------------------------SSSS
251 if(id.length==0 || metadataName.length==0){
252 generateEmptyGraph("SSSS",WORST,condition);
253 generateEmptyGraph("SSSS",BEST,condition);
254 }
255
256 else{
257 new_dataList = transformDataList(dataList,id);
258 WriteHTML("SSSS",new_dataList,id,metadataName,WORST,"Show completed graph",condition);
259 WriteHTML("SSSS",new_dataList,id,metadataName,BEST,"Show completed graph",condition);
260 }
261 //-----------------------------------------------------------------------SSSH
262 detailList = getHideFullDocumentDetail(dataList,id,metadataName);
263 if(detailList.size()==0){
264 generateEmptyGraph("SSSH",WORST,condition);
265 generateEmptyGraph("SSSH",BEST,condition);
266 }
267 else{
268 alist = (ArrayList)detailList.get(0);
269 idList = (String[])detailList.get(1);
270 metadataNameList = (String[])detailList.get(2);
271 new_dataList = transformDataList(alist,idList);
272 WriteHTML("SSSH",new_dataList,idList,metadataNameList,WORST,title4,condition);
273 WriteHTML("SSSH",new_dataList,idList,metadataNameList,BEST,title4,condition);
274 }
275 //-----------------------------------------------------------------------SSHS
276 detailList = getHideEmptyDocumentDetail(dataList,id,metadataName);
277 if(detailList.size()==0){
278 generateEmptyGraph("SSHS",WORST,condition);
279 generateEmptyGraph("SSHS",BEST,condition);
280 generateEmptyGraph("SSHH",WORST,condition);
281 generateEmptyGraph("SSHH",BEST,condition);
282 }
283 else{
284 alist = (ArrayList)detailList.get(0);
285 idList = (String[])detailList.get(1);
286 metadataNameList = (String[])detailList.get(2);
287 new_dataList = transformDataList(alist,idList);
288 WriteHTML("SSHS",new_dataList,idList,metadataNameList,WORST,title3,condition);
289 WriteHTML("SSHS",new_dataList,idList,metadataNameList,BEST,title3,condition);
290 //-----------------------------------------------------------------------SSHH
291 detailList = getHideFullDocumentDetail(alist,idList,metadataNameList);
292 if(detailList.size()==0){
293 generateEmptyGraph("SSHH",WORST,condition);
294 generateEmptyGraph("SSHH",BEST,condition);
295 }
296 else{
297 alist = (ArrayList)detailList.get(0);
298 idList = (String[])detailList.get(1);
299 metadataNameList = (String[])detailList.get(2);
300 new_dataList = transformDataList(alist,idList);
301 WriteHTML("SSHH",new_dataList,idList,metadataNameList,WORST,title3+title4,condition);
302 WriteHTML("SSHH",new_dataList,idList,metadataNameList,BEST,title3+title4,condition);
303 }
304 }
305
306 //---------------------------------------------------------------------------------------------------------SH
307 ArrayList xList = new ArrayList();
308 String[] idListCopy;
309 String[] metadataNameListCopy;
310
311 //-----------------------------------------------------------------------SHSS
312 detailList = getHideFullMetadataDetail(dataList,id,metadataName);
313
314 if(detailList.size()==0){
315 generateEmptyGraph("SHSS",WORST,condition);
316 generateEmptyGraph("SHSS",BEST,condition);
317
318 generateEmptyGraph("SHSH",WORST,condition);
319 generateEmptyGraph("SHSH",BEST,condition);
320
321 generateEmptyGraph("SHHS",WORST,condition);
322 generateEmptyGraph("SHHS",BEST,condition);
323
324 generateEmptyGraph("SHHH",WORST,condition);
325 generateEmptyGraph("SHHH",BEST,condition);
326 }
327 else{
328 alist = (ArrayList)detailList.get(0);
329 idList = (String[])detailList.get(1);
330 metadataNameList = (String[])detailList.get(2);
331
332 xList = (ArrayList)alist.clone();
333 idListCopy = (String[]) idList.clone();
334 metadataNameListCopy = (String[])metadataNameList.clone();
335 new_dataList = transformDataList(alist,idList);
336 WriteHTML("SHSS",new_dataList,idList,metadataNameList,WORST,title2,condition);
337 WriteHTML("SHSS",new_dataList,idList,metadataNameList,BEST,title2,condition);
338
339 //-----------------------------------------------------------------------SHHS
340 detailList = getHideEmptyDocumentDetail((ArrayList)xList.clone(),(String[])idListCopy.clone(),(String[])metadataNameListCopy.clone());
341 if(detailList.size()==0){
342 generateEmptyGraph("SHHS",WORST,condition);
343 generateEmptyGraph("SHHS",BEST,condition);
344 generateEmptyGraph("SHHH",WORST,condition);
345 generateEmptyGraph("SHHH",BEST,condition);
346 }
347 else{
348 alist = (ArrayList)detailList.get(0);
349 idList = (String[])detailList.get(1);
350 metadataNameList = (String[])detailList.get(2);
351 new_dataList = transformDataList(alist,idList);
352 WriteHTML("SHHS",new_dataList,idList,metadataNameList,WORST,title2+title3,condition);
353 WriteHTML("SHHS",new_dataList,idList,metadataNameList,BEST,title2+title3,condition);
354
355 //-----------------------------------------------------------------------SHHH
356 detailList = getHideFullDocumentDetail(alist,idList,metadataNameList);
357 if(detailList.size()==0){
358 generateEmptyGraph("SHHH",WORST,condition);
359 generateEmptyGraph("SHHH",BEST,condition);
360 }
361 else{
362 alist = (ArrayList)detailList.get(0);
363 idList = (String[])detailList.get(1);
364 metadataNameList = (String[])detailList.get(2);
365 new_dataList = transformDataList(alist,idList);
366 WriteHTML("SHHH",new_dataList,idList,metadataNameList,WORST,title2+title3+title4,condition);
367 WriteHTML("SHHH",new_dataList,idList,metadataNameList,BEST,title2+title3+title4,condition);
368 }
369 }
370
371 //-----------------------------------------------------------------------SHSH
372 detailList = getHideFullDocumentDetail((ArrayList)xList.clone(),(String[])idListCopy.clone(),(String[])metadataNameListCopy.clone());
373 if(detailList.size()==0){
374 generateEmptyGraph("SHSH",WORST,condition);
375 generateEmptyGraph("SHSH",BEST,condition);
376 }
377 else{
378 alist = (ArrayList)detailList.get(0);
379 idList = (String[])detailList.get(1);
380 metadataNameList = (String[])detailList.get(2);
381 new_dataList = transformDataList(alist,idList);
382 WriteHTML("SHSH",new_dataList,idList,metadataNameList,WORST,title2+title4,condition);
383 WriteHTML("SHSH",new_dataList,idList,metadataNameList,BEST,title2+title4,condition);
384 }
385 }
386
387 //---------------------------------------------------------------------------------------------------------HS
388 ArrayList xList1 = new ArrayList();
389 String[] idListCopy1;
390 String[] metadataNameListCopy1;
391
392 //-----------------------------------------------------------------------HSSS
393 detailList = getHideEmptyMetadataDetail(dataList,id,metadataName);
394
395 if(detailList.size()==0){
396 generateEmptyGraph("HSSS",WORST,condition);
397 generateEmptyGraph("HSSS",BEST,condition);
398
399 generateEmptyGraph("HSHS",WORST,condition);
400 generateEmptyGraph("HSHS",BEST,condition);
401
402 generateEmptyGraph("HSHH",WORST,condition);
403 generateEmptyGraph("HSHH",BEST,condition);
404
405 generateEmptyGraph("HSSH",WORST,condition);
406 generateEmptyGraph("HSSH",BEST,condition);
407 }
408 else{
409 alist = (ArrayList)detailList.get(0);
410 idList = (String[])detailList.get(1);
411 metadataNameList = (String[])detailList.get(2);
412
413 xList1 = (ArrayList)alist.clone();
414 idListCopy1 = (String[]) idList.clone();
415 metadataNameListCopy1 = (String[])metadataNameList.clone();
416
417 new_dataList = transformDataList(alist,idList);
418 WriteHTML("HSSS",new_dataList,idList,metadataNameList,WORST,title1,condition);
419 WriteHTML("HSSS",new_dataList,idList,metadataNameList,BEST,title1,condition);
420
421 //-----------------------------------------------------------------------HSHS
422 detailList = getHideEmptyDocumentDetail((ArrayList)xList1.clone(),(String[])idListCopy1.clone(),(String[])metadataNameListCopy1.clone());
423 if(detailList.size()==0){
424 generateEmptyGraph("HSHS",WORST,condition);
425 generateEmptyGraph("HSHS",BEST,condition);
426
427 generateEmptyGraph("HSHH",WORST,condition);
428 generateEmptyGraph("HSHH",BEST,condition);
429 }
430 else{
431 alist = (ArrayList)detailList.get(0);
432 idList = (String[])detailList.get(1);
433 metadataNameList = (String[])detailList.get(2);
434
435 new_dataList = transformDataList(alist,idList);
436 WriteHTML("HSHS",new_dataList,idList,metadataNameList,WORST,title1+title3,condition);
437 WriteHTML("HSHS",new_dataList,idList,metadataNameList,BEST,title1+title3,condition);
438
439 //-----------------------------------------------------------------------HSHH
440 detailList = getHideFullDocumentDetail(alist,idList,metadataNameList);
441 if(detailList.size()==0){
442 generateEmptyGraph("HSHH",WORST,condition);
443 generateEmptyGraph("HSHH",BEST,condition);
444 }
445 else{
446 alist = (ArrayList)detailList.get(0);
447 idList = (String[])detailList.get(1);
448 metadataNameList = (String[])detailList.get(2);
449
450 new_dataList = transformDataList(alist,idList);
451 WriteHTML("HSHH",new_dataList,idList,metadataNameList,WORST,title1+title3+title4,condition);
452 WriteHTML("HSHH",new_dataList,idList,metadataNameList,BEST,title1+title3+title4,condition);
453 }
454 }
455
456 //-----------------------------------------------------------------------HSSH
457 detailList = getHideFullDocumentDetail((ArrayList)xList1.clone(),(String[])idListCopy1.clone(),(String[])metadataNameListCopy1.clone());
458 if(detailList.size()==0){
459 generateEmptyGraph("HSSH",WORST,condition);
460 generateEmptyGraph("HSSH",BEST,condition);
461 }
462 else{
463 alist = (ArrayList)detailList.get(0);
464 idList = (String[])detailList.get(1);
465 metadataNameList = (String[])detailList.get(2);
466
467 new_dataList = transformDataList(alist,idList);
468 WriteHTML("HSSH",new_dataList,idList,metadataNameList,WORST,title1+title4,condition);
469 WriteHTML("HSSH",new_dataList,idList,metadataNameList,BEST,title1+title4,condition);
470 }
471 }
472
473 //---------------------------------------------------------------------------------------------------------HH
474 ArrayList xList2 = new ArrayList();
475 String[] idListCopy2;
476 String[] metadataNameListCopy2;
477
478 //-----------------------------------------------------------------------HHSS
479 detailList = getHideEmptyMetadataDetail(dataList,id,metadataName);
480 if(detailList.size()==0){
481 generateEmptyGraph("HHSS",WORST,condition);
482 generateEmptyGraph("HHSS",BEST,condition);
483
484 generateEmptyGraph("HHHS",WORST,condition);
485 generateEmptyGraph("HHHS",BEST,condition);
486
487 generateEmptyGraph("HHHH",WORST,condition);
488 generateEmptyGraph("HHHH",BEST,condition);
489
490 generateEmptyGraph("HHSH",WORST,condition);
491 generateEmptyGraph("HHSH",BEST,condition);
492 }
493 else{
494 alist = (ArrayList)detailList.get(0);
495 idList = (String[])detailList.get(1);
496 metadataNameList = (String[])detailList.get(2);
497
498 detailList = getHideFullMetadataDetail(alist,idList,metadataNameList);
499
500 if(detailList.size()==0){
501 generateEmptyGraph("HHSS",WORST,condition);
502 generateEmptyGraph("HHSS",BEST,condition);
503
504 generateEmptyGraph("HHHS",WORST,condition);
505 generateEmptyGraph("HHHS",BEST,condition);
506
507 generateEmptyGraph("HHHH",WORST,condition);
508 generateEmptyGraph("HHHH",BEST,condition);
509
510 generateEmptyGraph("HHSH",WORST,condition);
511 generateEmptyGraph("HHSH",BEST,condition);
512 }
513 else{
514 alist = (ArrayList)detailList.get(0);
515 idList = (String[])detailList.get(1);
516 metadataNameList = (String[])detailList.get(2);
517
518 xList2 = (ArrayList)alist.clone();
519 idListCopy2 = (String[]) idList.clone();
520 metadataNameListCopy2 = (String[])metadataNameList.clone();
521
522 new_dataList = transformDataList(alist,idList);
523 WriteHTML("HHSS",new_dataList,idList,metadataNameList,WORST,title1+title2,condition);
