source: gs3-extensions/mat/trunk/src/org/greenstone/mat/servlet/PrintHTML.java@ 21927

Last change on this file since 21927 was 21927, checked in by sjm84, 14 years ago

Renamed package to org.greenstone.mat from org.greenstone.gsdl3_extension.mat
File size: 107.5 KB
Line 
1package org.greenstone.mat.servlet;
2
3import java.io.*;
4import java.text.SimpleDateFormat;
5import java.util.ArrayList;
6import java.util.Calendar;
7import java.util.Collection;
8import java.util.HashMap;
9import java.util.Iterator;
10import java.util.Map;
11import java.util.Set;
12
13import javax.xml.parsers.DocumentBuilder;
14import javax.xml.parsers.DocumentBuilderFactory;
15
16import org.w3c.dom.Document;
17import org.w3c.dom.Element;
18import org.w3c.dom.Node;
19import org.w3c.dom.NodeList;
20import org.greenstone.gsdl3.*;
21import org.greenstone.gsdl3.util.GlobalProperties;
22
23public class PrintHTML {
24
25 private final String NORMAL = "normal";
26 private final String WORST = "worst";
27 private final String BEST = "best";
28 private final String title1 ="Hide Empty Metadata ,";
29 private final String title2 ="Hide Completed Metadata ,";
30 private final String title3 ="Hide Documents with Empty Metadata ,";
31 private final String title4 ="Hide Documents with Completed Metadata ,";
32 private final String title5 ="No Available Graph";
33 private final String metadataSet1 = "Dublin Core";
34 private final String metadataSet2 = "Extracted";
35 private final String spaceLeft = "<I>&laquo;";
36 private final String spaceRight = "&raquo;</I>";
37 private final String htmlSpace = "&nbsp;";
38 private final String oddChar ="odd Character";
39 private final String destination;
40 private final String space ="space";
41 private String hostName;
42 private String cssString;
43 private int port_number;
44 private GlobalProperties globalProperty = null;
45 private String fileSeparator = File.separator;
46
47 String linkIdentifier = "";
48
49
50
51 private HashMap url;
52
53 MetadataStats mds;
54 DataMaker dm;
55 Calendar cl;
56 SimpleDateFormat sdf;
57 PrintWriter printWriter;
58
59 public PrintHTML(MetadataStats ms, PrintWriter arg, int port){
60
61 port_number = port;
62 mds = ms;
63 printWriter = arg;
64 dm = new DataMaker(mds);
65 cl=Calendar.getInstance();
66 linkIdentifier = "dc.Identifier";
67 destination = mds.HTMLDirectory;
68 sdf = new SimpleDateFormat(" dd MMM yyyy 'at' HH:mm:ss z 'GMT'Z");
69 setupMetadataLink("metadataElementURL");
70
71
72 try {
73 java.net.InetAddress localMachine = java.net.InetAddress.getLocalHost();
74 hostName = localMachine.getHostName();
75 }catch (java.net.UnknownHostException uhe) {
76 uhe.printStackTrace();
77 }
78
79 cssString = "<link rel=\"stylesheet\" href=\"http://"+hostName+":"+port_number+"/greenstone3/mat/script/mat.css\" type=\"text/css\" >";
80 }
81
82 private ArrayList getHideEmptyMetadataDetail(ArrayList dataset, String[] id, String[] metadataName){
83
84 printWriter.write("<!-- generating all possible page -->");
85 printWriter.flush();
86
87 if(id.length==0 || metadataName.length==0){
88 return new ArrayList();
89 }
90
91 ArrayList detailList = new ArrayList();
92 ArrayList alist = new ArrayList();
93 ArrayList nameList = new ArrayList();
94
95 for(int i = 0; i< metadataName.length; i++){
96 if(!dm.IsElementEmpty(metadataName[i])){
97 alist.add(dm.getMetadataRows(metadataName[i]));
98 nameList.add(metadataName[i]);
99 }
100 }
101
102 String[] metaDataElementName = new String[nameList.size()];
103
104 for(int i = 0; i< nameList.size(); i++){
105 metaDataElementName[i] = nameList.get(i).toString();
106 }
107
108 if(id.length==0 || metaDataElementName.length==0){
109 return new ArrayList();
110 }
111
112 detailList.add(alist);
113 detailList.add(id);
114 detailList.add(metaDataElementName);
115 printWriter.write("<!-- generating all possible page-->");
116 printWriter.flush();
117
118 return detailList;
119 }
120
121 private ArrayList getHideFullMetadataDetail(ArrayList dataset, String[] id, String[] metadataName){
122
123 printWriter.write("<!-- generating all possible page-->");
124 printWriter.flush();
125
126 if(id.length==0 || metadataName.length==0){
127 return new ArrayList();
128 }
129
130 ArrayList detailList = new ArrayList();
131 ArrayList alist = new ArrayList();
132 ArrayList nameList = new ArrayList();
133
134 for(int i = 0; i< metadataName.length; i++){
135 if(!dm.IsElementFull(metadataName[i])){
136 alist.add(dm.getMetadataRows(metadataName[i]));
137 nameList.add(metadataName[i]);
138 }
139 }
140
141 String[] metaDataElementName = new String[nameList.size()];
142
143 for(int i = 0; i< nameList.size(); i++){
144 metaDataElementName[i] = nameList.get(i).toString();
145 }
146
147 if(id.length==0 || metaDataElementName.length==0){
148 return new ArrayList();
149 }
150
151 detailList.add(alist);
152 detailList.add(id);
153 detailList.add(metaDataElementName);
154
155 printWriter.write("<!-- generating all possible page-->");
156 printWriter.flush();
157
158 return detailList;
159 }
160
161 private ArrayList getHideEmptyDocumentDetail(ArrayList dataset, String[] id, String[] metadataName){
162
163 printWriter.write("<!-- generating all possible page-->");
164 printWriter.flush();
165
166 if(id.length==0 || metadataName.length==0){
167 return new ArrayList();
168 }
169
170 ArrayList detailList = new ArrayList();
171 ArrayList alist = dm.removeDocument(dataset, id, 0);
172 ArrayList idList = dm.getRemovedID();
173 ArrayList newIDList = new ArrayList();
174
175 for(int i = 0; i< id.length; i++){
176 if(!idList.contains(id[i])){
177 newIDList.add(id[i]);
178 }
179 }
180
181 id = new String[newIDList.size()];
182 for(int i = 0; i<newIDList.size(); i++){
183 id[i] = newIDList.get(i).toString();
184 }
185
186 if(id.length==0 || metadataName.length==0){
187 return new ArrayList();
188 }
189
190 detailList.add(alist);
191 detailList.add(id);
192 detailList.add(metadataName);
193
194 printWriter.write("<!-- generating all possible page-->");
195 printWriter.flush();
196
197 return detailList;
198 }
199
200 private ArrayList getHideFullDocumentDetail(ArrayList dataset, String[] id, String[] metadataName){
201
202 printWriter.write("<!-- generating all possible page-->");
203 printWriter.flush();
204
205 if(id.length==0 || metadataName.length==0){
206 return new ArrayList();
207 }
208
209 ArrayList detailList = new ArrayList();
210 ArrayList alist = dm.removeDocument(dataset, id ,1);
211 ArrayList idList = dm.getRemovedID();
212 ArrayList newIDList = new ArrayList();
213
214 for(int i = 0; i< id.length; i++){
215 if(!idList.contains(id[i])){
216 newIDList.add(id[i]);
217 }
218 }
219
220 id = new String[newIDList.size()];
221 for(int i = 0; i<newIDList.size(); i++){
222 id[i] = newIDList.get(i).toString();
223 }
224
225 if(id.length==0 || metadataName.length==0){
226 return new ArrayList();
227 }
228
229 detailList.add(alist);
230 detailList.add(id);
231 detailList.add(metadataName);
232
233 printWriter.write("<!-- generating all possible page-->");
234 printWriter.flush();
235
236 return detailList;
237 }
238
239 public void generateAllPossibleGraph(ArrayList Namelist,String[]ids,String[] names,String condition){
240
241 printWriter.write("<!-- generating all possible page-->");
242 printWriter.flush();
243
244 ArrayList dataList= Namelist;
245 String[] id = ids;
246 String[] metadataName = names;
247
248 ArrayList alist = new ArrayList();
249 ArrayList detailList = new ArrayList();
250 ArrayList new_dataList = new ArrayList();
251 String[] idList;
252 String[] metadataNameList;
253
254 //---------------------------------------------------------------------------------------------------------SS
255 //-----------------------------------------------------------------------SSSS
256 if(id.length==0 || metadataName.length==0){
257 generateEmptyGraph("SSSS",WORST,condition);
258 generateEmptyGraph("SSSS",BEST,condition);
259 }
260
261 else{
262 new_dataList = transformDataList(dataList,id);
263 WriteHTML("SSSS",new_dataList,id,metadataName,WORST,"Show completed graph",condition);
264 WriteHTML("SSSS",new_dataList,id,metadataName,BEST,"Show completed graph",condition);
265 }
266 //-----------------------------------------------------------------------SSSH
267 detailList = getHideFullDocumentDetail(dataList,id,metadataName);
268 if(detailList.size()==0){
269 generateEmptyGraph("SSSH",WORST,condition);
270 generateEmptyGraph("SSSH",BEST,condition);
271 }
272 else{
273 alist = (ArrayList)detailList.get(0);
274 idList = (String[])detailList.get(1);
275 metadataNameList = (String[])detailList.get(2);
276 new_dataList = transformDataList(alist,idList);
277 WriteHTML("SSSH",new_dataList,idList,metadataNameList,WORST,title4,condition);
278 WriteHTML("SSSH",new_dataList,idList,metadataNameList,BEST,title4,condition);
279 }
280 //-----------------------------------------------------------------------SSHS
281 detailList = getHideEmptyDocumentDetail(dataList,id,metadataName);
282 if(detailList.size()==0){
283 generateEmptyGraph("SSHS",WORST,condition);
284 generateEmptyGraph("SSHS",BEST,condition);
285 generateEmptyGraph("SSHH",WORST,condition);
286 generateEmptyGraph("SSHH",BEST,condition);
287 }
288 else{
289 alist = (ArrayList)detailList.get(0);
290 idList = (String[])detailList.get(1);
291 metadataNameList = (String[])detailList.get(2);
292 new_dataList = transformDataList(alist,idList);
293 WriteHTML("SSHS",new_dataList,idList,metadataNameList,WORST,title3,condition);
294 WriteHTML("SSHS",new_dataList,idList,metadataNameList,BEST,title3,condition);
295 //-----------------------------------------------------------------------SSHH
296 detailList = getHideFullDocumentDetail(alist,idList,metadataNameList);
297 if(detailList.size()==0){
298 generateEmptyGraph("SSHH",WORST,condition);
299 generateEmptyGraph("SSHH",BEST,condition);
300 }
301 else{
302 alist = (ArrayList)detailList.get(0);
303 idList = (String[])detailList.get(1);
304 metadataNameList = (String[])detailList.get(2);
305 new_dataList = transformDataList(alist,idList);
306 WriteHTML("SSHH",new_dataList,idList,metadataNameList,WORST,title3+title4,condition);
307 WriteHTML("SSHH",new_dataList,idList,metadataNameList,BEST,title3+title4,condition);
308 }
309 }
310
311 //---------------------------------------------------------------------------------------------------------SH
312 ArrayList xList = new ArrayList();
313 String[] idListCopy;
314 String[] metadataNameListCopy;
315
316 //-----------------------------------------------------------------------SHSS
317 detailList = getHideFullMetadataDetail(dataList,id,metadataName);
318
319 if(detailList.size()==0){
320 generateEmptyGraph("SHSS",WORST,condition);
321 generateEmptyGraph("SHSS",BEST,condition);
322
323 generateEmptyGraph("SHSH",WORST,condition);
324 generateEmptyGraph("SHSH",BEST,condition);
325
326 generateEmptyGraph("SHHS",WORST,condition);
327 generateEmptyGraph("SHHS",BEST,condition);
328
329 generateEmptyGraph("SHHH",WORST,condition);
330 generateEmptyGraph("SHHH",BEST,condition);
331 }
332 else{
333 alist = (ArrayList)detailList.get(0);
334 idList = (String[])detailList.get(1);
335 metadataNameList = (String[])detailList.get(2);
336
337 xList = (ArrayList)alist.clone();
338 idListCopy = (String[]) idList.clone();
339 metadataNameListCopy = (String[])metadataNameList.clone();
340 new_dataList = transformDataList(alist,idList);
341 WriteHTML("SHSS",new_dataList,idList,metadataNameList,WORST,title2,condition);
342 WriteHTML("SHSS",new_dataList,idList,metadataNameList,BEST,title2,condition);
343
344 //-----------------------------------------------------------------------SHHS
345 detailList = getHideEmptyDocumentDetail((ArrayList)xList.clone(),(String[])idListCopy.clone(),(String[])metadataNameListCopy.clone());
346 if(detailList.size()==0){
347 generateEmptyGraph("SHHS",WORST,condition);
348 generateEmptyGraph("SHHS",BEST,condition);
349 generateEmptyGraph("SHHH",WORST,condition);
350 generateEmptyGraph("SHHH",BEST,condition);
351 }
352 else{
353 alist = (ArrayList)detailList.get(0);
354 idList = (String[])detailList.get(1);