524 WriteHTML("HHSS",new_dataList,idList,metadataNameList,BEST,title1+title2,condition);
525
526 //---------------------------------------------------------------------- HHSH
527 detailList = getHideFullDocumentDetail((ArrayList)xList2.clone(),(String[])idListCopy2.clone(),(String[])metadataNameListCopy2.clone());
528 if(detailList.size()==0){
529 generateEmptyGraph("HHSH",WORST,condition);
530 generateEmptyGraph("HHSH",BEST,condition);
531 }
532 else{
533 alist = (ArrayList)detailList.get(0);
534 idList = (String[])detailList.get(1);
535 metadataNameList = (String[])detailList.get(2);
536
537 new_dataList = transformDataList(alist,idList);
538
539 WriteHTML("HHSH",new_dataList,idList,metadataNameList,WORST,title1+title2+title4,condition);
540 WriteHTML("HHSH",new_dataList,idList,metadataNameList,BEST,title1+title2+title4,condition);
541 }
542 //----------------------------------------------------------------------HHHS
543 detailList = getHideEmptyDocumentDetail((ArrayList)xList2.clone(),(String[])idListCopy2.clone(),(String[])metadataNameListCopy2.clone());
544 if(detailList.size()==0){
545 generateEmptyGraph("HHHS",WORST,condition);
546 generateEmptyGraph("HHHS",BEST,condition);
547
548 generateEmptyGraph("HHHH",WORST,condition);
549 generateEmptyGraph("HHHH",BEST,condition);
550 }
551 else{
552 alist = (ArrayList)detailList.get(0);
553 idList = (String[])detailList.get(1);
554 metadataNameList = (String[])detailList.get(2);
555
556 new_dataList = transformDataList(alist,idList);
557
558 WriteHTML("HHHS",new_dataList,idList,metadataNameList,WORST,title1+title2+title3,condition);
559 WriteHTML("HHHS",new_dataList,idList,metadataNameList,BEST,title1+title2+title3,condition);
560
561 //----------------------------------------------------------------------HHHH
562 detailList = getHideFullDocumentDetail((ArrayList)alist.clone(),(String[])idList.clone(),(String[])metadataNameList.clone());
563 if(detailList.size()==0){
564 generateEmptyGraph("HHHH",WORST,condition);
565 generateEmptyGraph("HHHH",BEST,condition);
566 }
567 else{
568 alist = (ArrayList)detailList.get(0);
569 idList = (String[])detailList.get(1);
570 metadataNameList = (String[])detailList.get(2);
571
572 new_dataList = transformDataList(alist,idList);
573
574 WriteHTML("HHHH",new_dataList,idList,metadataNameList,WORST,title1+title2+title3+title4,condition);
575 WriteHTML("HHHH",new_dataList,idList,metadataNameList,BEST,title1+title2+title3+title4,condition);
576 }
577 }
578 }
579 }
580 }
581
582 private ArrayList transformDataList(ArrayList list, String[] ids){
583
584 printWriter.write("<!-- generating all possible page-->");
585 printWriter.flush();
586
587 ArrayList wholeList = new ArrayList();
588
589 for(int i = 0; i< ids.length; i++){
590 ArrayList idList = new ArrayList();
591 for(int j = 0; j<list.size();j++){
592 int[] datarows = (int[]) list.get(j);
593 idList.add(new Integer (datarows[i]));
594 }
595 wholeList.add(idList);
596 }
597
598 for(int i = 0; i< wholeList.size(); i++){
599 ArrayList alist = (ArrayList)wholeList.get(i);
600 int[] rows = new int[alist.size()];
601 for(int j = 0; j< alist.size(); j++){
602 rows[j] = ((Integer)alist.get(j)).intValue();
603 }
604 wholeList.remove(i);
605 wholeList.add(i,rows);
606 }
607
608 printWriter.write("<!-- generating all possible page-->");
609 printWriter.flush();
610
611 return wholeList;
612
613 }
614
615 private void generateEmptyGraph(String fileName,String condition,String suffix){
616
617 printWriter.write("<!-- generating all possible page-->");
618 printWriter.flush();
619
620 String suf = suffix;
621 if(!suffix.equalsIgnoreCase("dublin")){
622 suf = "other";
623 }
624
625 String cases = condition;
626 String collectionFolder = mds.getCollectionName();
627
628 try{
629 FileWriter fstream = new FileWriter(destination+suf+"_"+fileName+"_"+cases+".html");
630 BufferedWriter out = new BufferedWriter(fstream);
631 out.write("<!-- This comment keeps IE6/7 in the reliable quirks mode -->\r\n");
632 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">\r\n");
633 out.write("<html>\r\n");
634 out.write("<head>\r\n<title> No Available Chart</title>\r\n");
635 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
636 out.write("<link rel=\"stylesheet\" href=\"../script/doby2.css\" type=\"text/css\"/>\r\n");
637 out.write("</head>\r\n");
638 out.write("<p><a href=\" Overall.html \">Summary</a></p>");
639 out.write("<body>\r\n");
640 out.write("<p>No data available to render chart.</p>\r\n");
641 out.write("<p>Reason: Document number is zero or Metadata element number is zero </p>\r\n");
642 out.write("</body></html>\r\n");
643 out.close();
644 fstream.close();
645
646 }catch(Exception e){}
647
648 printWriter.write("<!-- generating all possible page-->");
649 printWriter.flush();
650 }
651
652 private void WriteHTML(String fileName, ArrayList dataset, String[] ids, String[] metadataName, String condition, String title, String suffix){
653
654 int blueDot = 0;
655 String cases = condition;
656 String suf = suffix;
657 if(!suffix.equalsIgnoreCase("dublin")){
658 suf = "other";
659 }
660 try{
661 printWriter.write("<!-- generating all possible page-->");
662 printWriter.flush();
663
664 ArrayList tempList = mds.getMetadataNameList();
665 ArrayList urlIDList;
666 if(tempList.contains("dc.Identifier") && linkIdentifier.equals("dc.Identifier")){
667 urlIDList = dm.getURLMap("dc.Identifier");
668 }
669 else if(tempList.contains("nzir_internal.Link") && linkIdentifier.equals("nzir_internal.Link")){
670 urlIDList = dm.getURLMap("nzir_internal.Link");
671 }
672 else{
673 urlIDList = new ArrayList();
674 }
675
676 FileWriter fstream = new FileWriter(destination+suf+"_"+fileName+"_"+cases+".html");
677 BufferedWriter out = new BufferedWriter(fstream);
678
679 out.write("<!-- This comment keeps IE6/7 in the reliable quirks mode -->\r\n");
680 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">\r\n");
681 out.write("<html>\r\n");
682 out.write("<head>\r\n<title>"+title+"</title>\r\n");
683 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
684 out.write("<link rel=\"stylesheet\" href=\"../script/doby2.css\" type=\"text/css\"/>\r\n");
685 out.write("<script type=\"text/javascript\" src=\"../script/getInfomation.js\"></script>\r\n");
686 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/yahoo/yahoo-min.js\"></script>\r\n");
687 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/event/event-min.js\"></script>\r\n");
688 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/connection/connection-min.js\"></script>\r\n");
689 out.write("</head>\r\n");
690 out.write ("<p><a href=\" Overall.html \">Summary</a></p>");
691 out.write("<body id=\""+mds.getCollectionName()+"\" onLoad=\"reconfig()\">\r\n");
692 out.write("<div id=\"container\">\r\n");
693 out.write("<div class=\"tableContainer\">\r\n");
694 out.write ("<table cellspacing=\"0\">\r\n");
695 out.write ("<thead>\r\n");
696 out.write ("<tr>\r\n");
697 out.write("<td class=\"qh\">Info\r\n");
698 out.write("<td class=\"qh\">URL\r\n");
699
700 for(int a = 0; a< metadataName.length; a++){
701 out.write("<td>"+ metadataName[a]+"\r\n");
702 }
703
704 out.write ("</thead>\r\n<tfoot>\r\n<tr>\r\n");
705 out.write("<td class=\"qh\">&nbsp;\r\n");
706 out.write("<td class=\"qh\">&nbsp;\r\n");
707
708 for(int a = 0; a< metadataName.length; a++){
709 out.write("<td>"+ dm.Mean(metadataName[a])+"%\r\n");
710 }
711
712 out.write ("</tfoot>\r\n<tbody>\r\n");
713
714 if(cases.equals("normal")){
715 for(int i = ids.length; i>0; i--){
716 out.write ("<tr>");
717 int[] datarows = (int[])dataset.get(i-1);
718
719 for(int j = 0; j<datarows.length; j++){
720
721 if(datarows[j]==1){
722 out.write ("<td class=\"b\">");
723 blueDot++;
724 }
725 else {
726 out.write ("<td>&nbsp;");
727 }
728 }
729 }
730 }
731 else{
732 ArrayList idList = new ArrayList();
733 for(int i = 0; i<ids.length;i++){
734 idList.add(ids[i]);
735 }
736
737
738 HashMap hp = new HashMap();
739
740 for(int i = ids.length; i>0; i--){
741 String idValue = ids[i-1];
742 int dots = 0;
743 int[] datarows = (int[])dataset.get(i-1);
744 for(int j = 0; j<datarows.length; j++){
745 if(datarows[j]==1){
746 dots++;
747 }
748 }
749 hp.put(idValue,new Integer(dots));
750
751 }
752 ArrayList alist = dm.sortMap(hp);
753 String[] idIntValue = new String[alist.size()];
754 for(int i = 0; i< alist.size(); i++){
755 Map.Entry entry = (Map.Entry) alist.get(i);
756 String idElement = ((String) entry.getKey());
757 idIntValue[i] = idElement;
758 }
759
760 if(cases.equals("worst")){
761
762 for(int i = 0; i<idIntValue.length; i++){
763
764 printWriter.write("<!-- generating all possible page -->");
765 printWriter.flush();
766
767 int value = idList.indexOf(idIntValue[i]);
768 int[] datarows = (int[])dataset.get(value);
769
770 out.write ("<tr id=\""+idIntValue[i].substring(4)+"\">\r\n");
771 out.write("<td class=\"E\" onclick=\"GD(this)\">&nbsp;\r\n");
772
773 if(urlIDList.contains(idIntValue[i])){
774 out.write("<td class=\"qh\" onclick=\"GX(this)\"><span title=\"open URL in new window\">open</span>\r\n");
775 }
776 else{
777 out.write("<td>\r\n");
778 }
779
780 for(int j = 0; j<datarows.length; j++){
781
782 if(datarows[j]==1){
783 out.write ("<td class=\"b\">\r\n");
784 blueDot++;
785 }
786 else {
787 out.write ("<td class=\"w\">\r\n");
788 }
789 }
790 }
791 }
792
793 if(cases.equals("best")){
794
795 for(int i = idIntValue.length; i>0; i--){
796 printWriter.write("<!-- generating all possible page -->");
797 printWriter.flush();
798
799 int value = idList.indexOf(idIntValue[i-1]);
800 int[] datarows = (int[])dataset.get(value);
801
802 out.write ("<tr id=\""+idIntValue[i-1].substring(4)+"\">\r\n");
803 out.write("<td class=\"E\" onclick=\"GD(this)\">&nbsp;\r\n");
804
805 if(urlIDList.contains(idIntValue[i-1])){
806 out.write("<td class=\"qh\" onclick=\"GX(this)\"><span title=\"open URL in new window\">open</span>\r\n");
807 }
808 else{
809 out.write("<td>\r\n");
810 }
811
812 for(int j = 0; j<datarows.length; j++){
813
814 if(datarows[j]==1){
815 out.write ("<td class=\"b\">\r\n");
816 blueDot++;
817 }
818 else {
819 out.write ("<td class=\"w\">\r\n");
820 }
821 }
822 }
823 }
824 }
825
826 out.write ("</tbody>\r\n");
827 out.write ("</table>\r\n</div></div>\r\n");
828
829 int t1 = ids.length;
830 int t3 = metadataName.length;
831 int t4 = t1*t3;
832
833 out.write("<table>\r\n<tbody class=\"table1\">");
834 out.write("<tr>\r\n");
835
836 HashMap tempMap = mds.getMetadataSetMap();
837 MetadataSet ms = (MetadataSet)tempMap.get(suffix);
838
839 out.write("<td class=\"bfont\">This subset shows "+t1+" out of "+mds.getDocNum()+" documents");
840 out.write("<td class=\"bfont\">"+blueDot+" out of "+(mds.getDocNum()*t3)+" metadata items are defined");
841 out.write("<tr>\r\n");
842 out.write("<td class=\"bfont\">This subset shows "+t3+" out of "+(ms.getIndexsList().size())+" metadata elements");
843 out.write("<td class=\"bfont\">Subset completeness: "+dm.round((double)(blueDot*100/t4),5)+"%");
844 out.write("</table>\r\n");
845 out.write ("<p> <a href=\"Overall.html \">Summary</a></p>");
846 out.write ("</body></html>");
847 //Close the output stream
848
849 out.close();
850 fstream.close();
851 }catch (Exception e){
852 e.printStackTrace(printWriter);
853 }
854
855 printWriter.write("<!-- generating all possible page-->");
856 printWriter.flush();
857 }
858
859
860 public void generateOverallStatisticsPage(HashMap MetadataSetMap){
861
862 String fileName = "Overall";
863 ArrayList wholeList = new ArrayList();
864 HashMap hp = mds.getMetadataSetMap();
865 Collection c = hp.values();
866 Iterator i = c.iterator();
867 int counter = 0;
868 String[][] MetadataData = new String[c.size()][2];
869
870 while(i.hasNext()){
871 MetadataSet mds = (MetadataSet)i.next();
872 wholeList.add(mds);
873 ArrayList newMDS = new ArrayList();
874 newMDS.add(mds);
875
876 MetadataData[counter][0] = mds.getName();
877 MetadataData[counter][1] = dm.getSingleMetadataSetCompleteness(newMDS)+"%";
878
879 counter++;
880 }
881
882 try{
883 printWriter.write("<!-- generating all possible page-->");
884 printWriter.flush();
885
886 String str = sdf.format(cl.getTime());
887 String collectionFolder = mds.getCollectionName();
888
889 FileWriter fstream = new FileWriter(destination+fileName+".html");
890 BufferedWriter out = new BufferedWriter(fstream);
891
892 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
893 out.write("<html>\r\n");
894 out.write("<head><title>Summary</title>\r\n");
895 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
896 out.write("<style type=\"text/css\">\r\n");
897 out.write(".tam{height:20; text-align:right}\r\n");
898 out.write("th{height:20; text-align:left}\r\n");
899 out.write("body{font-family: Arial;}\r\n");
900 out.write("</style>\r\n");
901 out.write("<script type=\"text/javascript\" src=\"../script/status3.js\"></script>\r\n");
902 out.write("</head>\r\n");
903 out.write("<body>\r\n");
904 out.write ("<td> <div style=\"float:left;\"> <a href=\"http://"+hostName+":"+port_number+"/greenstone3/mat\">Mat Home</a></div>");
905 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>");
906 out.write("<h3 align=\"center\">Summary</h3>\r\n");
907 out.write("<table border=1 align=\"center\">\r\n");
908 out.write("<colgroup width=\"300\" span=\"2\">\r\n");
909 out.write("<tr>\r\n");
910 out.write("<th>OAI URL: </th>\r\n");
911
912 if(mds.getOAIURL().length()>50){
913 out.write("<td class=\"tam\"><font size=\"2px\"><a href=\""+mds.getOAIURL()+"\">"+mds.getOAIURL().substring(0,47)+"....</a></font>\r\n");
914 }
915 else{
916 out.write("<td class=\"tam\"><font size=\"2px\"><a href=\""+mds.getOAIURL()+"\">"+mds.getOAIURL()+"</a></font>\r\n");
917 }
918
919 out.write("</tr>\r\n");
920 out.write("<tr>\r\n");
921 out.write("<th>Number of Records:\r\n");
922 out.write("<td class=\"tam\">"+mds.getDocNum()+"\r\n");
923 out.write("</tr>\r\n");
924 out.write("<table border=1 align=\"center\">\r\n");
925 out.write("<colgroup width=\"300\" span=\"2\">\r\n");
926 out.write("<tr>\r\n");
927 out.write("<th>Metadata:\r\n");
928 out.write("<td class=\"tam\"><b>Completeness</b>\r\n");
929 out.write("</tr><br>\r\n");
930
931 for(int a = 0; a<MetadataData.length; a++){
932 out.write ("<tr>\r\n");
933 if(MetadataData[a][0].equalsIgnoreCase("dublin")){
934 out.write ("<th><a href =\""+MetadataData[a][0]+".html\">"+metadataSet1+"</a>");
935 }
936 else if (MetadataData[a][0].equalsIgnoreCase("extracted")){
937 out.write ("<th><a href =\""+MetadataData[a][0]+".html\">"+metadataSet2+"</a>");
938 }
939 else{
940 out.write ("<th><a href =\""+MetadataData[a][0]+".html\">"+MetadataData[a][0]+"</a>");
941 }
942 out.write ("<td class=\"tam\"> "+MetadataData[a][1]);
943 out.write ("</tr>\r\n");
944 }
945
946 out.write ("</table>");
947 out.write("<br>\r\n");
948 out.write("<FORM name=\"test\" onsubmit=\"checkStatus()\" action=\"\">");
949 out.write("<table border=1 align=\"center\">\r\n");
950 out.write ("<colgroup width=\"606\" span=\"1\">\r\n");
951 out.write("<tr>\r\n<th>Customize Visualization");
952 out.write("<tr>\r\n<td><LABEL FOR=\"H1\"><INPUT align=\"left\" TYPE=\"checkbox\" ID=\"H1\">Hide Empty Metadata Elements</LABEL>");
953 out.write("<tr>\r\n<td><LABEL FOR=\"H2\"><INPUT align=\"left\" TYPE=\"checkbox\" ID=\"H2\">Hide Completed Metadata Elements</LABEL>");
954 out.write("<tr>\r\n<td><LABEL FOR=\"H3\"><INPUT align=\"left\" TYPE=\"checkbox\" ID=\"H3\">Hide Documents with Empty Metadata Elements</LABEL>");
955 out.write("<tr>\r\n<td><LABEL FOR=\"H4\"><INPUT align=\"left\" TYPE=\"checkbox\" ID=\"H4\">Hide Documents with Completed Metadata Elements</LABEL>");
956 out.write("<tr>\r\n<th>Metadata: ");
957
958 if(MetadataData.length==1){
959 if(MetadataData[0][0].equalsIgnoreCase("dublin")){
960 out.write("<tr>\r\n<td><LABEL FOR=\"C1\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C1\" checked> Dublin Core </LABEL>");
961 }
962 else {
963 out.write("<tr>\r\n<td><LABEL FOR=\"C4\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C4\" checked>"+MetadataData[0][0]+"</LABEL>");
964 }
965 }
966 else{
967 for(int a = 0; a<MetadataData.length; a++){
968 if(a==0){
969 if(MetadataData[a][0].equalsIgnoreCase("dublin")){
970 out.write("<tr>\r\n<td><LABEL FOR=\"C1\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C1\" checked>Dublin Core</LABEL>");
971 }
972 else {
973 out.write("<tr>\r\n<td><LABEL FOR=\"C4\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C4\" checked>"+MetadataData[a][0]+"</LABEL>");
974 }
975 }
976 else{
977 if(MetadataData[a][0].equalsIgnoreCase("dublin")){
978 out.write("<tr>\r\n<td><LABEL FOR=\"C1\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C1\">"+MetadataData[a][0]+"</LABEL>");
979 }
980 else{
981 out.write("<tr>\r\n<td><LABEL FOR=\"C4\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C4\">"+MetadataData[a][0]+"</LABEL>");
982 }
983 }
984 }
985 }
986
987 out.write("<tr>\r\n<th>Order By Completeness : ");
988 out.write("<tr>\r\n<td><LABEL FOR=\"R1\"><INPUT align=\"left\" TYPE=\"radio\" name = \"order\" ID=\"R1\">Best Case to Worst Case</LABEL>");
989 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>");
990 out.write("</table><p align=\"center\"><INPUT TYPE=\"button\" VALUE=\"Show Visualization\" onClick=\"checkStatus()\"> </FORM>");
991 out.write("<p align=\"center\">"+str+"</p>");
992 out.write("</body></html>");
993
994 out.close();
995 fstream.close();
996 }catch (Exception e){
997 System.err.println("Error: " + e.getMessage());
998 }
999 printWriter.write("<!-- generating all possible page-->");
1000 printWriter.flush();
1001 }
1002
1003 public void WriteMetadataSetDetailHTML(MetadataSet mdset){
1004
1005 String fileName = mdset.getName();
1006
1007 try{
1008 printWriter.write("<!-- generating all possible page-->");
1009 printWriter.flush();
1010
1011 String str = sdf.format(cl.getTime());
1012 String collectionFolder = mds.getCollectionName();
1013
1014 FileWriter fstream = new FileWriter(destination+fileName+".html");
1015 BufferedWriter out = new BufferedWriter(fstream);
1016
1017 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
1018 out.write("<html>\r\n");
1019 out.write("<head>\r\n<title>Metadata Detail</title>\r\n");
1020 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
1021 out.write("<style type=\"text/css\">\r\n");
1022 out.write("td{height:20;text-align:right;}\r\n");
1023 out.write("th{height:20;text-align:left;}\r\n");
1024 out.write("body{font-family: Arial;}\r\n");
1025 out.write("</style>\r\n");
1026 out.write( "</head>\r\n");
1027 out.write ("<body>\r\n");
1028 out.write ("<td> <div style=\"float:left;\"> <a href=\" Overall.html \">Summary</a></div>");
1029 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>");
1030 if(fileName.equalsIgnoreCase("dublin")){
1031 out.write ("<h3 align=\"center\">Metadata Detail: "+metadataSet1+"</h3>\r\n");
1032 }
1033 else if (fileName.equalsIgnoreCase("extracted")){
1034 out.write ("<h3 align=\"center\">Metadata Detail: "+metadataSet2+"</h3>\r\n");
1035 }
1036 else{
1037 out.write ("<h3 align=\"center\">Metadata Detail: "+fileName+"</h3>\r\n");
1038 }
1039
1040 out.write ("<table border=\"1\" align=\"center\">\r\n");
1041 out.write ("<colgroup width=\"200\" span=\"2\">\r\n");
1042 out.write ("<tr>\r\n");
1043 out.write ("<th>Elements:\r\n");
1044 out.write ("<td><b>Completeness</b>\r\n");
1045 out.write ("</tr>\r\n");
1046
1047 ArrayList elementList = mdset.getIndexsList();
1048 HashMap hp = new HashMap();
1049 int num = elementList.size();
1050
1051 for(int i = 0; i<num; i++){
1052 String elementName = (String)elementList.get(i);
1053 Double elementValue = new Double(dm.Mean(elementName));
1054 hp.put(elementName, elementValue);
1055 }
1056
1057 elementList = new ArrayList();
1058 elementList = dm.sortMap(hp);
1059 num = elementList.size();
1060
1061 for(int i = 0; i<num; i++){
1062 Map.Entry entry = (Map.Entry) elementList.get(i);
1063 String elementName = (String) entry.getKey();
1064 out.write ("<tr>\r\n");
1065 out.write ("<th><a href=\""+ elementName +".html\">"+elementName+"</a>\r\n");
1066 out.write ("<td>"+ dm.Mean(elementName)+"%\r\n");
1067 out.write ("</tr>\r\n");
1068 }
1069
1070 out.write("</table>\r\n");
1071 out.write ("<p align=\"center\"> <a href=\" Overall.html \">Summary</a></p>");
1072 out.write("<p align=\"center\">"+str+"</p>");
1073 out.write ("</body></html>\r\n");
1074
1075 out.close();
1076 fstream.close();
1077 }catch (Exception e){
1078 e.printStackTrace(printWriter);
1079 }
1080
1081 printWriter.write("<!-- generating all possible page-->");
1082 printWriter.flush();
1083
1084 }
1085
1086 public void generateMetadataElementDetailPage(MetadataSet mds,PrintWriter out){
1087
1088 ArrayList nameList = mds.getIndexsList();
1089
1090 for(int i = 0; i<nameList.size();i++){
1091 WriteMetadataElementDetailHTML((String)nameList.get(i),mds.getName(),out);
1092 }
1093 }
1094
1095 public void WriteMetadataElementDetailHTML(String name, String linkName,PrintWriter outx){
1096
1097 String fileName = name;
1098
1099 try{
1100 printWriter.write("<!-- generating all possible page-->");
1101 printWriter.flush();
1102
1103 String collectionFolder = mds.getCollectionName();
1104
1105 FileWriter fstream = new FileWriter(destination+fileName+".html");
1106 BufferedWriter out = new BufferedWriter(fstream);
1107
1108 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
1109 out.write ("<html>\r\n");
1110 out.write("<head>\r\n<title> "+ name +" </title>\r\n");
1111 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
1112 out.write("<style type=\"text/css\">\r\n");
1113 out.write("td{height:20; text-align:left;}\r\n");
1114 out.write(".tam{height:20; text-align:center}\r\n");
1115 out.write("th{height:20; text-align:left;}\r\n");
1116 out.write("body{font-family: Arial;}\r\n");
1117 out.write("</style>\r\n");
1118 out.write("<script type=\"text/javascript\" src=\"http://"+hostName+":"+port_number+"/greenstone3/mat/script/status3.js\"></script>");
1119 out.write("</head>\r\n");
1120 out.write("<body>\r\n");
1121 out.write ("<td> <div style=\"float:left;\"> <a href=\" Overall.html \">Summary</a></div>");
1122 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>");
1123 out.write("&raquo;");
1124
1125 if(linkName.equalsIgnoreCase("dublin")){
1126 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+metadataSet1+")</a>");
1127 }
1128 else if (linkName.equalsIgnoreCase("extracted")){
1129 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+metadataSet2+")</a>");
1130 }
1131 else{
1132 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+linkName+")</a>");
1133 }
1134
1135 if(name.startsWith("dc.")){
1136 int dot = name.lastIndexOf('.');
1137 dot++;
1138 String nameReplace = name.substring(dot);
1139
1140 if(url.containsKey(nameReplace)){
1141 out.write ("<h3 align=\"center\">Metadata Element Detail:<a href=\""+ url.get(nameReplace) +"\">"+ name +"</a></h3>\r\n");
1142 }
1143 else{
1144 out.write ("<h3 align=\"center\">Metadata Element Detail: "+ name +"</a></h3>\r\n");
1145 }
1146 }
1147 else{
1148 out.write ("<h3 align=\"center\">Metadata Element Detail: "+ name +"</h3>\r\n");
1149 }
1150
1151 out.write ("<table border=\"1\" align=\"center\" width=1000>\r\n");
1152 out.write ("<colgroup width=\"500\" span=\"2\">\r\n");
1153 out.write ("<tr>\r\n");
1154 out.write ("<td><b>Total Number of Records</b>\r\n");