355 metadataNameList = (String[])detailList.get(2);
356 new_dataList = transformDataList(alist,idList);
357 WriteHTML("SHHS",new_dataList,idList,metadataNameList,WORST,title2+title3,condition);
358 WriteHTML("SHHS",new_dataList,idList,metadataNameList,BEST,title2+title3,condition);
359
360 //-----------------------------------------------------------------------SHHH
361 detailList = getHideFullDocumentDetail(alist,idList,metadataNameList);
362 if(detailList.size()==0){
363 generateEmptyGraph("SHHH",WORST,condition);
364 generateEmptyGraph("SHHH",BEST,condition);
365 }
366 else{
367 alist = (ArrayList)detailList.get(0);
368 idList = (String[])detailList.get(1);
369 metadataNameList = (String[])detailList.get(2);
370 new_dataList = transformDataList(alist,idList);
371 WriteHTML("SHHH",new_dataList,idList,metadataNameList,WORST,title2+title3+title4,condition);
372 WriteHTML("SHHH",new_dataList,idList,metadataNameList,BEST,title2+title3+title4,condition);
373 }
374 }
375
376 //-----------------------------------------------------------------------SHSH
377 detailList = getHideFullDocumentDetail((ArrayList)xList.clone(),(String[])idListCopy.clone(),(String[])metadataNameListCopy.clone());
378 if(detailList.size()==0){
379 generateEmptyGraph("SHSH",WORST,condition);
380 generateEmptyGraph("SHSH",BEST,condition);
381 }
382 else{
383 alist = (ArrayList)detailList.get(0);
384 idList = (String[])detailList.get(1);
385 metadataNameList = (String[])detailList.get(2);
386 new_dataList = transformDataList(alist,idList);
387 WriteHTML("SHSH",new_dataList,idList,metadataNameList,WORST,title2+title4,condition);
388 WriteHTML("SHSH",new_dataList,idList,metadataNameList,BEST,title2+title4,condition);
389 }
390 }
391
392 //---------------------------------------------------------------------------------------------------------HS
393 ArrayList xList1 = new ArrayList();
394 String[] idListCopy1;
395 String[] metadataNameListCopy1;
396
397 //-----------------------------------------------------------------------HSSS
398 detailList = getHideEmptyMetadataDetail(dataList,id,metadataName);
399
400 if(detailList.size()==0){
401 generateEmptyGraph("HSSS",WORST,condition);
402 generateEmptyGraph("HSSS",BEST,condition);
403
404 generateEmptyGraph("HSHS",WORST,condition);
405 generateEmptyGraph("HSHS",BEST,condition);
406
407 generateEmptyGraph("HSHH",WORST,condition);
408 generateEmptyGraph("HSHH",BEST,condition);
409
410 generateEmptyGraph("HSSH",WORST,condition);
411 generateEmptyGraph("HSSH",BEST,condition);
412 }
413 else{
414 alist = (ArrayList)detailList.get(0);
415 idList = (String[])detailList.get(1);
416 metadataNameList = (String[])detailList.get(2);
417
418 xList1 = (ArrayList)alist.clone();
419 idListCopy1 = (String[]) idList.clone();
420 metadataNameListCopy1 = (String[])metadataNameList.clone();
421
422 new_dataList = transformDataList(alist,idList);
423 WriteHTML("HSSS",new_dataList,idList,metadataNameList,WORST,title1,condition);
424 WriteHTML("HSSS",new_dataList,idList,metadataNameList,BEST,title1,condition);
425
426 //-----------------------------------------------------------------------HSHS
427 detailList = getHideEmptyDocumentDetail((ArrayList)xList1.clone(),(String[])idListCopy1.clone(),(String[])metadataNameListCopy1.clone());
428 if(detailList.size()==0){
429 generateEmptyGraph("HSHS",WORST,condition);
430 generateEmptyGraph("HSHS",BEST,condition);
431
432 generateEmptyGraph("HSHH",WORST,condition);
433 generateEmptyGraph("HSHH",BEST,condition);
434 }
435 else{
436 alist = (ArrayList)detailList.get(0);
437 idList = (String[])detailList.get(1);
438 metadataNameList = (String[])detailList.get(2);
439
440 new_dataList = transformDataList(alist,idList);
441 WriteHTML("HSHS",new_dataList,idList,metadataNameList,WORST,title1+title3,condition);
442 WriteHTML("HSHS",new_dataList,idList,metadataNameList,BEST,title1+title3,condition);
443
444 //-----------------------------------------------------------------------HSHH
445 detailList = getHideFullDocumentDetail(alist,idList,metadataNameList);
446 if(detailList.size()==0){
447 generateEmptyGraph("HSHH",WORST,condition);
448 generateEmptyGraph("HSHH",BEST,condition);
449 }
450 else{
451 alist = (ArrayList)detailList.get(0);
452 idList = (String[])detailList.get(1);
453 metadataNameList = (String[])detailList.get(2);
454
455 new_dataList = transformDataList(alist,idList);
456 WriteHTML("HSHH",new_dataList,idList,metadataNameList,WORST,title1+title3+title4,condition);
457 WriteHTML("HSHH",new_dataList,idList,metadataNameList,BEST,title1+title3+title4,condition);
458 }
459 }
460
461 //-----------------------------------------------------------------------HSSH
462 detailList = getHideFullDocumentDetail((ArrayList)xList1.clone(),(String[])idListCopy1.clone(),(String[])metadataNameListCopy1.clone());
463 if(detailList.size()==0){
464 generateEmptyGraph("HSSH",WORST,condition);
465 generateEmptyGraph("HSSH",BEST,condition);
466 }
467 else{
468 alist = (ArrayList)detailList.get(0);
469 idList = (String[])detailList.get(1);
470 metadataNameList = (String[])detailList.get(2);
471
472 new_dataList = transformDataList(alist,idList);
473 WriteHTML("HSSH",new_dataList,idList,metadataNameList,WORST,title1+title4,condition);
474 WriteHTML("HSSH",new_dataList,idList,metadataNameList,BEST,title1+title4,condition);
475 }
476 }
477
478 //---------------------------------------------------------------------------------------------------------HH
479 ArrayList xList2 = new ArrayList();
480 String[] idListCopy2;
481 String[] metadataNameListCopy2;
482
483 //-----------------------------------------------------------------------HHSS
484 detailList = getHideEmptyMetadataDetail(dataList,id,metadataName);
485 if(detailList.size()==0){
486 generateEmptyGraph("HHSS",WORST,condition);
487 generateEmptyGraph("HHSS",BEST,condition);
488
489 generateEmptyGraph("HHHS",WORST,condition);
490 generateEmptyGraph("HHHS",BEST,condition);
491
492 generateEmptyGraph("HHHH",WORST,condition);
493 generateEmptyGraph("HHHH",BEST,condition);
494
495 generateEmptyGraph("HHSH",WORST,condition);
496 generateEmptyGraph("HHSH",BEST,condition);
497 }
498 else{
499 alist = (ArrayList)detailList.get(0);
500 idList = (String[])detailList.get(1);
501 metadataNameList = (String[])detailList.get(2);
502
503 detailList = getHideFullMetadataDetail(alist,idList,metadataNameList);
504
505 if(detailList.size()==0){
506 generateEmptyGraph("HHSS",WORST,condition);
507 generateEmptyGraph("HHSS",BEST,condition);
508
509 generateEmptyGraph("HHHS",WORST,condition);
510 generateEmptyGraph("HHHS",BEST,condition);
511
512 generateEmptyGraph("HHHH",WORST,condition);
513 generateEmptyGraph("HHHH",BEST,condition);
514
515 generateEmptyGraph("HHSH",WORST,condition);
516 generateEmptyGraph("HHSH",BEST,condition);
517 }
518 else{
519 alist = (ArrayList)detailList.get(0);
520 idList = (String[])detailList.get(1);
521 metadataNameList = (String[])detailList.get(2);
522
523 xList2 = (ArrayList)alist.clone();
524 idListCopy2 = (String[]) idList.clone();
525 metadataNameListCopy2 = (String[])metadataNameList.clone();
526
527 new_dataList = transformDataList(alist,idList);
528 WriteHTML("HHSS",new_dataList,idList,metadataNameList,WORST,title1+title2,condition);
529 WriteHTML("HHSS",new_dataList,idList,metadataNameList,BEST,title1+title2,condition);
530
531 //---------------------------------------------------------------------- HHSH
532 detailList = getHideFullDocumentDetail((ArrayList)xList2.clone(),(String[])idListCopy2.clone(),(String[])metadataNameListCopy2.clone());
533 if(detailList.size()==0){
534 generateEmptyGraph("HHSH",WORST,condition);
535 generateEmptyGraph("HHSH",BEST,condition);
536 }
537 else{
538 alist = (ArrayList)detailList.get(0);
539 idList = (String[])detailList.get(1);
540 metadataNameList = (String[])detailList.get(2);
541
542 new_dataList = transformDataList(alist,idList);
543
544 WriteHTML("HHSH",new_dataList,idList,metadataNameList,WORST,title1+title2+title4,condition);
545 WriteHTML("HHSH",new_dataList,idList,metadataNameList,BEST,title1+title2+title4,condition);
546 }
547 //----------------------------------------------------------------------HHHS
548 detailList = getHideEmptyDocumentDetail((ArrayList)xList2.clone(),(String[])idListCopy2.clone(),(String[])metadataNameListCopy2.clone());
549 if(detailList.size()==0){
550 generateEmptyGraph("HHHS",WORST,condition);
551 generateEmptyGraph("HHHS",BEST,condition);
552
553 generateEmptyGraph("HHHH",WORST,condition);
554 generateEmptyGraph("HHHH",BEST,condition);
555 }
556 else{
557 alist = (ArrayList)detailList.get(0);
558 idList = (String[])detailList.get(1);
559 metadataNameList = (String[])detailList.get(2);
560
561 new_dataList = transformDataList(alist,idList);
562
563 WriteHTML("HHHS",new_dataList,idList,metadataNameList,WORST,title1+title2+title3,condition);
564 WriteHTML("HHHS",new_dataList,idList,metadataNameList,BEST,title1+title2+title3,condition);
565
566 //----------------------------------------------------------------------HHHH
567 detailList = getHideFullDocumentDetail((ArrayList)alist.clone(),(String[])idList.clone(),(String[])metadataNameList.clone());
568 if(detailList.size()==0){
569 generateEmptyGraph("HHHH",WORST,condition);
570 generateEmptyGraph("HHHH",BEST,condition);
571 }
572 else{
573 alist = (ArrayList)detailList.get(0);
574 idList = (String[])detailList.get(1);
575 metadataNameList = (String[])detailList.get(2);
576
577 new_dataList = transformDataList(alist,idList);
578
579 WriteHTML("HHHH",new_dataList,idList,metadataNameList,WORST,title1+title2+title3+title4,condition);
580 WriteHTML("HHHH",new_dataList,idList,metadataNameList,BEST,title1+title2+title3+title4,condition);
581 }
582 }
583 }
584 }
585 }
586
587 private ArrayList transformDataList(ArrayList list, String[] ids){
588
589 printWriter.write("<!-- generating all possible page-->");
590 printWriter.flush();
591
592 ArrayList wholeList = new ArrayList();
593
594 for(int i = 0; i< ids.length; i++){
595 ArrayList idList = new ArrayList();
596 for(int j = 0; j<list.size();j++){
597 int[] datarows = (int[]) list.get(j);
598 idList.add(new Integer (datarows[i]));
599 }
600 wholeList.add(idList);
601 }
602
603 for(int i = 0; i< wholeList.size(); i++){
604 ArrayList alist = (ArrayList)wholeList.get(i);
605 int[] rows = new int[alist.size()];
606 for(int j = 0; j< alist.size(); j++){
607 rows[j] = ((Integer)alist.get(j)).intValue();
608 }
609 wholeList.remove(i);
610 wholeList.add(i,rows);
611 }
612
613 printWriter.write("<!-- generating all possible page-->");
614 printWriter.flush();
615
616 return wholeList;
617
618 }
619
620 private void generateEmptyGraph(String fileName,String condition,String suffix){
621
622 printWriter.write("<!-- generating all possible page-->");
623 printWriter.flush();
624
625 String suf = suffix;
626 if(!suffix.equalsIgnoreCase("dublin")){
627 suf = "other";
628 }
629
630 String cases = condition;
631 String collectionFolder = mds.getCollectionName();
632
633 try{
634 FileWriter fstream = new FileWriter(destination+suf+"_"+fileName+"_"+cases+".html");
635 BufferedWriter out = new BufferedWriter(fstream);
636 out.write("<!-- This comment keeps IE6/7 in the reliable quirks mode -->\r\n");
637 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">\r\n");
638 out.write("<html>\r\n");
639 out.write("<head>\r\n<title> No Available Chart</title>\r\n");
640 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
641 out.write("<link rel=\"stylesheet\" href=\"../script/doby2.css\" type=\"text/css\"/>\r\n");
642 out.write("</head>\r\n");
643 out.write("<p><a href=\" Overall.html \">Summary</a></p>");
644 out.write("<body>\r\n");
645 out.write("<p>No data available to render chart.</p>\r\n");
646 out.write("<p>Reason: Document number is zero or Metadata element number is zero </p>\r\n");
647 out.write("</body></html>\r\n");
648 out.close();
649 fstream.close();
650
651 }catch(Exception e){}
652
653 printWriter.write("<!-- generating all possible page-->");
654 printWriter.flush();
655 }
656
657 private void WriteHTML(String fileName, ArrayList dataset, String[] ids, String[] metadataName, String condition, String title, String suffix){
658
659 int blueDot = 0;
660 String cases = condition;
661 String suf = suffix;
662 if(!suffix.equalsIgnoreCase("dublin")){
663 suf = "other";
664 }
665 try{
666 printWriter.write("<!-- generating all possible page-->");
667 printWriter.flush();
668
669 ArrayList tempList = mds.getMetadataNameList();
670 ArrayList urlIDList;
671 if(tempList.contains("dc.Identifier") && linkIdentifier.equals("dc.Identifier")){
672 urlIDList = dm.getURLMap("dc.Identifier");
673 }
674 else if(tempList.contains("nzir_internal.Link") && linkIdentifier.equals("nzir_internal.Link")){
675 urlIDList = dm.getURLMap("nzir_internal.Link");
676 }
677 else{
678 urlIDList = new ArrayList();
679 }
680
681 FileWriter fstream = new FileWriter(destination+suf+"_"+fileName+"_"+cases+".html");
682 BufferedWriter out = new BufferedWriter(fstream);
683
684 out.write("<!-- This comment keeps IE6/7 in the reliable quirks mode -->\r\n");
685 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">\r\n");
686 out.write("<html>\r\n");
687 out.write("<head>\r\n<title>"+title+"</title>\r\n");
688 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
689 out.write("<link rel=\"stylesheet\" href=\"../script/doby2.css\" type=\"text/css\"/>\r\n");
690 out.write("<script type=\"text/javascript\" src=\"../script/getInfomation.js\"></script>\r\n");
691 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/yahoo/yahoo-min.js\"></script>\r\n");
692 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/event/event-min.js\"></script>\r\n");
693 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/connection/connection-min.js\"></script>\r\n");
694 out.write("</head>\r\n");
695 out.write ("<p><a href=\" Overall.html \">Summary</a></p>");
696 out.write("<body id=\""+mds.getCollectionName()+"\" onLoad=\"reconfig()\">\r\n");
697 out.write("<div id=\"container\">\r\n");
698 out.write("<div class=\"tableContainer\">\r\n");
699 out.write ("<table cellspacing=\"0\">\r\n");
700 out.write ("<thead>\r\n");
701 out.write ("<tr>\r\n");
702 out.write("<td class=\"qh\">Info\r\n");
703 out.write("<td class=\"qh\">URL\r\n");
704
705 for(int a = 0; a< metadataName.length; a++){
706 out.write("<td>"+ metadataName[a]+"\r\n");
707 }
708
709 out.write ("</thead>\r\n<tfoot>\r\n<tr>\r\n");
710 out.write("<td class=\"qh\">&nbsp;\r\n");
711 out.write("<td class=\"qh\">&nbsp;\r\n");
712
713 for(int a = 0; a< metadataName.length; a++){
714 out.write("<td>"+ dm.Mean(metadataName[a])+"%\r\n");
715 }
716
717 out.write ("</tfoot>\r\n<tbody>\r\n");
718
719 if(cases.equals("normal")){
720 for(int i = ids.length; i>0; i--){
721 out.write ("<tr>");
722 int[] datarows = (int[])dataset.get(i-1);
723
724 for(int j = 0; j<datarows.length; j++){
725
726 if(datarows[j]==1){
727 out.write ("<td class=\"b\">");
728 blueDot++;
729 }
730 else {
731 out.write ("<td>&nbsp;");
732 }
733 }
734 }
735 }
736 else{
737 ArrayList idList = new ArrayList();
738 for(int i = 0; i<ids.length;i++){
739 idList.add(ids[i]);
740 }
741
742
743 HashMap hp = new HashMap();
744
745 for(int i = ids.length; i>0; i--){
746 String idValue = ids[i-1];
747 int dots = 0;
748 int[] datarows = (int[])dataset.get(i-1);
749 for(int j = 0; j<datarows.length; j++){
750 if(datarows[j]==1){
751 dots++;
752 }
753 }
754 hp.put(idValue,new Integer(dots));
755
756 }
757 ArrayList alist = dm.sortMap(hp);
758 String[] idIntValue = new String[alist.size()];
759 for(int i = 0; i< alist.size(); i++){
760 Map.Entry entry = (Map.Entry) alist.get(i);
761 String idElement = ((String) entry.getKey());
762 idIntValue[i] = idElement;
763 }
764
765 if(cases.equals("worst")){
766
767 for(int i = 0; i<idIntValue.length; i++){
768
769 printWriter.write("<!-- generating all possible page -->");
770 printWriter.flush();
771
772 int value = idList.indexOf(idIntValue[i]);
773 int[] datarows = (int[])dataset.get(value);
774
775 out.write ("<tr id=\""+idIntValue[i].substring(4)+"\">\r\n");
776 out.write("<td class=\"E\" onclick=\"GD(this)\">&nbsp;\r\n");
777
778 if(urlIDList.contains(idIntValue[i])){
779 out.write("<td class=\"qh\" onclick=\"GX(this)\"><span title=\"open URL in new window\">open</span>\r\n");
780 }
781 else{
782 out.write("<td>\r\n");
783 }
784
785 for(int j = 0; j<datarows.length; j++){
786
787 if(datarows[j]==1){
788 out.write ("<td class=\"b\">\r\n");
789 blueDot++;
790 }
791 else {
792 out.write ("<td class=\"w\">\r\n");
793 }
794 }
795 }
796 }
797
798 if(cases.equals("best")){
799
800 for(int i = idIntValue.length; i>0; i--){
801 printWriter.write("<!-- generating all possible page -->");
802 printWriter.flush();
803
804 int value = idList.indexOf(idIntValue[i-1]);
805 int[] datarows = (int[])dataset.get(value);
806
807 out.write ("<tr id=\""+idIntValue[i-1].substring(4)+"\">\r\n");
808 out.write("<td class=\"E\" onclick=\"GD(this)\">&nbsp;\r\n");
809
810 if(urlIDList.contains(idIntValue[i-1])){
811 out.write("<td class=\"qh\" onclick=\"GX(this)\"><span title=\"open URL in new window\">open</span>\r\n");
812 }
813 else{
814 out.write("<td>\r\n");
815 }
816
817 for(int j = 0; j<datarows.length; j++){
818
819 if(datarows[j]==1){
820 out.write ("<td class=\"b\">\r\n");
821 blueDot++;
822 }
823 else {
824 out.write ("<td class=\"w\">\r\n");
825 }
826 }
827 }
828 }
829 }
830
831 out.write ("</tbody>\r\n");
832 out.write ("</table>\r\n</div></div>\r\n");
833
834 int t1 = ids.length;
835 int t3 = metadataName.length;
836 int t4 = t1*t3;
837
838 out.write("<table>\r\n<tbody class=\"table1\">");
839 out.write("<tr>\r\n");
840
841 HashMap tempMap = mds.getMetadataSetMap();
842 MetadataSet ms = (MetadataSet)tempMap.get(suffix);
843
844 out.write("<td class=\"bfont\">This subset shows "+t1+" out of "+mds.getDocNum()+" documents");
845 out.write("<td class=\"bfont\">"+blueDot+" out of "+(mds.getDocNum()*t3)+" metadata items are defined");
846 out.write("<tr>\r\n");
847 out.write("<td class=\"bfont\">This subset shows "+t3+" out of "+(ms.getIndexsList().size())+" metadata elements");
848 out.write("<td class=\"bfont\">Subset completeness: "+dm.round((double)(blueDot*100/t4),5)+"%");
849 out.write("</table>\r\n");
850 out.write ("<p> <a href=\"Overall.html \">Summary</a></p>");
851 out.write ("</body></html>");
852 //Close the output stream
853
854 out.close();
855 fstream.close();
856 }catch (Exception e){
857 e.printStackTrace(printWriter);
858 }
859
860 printWriter.write("<!-- generating all possible page-->");
861 printWriter.flush();
862 }
863
864
865 public void generateOverallStatisticsPage(HashMap MetadataSetMap){
866
867 String fileName = "Overall";
868 ArrayList wholeList = new ArrayList();
869 HashMap hp = mds.getMetadataSetMap();
870 Collection c = hp.values();
871 Iterator i = c.iterator();
872 int counter = 0;
873 String[][] MetadataData = new String[c.size()][2];
874
875 while(i.hasNext()){
876 MetadataSet mds = (MetadataSet)i.next();
877 wholeList.add(mds);
878 ArrayList newMDS = new ArrayList();
879 newMDS.add(mds);
880
881 MetadataData[counter][0] = mds.getName();
882 MetadataData[counter][1] = dm.getSingleMetadataSetCompleteness(newMDS)+"%";
883
884 counter++;
885 }
886
887 try{
888 printWriter.write("<!-- generating all possible page-->");
889 printWriter.flush();
890
891 String str = sdf.format(cl.getTime());
892 String collectionFolder = mds.getCollectionName();
893
894 FileWriter fstream = new FileWriter(destination+fileName+".html");
895 BufferedWriter out = new BufferedWriter(fstream);
896
897 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
898 out.write("<html>\r\n");
899 out.write("<head><title>Summary</title>\r\n");
900 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
901 out.write("<style type=\"text/css\">\r\n");
902 out.write(".tam{height:20; text-align:right}\r\n");
903 out.write("th{height:20; text-align:left}\r\n");
904 out.write("body{font-family: Arial;}\r\n");
905 out.write("</style>\r\n");
906 out.write("<script type=\"text/javascript\" src=\"../script/status3.js\"></script>\r\n");
907 out.write("</head>\r\n");
908 out.write("<body>\r\n");
909 out.write ("<td> <div style=\"float:left;\"> <a href=\"http://"+hostName+":"+port_number+"/greenstone3/mat\">Mat Home</a></div>");
910 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>");
911 out.write("<h3 align=\"center\">Summary</h3>\r\n");
912 out.write("<table border=1 align=\"center\">\r\n");
913 out.write("<colgroup width=\"300\" span=\"2\">\r\n");
914 out.write("<tr>\r\n");
915 out.write("<th>OAI URL: </th>\r\n");
916
917 if(mds.getOAIURL().length()>50){
918 out.write("<td class=\"tam\"><font size=\"2px\"><a href=\""+mds.getOAIURL()+"\">"+mds.getOAIURL().substring(0,47)+"....</a></font>\r\n");
919 }
920 else{
921 out.write("<td class=\"tam\"><font size=\"2px\"><a href=\""+mds.getOAIURL()+"\">"+mds.getOAIURL()+"</a></font>\r\n");
922 }
923
924 out.write("</tr>\r\n");
925 out.write("<tr>\r\n");
926 out.write("<th>Number of Records:\r\n");
927 out.write("<td class=\"tam\">"+mds.getDocNum()+"\r\n");
928 out.write("</tr>\r\n");
929 out.write("<table border=1 align=\"center\">\r\n");
930 out.write("<colgroup width=\"300\" span=\"2\">\r\n");
931 out.write("<tr>\r\n");
932 out.write("<th>Metadata:\r\n");
933 out.write("<td class=\"tam\"><b>Completeness</b>\r\n");
934 out.write("</tr><br>\r\n");
935
936 for(int a = 0; a<MetadataData.length; a++){
937 out.write ("<tr>\r\n");
938 if(MetadataData[a][0].equalsIgnoreCase("dublin")){
939 out.write ("<th><a href =\""+MetadataData[a][0]+".html\">"+metadataSet1+"</a>");
940 }
941 else if (MetadataData[a][0].equalsIgnoreCase("extracted")){
942 out.write ("<th><a href =\""+MetadataData[a][0]+".html\">"+metadataSet2+"</a>");
943 }
944 else{
945 out.write ("<th><a href =\""+MetadataData[a][0]+".html\">"+MetadataData[a][0]+"</a>");
946 }
947 out.write ("<td class=\"tam\"> "+MetadataData[a][1]);
948 out.write ("</tr>\r\n");
949 }
950
951 out.write ("</table>");
952 out.write("<br>\r\n");
953 out.write("<FORM name=\"test\" onsubmit=\"checkStatus()\" action=\"\">");
954 out.write("<table border=1 align=\"center\">\r\n");
955 out.write ("<colgroup width=\"606\" span=\"1\">\r\n");
956 out.write("<tr>\r\n<th>Customize Visualization");
957 out.write("<tr>\r\n<td><LABEL FOR=\"H1\"><INPUT align=\"left\" TYPE=\"checkbox\" ID=\"H1\">Hide Empty Metadata Elements</LABEL>");
958 out.write("<tr>\r\n<td><LABEL FOR=\"H2\"><INPUT align=\"left\" TYPE=\"checkbox\" ID=\"H2\">Hide Completed Metadata Elements</LABEL>");
959 out.write("<tr>\r\n<td><LABEL FOR=\"H3\"><INPUT align=\"left\" TYPE=\"checkbox\" ID=\"H3\">Hide Documents with Empty Metadata Elements</LABEL>");
960 out.write("<tr>\r\n<td><LABEL FOR=\"H4\"><INPUT align=\"left\" TYPE=\"checkbox\" ID=\"H4\">Hide Documents with Completed Metadata Elements</LABEL>");
961 out.write("<tr>\r\n<th>Metadata: ");
962
963 if(MetadataData.length==1){
964 if(MetadataData[0][0].equalsIgnoreCase("dublin")){
965 out.write("<tr>\r\n<td><LABEL FOR=\"C1\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C1\" checked> Dublin Core </LABEL>");
966 }
967 else {
968 out.write("<tr>\r\n<td><LABEL FOR=\"C4\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C4\" checked>"+MetadataData[0][0]+"</LABEL>");
969 }
970 }
971 else{
972 for(int a = 0; a<MetadataData.length; a++){
973 if(a==0){
974 if(MetadataData[a][0].equalsIgnoreCase("dublin")){
975 out.write("<tr>\r\n<td><LABEL FOR=\"C1\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C1\" checked>Dublin Core</LABEL>");
976 }
977 else {
978 out.write("<tr>\r\n<td><LABEL FOR=\"C4\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C4\" checked>"+MetadataData[a][0]+"</LABEL>");
979 }
980 }
981 else{
982 if(MetadataData[a][0].equalsIgnoreCase("dublin")){
983 out.write("<tr>\r\n<td><LABEL FOR=\"C1\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C1\">"+MetadataData[a][0]+"</LABEL>");
984 }
985 else{
986 out.write("<tr>\r\n<td><LABEL FOR=\"C4\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C4\">"+MetadataData[a][0]+"</LABEL>");
987 }
988 }
989 }
990 }
991
992 out.write("<tr>\r\n<th>Order By Completeness : ");