1155 out.write ("<td class=\"tam\">"+mds.getDocNum()+"\r\n");
1156 out.write ("</tr>\r\n");
1157 out.write ("<tr>\r\n");
1158 out.write ("<td><b>Unique Values</b>\r\n");
1159 out.write ("<td class=\"tam\">"+dm.getDistinctNumber(name)+"\r\n");
1160 out.write ("</tr>\r\n");
1161 out.write ("<tr>\r\n");
1162 out.write ("<td><b>Total times element used</b>\r\n");
1163 out.write ("<td class=\"tam\">"+dm.getFrequency(name) +"\r\n");
1164 out.write ("</tr>\r\n");
1165 out.write ("<tr>\r\n");
1166 out.write ("<td><b>No. of records containing element</b>\r\n");
1167 out.write ("<td class=\"tam\"> "+dm.getDocumentUsedElement(name)+"\r\n");
1168 out.write ("</tr>\r\n");
1169 out.write ("<tr>\r\n");
1170
1171 double percentage = dm.Mean(name);
1172
1173 out.write ("<th title = \"The completeness means the arithmetic average \"> <div style=\"float:left;\"> Completeness</div>");
1174 out.write ("<td class=\"tam\"> "+dm.Mean(name) +"%\r\n");
1175 out.write ("</tr>\r\n");
1176 out.write ("<tr>\r\n");
1177 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");
1178 out.write ("<td class=\"tam\"> "+dm.getMinRange(name) +"\r\n");
1179 out.write ("</tr>\r\n");
1180
1181 out.write ("<tr>\r\n");
1182 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");
1183 out.write ("<td class=\"tam\"> "+dm.getMaxRange(name) +"\r\n");
1184 out.write ("</tr>\r\n");
1185
1186 out.write ("<tr>\r\n");
1187 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");
1188 out.write ("<td class=\"tam\"> "+dm.Average(name) +"\r\n");
1189 out.write ("</tr>\r\n");
1190
1191 out.write ("<tr>\r\n");
1192 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");
1193 out.write ("<td class=\"tam\"> "+dm.getMode(name) +"\r\n");
1194 out.write ("</tr>\r\n");
1195
1196 out.write ("<tr>\r\n");
1197 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");
1198 out.write ("<td class=\"tam\">"+dm.ModeFrequency(name) +"%\r\n");
1199 out.write ("</tr>\r\n");
1200
1201 if(mds.getOaiPrefix().equalsIgnoreCase("oai_dc")){
1202
1203 HashMap suggestionMap = generateMetadataElementSortList(fileName,"ASCII",linkName,outx);
1204
1205 if(suggestionMap.size()>0){
1206 boolean status = compareElement(fileName,collectionFolder,suggestionMap,linkName);
1207 if(status){
1208 out.write("<tr><td class=\"tam\"><a href =\""+fileName+"_Suggestion.html\">View Potential Duplicate List</a>");
1209 if(percentage<100){
1210 out.write ("<td class=\"tam\"><a href =\""+fileName+"_IncompletedList.html\">Records missing "+fileName+"</a></td>");
1211 createIncompletedList(fileName,linkName,collectionFolder,outx);
1212 }
1213 else{
1214 out.write("<td class=\"tam\">No Records Missing "+ fileName);
1215 }
1216 }
1217 else{
1218 out.write("<tr><td class=\"tam\">No Potential Duplicates");
1219 if(percentage<100){
1220 out.write ("<td class=\"tam\"><a href =\""+fileName+"_IncompletedList.html\">Records missing "+fileName+"</a></td>");
1221 createIncompletedList(fileName,linkName,collectionFolder,outx);
1222 }
1223 else{
1224 out.write("<td class=\"tam\">No Records Missing "+ fileName);
1225 }
1226 }
1227 }
1228 else{
1229 out.write("<tr><td class=\"tam\">No Potential Duplicates");
1230 if(percentage<100){
1231 out.write ("<td class=\"tam\"><a href =\""+fileName+"_IncompletedList.html\">Records missing "+fileName+"</a></td>");
1232 createIncompletedList(fileName,linkName,collectionFolder,outx);
1233 }
1234 else{
1235 out.write("<td class=\"tam\">No Records Missing "+ fileName);
1236 }
1237 }
1238 generateMetadataElementSortList(fileName,"Frequency-based",linkName,outx);
1239 }
1240 else{
1241 HashMap suggestionMap = generateMetadataElementSortListNoIdentifier(fileName,"ASCII",linkName,outx);
1242 generateMetadataElementSortListNoIdentifier(fileName,"Frequency-based",linkName,outx);
1243 }
1244
1245 out.write ("<tr>\r\n");
1246 out.write ("<td class=\"tam\"><a href =\""+fileName+"_Frequency-based.html\">"+"View Full Frequency Sorted list</a>" +
1247 " <td class=\"tam\"><a href =\""+fileName+"_ASCII.html\">"+"View Full ASCII Sorted list</a></td>");
1248 out.write ("</tr>\r\n");
1249 out.write ("</table><br>\r\n");
1250 out.write ("<table border=\"1\" align=\"center\" width=1000>\r\n");
1251 out.write ("<colgroup width=\"500\" span=\"2\">\r\n");
1252 out.write ("<tr>\r\n");
1253 out.write ("<th>ASCII-Based\r\n");
1254 out.write ("<th>First Five\r\n");
1255 out.write ("</tr>\r\n");
1256
1257 String[] temp = dm.getSortList(name,"ASCII");
1258 String[] temp2 = {"&nbsp;","&nbsp;","&nbsp;","&nbsp;","&nbsp;"};
1259 int length = 0;
1260 if(temp.length>=5){
1261 length=5;
1262 }
1263 else if(temp.length<5){
1264 length = temp.length;
1265 }
1266
1267 for(int i =0; i<length; i++){
1268 temp2[i] = temp[i];
1269 }
1270
1271 int x = temp2.length;
1272 for(int a = 0; a<x; a++){
1273 out.write ("<tr>\r\n");
1274 if(!temp2[a].equals("&nbsp;")){
1275 out.write ("<th>"+(a+1)+"\r\n");
1276 }
1277 else{
1278 out.write ("<th>&nbsp;\r\n");
1279 }
1280 if(temp2[a].startsWith("http://")){
1281 if(temp2[a].length()>60){
1282 out.write ("<td><a href = \""+temp2[a]+"\">"+temp2[a].substring(0, 60)+"...</a>");
1283 }
1284 else {
1285 out.write ("<td><a href = \""+temp2[a]+"\">"+temp2[a]+"</a>");
1286 }
1287 }
1288 else{
1289 char singleChar = temp2[a].charAt(0);
1290 if(temp2[a].length()>61 ){
1291 if(temp2[a].startsWith(" ") && ((int)singleChar!=65279)){
1292 temp2[a] = temp2[a].substring(1);
1293 out.write ("<td>"+spaceLeft+space+spaceRight+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#topFive\"> ... </a>\r\n");
1294 }
1295 else if (((int)singleChar==65279)){
1296 temp2[a] = temp2[a].substring(1);
1297 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#topFive\"> ... </a>\r\n");
1298 }
1299 else{
1300 out.write ("<td>"+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#topFive\"> ... </a>\r\n");
1301 }
1302 }
1303 else {
1304 if(temp2[a].startsWith(" ")&& ((int)singleChar!=65279)){
1305 temp2[a] = temp2[a].substring(1);
1306 out.write ("<td>"+spaceLeft+space+spaceRight+temp2[a]+"\r\n");
1307 }
1308 else if (((int)singleChar==65279)){
1309 temp2[a] = temp2[a].substring(1);
1310 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp2[a]+"\r\n");
1311 }
1312 else{
1313 out.write ("<td>"+temp2[a]+"\r\n");
1314 }
1315 }
1316 }
1317 out.write ("</tr>\r\n");
1318 }
1319
1320 String[] temp3 ={"&nbsp;","&nbsp;","&nbsp;","&nbsp;","&nbsp;"};
1321 length = 0;
1322 int start = temp.length;
1323
1324 if(temp.length>=5){
1325 length= 5;
1326 }
1327 else if(temp.length<5){
1328 length = temp.length;
1329 }
1330
1331 for(int i = length; i>0; i--){
1332 temp3[i-1] = temp[start-1];
1333 start--;
1334 }
1335
1336 out.write ("<tr>\r\n");
1337 out.write ("<th>......\r\n");
1338 out.write ("<th>Last Five\r\n");
1339 out.write ("</tr>\r\n");
1340
1341 int counter = temp.length;
1342 start = temp.length;
1343 x = temp3.length;
1344
1345 for(int a = 0; a<x; a++){
1346 out.write ("<tr>\r\n");
1347
1348 if(!temp3[a].equals("&nbsp;")){
1349 out.write ("<th>"+(start-length+1+a));
1350 }
1351 else{
1352 out.write ("<th>&nbsp;");
1353 }
1354 if(temp3[a].startsWith("http://")){
1355 if(temp3[a].length()>60){
1356 out.write ("<td><a href = \""+temp3[a]+"\">"+temp3[a].substring(0, 60)+"...</a>");
1357 }
1358 else {
1359 out.write ("<td><a href = \""+temp3[a]+"\">"+temp3[a]+"</a>");
1360 }
1361 }
1362 else{
1363 char singleChar = temp3[a].charAt(0);
1364 if(temp3[a].length()>61){
1365 if(temp3[a].startsWith(" ") && (int)singleChar!=65279){
1366 temp3[a] = temp3[a].substring(1);
1367 out.write ("<td>"+spaceLeft+space+spaceRight+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#lastFive\"> ... </a>\r\n");
1368 }
1369 else if((int)singleChar==65279){
1370 temp3[a] = temp3[a].substring(1);
1371 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#lastFive\"> ... </a>\r\n");
1372
1373 }
1374 else{
1375 out.write ("<td>"+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#lastFive\"> ... </a>\r\n");
1376 }
1377 }
1378 else{
1379 if(temp3[a].startsWith(" ")&& (int)singleChar!=65279){
1380 temp3[a] = temp3[a].substring(1);
1381 out.write ("<td>"+spaceLeft+space+spaceRight+temp3[a]+"\r\n");
1382 }
1383 else if((int)singleChar==65279){
1384 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp3[a]+"\r\n");
1385 }
1386 else{
1387 out.write ("<td>"+temp3[a]+"\r\n");
1388 }
1389 }
1390 out.write ("</tr>\r\n");
1391 }
1392 }
1393 out.write ("</table><br>\r\n");
1394 out.write ("<table border=\"1\" align=\"center\" width=1000>\r\n");
1395 out.write ("<colgroup width=\"500\" span=\"2\">\r\n");
1396 out.write ("<tr>\r\n");
1397 out.write ("<th>Frequency-Based:\r\n");
1398 out.write ("<th>First Five\r\n");
1399 out.write ("</tr>\r\n");
1400
1401 HashMap xMap = dm.getDistinctValueMap(name);
1402 temp = dm.getSortList(name,"Frequency-based");
1403 temp2 = new String[] {"&nbsp;","&nbsp;","&nbsp;","&nbsp;","&nbsp;"};
1404 length = 0;
1405
1406 if(temp.length>=5){
1407 length=5;
1408 }
1409 else if(temp.length<5){length = temp.length;}
1410
1411 for(int i =0; i<length; i++){
1412 temp2[i] = temp[i];
1413 }
1414
1415 ArrayList tempList = new ArrayList();
1416
1417 for(int i =0; i<length; i++){
1418 if(((Integer)xMap.get(temp2[i])).intValue()==1){
1419 tempList.add(temp2[i]);
1420 }
1421 }
1422
1423 x = temp2.length;
1424 for(int a = 0; a<x; a++){
1425 out.write ("<tr>\r\n");
1426
1427 if(!temp2[a].equals("&nbsp;")){
1428 char singleChar = temp2[a].charAt(0);
1429 out.write ("<th>"+(a+1)+". (No. of occurrences: "+((Integer)xMap.get(temp2[a])).toString()+")\r\n");
1430 if(temp2[a].startsWith("http://") && (int)singleChar != 65279){
1431 if(temp2[a].length()>61){
1432 out.write ("<td><a href = \""+temp2[a]+"\">"+temp2[a].substring(0, 60)+"...</a>");
1433 }
1434 else {
1435 out.write ("<td><a href = \""+temp2[a]+"\">"+temp2[a]+"</a>");
1436 }
1437 }
1438
1439 else{
1440 if(temp2[a].length()>61){
1441
1442 if(temp2[a].startsWith(" ")){
1443 temp2[a] = temp2[a].substring(1);
1444 out.write ("<td>"+spaceLeft+space+spaceRight+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#topFive\"> ... </a>\r\n");
1445 }
1446
1447 else if((int)singleChar == 65279){
1448 temp2[a] = temp2[a].substring(1);
1449 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#topFive\"> ... </a>\r\n");
1450 }
1451 else{
1452 out.write ("<td>"+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#topFive\"> ... </a>\r\n");
1453 }
1454
1455 }
1456 else{
1457 if(temp2[a].startsWith(" ")){
1458 temp2[a] = temp2[a].substring(1);
1459 out.write ("<td>"+spaceLeft+space+spaceRight+temp2[a]+"\r\n");
1460 }
1461 else if((int)singleChar == 65279){
1462 temp2[a] = temp2[a].substring(1);
1463 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp2[a]+"\r\n");
1464 }
1465 else{
1466 out.write ("<td>"+temp2[a]+"\r\n");
1467 }
1468 }
1469 }
1470 }
1471 else{
1472 out.write ("<th>&nbsp;\r\n");
1473 out.write ("<td>\r\n");
1474 }
1475 out.write ("</tr>\r\n");
1476 }
1477
1478 temp3 = new String[]{"&nbsp;","&nbsp;","&nbsp;","&nbsp;","&nbsp;"};
1479 length = 0;
1480 start = temp.length;
1481 if(temp.length>=5){length= 5;}
1482 else if(temp.length<5){length = temp.length;}
1483
1484 for(int i = length; i>0; i--){
1485 temp3[i-1] = temp[start-1];
1486 start--;
1487 }
1488
1489 out.write ("<tr>\r\n");
1490 out.write ("<th>......\r\n");
1491 out.write ("<th>Last Five\r\n");
1492 out.write ("</tr>\r\n");
1493
1494 x = temp3.length;
1495 start = temp.length;
1496 for(int a = 0; a<x; a++){
1497 out.write ("<tr>\r\n");
1498 if(!temp3[a].equals("&nbsp;")){
1499
1500
1501 out.write ("<th>"+(start-length+1+a)+". (No. of occurrences: "+((Integer)xMap.get(temp3[a])).toString()+")\r\n");
1502