993 out.write("<tr>\r\n<td><LABEL FOR=\"R1\"><INPUT align=\"left\" TYPE=\"radio\" name = \"order\" ID=\"R1\">Best Case to Worst Case</LABEL>");
994 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>");
995 out.write("</table><p align=\"center\"><INPUT TYPE=\"button\" VALUE=\"Show Visualization\" onClick=\"checkStatus()\"> </FORM>");
996 out.write("<p align=\"center\">"+str+"</p>");
997 out.write("</body></html>");
998
999 out.close();
1000 fstream.close();
1001 }catch (Exception e){
1002 System.err.println("Error: " + e.getMessage());
1003 }
1004 printWriter.write("<!-- generating all possible page-->");
1005 printWriter.flush();
1006 }
1007
1008 public void WriteMetadataSetDetailHTML(MetadataSet mdset){
1009
1010 String fileName = mdset.getName();
1011
1012 try{
1013 printWriter.write("<!-- generating all possible page-->");
1014 printWriter.flush();
1015
1016 String str = sdf.format(cl.getTime());
1017 String collectionFolder = mds.getCollectionName();
1018
1019 FileWriter fstream = new FileWriter(destination+fileName+".html");
1020 BufferedWriter out = new BufferedWriter(fstream);
1021
1022 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
1023 out.write("<html>\r\n");
1024 out.write("<head>\r\n<title>Metadata Detail</title>\r\n");
1025 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
1026 out.write("<style type=\"text/css\">\r\n");
1027 out.write("td{height:20;text-align:right;}\r\n");
1028 out.write("th{height:20;text-align:left;}\r\n");
1029 out.write("body{font-family: Arial;}\r\n");
1030 out.write("</style>\r\n");
1031 out.write( "</head>\r\n");
1032 out.write ("<body>\r\n");
1033 out.write ("<td> <div style=\"float:left;\"> <a href=\" Overall.html \">Summary</a></div>");
1034 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>");
1035 if(fileName.equalsIgnoreCase("dublin")){
1036 out.write ("<h3 align=\"center\">Metadata Detail: "+metadataSet1+"</h3>\r\n");
1037 }
1038 else if (fileName.equalsIgnoreCase("extracted")){
1039 out.write ("<h3 align=\"center\">Metadata Detail: "+metadataSet2+"</h3>\r\n");
1040 }
1041 else{
1042 out.write ("<h3 align=\"center\">Metadata Detail: "+fileName+"</h3>\r\n");
1043 }
1044
1045 out.write ("<table border=\"1\" align=\"center\">\r\n");
1046 out.write ("<colgroup width=\"200\" span=\"2\">\r\n");
1047 out.write ("<tr>\r\n");
1048 out.write ("<th>Elements:\r\n");
1049 out.write ("<td><b>Completeness</b>\r\n");
1050 out.write ("</tr>\r\n");
1051
1052 ArrayList elementList = mdset.getIndexsList();
1053 HashMap hp = new HashMap();
1054 int num = elementList.size();
1055
1056 for(int i = 0; i<num; i++){
1057 String elementName = (String)elementList.get(i);
1058 Double elementValue = new Double(dm.Mean(elementName));
1059 hp.put(elementName, elementValue);
1060 }
1061
1062 elementList = new ArrayList();
1063 elementList = dm.sortMap(hp);
1064 num = elementList.size();
1065
1066 for(int i = 0; i<num; i++){
1067 Map.Entry entry = (Map.Entry) elementList.get(i);
1068 String elementName = (String) entry.getKey();
1069 out.write ("<tr>\r\n");
1070 out.write ("<th><a href=\""+ elementName +".html\">"+elementName+"</a>\r\n");
1071 out.write ("<td>"+ dm.Mean(elementName)+"%\r\n");
1072 out.write ("</tr>\r\n");
1073 }
1074
1075 out.write("</table>\r\n");
1076 out.write ("<p align=\"center\"> <a href=\" Overall.html \">Summary</a></p>");
1077 out.write("<p align=\"center\">"+str+"</p>");
1078 out.write ("</body></html>\r\n");
1079
1080 out.close();
1081 fstream.close();
1082 }catch (Exception e){
1083 e.printStackTrace(printWriter);
1084 }
1085
1086 printWriter.write("<!-- generating all possible page-->");
1087 printWriter.flush();
1088
1089 }
1090
1091 public void generateMetadataElementDetailPage(MetadataSet mds,PrintWriter out){
1092
1093 ArrayList nameList = mds.getIndexsList();
1094
1095 for(int i = 0; i<nameList.size();i++){
1096 WriteMetadataElementDetailHTML((String)nameList.get(i),mds.getName(),out);
1097 }
1098 }
1099
1100 public void WriteMetadataElementDetailHTML(String name, String linkName,PrintWriter outx){
1101
1102 String fileName = name;
1103
1104 try{
1105 printWriter.write("<!-- generating all possible page-->");
1106 printWriter.flush();
1107
1108 String collectionFolder = mds.getCollectionName();
1109
1110 FileWriter fstream = new FileWriter(destination+fileName+".html");
1111 BufferedWriter out = new BufferedWriter(fstream);
1112
1113 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
1114 out.write ("<html>\r\n");
1115 out.write("<head>\r\n<title> "+ name +" </title>\r\n");
1116 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
1117 out.write("<style type=\"text/css\">\r\n");
1118 out.write("td{height:20; text-align:left;}\r\n");
1119 out.write(".tam{height:20; text-align:center}\r\n");
1120 out.write("th{height:20; text-align:left;}\r\n");
1121 out.write("body{font-family: Arial;}\r\n");
1122 out.write("</style>\r\n");
1123 out.write("<script type=\"text/javascript\" src=\"http://"+hostName+":"+port_number+"/greenstone3/mat/script/status3.js\"></script>");
1124 out.write("</head>\r\n");
1125 out.write("<body>\r\n");
1126 out.write ("<td> <div style=\"float:left;\"> <a href=\" Overall.html \">Summary</a></div>");
1127 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>");
1128 out.write("&raquo;");
1129
1130 if(linkName.equalsIgnoreCase("dublin")){
1131 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+metadataSet1+")</a>");
1132 }
1133 else if (linkName.equalsIgnoreCase("extracted")){
1134 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+metadataSet2+")</a>");
1135 }
1136 else{
1137 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+linkName+")</a>");
1138 }
1139
1140 if(name.startsWith("dc.")){
1141 int dot = name.lastIndexOf('.');
1142 dot++;
1143 String nameReplace = name.substring(dot);
1144
1145 if(url.containsKey(nameReplace)){
1146 out.write ("<h3 align=\"center\">Metadata Element Detail:<a href=\""+ url.get(nameReplace) +"\">"+ name +"</a></h3>\r\n");
1147 }
1148 else{
1149 out.write ("<h3 align=\"center\">Metadata Element Detail: "+ name +"</a></h3>\r\n");
1150 }
1151 }
1152 else{
1153 out.write ("<h3 align=\"center\">Metadata Element Detail: "+ name +"</h3>\r\n");
1154 }
1155
1156 out.write ("<table border=\"1\" align=\"center\" width=1000>\r\n");
1157 out.write ("<colgroup width=\"500\" span=\"2\">\r\n");
1158 out.write ("<tr>\r\n");
1159 out.write ("<td><b>Total Number of Records</b>\r\n");
1160 out.write ("<td class=\"tam\">"+mds.getDocNum()+"\r\n");
1161 out.write ("</tr>\r\n");
1162 out.write ("<tr>\r\n");
1163 out.write ("<td><b>Unique Values</b>\r\n");
1164 out.write ("<td class=\"tam\">"+dm.getDistinctNumber(name)+"\r\n");
1165 out.write ("</tr>\r\n");
1166 out.write ("<tr>\r\n");
1167 out.write ("<td><b>Total times element used</b>\r\n");
1168 out.write ("<td class=\"tam\">"+dm.getFrequency(name) +"\r\n");
1169 out.write ("</tr>\r\n");
1170 out.write ("<tr>\r\n");
1171 out.write ("<td><b>No. of records containing element</b>\r\n");
1172 out.write ("<td class=\"tam\"> "+dm.getDocumentUsedElement(name)+"\r\n");
1173 out.write ("</tr>\r\n");
1174 out.write ("<tr>\r\n");
1175
1176 double percentage = dm.Mean(name);
1177
1178 out.write ("<th title = \"The completeness means the arithmetic average \"> <div style=\"float:left;\"> Completeness</div>");
1179 out.write ("<td class=\"tam\"> "+dm.Mean(name) +"%\r\n");
1180 out.write ("</tr>\r\n");
1181 out.write ("<tr>\r\n");
1182 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");
1183 out.write ("<td class=\"tam\"> "+dm.getMinRange(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;\">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");
1188 out.write ("<td class=\"tam\"> "+dm.getMaxRange(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;\">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");
1193 out.write ("<td class=\"tam\"> "+dm.Average(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;\">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");
1198 out.write ("<td class=\"tam\"> "+dm.getMode(name) +"\r\n");
1199 out.write ("</tr>\r\n");
1200
1201 out.write ("<tr>\r\n");
1202 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");
1203 out.write ("<td class=\"tam\">"+dm.ModeFrequency(name) +"%\r\n");
1204 out.write ("</tr>\r\n");
1205
1206 if(mds.getOaiPrefix().equalsIgnoreCase("oai_dc")){
1207
1208 HashMap suggestionMap = generateMetadataElementSortList(fileName,"ASCII",linkName,outx);
1209
1210 if(suggestionMap.size()>0){
1211 boolean status = compareElement(fileName,collectionFolder,suggestionMap,linkName);
1212 if(status){
1213 out.write("<tr><td class=\"tam\"><a href =\""+fileName+"_Suggestion.html\">View Potential Duplicate List</a>");
1214 if(percentage<100){
1215 out.write ("<td class=\"tam\"><a href =\""+fileName+"_IncompletedList.html\">Records missing "+fileName+"</a></td>");
1216 createIncompletedList(fileName,linkName,collectionFolder,outx);
1217 }
1218 else{
1219 out.write("<td class=\"tam\">No Records Missing "+ fileName);
1220 }
1221 }
1222 else{
1223 out.write("<tr><td class=\"tam\">No Potential Duplicates");
1224 if(percentage<100){
1225 out.write ("<td class=\"tam\"><a href =\""+fileName+"_IncompletedList.html\">Records missing "+fileName+"</a></td>");
1226 createIncompletedList(fileName,linkName,collectionFolder,outx);
1227 }
1228 else{
1229 out.write("<td class=\"tam\">No Records Missing "+ fileName);
1230 }
1231 }
1232 }
1233 else{
1234 out.write("<tr><td class=\"tam\">No Potential Duplicates");
1235 if(percentage<100){
1236 out.write ("<td class=\"tam\"><a href =\""+fileName+"_IncompletedList.html\">Records missing "+fileName+"</a></td>");
1237 createIncompletedList(fileName,linkName,collectionFolder,outx);
1238 }
1239 else{
1240 out.write("<td class=\"tam\">No Records Missing "+ fileName);
1241 }
1242 }
1243 generateMetadataElementSortList(fileName,"Frequency-based",linkName,outx);
1244 }
1245 else{
1246 HashMap suggestionMap = generateMetadataElementSortListNoIdentifier(fileName,"ASCII",linkName,outx);
1247 generateMetadataElementSortListNoIdentifier(fileName,"Frequency-based",linkName,outx);
1248 }
1249
1250 out.write ("<tr>\r\n");
1251 out.write ("<td class=\"tam\"><a href =\""+fileName+"_Frequency-based.html\">"+"View Full Frequency Sorted list</a>" +
1252 " <td class=\"tam\"><a href =\""+fileName+"_ASCII.html\">"+"View Full ASCII Sorted list</a></td>");
1253 out.write ("</tr>\r\n");
1254 out.write ("</table><br>\r\n");
1255 out.write ("<table border=\"1\" align=\"center\" width=1000>\r\n");
1256 out.write ("<colgroup width=\"500\" span=\"2\">\r\n");
1257 out.write ("<tr>\r\n");
1258 out.write ("<th>ASCII-Based\r\n");
1259 out.write ("<th>First Five\r\n");
1260 out.write ("</tr>\r\n");
1261
1262 String[] temp = dm.getSortList(name,"ASCII");
1263 String[] temp2 = {"&nbsp;","&nbsp;","&nbsp;","&nbsp;","&nbsp;"};
1264 int length = 0;
1265 if(temp.length>=5){
1266 length=5;
1267 }
1268 else if(temp.length<5){
1269 length = temp.length;
1270 }
1271
1272 for(int i =0; i<length; i++){
1273 temp2[i] = temp[i];
1274 }
1275
1276 int x = temp2.length;
1277 for(int a = 0; a<x; a++){
1278 out.write ("<tr>\r\n");
1279 if(!temp2[a].equals("&nbsp;")){
1280 out.write ("<th>"+(a+1)+"\r\n");
1281 }
1282 else{
1283 out.write ("<th>&nbsp;\r\n");
1284 }
1285 if(temp2[a].startsWith("http://")){
1286 if(temp2[a].length()>60){
1287 out.write ("<td><a href = \""+temp2[a]+"\">"+temp2[a].substring(0, 60)+"...</a>");
1288 }
1289 else {
1290 out.write ("<td><a href = \""+temp2[a]+"\">"+temp2[a]+"</a>");
1291 }
1292 }
1293 else{
1294 char singleChar = temp2[a].charAt(0);
1295 if(temp2[a].length()>61 ){
1296 if(temp2[a].startsWith(" ") && ((int)singleChar!=65279)){
1297 temp2[a] = temp2[a].substring(1);
1298 out.write ("<td>"+spaceLeft+space+spaceRight+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#topFive\"> ... </a>\r\n");
1299 }
1300 else if (((int)singleChar==65279)){
1301 temp2[a] = temp2[a].substring(1);
1302 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#topFive\"> ... </a>\r\n");
1303 }
1304 else{
1305 out.write ("<td>"+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#topFive\"> ... </a>\r\n");
1306 }
1307 }
1308 else {
1309 if(temp2[a].startsWith(" ")&& ((int)singleChar!=65279)){
1310 temp2[a] = temp2[a].substring(1);
1311 out.write ("<td>"+spaceLeft+space+spaceRight+temp2[a]+"\r\n");
1312 }
1313 else if (((int)singleChar==65279)){
1314 temp2[a] = temp2[a].substring(1);
1315 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp2[a]+"\r\n");
1316 }
1317 else{
1318 out.write ("<td>"+temp2[a]+"\r\n");
1319 }
1320 }
1321 }
1322 out.write ("</tr>\r\n");
1323 }
1324
1325 String[] temp3 ={"&nbsp;","&nbsp;","&nbsp;","&nbsp;","&nbsp;"};
1326 length = 0;
1327 int start = temp.length;
1328
1329 if(temp.length>=5){
1330 length= 5;
1331 }
1332 else if(temp.length<5){
1333 length = temp.length;
1334 }
1335
1336 for(int i = length; i>0; i--){
1337 temp3[i-1] = temp[start-1];
1338 start--;
1339 }
1340
1341 out.write ("<tr>\r\n");
1342 out.write ("<th>......\r\n");
1343 out.write ("<th>Last Five\r\n");
1344 out.write ("</tr>\r\n");
1345
1346 int counter = temp.length;
1347 start = temp.length;
1348 x = temp3.length;
1349
1350 for(int a = 0; a<x; a++){
1351 out.write ("<tr>\r\n");
1352
1353 if(!temp3[a].equals("&nbsp;")){
1354 out.write ("<th>"+(start-length+1+a));
1355 }
1356 else{
1357 out.write ("<th>&nbsp;");
1358 }
1359 if(temp3[a].startsWith("http://")){
1360 if(temp3[a].length()>60){
1361 out.write ("<td><a href = \""+temp3[a]+"\">"+temp3[a].substring(0, 60)+"...</a>");
1362 }
1363 else {
1364 out.write ("<td><a href = \""+temp3[a]+"\">"+temp3[a]+"</a>");
1365 }
1366 }
1367 else{
1368 char singleChar = temp3[a].charAt(0);
1369 if(temp3[a].length()>61){
1370 if(temp3[a].startsWith(" ") && (int)singleChar!=65279){
1371 temp3[a] = temp3[a].substring(1);
1372 out.write ("<td>"+spaceLeft+space+spaceRight+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#lastFive\"> ... </a>\r\n");
1373 }
1374 else if((int)singleChar==65279){
1375 temp3[a] = temp3[a].substring(1);
1376 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#lastFive\"> ... </a>\r\n");
1377
1378 }
1379 else{
1380 out.write ("<td>"+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#lastFive\"> ... </a>\r\n");
1381 }
1382 }
1383 else{
1384 if(temp3[a].startsWith(" ")&& (int)singleChar!=65279){
1385 temp3[a] = temp3[a].substring(1);
1386 out.write ("<td>"+spaceLeft+space+spaceRight+temp3[a]+"\r\n");
1387 }
1388 else if((int)singleChar==65279){
1389 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp3[a]+"\r\n");
1390 }
1391 else{
1392 out.write ("<td>"+temp3[a]+"\r\n");
1393 }
1394 }
1395 out.write ("</tr>\r\n");
1396 }
1397 }
1398 out.write ("</table><br>\r\n");
1399 out.write ("<table border=\"1\" align=\"center\" width=1000>\r\n");
1400 out.write ("<colgroup width=\"500\" span=\"2\">\r\n");
1401 out.write ("<tr>\r\n");
1402 out.write ("<th>Frequency-Based:\r\n");
1403 out.write ("<th>First Five\r\n");
1404 out.write ("</tr>\r\n");
1405
1406 HashMap xMap = dm.getDistinctValueMap(name);
1407 temp = dm.getSortList(name,"Frequency-based");
1408 temp2 = new String[] {"&nbsp;","&nbsp;","&nbsp;","&nbsp;","&nbsp;"};
1409 length = 0;
1410
1411 if(temp.length>=5){
1412 length=5;
1413 }
1414 else if(temp.length<5){length = temp.length;}
1415
1416 for(int i =0; i<length; i++){
1417 temp2[i] = temp[i];
1418 }
1419
1420 ArrayList tempList = new ArrayList();
1421
1422 for(int i =0; i<length; i++){
1423 if(((Integer)xMap.get(temp2[i])).intValue()==1){
1424 tempList.add(temp2[i]);
1425 }
1426 }
1427
1428 x = temp2.length;
1429 for(int a = 0; a<x; a++){
1430 out.write ("<tr>\r\n");
1431
1432 if(!temp2[a].equals("&nbsp;")){
1433 char singleChar = temp2[a].charAt(0);
1434 out.write ("<th>"+(a+1)+". (No. of occurrences: "+((Integer)xMap.get(temp2[a])).toString()+")\r\n");
1435 if(temp2[a].startsWith("http://") && (int)singleChar != 65279){
1436 if(temp2[a].length()>61){
1437 out.write ("<td><a href = \""+temp2[a]+"\">"+temp2[a].substring(0, 60)+"...</a>");
1438 }
1439 else {
1440 out.write ("<td><a href = \""+temp2[a]+"\">"+temp2[a]+"</a>");
1441 }
1442 }
1443
1444 else{
1445 if(temp2[a].length()>61){
1446
1447 if(temp2[a].startsWith(" ")){
1448 temp2[a] = temp2[a].substring(1);
1449 out.write ("<td>"+spaceLeft+space+spaceRight+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#topFive\"> ... </a>\r\n");
1450 }
1451
1452 else if((int)singleChar == 65279){
1453 temp2[a] = temp2[a].substring(1);
1454 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#topFive\"> ... </a>\r\n");
1455 }
1456 else{
1457 out.write ("<td>"+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#topFive\"> ... </a>\r\n");
1458 }
1459
1460 }
1461 else{
1462 if(temp2[a].startsWith(" ")){
1463 temp2[a] = temp2[a].substring(1);
1464 out.write ("<td>"+spaceLeft+space+spaceRight+temp2[a]+"\r\n");
1465 }
1466 else if((int)singleChar == 65279){
1467 temp2[a] = temp2[a].substring(1);
1468 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp2[a]+"\r\n");
1469 }
1470 else{
1471 out.write ("<td>"+temp2[a]+"\r\n");
1472 }
1473 }
1474 }
1475 }
1476 else{
1477 out.write ("<th>&nbsp;\r\n");
1478 out.write ("<td>\r\n");
1479 }
1480 out.write ("</tr>\r\n");
1481 }
1482
1483 temp3 = new String[]{"&nbsp;","&nbsp;","&nbsp;","&nbsp;","&nbsp;"};
1484 length = 0;
1485 start = temp.length;
1486 if(temp.length>=5){length= 5;}
1487 else if(temp.length<5){length = temp.length;}
1488
1489 for(int i = length; i>0; i--){
1490 temp3[i-1] = temp[start-1];
1491 start--;
1492 }
1493
1494 out.write ("<tr>\r\n");
1495 out.write ("<th>......\r\n");
1496 out.write ("<th>Last Five\r\n");
1497 out.write ("</tr>\r\n");
1498
1499 x = temp3.length;
1500 start = temp.length;
1501 for(int a = 0; a<x; a++){
1502 out.write ("<tr>\r\n");
1503 if(!temp3[a].equals("&nbsp;")){
1504
1505
1506 out.write ("<th>"+(start-length+1+a)+". (No. of occurrences: "+((Integer)xMap.get(temp3[a])).toString()+")\r\n");
1507
1508 char singleChar = temp3[a].charAt(0);
1509 if(temp3[a].startsWith("http://")){
1510 if(temp3[a].length()>60){
1511 out.write ("<td><a href = \""+temp3[a]+"\">"+temp3[a].substring(0, 60)+"...</a>");
1512 }
1513 else {
1514 out.write ("<td><a href = \""+temp3[a]+"\">"+temp3[a]+"</a>");
1515 }
1516 }
1517 else{
1518 if(temp3[a].length()>61){
1519
1520 if(temp3[a].startsWith(" ")){
1521 temp3[a] = temp3[a].substring(1);
1522 out.write ("<td>"+spaceLeft+space+spaceRight+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#lastFive\"> ... </a>\r\n");
1523 }
1524 else if((int)singleChar == 65279){
1525 temp3[a] = temp3[a].substring(1);
1526 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#lastFive\"> ... </a>\r\n");
1527
1528 }
1529 else{
1530 out.write ("<td>"+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#lastFive\"> ... </a>\r\n");
1531 }
1532
1533 }
1534 else{
1535 if(temp3[a].startsWith(" ")){
1536 out.write ("<td>"+spaceLeft+space+spaceRight+temp3[a]+"\r\n");
1537 }
1538
1539 else if((int)singleChar == 65279){
1540 temp3[a] = temp3[a].substring(1);
1541 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp3[a]+"\r\n");
1542
1543 }
1544
1545 else{
1546 out.write ("<td>"+temp3[a]+"\r\n");
1547 }
1548 }
1549 }
1550
1551 }
1552 else{
1553 //out.write ("<th>"+(a+1)+"\r\n");
1554 out.write ("<th>&nbsp;\r\n");
1555 out.write ("<th>\r\n");
1556 }
1557 out.write ("</tr>\r\n");
1558 }
1559
1560
1561 out.write ("</table>\r\n");
1562
1563 ///////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
1564 out.write ("<p> <a href=\" Overall.html \">Summary</a>");
1565 out.write ("&raquo;");
1566
1567 if(linkName.equalsIgnoreCase("dublin")){
1568 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+metadataSet1+")</a>");
1569
1570 }
1571 else if (linkName.equalsIgnoreCase("extracted")){
1572 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+metadataSet2+")</a>");
1573
1574 }
1575 else{
1576 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+linkName+")</a>");
1577 }
1578 //out.write(str+"</p>");
1579 out.write ("</body></html>\r\n");
1580
1581 //Close the output stream
1582 out.close();
1583 fstream.close();
1584 }catch (Exception e){
1585
1586 e.printStackTrace(printWriter);
1587 }
1588
1589 printWriter.write("<!-- generating all possible page-->");
1590 printWriter.flush();
1591 //System.out.println("detail ends");
1592
1593 }
1594
1595
1596 private HashMap generateMetadataElementSortList(String title, String sort, String metadataSetName,PrintWriter outx){
1597
1598 String fileName = title+"_"+sort;
1599 String collectionFolder = mds.getCollectionName();
1600
1601 String IDENTIFIER = linkIdentifier;
1602 SearchLink sl = new SearchLink(mds.StatsDirectory);
1603 HashMap suggestionMap = new HashMap();
1604
1605 HashMap valueMap = sl.createValueMap(title,collectionFolder);
1606
1607 HashMap linkMap = sl.createLinkMap(IDENTIFIER,collectionFolder);
1608 HashMap internalIDMap = dm.getInternalIdentifier(title);
1609 boolean status = false;
1610
1611 try{
1612 printWriter.write("<!-- generating all possible page-->");
1613 printWriter.flush();
1614
1615 FileWriter fstream = new FileWriter(destination+fileName+".html");
1616 BufferedWriter out = new BufferedWriter(fstream);
1617
1618 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
1619 out.write ("<html>\r\n");
1620 out.write("<head>\r\n<title>Metadata Element Sort List</title>\r\n");
1621 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
1622 out.write("<script type=\"text/javascript\" src=\"../script/getInfomation.js\"></script>\r\n");
1623 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/yahoo/yahoo-min.js\"></script>\r\n");
1624 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/event/event-min.js\"></script>\r\n");
1625 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/connection/connection-min.js\"></script>\r\n");
1626
1627 out.write("<style type=\"text/css\">\r\n");
1628 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
1629 out.write(".tam{height:20; text-align:center}\r\n");
1630 out.write("th{height:20; text-align:center;}\r\n");
1631 out.write("body{font-family: Arial;}\r\n");
1632 out.write("</style>\r\n");
1633 out.write("</head><body id=\""+collectionFolder+"\">\r\n");
1634
1635 out.write("<td><div style=\"float:left;\"><a href=\" Overall.html \">Summary</a>");
1636 out.write("&raquo;");
1637 ////////////////////////////////////////////////////////////////////////////////////
1638 if(metadataSetName.equalsIgnoreCase("dublin")){
1639 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
1640 }
1641 else if (metadataSetName.equalsIgnoreCase("extracted")){
1642 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
1643 }
1644 else{
1645 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
1646 }
1647 out.write("&raquo;");
1648 out.write("<a href=\""+title+".html\">"+title+"</a></div>");
1649 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>");
1650
1651
1652 String[] list = dm.getSortList(title, sort);
1653
1654
1655 if(list.length>=1){
1656
1657
1658 if(sort.equals("ASCII")){
1659
1660 out.write("<table border=\"1\" align=\"center\" width=\"800\">\r\n");
1661 out.write ("<h2 align=\"center\">"+ title+"</h2>\r\n");
1662 out.write ("<th class=\"tam\">ASCII Sort\r\n");
1663 out.write ("<th class=\"tam\">Element Values\r\n");
1664 out.write ("<th class=\"tam\">Source Documents\r\n");
1665 out.write ("<th class=\"tam\">Internal Link\r\n");
1666
1667 out.write ("<a name='topFive'>\r\n");
1668
1669 int counter = 0;
1670
1671 for(int i = 0; i<list.length; i++){
1672 if(list.length<=5 && i==0){
1673 out.write ("<a name='lastFive'>\r\n");
1674 }
1675 else if((list.length>5) && (list.length-5==i)){
1676 out.write ("<a name='lastFive'>\r\n");
1677 }
1678
1679
1680 InternalLink il= (InternalLink)internalIDMap.get(list[i]);
1681 ArrayList alist2 = il.retrieveList();
1682 String id = (String)alist2.get(0);
1683 id = id.substring(4);
1684 out.write("<tr id=\""+id+"\" >\r\n");
1685
1686
1687 if(list[i].length()>=201){
1688 if(list[i].startsWith("http://")){
1689
1690 if(title.equals(IDENTIFIER)){
1691 if(list[i].length()>=100){
1692 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1693
1694 }
1695 else{
1696 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1697 }