1503 char singleChar = temp3[a].charAt(0);
1504 if(temp3[a].startsWith("http://")){
1505 if(temp3[a].length()>60){
1506 out.write ("<td><a href = \""+temp3[a]+"\">"+temp3[a].substring(0, 60)+"...</a>");
1507 }
1508 else {
1509 out.write ("<td><a href = \""+temp3[a]+"\">"+temp3[a]+"</a>");
1510 }
1511 }
1512 else{
1513 if(temp3[a].length()>61){
1514
1515 if(temp3[a].startsWith(" ")){
1516 temp3[a] = temp3[a].substring(1);
1517 out.write ("<td>"+spaceLeft+space+spaceRight+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#lastFive\"> ... </a>\r\n");
1518 }
1519 else if((int)singleChar == 65279){
1520 temp3[a] = temp3[a].substring(1);
1521 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#lastFive\"> ... </a>\r\n");
1522
1523 }
1524 else{
1525 out.write ("<td>"+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#lastFive\"> ... </a>\r\n");
1526 }
1527
1528 }
1529 else{
1530 if(temp3[a].startsWith(" ")){
1531 out.write ("<td>"+spaceLeft+space+spaceRight+temp3[a]+"\r\n");
1532 }
1533
1534 else if((int)singleChar == 65279){
1535 temp3[a] = temp3[a].substring(1);
1536 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp3[a]+"\r\n");
1537
1538 }
1539
1540 else{
1541 out.write ("<td>"+temp3[a]+"\r\n");
1542 }
1543 }
1544 }
1545
1546 }
1547 else{
1548 //out.write ("<th>"+(a+1)+"\r\n");
1549 out.write ("<th>&nbsp;\r\n");
1550 out.write ("<th>\r\n");
1551 }
1552 out.write ("</tr>\r\n");
1553 }
1554
1555
1556 out.write ("</table>\r\n");
1557
1558 ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
1559 out.write ("<p> <a href=\" Overall.html \">Summary</a>");
1560 out.write ("&raquo;");
1561
1562 if(linkName.equalsIgnoreCase("dublin")){
1563 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+metadataSet1+")</a>");
1564
1565 }
1566 else if (linkName.equalsIgnoreCase("extracted")){
1567 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+metadataSet2+")</a>");
1568
1569 }
1570 else{
1571 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+linkName+")</a>");
1572 }
1573 //out.write(str+"</p>");
1574 out.write ("</body></html>\r\n");
1575
1576 //Close the output stream
1577 out.close();
1578 fstream.close();
1579 }catch (Exception e){
1580
1581 e.printStackTrace(printWriter);
1582 }
1583
1584 printWriter.write("<!-- generating all possible page-->");
1585 printWriter.flush();
1586 //System.out.println("detail ends");
1587
1588 }
1589
1590
1591 private HashMap generateMetadataElementSortList(String title, String sort, String metadataSetName,PrintWriter outx){
1592
1593 String fileName = title+"_"+sort;
1594 String collectionFolder = mds.getCollectionName();
1595
1596 String IDENTIFIER = linkIdentifier;
1597 SearchLink sl = new SearchLink(mds.StatsDirectory);
1598 HashMap suggestionMap = new HashMap();
1599
1600 HashMap valueMap = sl.createValueMap(title,collectionFolder);
1601
1602 HashMap linkMap = sl.createLinkMap(IDENTIFIER,collectionFolder);
1603 HashMap internalIDMap = dm.getInternalIdentifier(title);
1604 boolean status = false;
1605
1606 try{
1607 printWriter.write("<!-- generating all possible page-->");
1608 printWriter.flush();
1609
1610 FileWriter fstream = new FileWriter(destination+fileName+".html");
1611 BufferedWriter out = new BufferedWriter(fstream);
1612
1613 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
1614 out.write ("<html>\r\n");
1615 out.write("<head>\r\n<title>Metadata Element Sort List</title>\r\n");
1616 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
1617 out.write("<script type=\"text/javascript\" src=\"../script/getInfomation.js\"></script>\r\n");
1618 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/yahoo/yahoo-min.js\"></script>\r\n");
1619 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/event/event-min.js\"></script>\r\n");
1620 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/connection/connection-min.js\"></script>\r\n");
1621
1622 out.write("<style type=\"text/css\">\r\n");
1623 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
1624 out.write(".tam{height:20; text-align:center}\r\n");
1625 out.write("th{height:20; text-align:center;}\r\n");
1626 out.write("body{font-family: Arial;}\r\n");
1627 out.write("</style>\r\n");
1628 out.write("</head><body id=\""+collectionFolder+"\">\r\n");
1629
1630 out.write("<td><div style=\"float:left;\"><a href=\" Overall.html \">Summary</a>");
1631 out.write("&raquo;");
1632 ////////////////////////////////////////////////////////////////////////////////////
1633 if(metadataSetName.equalsIgnoreCase("dublin")){
1634 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
1635 }
1636 else if (metadataSetName.equalsIgnoreCase("extracted")){
1637 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
1638 }
1639 else{
1640 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
1641 }
1642 out.write("&raquo;");
1643 out.write("<a href=\""+title+".html\">"+title+"</a></div>");
1644 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>");
1645
1646
1647 String[] list = dm.getSortList(title, sort);
1648
1649
1650 if(list.length>=1){
1651
1652
1653 if(sort.equals("ASCII")){
1654
1655 out.write("<table border=\"1\" align=\"center\" width=\"800\">\r\n");
1656 out.write ("<h2 align=\"center\">"+ title+"</h2>\r\n");
1657 out.write ("<th class=\"tam\">ASCII Sort\r\n");
1658 out.write ("<th class=\"tam\">Element Values\r\n");
1659 out.write ("<th class=\"tam\">Source Documents\r\n");
1660 out.write ("<th class=\"tam\">Internal Link\r\n");
1661
1662 out.write ("<a name='topFive'>\r\n");
1663
1664 int counter = 0;
1665
1666 for(int i = 0; i<list.length; i++){
1667 if(list.length<=5 && i==0){
1668 out.write ("<a name='lastFive'>\r\n");
1669 }
1670 else if((list.length>5) && (list.length-5==i)){
1671 out.write ("<a name='lastFive'>\r\n");
1672 }
1673
1674
1675 InternalLink il= (InternalLink)internalIDMap.get(list[i]);
1676 ArrayList alist2 = il.retrieveList();
1677 String id = (String)alist2.get(0);
1678 id = id.substring(4);
1679 out.write("<tr id=\""+id+"\" >\r\n");
1680
1681
1682 if(list[i].length()>=201){
1683 if(list[i].startsWith("http://")){
1684
1685 if(title.equals(IDENTIFIER)){
1686 if(list[i].length()>=100){
1687 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1688
1689 }
1690 else{
1691 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1692 }
1693 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
1694 }
1695 else{
1696 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1697
1698 if(list[i].length()>=100){
1699 String url = (String)alist.get(0);
1700 if(alist.size()==1){
1701 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1702 if(url.startsWith("http://")){
1703 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1704 }else{
1705 out.write("<td>Source\r\n");
1706 }
1707 }
1708 else if(alist.size()>1){
1709 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1710 if(url.startsWith("http://")){
1711 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1712 }else{
1713 out.write("<td>Source\r\n");
1714 }
1715
1716 }
1717 else{
1718 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1719 out.write("<td>No Source Available\r\n");
1720 }
1721 suggestionMap.put(list[i], url);
1722 }
1723 else{
1724
1725 String url = (String)alist.get(0);
1726 if(alist.size()==1){
1727 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1728 //out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1729 if(url.startsWith("http://")){
1730 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1731 }else{
1732 out.write("<td>Source\r\n");
1733 }
1734 }
1735 else if(alist.size()>1){
1736 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1737 //out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1738 if(url.startsWith("http://")){
1739 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1740 }else{
1741 out.write("<td>Source\r\n");
1742 }
1743 }
1744 else{
1745 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1746 out.write("<td>No Source Available\r\n");
1747 }
1748 suggestionMap.put(list[i], url);
1749 }
1750 }
1751
1752
1753
1754 }
1755 else{
1756
1757 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1758
1759 //////////////////////////////////////
1760
1761 if(list[i].startsWith(" ")){
1762 String elements = list[i];
1763 list[i] = list[i].substring(1);
1764
1765 if(alist.size()==1){
1766 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i].substring(0, 200)+"...<td>"+
1767 "<a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1768 suggestionMap.put(elements, (String)alist.get(0));
1769 }
1770 else if(alist.size()>1){
1771 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i].substring(0, 200)+"...<td>"+
1772 "<a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1773 suggestionMap.put(elements, (String)alist.get(0));
1774 }
1775 else{
1776 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i].substring(0, 200)+"...<td>No Source Available \r\n");
1777 suggestionMap.put(elements, "No Source Available");
1778 }
1779 }
1780 else{
1781
1782 if(alist.size()==1){
1783 out.write("<td>"+(counter+1)+"<td>"+list[i].substring(0, 200)+"...<td>" +
1784 "<a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1785 suggestionMap.put(list[i], (String)alist.get(0));
1786 }
1787 else if(alist.size()>1){
1788 out.write("<td>"+(counter+1)+"<td>"+list[i].substring(0, 200)+"...<td>" +
1789 "<a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1790 suggestionMap.put(list[i], (String)alist.get(0));
1791 }
1792 else{
1793 out.write("<td>"+(counter+1)+"<td>"+list[i].substring(0, 200)+"... <td>No Source Available\r\n");
1794 suggestionMap.put(list[i],"No Source Available");
1795 }
1796 }
1797 }
1798
1799 }
1800 else{
1801 if(list[i].startsWith("http://")){
1802
1803 if(title.equals(IDENTIFIER)){
1804 if(list[i].length()>=100){
1805 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1806
1807 }
1808 else{
1809 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1810 }
1811 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
1812 }
1813 else{
1814 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1815
1816 if(list[i].length()>=100){
1817 if(alist.size()==1){
1818 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1819 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1820 }
1821 else 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{
1826 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1827 out.write("<td>No Source Available\r\n");
1828 }
1829 }
1830 else{
1831 if(alist.size()==1){
1832 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1833 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1834 suggestionMap.put(list[i], (String)alist.get(0));
1835 }
1836 else if(alist.size()>1){
1837 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1838 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1839 suggestionMap.put(list[i], (String)alist.get(0));
1840 }
1841 else{
1842 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1843 out.write("<td>No Source Available\r\n");
1844 suggestionMap.put(list[i],"No Source Available");
1845 }
1846 }
1847 }
1848
1849 }
1850
1851
1852 else{
1853
1854 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1855
1856 if(list[i].startsWith(" ")){
1857 String elements = list[i];
1858 list[i] = list[i].substring(1);
1859
1860 if(alist.size()==1){
1861 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i]+
1862 "<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1863 suggestionMap.put(list[i], (String)alist.get(0));
1864 }
1865 else if(alist.size()>1){
1866 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i]+
1867 "<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1868 suggestionMap.put(list[i], (String)alist.get(0));
1869 }
1870 else{