1698 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
1699 }
1700 else{
1701 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1702
1703 if(list[i].length()>=100){
1704 String url = (String)alist.get(0);
1705 if(alist.size()==1){
1706 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1707 if(url.startsWith("http://")){
1708 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1709 }else{
1710 out.write("<td>Source\r\n");
1711 }
1712 }
1713 else if(alist.size()>1){
1714 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1715 if(url.startsWith("http://")){
1716 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1717 }else{
1718 out.write("<td>Source\r\n");
1719 }
1720
1721 }
1722 else{
1723 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1724 out.write("<td>No Source Available\r\n");
1725 }
1726 suggestionMap.put(list[i], url);
1727 }
1728 else{
1729
1730 String url = (String)alist.get(0);
1731 if(alist.size()==1){
1732 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1733 //out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1734 if(url.startsWith("http://")){
1735 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1736 }else{
1737 out.write("<td>Source\r\n");
1738 }
1739 }
1740 else if(alist.size()>1){
1741 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1742 //out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1743 if(url.startsWith("http://")){
1744 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1745 }else{
1746 out.write("<td>Source\r\n");
1747 }
1748 }
1749 else{
1750 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1751 out.write("<td>No Source Available\r\n");
1752 }
1753 suggestionMap.put(list[i], url);
1754 }
1755 }
1756
1757
1758
1759 }
1760 else{
1761
1762 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1763
1764 //////////////////////////////////////
1765
1766 if(list[i].startsWith(" ")){
1767 String elements = list[i];
1768 list[i] = list[i].substring(1);
1769
1770 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 if(alist.size()>1){
1776 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i].substring(0, 200)+"...<td>"+
1777 "<a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1778 suggestionMap.put(elements, (String)alist.get(0));
1779 }
1780 else{
1781 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i].substring(0, 200)+"...<td>No Source Available \r\n");
1782 suggestionMap.put(elements, "No Source Available");
1783 }
1784 }
1785 else{
1786
1787 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 if(alist.size()>1){
1793 out.write("<td>"+(counter+1)+"<td>"+list[i].substring(0, 200)+"...<td>" +
1794 "<a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1795 suggestionMap.put(list[i], (String)alist.get(0));
1796 }
1797 else{
1798 out.write("<td>"+(counter+1)+"<td>"+list[i].substring(0, 200)+"... <td>No Source Available\r\n");
1799 suggestionMap.put(list[i],"No Source Available");
1800 }
1801 }
1802 }
1803
1804 }
1805 else{
1806 if(list[i].startsWith("http://")){
1807
1808 if(title.equals(IDENTIFIER)){
1809 if(list[i].length()>=100){
1810 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1811
1812 }
1813 else{
1814 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1815 }
1816 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
1817 }
1818 else{
1819 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1820
1821 if(list[i].length()>=100){
1822 if(alist.size()==1){
1823 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1824 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1825 }
1826 else if(alist.size()>1){
1827 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1828 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1829 }
1830 else{
1831 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1832 out.write("<td>No Source Available\r\n");
1833 }
1834 }
1835 else{
1836 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 if(alist.size()>1){
1842 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1843 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1844 suggestionMap.put(list[i], (String)alist.get(0));
1845 }
1846 else{
1847 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1848 out.write("<td>No Source Available\r\n");
1849 suggestionMap.put(list[i],"No Source Available");
1850 }
1851 }
1852 }
1853
1854 }
1855
1856
1857 else{
1858
1859 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1860
1861 if(list[i].startsWith(" ")){
1862 String elements = list[i];
1863 list[i] = list[i].substring(1);
1864
1865 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 if(alist.size()>1){
1871 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i]+
1872 "<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1873 suggestionMap.put(list[i], (String)alist.get(0));
1874 }
1875 else{
1876 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i]+"<td>No Source Available\r\n");
1877 suggestionMap.put(elements,"No Source Available");
1878 }
1879
1880 }
1881 else{
1882 if(alist.size()==1){
1883 out.write("<td>"+(counter+1)+"<td>"+list[i]+"<td><a href=\""+alist.get(0)+"\">Source</a>\r\n");
1884 suggestionMap.put(list[i],(String)alist.get(0));
1885 }
1886 else if(alist.size()>1){
1887 out.write("<td>"+(counter+1)+"<td>"+list[i]+"<td><a href=\""+alist.get(0)+"\">Source</a>...\r\n");
1888 suggestionMap.put(list[i],(String)alist.get(0));
1889 }
1890 else{
1891 out.write("<td>"+(counter+1)+"<td>"+list[i]+"<td>No Source Available\r\n");
1892 suggestionMap.put(list[i],"No Source Available");
1893 }
1894 //out.write("<td>"+(counter+1)+"<td>"+list[i]+"\r\n");
1895 }
1896 }
1897 }
1898 counter++;
1899 out.write("<td onclick=\"GD(this)\">View");
1900 }
1901 if(suggestionMap.size()>1 && !title.equals(IDENTIFIER)){
1902 status = true;
1903 }
1904 }
1905 else{
1906
1907
1908 out.write("<table border=\"1\" align=\"center\" width=\"800\">\r\n");
1909 out.write ("<h2 align=\"center\">"+ title+"</h2>\r\n");
1910 out.write ("<th class=\"tam\">&nbsp;\r\n");
1911 out.write ("<th class=\"tam\">Frequency\r\n");
1912 out.write ("<th class=\"tam\">Element Values\r\n");
1913 out.write ("<th class=\"tam\">Source Documents\r\n");
1914 out.write ("<th class=\"tam\">Internal Link\r\n");
1915 out.write ("<a name='topFive'>\r\n");
1916
1917 HashMap xMap = dm.getDistinctValueMap(title);
1918 int counter = 0;
1919 for(int i = 0; i<list.length; i++){
1920 if(list.length<=5 && i==0){
1921 out.write ("<a name='lastFive'>\r\n");
1922 }
1923 else if((list.length>5) && (list.length-5==i)){
1924 out.write ("<a name='lastFive'>\r\n");
1925 }
1926
1927 InternalLink il= (InternalLink)internalIDMap.get(list[i]);
1928 ArrayList alist2 = il.retrieveList();
1929 String id = (String)alist2.get(0);
1930 id = id.substring(4);
1931 out.write("<tr id=\""+id+"\" >\r\n");
1932
1933 out.write("<td>"+(counter+1)+"<th> "+((Integer)xMap.get(list[i])).toString()+"\r\n");
1934
1935 if(list[i].length()>=201){
1936
1937 if(list[i].startsWith("http://")){
1938 if(title.equals(IDENTIFIER)){
1939 if(list[i].length()>=100){
1940 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1941 }
1942 else{
1943 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1944 }
1945
1946 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
1947 }
1948 else{
1949 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1950 if(list[i].length()>=100){
1951 if(alist.size()==1){
1952 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1953 out.write("<td><a href=\""+alist.get(0)+"\">Source</a>\r\n");
1954 }
1955 else if (alist.size()>1){
1956 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1957 out.write("<td><a href=\""+alist.get(0)+"\">Source</a>...\r\n");
1958 }
1959 else{
1960 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1961 out.write("<td>No Source Available\r\n");
1962 }
1963
1964 }
1965 else{
1966 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1967 }
1968 }
1969 }
1970
1971
1972
1973
1974 else{
1975 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1976
1977 if(list[i].startsWith(" ")){
1978 list[i]=list[i].substring(1);
1979 if(alist.size()==1){
1980 out.write("<td>"+spaceLeft+space+spaceRight+list[i].substring(0,200)+"...<td>"+
1981 "<a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1982 }
1983 else if(alist.size()>1){
1984 out.write("<td>"+spaceLeft+space+spaceRight+list[i].substring(0,200)+"...<td>"+
1985 "<a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1986 }
1987 else{
1988 out.write("<td>"+spaceLeft+space+spaceRight+list[i].substring(0,200)+"...<td>No Source Available\r\n");
1989 }
1990 }
1991 else{
1992
1993 if(alist.size()==1){
1994 out.write("<td>"+list[i].substring(0,200)+"...<td>" +
1995 "<a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1996 }
1997 else if(alist.size()>1){
1998 out.write("<td>"+list[i].substring(0,200)+"...<td>"+
1999 "<a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
2000 }
2001 else{
2002 out.write("<td>"+list[i].substring(0,200)+"...<td>No Source Available\r\n");
2003 }
2004 }
2005 }
2006 }
2007 else{
2008 if(list[i].startsWith("http://")){
2009 if(title.equals(IDENTIFIER)){
2010 if(list[i].length()>=100){
2011 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2012 }
2013 else{
2014 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2015 }
2016 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2017 }
2018 else{
2019 //ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],"dc.Identifier", collectionFolder,valueMap,linkMap);
2020 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
2021
2022 if(list[i].length()>=100){
2023 if(alist.size()==1){
2024 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2025 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
2026 }
2027 else if(alist.size()>1){
2028 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2029 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
2030 }
2031 else{
2032 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2033 out.write("<td>No Source Available\r\n");
2034 }
2035 }
2036 else{
2037 if(alist.size()==1){
2038 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2039 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
2040 }
2041 else if(alist.size()>1){
2042 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2043 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
2044 }
2045 else{
2046 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2047 out.write("<td>No Source Available\r\n");
2048 }
2049 }
2050
2051 }
2052
2053 }
2054 else{
2055
2056 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
2057
2058 if(list[i].startsWith(" ")){
2059 list[i]=list[i].substring(1);
2060 if(alist.size()==1){
2061 out.write("<td>"+spaceLeft+space+spaceRight+list[i]+
2062 "<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
2063 }
2064 else if(alist.size()>1){
2065 out.write("<td>"+spaceLeft+space+spaceRight+list[i]+
2066 "<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
2067 }
2068 else{
2069 out.write("<td>"+spaceLeft+space+spaceRight+list[i]+"<td>No Source Available\r\n");
2070 }
2071 }
2072 else{
2073 if(alist.size()==1){
2074 out.write("<td>"+list[i]+"<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
2075 }
2076 else if(alist.size()>1){
2077 out.write("<td>"+list[i]+"<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
2078 }
2079 else{
2080 out.write("<td>"+list[i]+"<td>No Source Available\r\n");
2081 }
2082 }
2083 }
2084 }
2085 counter++;
2086 out.write("<td onclick=\"GD(this)\">View");
2087 }
2088 }
2089 out.write("</table>\r\n");
2090 }
2091 else{
2092 out.write("<p>Sorry! The list is empty</p>");
2093 }
2094 out.write("<tr>\r\n");
2095
2096
2097 out.write("<p><a href=\" Overall.html \">Summary</a>");
2098 out.write("&raquo;");
2099 ////////////////////////////////////////////////////////////////////////////////////
2100 if(metadataSetName.equalsIgnoreCase("dublin")){
2101 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2102 }
2103 else if (metadataSetName.equalsIgnoreCase("extracted")){
2104 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2105 }
2106 else{
2107 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2108 }