1871 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i]+"<td>No Source Available\r\n");
1872 suggestionMap.put(elements,"No Source Available");
1873 }
1874
1875 }
1876 else{
1877 if(alist.size()==1){
1878 out.write("<td>"+(counter+1)+"<td>"+list[i]+"<td><a href=\""+alist.get(0)+"\">Source</a>\r\n");
1879 suggestionMap.put(list[i],(String)alist.get(0));
1880 }
1881 else 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{
1886 out.write("<td>"+(counter+1)+"<td>"+list[i]+"<td>No Source Available\r\n");
1887 suggestionMap.put(list[i],"No Source Available");
1888 }
1889 //out.write("<td>"+(counter+1)+"<td>"+list[i]+"\r\n");
1890 }
1891 }
1892 }
1893 counter++;
1894 out.write("<td onclick=\"GD(this)\">View");
1895 }
1896 if(suggestionMap.size()>1 && !title.equals(IDENTIFIER)){
1897 status = true;
1898 }
1899 }
1900 else{
1901
1902
1903 out.write("<table border=\"1\" align=\"center\" width=\"800\">\r\n");
1904 out.write ("<h2 align=\"center\">"+ title+"</h2>\r\n");
1905 out.write ("<th class=\"tam\">&nbsp;\r\n");
1906 out.write ("<th class=\"tam\">Frequency\r\n");
1907 out.write ("<th class=\"tam\">Element Values\r\n");
1908 out.write ("<th class=\"tam\">Source Documents\r\n");
1909 out.write ("<th class=\"tam\">Internal Link\r\n");
1910 out.write ("<a name='topFive'>\r\n");
1911
1912 HashMap xMap = dm.getDistinctValueMap(title);
1913 int counter = 0;
1914 for(int i = 0; i<list.length; i++){
1915 if(list.length<=5 && i==0){
1916 out.write ("<a name='lastFive'>\r\n");
1917 }
1918 else if((list.length>5) && (list.length-5==i)){
1919 out.write ("<a name='lastFive'>\r\n");
1920 }
1921
1922 InternalLink il= (InternalLink)internalIDMap.get(list[i]);
1923 ArrayList alist2 = il.retrieveList();
1924 String id = (String)alist2.get(0);
1925 id = id.substring(4);
1926 out.write("<tr id=\""+id+"\" >\r\n");
1927
1928 out.write("<td>"+(counter+1)+"<th> "+((Integer)xMap.get(list[i])).toString()+"\r\n");
1929
1930 if(list[i].length()>=201){
1931
1932 if(list[i].startsWith("http://")){
1933 if(title.equals(IDENTIFIER)){
1934 if(list[i].length()>=100){
1935 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1936 }
1937 else{
1938 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1939 }
1940
1941 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
1942 }
1943 else{
1944 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1945 if(list[i].length()>=100){
1946 if(alist.size()==1){
1947 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1948 out.write("<td><a href=\""+alist.get(0)+"\">Source</a>\r\n");
1949 }
1950 else 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{
1955 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1956 out.write("<td>No Source Available\r\n");
1957 }
1958
1959 }
1960 else{
1961 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1962 }
1963 }
1964 }
1965
1966
1967
1968
1969 else{
1970 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1971
1972 if(list[i].startsWith(" ")){
1973 list[i]=list[i].substring(1);
1974 if(alist.size()==1){
1975 out.write("<td>"+spaceLeft+space+spaceRight+list[i].substring(0,200)+"...<td>"+
1976 "<a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1977 }
1978 else 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{
1983 out.write("<td>"+spaceLeft+space+spaceRight+list[i].substring(0,200)+"...<td>No Source Available\r\n");
1984 }
1985 }
1986 else{
1987
1988 if(alist.size()==1){
1989 out.write("<td>"+list[i].substring(0,200)+"...<td>" +
1990 "<a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1991 }
1992 else 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{
1997 out.write("<td>"+list[i].substring(0,200)+"...<td>No Source Available\r\n");
1998 }
1999 }
2000 }
2001 }
2002 else{
2003 if(list[i].startsWith("http://")){
2004 if(title.equals(IDENTIFIER)){
2005 if(list[i].length()>=100){
2006 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2007 }
2008 else{
2009 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2010 }
2011 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2012 }
2013 else{
2014 //ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],"dc.Identifier", collectionFolder,valueMap,linkMap);
2015 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
2016
2017 if(list[i].length()>=100){
2018 if(alist.size()==1){
2019 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2020 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
2021 }
2022 else 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{
2027 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2028 out.write("<td>No Source Available\r\n");
2029 }
2030 }
2031 else{
2032 if(alist.size()==1){
2033 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2034 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
2035 }
2036 else 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{
2041 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2042 out.write("<td>No Source Available\r\n");
2043 }
2044 }
2045
2046 }
2047
2048 }
2049 else{
2050
2051 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
2052
2053 if(list[i].startsWith(" ")){
2054 list[i]=list[i].substring(1);
2055 if(alist.size()==1){
2056 out.write("<td>"+spaceLeft+space+spaceRight+list[i]+
2057 "<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
2058 }
2059 else 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{
2064 out.write("<td>"+spaceLeft+space+spaceRight+list[i]+"<td>No Source Available\r\n");
2065 }
2066 }
2067 else{
2068 if(alist.size()==1){
2069 out.write("<td>"+list[i]+"<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
2070 }
2071 else if(alist.size()>1){
2072 out.write("<td>"+list[i]+"<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
2073 }
2074 else{
2075 out.write("<td>"+list[i]+"<td>No Source Available\r\n");
2076 }
2077 }
2078 }
2079 }
2080 counter++;
2081 out.write("<td onclick=\"GD(this)\">View");
2082 }
2083 }
2084 out.write("</table>\r\n");
2085 }
2086 else{
2087 out.write("<p>Sorry! The list is empty</p>");
2088 }
2089 out.write("<tr>\r\n");
2090
2091
2092 out.write("<p><a href=\" Overall.html \">Summary</a>");
2093 out.write("&raquo;");
2094 ////////////////////////////////////////////////////////////////////////////////////
2095 if(metadataSetName.equalsIgnoreCase("dublin")){
2096 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2097 }
2098 else if (metadataSetName.equalsIgnoreCase("extracted")){
2099 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2100 }
2101 else{
2102 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2103 }
2104 out.write("&raquo;");
2105 out.write("<a href=\""+title+".html\">"+title+"</a></p>");
2106 out.write("</body></html>\r\n");
2107 out.close();
2108 fstream.close();
2109 }catch(Exception e){
2110
2111 e.printStackTrace(outx);
2112 }
2113
2114 if(status){
2115 return suggestionMap;
2116 }
2117 else{
2118 return new HashMap();
2119 }
2120 }
2121
2122
2123 private void setupMetadataLink(String fileName){
2124
2125 try{
2126 DocumentBuilderFactory docBuilderFactory = DocumentBuilderFactory.newInstance();
2127 DocumentBuilder docBuilder = docBuilderFactory.newDocumentBuilder();
2128 Document doc = docBuilder.newDocument();
2129 doc = docBuilder.parse (new File("/research/cc108/greenstone3/web/mat/script/"+fileName+".xml"));
2130 Element rootNode = doc.getDocumentElement();
2131
2132 NodeList listOfName = rootNode.getElementsByTagName("metadataElement");
2133 url = new HashMap();
2134 for(int i = 0; i<listOfName.getLength(); i++){
2135 Node NameNode = listOfName.item(i);
2136 Element docElement = (Element)NameNode;
2137 NodeList valueList = docElement.getElementsByTagName("URL");
2138 Node urlParentNode = valueList.item(0);
2139 String urlText = urlParentNode.getChildNodes().item(0).getNodeValue();
2140
2141 valueList = docElement.getElementsByTagName("name");
2142 Node urlNameParentNode = valueList.item(0);
2143 String urlNameText = urlNameParentNode.getChildNodes().item(0).getNodeValue();
2144
2145 url.put(urlNameText, urlText);
2146
2147 }
2148
2149
2150 }catch(Exception e){
2151 e.printStackTrace(printWriter);
2152 }
2153 }
2154
2155 private void createIncompletedList(String fileName, String metadataSetName, String collectionFolder,PrintWriter outx){
2156
2157 HashMap hp = dm.getIdentifierLink(linkIdentifier);
2158 String[] ids = dm.getDocumentIDList(fileName);
2159
2160 for(int i = 0; i<ids.length; i++){
2161 hp.remove(ids[i]);
2162 }
2163
2164 try{
2165 FileWriter fstream = new FileWriter(destination+fileName+"_IncompletedList.html");
2166 BufferedWriter out = new BufferedWriter(fstream);
2167
2168 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
2169 out.write ("<html>\r\n");
2170 out.write("<head>\r\n<title>Incompleted Document List</title>\r\n");
2171 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
2172 out.write("<style type=\"text/css\">\r\n");
2173 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
2174 out.write(".tam{height:20; text-align:center}\r\n");
2175 out.write("th{height:20; text-align:center;}\r\n");
2176 out.write("body{font-family: Arial;}\r\n");
2177 out.write("</style>\r\n");
2178 out.write("</head><body>\r\n");
2179
2180 out.write("<p><a href=\" Overall.html \">Summary</a>");
2181 out.write("&raquo;");
2182 if(metadataSetName.equalsIgnoreCase("dublin")){
2183 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2184 }
2185 else if (metadataSetName.equalsIgnoreCase("extracted")){
2186 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2187 }
2188 else{
2189 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2190 }
2191 out.write("&raquo;");
2192 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
2193
2194 out.write("<h3 align=center>"+fileName+" does not appear in the following documents</h3>");
2195 int counter = 0;
2196 Set s = hp.keySet();
2197 Iterator i = s.iterator();
2198 out.write("<table border=1 align=center><tr><th>Document ID<th>Source Link\r\n");
2199 while(i.hasNext()){
2200 counter++;
2201 String keys = (String)i.next();
2202 InternalLink il = (InternalLink) hp.get(keys);
2203 ArrayList alist = il.retrieveList();
2204 String url = (String)alist.get(0);
2205 out.write("<tr><td>"+counter+"<td><a href=\""+url+"\">"+url+"</a>\r\n");
2206 }
2207 out.write("</table></body></html>\r\n");
2208
2209
2210 out.write("<p><a href=\" Overall.html \">Summary</a>");
2211 out.write("&raquo;");
2212 if(metadataSetName.equalsIgnoreCase("dublin")){
2213 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2214 }
2215 else if (metadataSetName.equalsIgnoreCase("extracted")){
2216 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2217 }
2218 else{
2219 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2220 }
2221 out.write("&raquo;");
2222 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
2223
2224 out.close();
2225 fstream.close();
2226 }catch(IOException ex){ex.printStackTrace();}
2227
2228
2229 }
2230
2231 public boolean compareElement(String fileName,String collectionFolder,HashMap suggestionMap, String metadataSetName){
2232
2233 boolean status = false;
2234
2235 Set kset = suggestionMap.keySet();
2236 ArrayList arrayList = new ArrayList();
2237 Iterator is = kset.iterator();
2238 while(is.hasNext()){
2239 arrayList.add((String)is.next());
2240 }
2241 HashMap distanceMap = new HashMap();
2242
2243 int totalLength = 0;
2244
2245 int arrayListLength = arrayList.size();
2246 int counter = 0;
2247
2248 printWriter.write("<!-- generating all possible page-->");
2249 printWriter.flush();
2250 double distance ;
2251 try{
2252 for(int i = 0; i<arrayListLength; i++){
2253 String keyword = (String)arrayList.get(i);
2254
2255 int spaceCounterFront1 = 0;
2256 int spaceCounterEnd1 = 0;
2257 String keywordClone = keyword;
2258 String _keywordClone = keyword;