2109 out.write("&raquo;");
2110 out.write("<a href=\""+title+".html\">"+title+"</a></p>");
2111 out.write("</body></html>\r\n");
2112 out.close();
2113 fstream.close();
2114 }catch(Exception e){
2115
2116 e.printStackTrace(outx);
2117 }
2118
2119 if(status){
2120 return suggestionMap;
2121 }
2122 else{
2123 return new HashMap();
2124 }
2125 }
2126
2127
2128 private void setupMetadataLink(String fileName){
2129
2130 String destination = globalProperty.getGSDL3Home()+fileSeparator+"mat"+fileSeparator+"script"+fileSeparator+fileName+".xml";
2131
2132 try{
2133 DocumentBuilderFactory docBuilderFactory = DocumentBuilderFactory.newInstance();
2134 DocumentBuilder docBuilder = docBuilderFactory.newDocumentBuilder();
2135 Document doc = docBuilder.newDocument();
2136 //doc = docBuilder.parse (new File("/research/cc108/greenstone3/web/mat/script/"+fileName+".xml"));
2137 doc = docBuilder.parse (destination);
2138 Element rootNode = doc.getDocumentElement();
2139
2140 NodeList listOfName = rootNode.getElementsByTagName("metadataElement");
2141 url = new HashMap();
2142 for(int i = 0; i<listOfName.getLength(); i++){
2143 Node NameNode = listOfName.item(i);
2144 Element docElement = (Element)NameNode;
2145 NodeList valueList = docElement.getElementsByTagName("URL");
2146 Node urlParentNode = valueList.item(0);
2147 String urlText = urlParentNode.getChildNodes().item(0).getNodeValue();
2148
2149 valueList = docElement.getElementsByTagName("name");
2150 Node urlNameParentNode = valueList.item(0);
2151 String urlNameText = urlNameParentNode.getChildNodes().item(0).getNodeValue();
2152
2153 url.put(urlNameText, urlText);
2154
2155 }
2156
2157
2158 }catch(Exception e){
2159 e.printStackTrace(printWriter);
2160 }
2161 }
2162
2163 private void createIncompletedList(String fileName, String metadataSetName, String collectionFolder,PrintWriter outx){
2164
2165 HashMap hp = dm.getIdentifierLink(linkIdentifier);
2166 String[] ids = dm.getDocumentIDList(fileName);
2167
2168 for(int i = 0; i<ids.length; i++){
2169 hp.remove(ids[i]);
2170 }
2171
2172 try{
2173 FileWriter fstream = new FileWriter(destination+fileName+"_IncompletedList.html");
2174 BufferedWriter out = new BufferedWriter(fstream);
2175
2176 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
2177 out.write ("<html>\r\n");
2178 out.write("<head>\r\n<title>Incompleted Document List</title>\r\n");
2179 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
2180 out.write("<style type=\"text/css\">\r\n");
2181 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
2182 out.write(".tam{height:20; text-align:center}\r\n");
2183 out.write("th{height:20; text-align:center;}\r\n");
2184 out.write("body{font-family: Arial;}\r\n");
2185 out.write("</style>\r\n");
2186 out.write("</head><body>\r\n");
2187
2188 out.write("<p><a href=\" Overall.html \">Summary</a>");
2189 out.write("&raquo;");
2190 if(metadataSetName.equalsIgnoreCase("dublin")){
2191 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2192 }
2193 else if (metadataSetName.equalsIgnoreCase("extracted")){
2194 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2195 }
2196 else{
2197 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2198 }
2199 out.write("&raquo;");
2200 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
2201
2202 out.write("<h3 align=center>"+fileName+" does not appear in the following documents</h3>");
2203 int counter = 0;
2204 Set s = hp.keySet();
2205 Iterator i = s.iterator();
2206 out.write("<table border=1 align=center><tr><th>Document ID<th>Source Link\r\n");
2207 while(i.hasNext()){
2208 counter++;
2209 String keys = (String)i.next();
2210 InternalLink il = (InternalLink) hp.get(keys);
2211 ArrayList alist = il.retrieveList();
2212 String url = (String)alist.get(0);
2213 out.write("<tr><td>"+counter+"<td><a href=\""+url+"\">"+url+"</a>\r\n");
2214 }
2215 out.write("</table></body></html>\r\n");
2216
2217
2218 out.write("<p><a href=\" Overall.html \">Summary</a>");
2219 out.write("&raquo;");
2220 if(metadataSetName.equalsIgnoreCase("dublin")){
2221 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2222 }
2223 else if (metadataSetName.equalsIgnoreCase("extracted")){
2224 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2225 }
2226 else{
2227 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2228 }
2229 out.write("&raquo;");
2230 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
2231
2232 out.close();
2233 fstream.close();
2234 }catch(IOException ex){ex.printStackTrace();}
2235
2236
2237 }
2238
2239 public boolean compareElement(String fileName,String collectionFolder,HashMap suggestionMap, String metadataSetName){
2240
2241 boolean status = false;
2242
2243 Set kset = suggestionMap.keySet();
2244 ArrayList arrayList = new ArrayList();
2245 Iterator is = kset.iterator();
2246 while(is.hasNext()){
2247 arrayList.add((String)is.next());
2248 }
2249 HashMap distanceMap = new HashMap();
2250
2251 int totalLength = 0;
2252
2253 int arrayListLength = arrayList.size();
2254 int counter = 0;
2255
2256 printWriter.write("<!-- generating all possible page-->");
2257 printWriter.flush();
2258 double distance ;
2259 try{
2260 for(int i = 0; i<arrayListLength; i++){
2261 String keyword = (String)arrayList.get(i);
2262
2263 int spaceCounterFront1 = 0;
2264 int spaceCounterEnd1 = 0;
2265 String keywordClone = keyword;
2266 String _keywordClone = keyword;
2267 String bkKeyWord = keyword;
2268
2269 keywordClone = removeUnusedCharacter(keywordClone);
2270
2271 while(true){
2272
2273 if(_keywordClone.length()>1){
2274 if(_keywordClone.charAt(0)==' '){
2275 _keywordClone = _keywordClone.substring(1);
2276 spaceCounterFront1++;
2277 }
2278 else{
2279 if(_keywordClone.charAt(_keywordClone.length()-1)==' '){
2280 _keywordClone = _keywordClone.substring(0,_keywordClone.length()-1);
2281 spaceCounterEnd1++;
2282 }
2283 else{
2284 break;
2285 }
2286 }
2287 }
2288 else{
2289 break;
2290 }
2291 }
2292
2293 int keywordLength = keywordClone.length();
2294 totalLength = totalLength + keywordLength;
2295 for(int j = i+1; j<arrayListLength; j++){
2296 String keyword2 = (String)arrayList.get(j);
2297 counter ++;
2298 double pre_cost = 0;
2299 String keywordClone2 = keyword2;
2300 String _keywordClone2 = keyword2;
2301 String bkKeyWord2 = keyword2;
2302 costModel cm = new costModel();
2303 cm = removeUnusedCharacter(keywordClone2,pre_cost);
2304 keywordClone2 = cm.getString();
2305 pre_cost = cm.getCost();
2306
2307 int spaceCounterFront2 = 0;
2308 int spaceCounterEnd2 = 0;
2309
2310 while(true){
2311 if(_keywordClone2.length()>1){
2312 if(_keywordClone2.charAt(0)==' '){
2313 _keywordClone2 = _keywordClone2.substring(1);
2314 spaceCounterFront2++;
2315 }
2316 else{
2317 if(_keywordClone2.charAt(_keywordClone2.length()-1)==' '){
2318 _keywordClone2 = _keywordClone2.substring(0,_keywordClone2.length()-1);
2319 spaceCounterEnd2++;
2320 }
2321 else{
2322 break;
2323 }
2324 }}
2325 else{
2326 break;
2327 }
2328 }
2329
2330 if(counter == 50000){
2331 printWriter.write("<!-- generating all possible page-->");
2332 printWriter.flush();
2333 counter = 0;
2334 }
2335 int keyword2Length = keywordClone2.length();
2336
2337 if(keywordLength>(keyword2Length+2) || (keywordLength+2)<keyword2Length ){}
2338 else{
2339
2340 distance = calculateEditDistance(keywordClone.toLowerCase().toCharArray(),keywordClone2.toLowerCase().toCharArray());
2341
2342 distance = distance + pre_cost;
2343
2344 if(distance<=2){
2345 if(distanceMap.containsKey(_keywordClone)){
2346 InternalLink il = (InternalLink)distanceMap.get(_keywordClone);
2347
2348 String keywordHolder = _keywordClone2;
2349
2350 while(true){
2351 if(keywordHolder.indexOf(" ")!=-1){
2352 keywordHolder = keywordHolder.replaceFirst(" ",spaceLeft+space+spaceRight+" ");
2353 }
2354 else{
2355 break;
2356 }
2357 }
2358
2359 for(int a = 0; a<spaceCounterFront2; a++){
2360 keywordHolder = spaceLeft+space+spaceRight+keywordHolder;
2361 }
2362 for(int a = 0; a<spaceCounterEnd2; a++){
2363 keywordHolder = keywordHolder+spaceLeft+space+spaceRight;
2364 }
2365
2366 InternalLink il2 = new InternalLink();
2367 il2.setValue(keywordHolder);
2368 if(suggestionMap.containsKey(bkKeyWord2)){
2369 il2.increaseElement((String)suggestionMap.get(bkKeyWord2));
2370 //System.out.println("input this url(3):"+(String)alistkw2.get(0));
2371 }
2372 else{
2373 il2.increaseElement("&nbsp;");
2374 }
2375
2376 il.increaseNode(il2);
2377
2378 distanceMap.put(_keywordClone, il);
2379
2380 }
2381 else{
2382 InternalLink il = new InternalLink();
2383 String keywordHolder = _keywordClone;
2384
2385 while(true){
2386 if(keywordHolder.indexOf(" ")!=-1){
2387 keywordHolder = keywordHolder.replaceFirst(" ",spaceLeft+space+spaceRight+" ");
2388 }
2389 else{
2390 break;
2391 }
2392 }
2393
2394 for(int a = 0; a<spaceCounterFront1; a++){
2395 keywordHolder = spaceLeft+space+spaceRight+keywordHolder;
2396 }
2397 for(int a = 0; a<spaceCounterEnd1; a++){
2398 keywordHolder = keywordHolder+spaceLeft+space+spaceRight;
2399 }
2400
2401 InternalLink il2 = new InternalLink();
2402 il2.setValue(keywordHolder);
2403 if(suggestionMap.containsKey(bkKeyWord)){
2404 il2.increaseElement((String)suggestionMap.get(bkKeyWord));
2405
2406 }
2407 else{
2408 il2.increaseElement("&nbsp;");
2409 }
2410
2411 il.increaseNode(il2);
2412
2413
2414 keywordHolder = _keywordClone2;
2415
2416 while(true){
2417 if(keywordHolder.indexOf(" ")!=-1){
2418 keywordHolder = keywordHolder.replaceFirst(" ",spaceLeft+space+spaceRight+" ");
2419 }
2420 else{
2421 break;
2422 }
2423 }
2424 for(int a = 0; a<spaceCounterFront2; a++){
2425 keywordHolder = spaceLeft+space+spaceRight+keywordHolder;
2426 }
2427 for(int a = 0; a<spaceCounterEnd2; a++){
2428 keywordHolder = keywordHolder+spaceLeft+space+spaceRight;
2429 }
2430
2431 InternalLink il3 = new InternalLink();
2432 il3.setValue(keywordHolder);
2433 if(suggestionMap.containsKey(bkKeyWord2)){
2434 il3.increaseElement((String)suggestionMap.get(bkKeyWord2));
2435
2436 }
2437 else{
2438 il3.increaseElement("&nbsp;");
2439 }
2440
2441 il.increaseNode(il3);
2442
2443 distanceMap.put(_keywordClone, il);
2444 }
2445 }
2446 }
2447
2448 }
2449
2450 }
2451
2452
2453 printWriter.write("<!-- generating all possible page-->");
2454 printWriter.flush();
2455
2456 if(distanceMap.size()!=0){
2457 generateHTML(distanceMap,fileName,metadataSetName);
2458 status = true;
2459 }
2460 }catch(Exception ex){ex.printStackTrace();}
2461 return status;
2462
2463 }
2464
2465
2466 private void generateHTML(HashMap distanceMap, String fileName,String metadataSetName){
2467 printWriter.write("<!-- generating all possible page-->");
2468 printWriter.flush();
2469
2470 String collectionFolder = mds.getCollectionName();
2471
2472 try{
2473 FileWriter fstream = new FileWriter(destination+fileName+"_Suggestion.html");
2474 BufferedWriter out = new BufferedWriter(fstream);
2475 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
2476 out.write ("<html>\r\n");
2477 out.write("<head>\r\n<title>Potential Duplicate List</title>\r\n");
2478 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
2479 out.write("<style type=\"text/css\">\r\n");
2480 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
2481 out.write(".tam{height:20; text-align:center}\r\n");
2482 out.write("th{height:20; text-align:left;}\r\n");
2483 out.write("body{font-family: Arial;}\r\n");
2484 out.write("</style>\r\n");
2485 out.write("</head><body>\r\n");
2486 out.write("<p><a href=\" Overall.html \">Summary</a>");
2487 out.write("&raquo;");
2488 if(metadataSetName.equalsIgnoreCase("dublin")){
2489 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2490 }
2491 else if (metadataSetName.equalsIgnoreCase("extracted")){
2492 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2493 }
2494 else{
2495 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2496 }
2497 out.write("&raquo;");
2498 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
2499 out.write("<h3 align=center>Potential Duplicate List</h3>");
2500 out.write("<table border=1 align=center>\r\n");
2501
2502 Set s = distanceMap.keySet();
2503 Iterator i = s.iterator();