2259 String bkKeyWord = keyword;
2260
2261 keywordClone = removeUnusedCharacter(keywordClone);
2262
2263 while(true){
2264
2265 if(_keywordClone.length()>1){
2266 if(_keywordClone.charAt(0)==' '){
2267 _keywordClone = _keywordClone.substring(1);
2268 spaceCounterFront1++;
2269 }
2270 else{
2271 if(_keywordClone.charAt(_keywordClone.length()-1)==' '){
2272 _keywordClone = _keywordClone.substring(0,_keywordClone.length()-1);
2273 spaceCounterEnd1++;
2274 }
2275 else{
2276 break;
2277 }
2278 }
2279 }
2280 else{
2281 break;
2282 }
2283 }
2284
2285 int keywordLength = keywordClone.length();
2286 totalLength = totalLength + keywordLength;
2287 for(int j = i+1; j<arrayListLength; j++){
2288 String keyword2 = (String)arrayList.get(j);
2289 counter ++;
2290 double pre_cost = 0;
2291 String keywordClone2 = keyword2;
2292 String _keywordClone2 = keyword2;
2293 String bkKeyWord2 = keyword2;
2294 costModel cm = new costModel();
2295 cm = removeUnusedCharacter(keywordClone2,pre_cost);
2296 keywordClone2 = cm.getString();
2297 pre_cost = cm.getCost();
2298
2299 int spaceCounterFront2 = 0;
2300 int spaceCounterEnd2 = 0;
2301
2302 while(true){
2303 if(_keywordClone2.length()>1){
2304 if(_keywordClone2.charAt(0)==' '){
2305 _keywordClone2 = _keywordClone2.substring(1);
2306 spaceCounterFront2++;
2307 }
2308 else{
2309 if(_keywordClone2.charAt(_keywordClone2.length()-1)==' '){
2310 _keywordClone2 = _keywordClone2.substring(0,_keywordClone2.length()-1);
2311 spaceCounterEnd2++;
2312 }
2313 else{
2314 break;
2315 }
2316 }}
2317 else{
2318 break;
2319 }
2320 }
2321
2322 if(counter == 50000){
2323 printWriter.write("<!-- generating all possible page-->");
2324 printWriter.flush();
2325 counter = 0;
2326 }
2327 int keyword2Length = keywordClone2.length();
2328
2329 if(keywordLength>(keyword2Length+2) || (keywordLength+2)<keyword2Length ){}
2330 else{
2331
2332 distance = calculateEditDistance(keywordClone.toLowerCase().toCharArray(),keywordClone2.toLowerCase().toCharArray());
2333
2334 distance = distance + pre_cost;
2335
2336 if(distance<=2){
2337 if(distanceMap.containsKey(_keywordClone)){
2338 InternalLink il = (InternalLink)distanceMap.get(_keywordClone);
2339
2340 String keywordHolder = _keywordClone2;
2341
2342 while(true){
2343 if(keywordHolder.contains(" ")){
2344 keywordHolder = keywordHolder.replaceFirst(" ",spaceLeft+space+spaceRight+" ");
2345 }
2346 else{
2347 break;
2348 }
2349 }
2350
2351 for(int a = 0; a<spaceCounterFront2; a++){
2352 keywordHolder = spaceLeft+space+spaceRight+keywordHolder;
2353 }
2354 for(int a = 0; a<spaceCounterEnd2; a++){
2355 keywordHolder = keywordHolder+spaceLeft+space+spaceRight;
2356 }
2357
2358 InternalLink il2 = new InternalLink();
2359 il2.setValue(keywordHolder);
2360 if(suggestionMap.containsKey(bkKeyWord2)){
2361 il2.increaseElement((String)suggestionMap.get(bkKeyWord2));
2362 //System.out.println("input this url(3):"+(String)alistkw2.get(0));
2363 }
2364 else{
2365 il2.increaseElement("&nbsp;");
2366 }
2367
2368 il.increaseNode(il2);
2369
2370 distanceMap.put(_keywordClone, il);
2371
2372 }
2373 else{
2374 InternalLink il = new InternalLink();
2375 String keywordHolder = _keywordClone;
2376
2377 while(true){
2378 if(keywordHolder.contains(" ")){
2379 keywordHolder = keywordHolder.replaceFirst(" ",spaceLeft+space+spaceRight+" ");
2380 }
2381 else{
2382 break;
2383 }
2384 }
2385
2386 for(int a = 0; a<spaceCounterFront1; a++){
2387 keywordHolder = spaceLeft+space+spaceRight+keywordHolder;
2388 }
2389 for(int a = 0; a<spaceCounterEnd1; a++){
2390 keywordHolder = keywordHolder+spaceLeft+space+spaceRight;
2391 }
2392
2393 InternalLink il2 = new InternalLink();
2394 il2.setValue(keywordHolder);
2395 if(suggestionMap.containsKey(bkKeyWord)){
2396 il2.increaseElement((String)suggestionMap.get(bkKeyWord));
2397
2398 }
2399 else{
2400 il2.increaseElement("&nbsp;");
2401 }
2402
2403 il.increaseNode(il2);
2404
2405
2406 keywordHolder = _keywordClone2;
2407
2408 while(true){
2409 if(keywordHolder.contains(" ")){
2410 keywordHolder = keywordHolder.replaceFirst(" ",spaceLeft+space+spaceRight+" ");
2411 }
2412 else{
2413 break;
2414 }
2415 }
2416 for(int a = 0; a<spaceCounterFront2; a++){
2417 keywordHolder = spaceLeft+space+spaceRight+keywordHolder;
2418 }
2419 for(int a = 0; a<spaceCounterEnd2; a++){
2420 keywordHolder = keywordHolder+spaceLeft+space+spaceRight;
2421 }
2422
2423 InternalLink il3 = new InternalLink();
2424 il3.setValue(keywordHolder);
2425 if(suggestionMap.containsKey(bkKeyWord2)){
2426 il3.increaseElement((String)suggestionMap.get(bkKeyWord2));
2427
2428 }
2429 else{
2430 il3.increaseElement("&nbsp;");
2431 }
2432
2433 il.increaseNode(il3);
2434
2435 distanceMap.put(_keywordClone, il);
2436 }
2437 }
2438 }
2439
2440 }
2441
2442 }
2443
2444
2445 printWriter.write("<!-- generating all possible page-->");
2446 printWriter.flush();
2447
2448 if(distanceMap.size()!=0){
2449 generateHTML(distanceMap,fileName,metadataSetName);
2450 status = true;
2451 }
2452 }catch(Exception ex){ex.printStackTrace();}
2453 return status;
2454
2455 }
2456
2457
2458 private void generateHTML(HashMap distanceMap, String fileName,String metadataSetName){
2459 printWriter.write("<!-- generating all possible page-->");
2460 printWriter.flush();
2461
2462 String collectionFolder = mds.getCollectionName();
2463
2464 try{
2465 FileWriter fstream = new FileWriter(destination+fileName+"_Suggestion.html");
2466 BufferedWriter out = new BufferedWriter(fstream);
2467 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
2468 out.write ("<html>\r\n");
2469 out.write("<head>\r\n<title>Potential Duplicate List</title>\r\n");
2470 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
2471 out.write("<style type=\"text/css\">\r\n");
2472 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
2473 out.write(".tam{height:20; text-align:center}\r\n");
2474 out.write("th{height:20; text-align:left;}\r\n");
2475 out.write("body{font-family: Arial;}\r\n");
2476 out.write("</style>\r\n");
2477 out.write("</head><body>\r\n");
2478 out.write("<p><a href=\" Overall.html \">Summary</a>");
2479 out.write("&raquo;");
2480 if(metadataSetName.equalsIgnoreCase("dublin")){
2481 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2482 }
2483 else if (metadataSetName.equalsIgnoreCase("extracted")){
2484 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2485 }
2486 else{
2487 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2488 }
2489 out.write("&raquo;");
2490 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
2491 out.write("<h3 align=center>Potential Duplicate List</h3>");
2492 out.write("<table border=1 align=center>\r\n");
2493
2494 Set s = distanceMap.keySet();
2495 Iterator i = s.iterator();
2496 while(i.hasNext()){
2497
2498 String keyword = (String)i.next();
2499 InternalLink il = (InternalLink)distanceMap.get(keyword);
2500
2501 ArrayList alist = il.retrieveNodeList();
2502
2503 out.write("<tr><th>Original Text<th>Source Link\r\n");
2504 for(int a = 0; a<alist.size(); a++){
2505 InternalLink il2 = new InternalLink();
2506 il2 = (InternalLink)alist.get(a);
2507 String url = (String)il2.retrieveList().get(0);
2508 out.write("<tr><td>"+il2.getValue()+"<td><a href=\""+url+"\">"+url+"</a>");
2509
2510 }
2511 out.write("</tr>");
2512
2513 }
2514 out.write("</table>\r\n");
2515
2516 out.write("<p><a href=\" Overall.html \">Summary</a>");
2517 out.write("&raquo;");
2518 if(metadataSetName.equalsIgnoreCase("dublin")){
2519 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2520 }
2521 else if (metadataSetName.equalsIgnoreCase("extracted")){
2522 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2523 }
2524 else{
2525 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2526 }
2527 out.write("&raquo;");
2528 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
2529 out.write("</body></html>\r\n");
2530 out.close();
2531 fstream.close();
2532 }catch(IOException ex){ex.printStackTrace();}
2533
2534 }
2535
2536 public int calculateEditDistance(char[] args1, char[] args2){
2537
2538 int n = args1.length;
2539 int m = args2.length;
2540 if (n == 0) {
2541 return m;
2542 }
2543 else if (m == 0) {
2544 return n;
2545 }
2546
2547 int[] p = new int[n + 1];
2548 int[] d = new int[n + 1];
2549 int[] _d;
2550
2551
2552 int i;
2553 int j;
2554
2555 int cost; // cost
2556
2557 for (i = 0; i <= n; i++) {
2558 p[i] = i;
2559 }
2560 for (j = 1; j <= m; j++) {
2561
2562 d[0] = j;
2563 for (i = 1; i <= n; i++) {
2564 cost = (args1[i-1] == args2[j-1]) ? 0 : 1;
2565 d[i] = Math.min(Math.min(d[i - 1] + 1, p[i] + 1),
2566 p[i - 1] + cost);
2567 }
2568
2569 // copy current distance counts to 'previous row' distance counts
2570 _d = p;
2571 p = d;
2572 d = _d;
2573 }
2574 // our last action in the above loop was to switch d and p, so p now
2575 // actually has the most recent cost counts
2576 return p[n];
2577 }
2578
2579 private String removeUnusedCharacter(String target){
2580 //printWriter.write("<!-- generating all possible page-->");
2581 //printWriter.flush();
2582
2583 // remove \n
2584 while(true){
2585 if(target.contains("\\n")){
2586 target = target.replaceFirst("\\\\n","");
2587 }
2588 else{
2589 break;
2590 }
2591 }
2592
2593 //remove spaces at the end of string
2594 if(target.length()>1){
2595 while(true){
2596 if(target.length()>1){
2597 if(target.charAt(target.length()-1)==' '){
2598 target = target.substring(0,target.length()-1);
2599 }
2600 else{
2601 break;
2602 }
2603 }
2604 else{
2605 break;
2606 }
2607 }
2608 }
2609 //remove leading spaces
2610 if(target.length()>1){
2611 while(true){
2612
2613 if(target.length()>1){
2614 if(target.charAt(0)==' '){
2615 target = target.substring(1,target.length());
2616 }
2617 else{
2618 break;
2619 }}
2620 else{
2621 break;
2622 }
2623 }
2624 }
2625 //remove multiple spaces between words
2626 while(true){
2627 if(target.contains(" ")){
2628 target = target.replaceFirst(" "," ");
2629 }
2630 else{
2631 break;
2632 }
2633 }
2634
2635 return target;
2636 }
2637
2638
2639 private costModel removeUnusedCharacter(String target, double cost){
2640
2641
2642 costModel cm = new costModel();
2643
2644 // remove \n plus cost
2645 while(true){
2646
2647 if(target.contains("\\n")){
2648 target = target.replaceFirst("\\\\n","");
2649 cost = cost + 0.2;
2650 }
2651 else{
2652 break;
2653 }
2654 }
2655
2656 //remove spaces at the end of string
2657 if(target.length()>1){
2658 while(true){
2659 if(target.length()>1){
2660 if(target.charAt(target.length()-1)==' '){
2661 target = target.substring(0,target.length()-1);
2662 cost = cost + 0.2;
2663 }
2664 else{
2665 break;
2666 }}
2667 else{
2668 break;
2669 }
2670 }
2671 }
2672 //remove leading spaces
2673 if(target.length()>1){
2674 while(true){
2675
2676 if(target.length()>1){
2677 if(target.charAt(0)==' '){
2678 target = target.substring(1,target.length());
2679 cost = cost + 0.2;
2680 }
2681 else{
2682 break;
2683 }}
2684 else{
2685 break;
2686 }
2687 }
2688 }
2689 //remove multiple spaces between words
2690 while(true){
2691
2692 if(target.contains(" ")){
2693 target = target.replaceFirst(" "," ");
2694 cost = cost + 0.2;
2695 }
2696 else{
2697 break;
2698 }
2699 }
2700
2701 cm.setCost(cost);
2702 cm.setString(target);
2703
2704 return cm;
2705 }
2706
2707
2708
2709 private HashMap generateMetadataElementSortListNoIdentifier(String title, String sort, String metadataSetName,PrintWriter outx){
2710
2711 String fileName = title+"_"+sort;
2712 String collectionFolder = mds.getCollectionName();
2713 System.out.println("level 3");
2714 SearchLink sl = new SearchLink(mds.StatsDirectory);
2715 System.out.println("level 4");
2716 HashMap suggestionMap = new HashMap();
2717
2718 HashMap internalIDMap = dm.getInternalIdentifier(title);
2719 boolean status = false;
2720
2721 try{
2722 printWriter.write("<!-- generating all possible page-->");
2723 printWriter.flush();
2724
2725 FileWriter fstream = new FileWriter(destination+fileName+".html");
2726 BufferedWriter out = new BufferedWriter(fstream);
2727
2728 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
2729 out.write ("<html>\r\n");
2730 out.write("<head>\r\n<title>Metadata Element Sort List</title>\r\n");
2731 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
2732 out.write("<script type=\"text/javascript\" src=\"../script/getInfomation.js\"></script>\r\n");
2733 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/yahoo/yahoo-min.js\"></script>\r\n");
2734 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/event/event-min.js\"></script>\r\n");
2735 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/connection/connection-min.js\"></script>\r\n");
2736
2737 out.write("<style type=\"text/css\">\r\n");
2738 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
2739 out.write(".tam{height:20; text-align:center}\r\n");
2740 out.write("th{height:20; text-align:center;}\r\n");
2741 out.write("body{font-family: Arial;}\r\n");
2742 out.write("</style>\r\n");
2743 out.write("</head><body id=\""+collectionFolder+"\">\r\n");
2744
2745 out.write("<td><div style=\"float:left;\"><a href=\" Overall.html \">Summary</a>");
2746 out.write("&raquo;");
2747
2748 if(metadataSetName.equalsIgnoreCase("dublin")){
2749 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2750 }
2751 else if (metadataSetName.equalsIgnoreCase("extracted")){
2752 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2753 }
2754 else{
2755 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2756 }
2757 out.write("&raquo;");
2758 out.write("<a href=\""+title+".html\">"+title+"</a></div>");
2759
2760 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>");
2761
2762
2763 String[] list = dm.getSortList(title, sort);
2764
2765
2766
2767 if(list.length>=1){
2768
2769
2770 if(sort.equals("ASCII")){
2771
2772 out.write("<table border=\"1\" align=\"center\" width=\"800\">\r\n");
2773 out.write ("<h2 align=\"center\">"+ title+"</h2>\r\n");
2774 out.write ("<th class=\"tam\">ASCII Sort\r\n");
2775 out.write ("<th class=\"tam\">Element Values\r\n");
2776 out.write ("<th class=\"tam\">Source Documents\r\n");
2777 out.write ("<th class=\"tam\">Internal Link\r\n");
2778
2779 out.write ("<a name='topFive'>\r\n");
2780
2781 int counter = 0;
2782
2783 for(int i = 0; i<list.length; i++){
2784
2785 if(list.length<=5 && i==0){
2786 out.write ("<a name='lastFive'>\r\n");
2787 }
2788 else if((list.length>5) && (list.length-5==i)){
2789 out.write ("<a name='lastFive'>\r\n");
2790 }
2791
2792
2793 InternalLink il= (InternalLink)internalIDMap.get(list[i]);
2794 ArrayList alist2 = il.retrieveList();
2795 String id = (String)alist2.get(0);
2796 id = id.substring(4);
2797 out.write("<tr id=\""+id+"\" >\r\n");
2798
2799 if(list[i].length()>=201){
2800 if(list[i].startsWith("http://")){
2801
2802 if(list[i].length()>=100){
2803 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2804 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2805 }
2806 else{
2807 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2808 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2809 }
2810
2811 }
2812 else{
2813
2814 if(list[i].startsWith(" ")){
2815 String elements = list[i];
2816 list[i] = list[i].substring(1);
2817 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i].substring(0, 200)+"...<td>No Source Available \r\n");
2818 }
2819 else{
2820 out.write("<td>"+(counter+1)+"<td>"+list[i].substring(0, 200)+"... <td>No Source Available\r\n");
2821 }
2822 }
2823
2824 }
2825 else{
2826
2827
2828 if(list[i].startsWith("http://")){
2829
2830 if(list[i].length()>=100){
2831 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
2832 }
2833 else{
2834 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2835 }
2836 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2837
2838 }
2839
2840
2841
2842 else{
2843
2844 if(list[i].startsWith(" ")){
2845 String elements = list[i];
2846 list[i] = list[i].substring(1);
2847
2848 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i]+"<td>No Source Available\r\n");
2849 suggestionMap.put(elements,"No Source Available");
2850
2851 }
2852 else{
2853 out.write("<td>"+(counter+1)+"<td>"+list[i]+"<td>No Source Available\r\n");
2854 }
2855 }
2856 }
2857 counter++;
2858 out.write("<td onclick=\"GD(this)\">View");
2859 }
2860
2861 }
2862 else{
2863
2864
2865 out.write("<table border=\"1\" align=\"center\" width=\"800\">\r\n");
2866 out.write ("<h2 align=\"center\">"+ title+"</h2>\r\n");
2867 out.write ("<th class=\"tam\">&nbsp;\r\n");
2868 out.write ("<th class=\"tam\">Frequency\r\n");
2869 out.write ("<th class=\"tam\">Element Values\r\n");
2870 out.write ("<th class=\"tam\">Source Documents\r\n");
2871 out.write ("<th class=\"tam\">Internal Link\r\n");
2872 out.write ("<a name='topFive'>\r\n");
2873
2874 HashMap xMap = dm.getDistinctValueMap(title);
2875 int counter = 0;
2876 for(int i = 0; i<list.length; i++){
2877 if(list.length<=5 && i==0){
2878 out.write ("<a name='lastFive'>\r\n");
2879 }
2880 else if((list.length>5) && (list.length-5==i)){
2881 out.write ("<a name='lastFive'>\r\n");
2882 }
2883
2884 InternalLink il= (InternalLink)internalIDMap.get(list[i]);
2885 ArrayList alist2 = il.retrieveList();
2886 String id = (String)alist2.get(0);
2887 id = id.substring(4);
2888 out.write("<tr id=\""+id+"\" >\r\n");
2889
2890 out.write("<td>"+(counter+1)+"<th> "+((Integer)xMap.get(list[i])).toString()+"\r\n");
2891
2892 if(list[i].length()>=201){
2893
2894 if(list[i].startsWith("http://")){
2895 if(list[i].length()>=100){
2896 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2897 }
2898 else{
2899 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2900 }
2901
2902 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2903 }
2904
2905
2906
2907
2908 else{
2909 if(list[i].startsWith(" ")){
2910 list[i]=list[i].substring(1);
2911
2912 out.write("<td>"+spaceLeft+space+spaceRight+list[i].substring(0,200)+"...<td>No Source Available\r\n");
2913
2914 }
2915 else{
2916
2917 out.write("<td>"+list[i].substring(0,200)+"...<td>No Source Available\r\n");
2918
2919 }
2920 }
2921 }
2922 else{
2923 if(list[i].startsWith("http://")){
2924 if(list[i].length()>=100){
2925 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2926 }
2927 else{
2928 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2929 }
2930 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2931
2932 }
2933 else{
2934
2935 if(list[i].startsWith(" ")){
2936 list[i]=list[i].substring(1);
2937
2938
2939 out.write("<td>"+spaceLeft+space+spaceRight+list[i]+"<td>No Source Available\r\n");
2940
2941 }
2942 else{
2943
2944 out.write("<td>"+list[i]+"<td>No Source Available\r\n");
2945
2946 }
2947 }
2948 }
2949 counter++;
2950 out.write("<td onclick=\"GD(this)\">View");
2951 }
2952 }
2953 out.write("</table>\r\n");
2954 }
2955 else{
2956 out.write("<p>Sorry! The list is empty</p>");
2957 }
2958 out.write("<tr>\r\n");
2959
2960
2961 out.write("<p><a href=\" Overall.html \">Summary</a>");
2962 out.write("&raquo;");
2963
2964 if(metadataSetName.equalsIgnoreCase("dublin")){
2965 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2966 }
2967 else if (metadataSetName.equalsIgnoreCase("extracted")){
2968 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2969 }
2970 else{
2971 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2972 }
2973 out.write("&raquo;");
2974 out.write("<a href=\""+title+".html\">"+title+"</a></p>");
2975 out.write("</body></html>\r\n");
2976 out.close();
2977 fstream.close();
2978 }catch(Exception e){
2979
2980 e.printStackTrace(outx);
2981 }
2982
2983 return new HashMap();
2984 }
2985
2986
2987
2988
2989
2990
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 private void createIncompletedListNoIdentifier(String fileName, String metadataSetName, String collectionFolder,PrintWriter outx){
3018 //HashMap hp = dm.getIdentifierLink("dc.Identifier");
3019 HashMap hp = dm.getIdentifierLinkNoIdentifier();
3020 String[] ids = dm.getDocumentIDList(fileName);
3021 System.out.println("Length: "+ids.length+" "+fileName+ " "+hp.size());
3022
3023 Set st = hp.keySet();
3024 Iterator it = st.iterator();
3025 while(it.hasNext()){
3026 System.out.println(it.next());
3027 }
3028 for(int i = 0; i<ids.length; i++){
3029 System.out.println("IDS:" +ids[i]);
3030 hp.remove(ids[i]);
3031 }
3032 System.out.println("Length: "+ids.length+" "+fileName+ " "+hp.size());
3033 //String collectionFolder = mds.getCollectionName();
3034 //System.out.println(hp.size());
3035 try{
3036 FileWriter fstream = new FileWriter(destination+fileName+"_IncompletedList.html");
3037 BufferedWriter out = new BufferedWriter(fstream);
3038
3039 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
3040 out.write ("<html>\r\n");
3041 out.write("<head>\r\n<title>Incompleted Document List</title>\r\n");
3042 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
3043 out.write("<style type=\"text/css\">\r\n");
3044 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
3045 out.write(".tam{height:20; text-align:center}\r\n");
3046 out.write("th{height:20; text-align:center;}\r\n");
3047 out.write("body{font-family: Arial;}\r\n");
3048 out.write("</style>\r\n");
3049 out.write("</head><body>\r\n");
3050
3051 out.write("<p><a href=\" Overall.html \">Summary</a>");
3052 out.write("&raquo;");
3053 if(metadataSetName.equalsIgnoreCase("dublin")){
3054 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
3055 }
3056 else if (metadataSetName.equalsIgnoreCase("extracted")){
3057 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
3058 }
3059 else{
3060 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
3061 }
3062 out.write("&raquo;");
3063 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
3064
3065 out.write("<h3 align=center>"+fileName+" does not appear in the following documents</h3>");
3066 int counter = 0;
3067 System.out.println("link.....");
3068 Set s = hp.keySet();
3069 Iterator i = s.iterator();
3070 System.out.println("link2.....");
3071 out.write("<table border=1 align=center><tr><th>Document ID<th>Source Link\r\n");
3072 while(i.hasNext()){
3073 //printWriter.write("<!-- generating all possible page-->");
3074 System.out.println("link.....3");
3075 counter++;
3076 String keys = (String)i.next();
3077 System.out.println("link.....4"+keys);
3078 InternalLink il = (InternalLink) hp.get(keys);
3079
3080 ArrayList alist = il.retrieveList();
3081 System.out.println("link.....5"+alist.size());
3082 String url = (String)alist.get(0);
3083 out.write("<tr><td>"+counter+"<td><a href=\""+url+"\">"+url+"</a>\r\n");
3084 }
3085 out.write("</table></body></html>\r\n");
3086
3087
3088 out.write("<p><a href=\" Overall.html \">Summary</a>");
3089 out.write("&raquo;");
3090 if(metadataSetName.equalsIgnoreCase("dublin")){
3091 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
3092 }
3093 else if (metadataSetName.equalsIgnoreCase("extracted")){
3094 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
3095 }
3096 else{
3097 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
3098 }
3099 out.write("&raquo;");
3100 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
3101
3102 out.close();
3103 fstream.close();
3104 }catch(IOException ex){ex.printStackTrace();}
3105
3106
3107 }
3108 ///////////////////////////////////////////////////////////////////////////////////////
3109}
3110
Note: See TracBrowser for help on using the repository browser.