2504 while(i.hasNext()){
2505
2506 String keyword = (String)i.next();
2507 InternalLink il = (InternalLink)distanceMap.get(keyword);
2508
2509 ArrayList alist = il.retrieveNodeList();
2510
2511 out.write("<tr><th>Original Text<th>Source Link\r\n");
2512 for(int a = 0; a<alist.size(); a++){
2513 InternalLink il2 = new InternalLink();
2514 il2 = (InternalLink)alist.get(a);
2515 String url = (String)il2.retrieveList().get(0);
2516 out.write("<tr><td>"+il2.getValue()+"<td><a href=\""+url+"\">"+url+"</a>");
2517
2518 }
2519 out.write("</tr>");
2520
2521 }
2522 out.write("</table>\r\n");
2523
2524 out.write("<p><a href=\" Overall.html \">Summary</a>");
2525 out.write("&raquo;");
2526 if(metadataSetName.equalsIgnoreCase("dublin")){
2527 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2528 }
2529 else if (metadataSetName.equalsIgnoreCase("extracted")){
2530 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2531 }
2532 else{
2533 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2534 }
2535 out.write("&raquo;");
2536 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
2537 out.write("</body></html>\r\n");
2538 out.close();
2539 fstream.close();
2540 }catch(IOException ex){ex.printStackTrace();}
2541
2542 }
2543
2544 public int calculateEditDistance(char[] args1, char[] args2){
2545
2546 int n = args1.length;
2547 int m = args2.length;
2548 if (n == 0) {
2549 return m;
2550 }
2551 else if (m == 0) {
2552 return n;
2553 }
2554
2555 int[] p = new int[n + 1];
2556 int[] d = new int[n + 1];
2557 int[] _d;
2558
2559
2560 int i;
2561 int j;
2562
2563 int cost; // cost
2564
2565 for (i = 0; i <= n; i++) {
2566 p[i] = i;
2567 }
2568 for (j = 1; j <= m; j++) {
2569
2570 d[0] = j;
2571 for (i = 1; i <= n; i++) {
2572 cost = (args1[i-1] == args2[j-1]) ? 0 : 1;
2573 d[i] = Math.min(Math.min(d[i - 1] + 1, p[i] + 1),
2574 p[i - 1] + cost);
2575 }
2576
2577 // copy current distance counts to 'previous row' distance counts
2578 _d = p;
2579 p = d;
2580 d = _d;
2581 }
2582 // our last action in the above loop was to switch d and p, so p now
2583 // actually has the most recent cost counts
2584 return p[n];
2585 }
2586
2587 private String removeUnusedCharacter(String target){
2588 //printWriter.write("<!-- generating all possible page-->");
2589 //printWriter.flush();
2590
2591 // remove \n
2592 while(true){
2593 if(target.indexOf("\\n")!=-1){
2594 target = target.replaceFirst("\\\\n","");
2595 }
2596 else{
2597 break;
2598 }
2599 }
2600
2601 //remove spaces at the end of string
2602 if(target.length()>1){
2603 while(true){
2604 if(target.length()>1){
2605 if(target.charAt(target.length()-1)==' '){
2606 target = target.substring(0,target.length()-1);
2607 }
2608 else{
2609 break;
2610 }
2611 }
2612 else{
2613 break;
2614 }
2615 }
2616 }
2617 //remove leading spaces
2618 if(target.length()>1){
2619 while(true){
2620
2621 if(target.length()>1){
2622 if(target.charAt(0)==' '){
2623 target = target.substring(1,target.length());
2624 }
2625 else{
2626 break;
2627 }}
2628 else{
2629 break;
2630 }
2631 }
2632 }
2633 //remove multiple spaces between words
2634 while(true){
2635 if(target.indexOf(" ")!=-1){
2636 target = target.replaceFirst(" "," ");
2637 }
2638 else{
2639 break;
2640 }
2641 }
2642
2643 return target;
2644 }
2645
2646
2647 private costModel removeUnusedCharacter(String target, double cost){
2648
2649
2650 costModel cm = new costModel();
2651
2652 // remove \n plus cost
2653 while(true){
2654
2655 if(target.indexOf("\\n")!=-1){
2656 target = target.replaceFirst("\\\\n","");
2657 cost = cost + 0.2;
2658 }
2659 else{
2660 break;
2661 }
2662 }
2663
2664 //remove spaces at the end of string
2665 if(target.length()>1){
2666 while(true){
2667 if(target.length()>1){
2668 if(target.charAt(target.length()-1)==' '){
2669 target = target.substring(0,target.length()-1);
2670 cost = cost + 0.2;
2671 }
2672 else{
2673 break;
2674 }}
2675 else{
2676 break;
2677 }
2678 }
2679 }
2680 //remove leading spaces
2681 if(target.length()>1){
2682 while(true){
2683
2684 if(target.length()>1){
2685 if(target.charAt(0)==' '){
2686 target = target.substring(1,target.length());
2687 cost = cost + 0.2;
2688 }
2689 else{
2690 break;
2691 }}
2692 else{
2693 break;
2694 }
2695 }
2696 }
2697 //remove multiple spaces between words
2698 while(true){
2699
2700 if(target.indexOf(" ")!=-1){
2701 target = target.replaceFirst(" "," ");
2702 cost = cost + 0.2;
2703 }
2704 else{
2705 break;
2706 }
2707 }
2708
2709 cm.setCost(cost);
2710 cm.setString(target);
2711
2712 return cm;
2713 }
2714
2715
2716
2717 private HashMap generateMetadataElementSortListNoIdentifier(String title, String sort, String metadataSetName,PrintWriter outx){
2718
2719 String fileName = title+"_"+sort;
2720 String collectionFolder = mds.getCollectionName();
2721 System.out.println("level 3");
2722 SearchLink sl = new SearchLink(mds.StatsDirectory);
2723 System.out.println("level 4");
2724 HashMap suggestionMap = new HashMap();
2725
2726 HashMap internalIDMap = dm.getInternalIdentifier(title);
2727 boolean status = false;
2728
2729 try{
2730 printWriter.write("<!-- generating all possible page-->");
2731 printWriter.flush();
2732
2733 FileWriter fstream = new FileWriter(destination+fileName+".html");
2734 BufferedWriter out = new BufferedWriter(fstream);
2735
2736 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
2737 out.write ("<html>\r\n");
2738 out.write("<head>\r\n<title>Metadata Element Sort List</title>\r\n");
2739 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
2740 out.write("<script type=\"text/javascript\" src=\"../script/getInfomation.js\"></script>\r\n");
2741 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/yahoo/yahoo-min.js\"></script>\r\n");
2742 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/event/event-min.js\"></script>\r\n");
2743 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/connection/connection-min.js\"></script>\r\n");
2744
2745 out.write("<style type=\"text/css\">\r\n");
2746 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
2747 out.write(".tam{height:20; text-align:center}\r\n");
2748 out.write("th{height:20; text-align:center;}\r\n");
2749 out.write("body{font-family: Arial;}\r\n");
2750 out.write("</style>\r\n");
2751 out.write("</head><body id=\""+collectionFolder+"\">\r\n");
2752
2753 out.write("<td><div style=\"float:left;\"><a href=\" Overall.html \">Summary</a>");
2754 out.write("&raquo;");
2755
2756 if(metadataSetName.equalsIgnoreCase("dublin")){
2757 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2758 }
2759 else if (metadataSetName.equalsIgnoreCase("extracted")){
2760 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2761 }
2762 else{
2763 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2764 }
2765 out.write("&raquo;");
2766 out.write("<a href=\""+title+".html\">"+title+"</a></div>");
2767
2768 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>");
2769
2770
2771 String[] list = dm.getSortList(title, sort);
2772
2773
2774
2775 if(list.length>=1){
2776
2777
2778 if(sort.equals("ASCII")){
2779
2780 out.write("<table border=\"1\" align=\"center\" width=\"800\">\r\n");
2781 out.write ("<h2 align=\"center\">"+ title+"</h2>\r\n");
2782 out.write ("<th class=\"tam\">ASCII Sort\r\n");
2783 out.write ("<th class=\"tam\">Element Values\r\n");
2784 out.write ("<th class=\"tam\">Source Documents\r\n");
2785 out.write ("<th class=\"tam\">Internal Link\r\n");
2786
2787 out.write ("<a name='topFive'>\r\n");
2788
2789 int counter = 0;
2790
2791 for(int i = 0; i<list.length; i++){
2792
2793 if(list.length<=5 && i==0){
2794 out.write ("<a name='lastFive'>\r\n");
2795 }
2796 else if((list.length>5) && (list.length-5==i)){
2797 out.write ("<a name='lastFive'>\r\n");
2798 }
2799
2800
2801 InternalLink il= (InternalLink)internalIDMap.get(list[i]);
2802 ArrayList alist2 = il.retrieveList();
2803 String id = (String)alist2.get(0);
2804 id = id.substring(4);
2805 out.write("<tr id=\""+id+"\" >\r\n");
2806
2807 if(list[i].length()>=201){
2808 if(list[i].startsWith("http://")){
2809
2810 if(list[i].length()>=100){
2811 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2812 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2813 }
2814 else{
2815 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2816 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2817 }
2818
2819 }
2820 else{
2821
2822 if(list[i].startsWith(" ")){
2823 String elements = list[i];
2824 list[i] = list[i].substring(1);
2825 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i].substring(0, 200)+"...<td>No Source Available \r\n");
2826 }
2827 else{
2828 out.write("<td>"+(counter+1)+"<td>"+list[i].substring(0, 200)+"... <td>No Source Available\r\n");
2829 }
2830 }
2831
2832 }
2833 else{
2834
2835
2836 if(list[i].startsWith("http://")){
2837
2838 if(list[i].length()>=100){
2839 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
2840 }
2841 else{
2842 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2843 }
2844 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2845
2846 }
2847
2848
2849
2850 else{
2851
2852 if(list[i].startsWith(" ")){
2853 String elements = list[i];
2854 list[i] = list[i].substring(1);
2855
2856 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i]+"<td>No Source Available\r\n");
2857 suggestionMap.put(elements,"No Source Available");
2858
2859 }
2860 else{
2861 out.write("<td>"+(counter+1)+"<td>"+list[i]+"<td>No Source Available\r\n");
2862 }
2863 }
2864 }
2865 counter++;
2866 out.write("<td onclick=\"GD(this)\">View");
2867 }
2868
2869 }
2870 else{
2871
2872
2873 out.write("<table border=\"1\" align=\"center\" width=\"800\">\r\n");
2874 out.write ("<h2 align=\"center\">"+ title+"</h2>\r\n");
2875 out.write ("<th class=\"tam\">&nbsp;\r\n");
2876 out.write ("<th class=\"tam\">Frequency\r\n");
2877 out.write ("<th class=\"tam\">Element Values\r\n");
2878 out.write ("<th class=\"tam\">Source Documents\r\n");
2879 out.write ("<th class=\"tam\">Internal Link\r\n");
2880 out.write ("<a name='topFive'>\r\n");
2881
2882 HashMap xMap = dm.getDistinctValueMap(title);
2883 int counter = 0;
2884 for(int i = 0; i<list.length; i++){
2885 if(list.length<=5 && i==0){
2886 out.write ("<a name='lastFive'>\r\n");
2887 }
2888 else if((list.length>5) && (list.length-5==i)){
2889 out.write ("<a name='lastFive'>\r\n");
2890 }
2891
2892 InternalLink il= (InternalLink)internalIDMap.get(list[i]);
2893 ArrayList alist2 = il.retrieveList();
2894 String id = (String)alist2.get(0);
2895 id = id.substring(4);
2896 out.write("<tr id=\""+id+"\" >\r\n");
2897
2898 out.write("<td>"+(counter+1)+"<th> "+((Integer)xMap.get(list[i])).toString()+"\r\n");
2899
2900 if(list[i].length()>=201){
2901
2902 if(list[i].startsWith("http://")){
2903 if(list[i].length()>=100){
2904 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2905 }
2906 else{
2907 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2908 }
2909
2910 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2911 }
2912
2913
2914
2915
2916 else{
2917 if(list[i].startsWith(" ")){
2918 list[i]=list[i].substring(1);
2919
2920 out.write("<td>"+spaceLeft+space+spaceRight+list[i].substring(0,200)+"...<td>No Source Available\r\n");
2921
2922 }
2923 else{
2924
2925 out.write("<td>"+list[i].substring(0,200)+"...<td>No Source Available\r\n");
2926
2927 }
2928 }
2929 }
2930 else{
2931 if(list[i].startsWith("http://")){
2932 if(list[i].length()>=100){
2933 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2934 }
2935 else{
2936 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2937 }
2938 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2939
2940 }
2941 else{
2942
2943 if(list[i].startsWith(" ")){
2944 list[i]=list[i].substring(1);
2945
2946
2947 out.write("<td>"+spaceLeft+space+spaceRight+list[i]+"<td>No Source Available\r\n");
2948
2949 }
2950 else{
2951
2952 out.write("<td>"+list[i]+"<td>No Source Available\r\n");
2953
2954 }
2955 }
2956 }
2957 counter++;
2958 out.write("<td onclick=\"GD(this)\">View");
2959 }
2960 }
2961 out.write("</table>\r\n");
2962 }
2963 else{
2964 out.write("<p>Sorry! The list is empty</p>");
2965 }
2966 out.write("<tr>\r\n");
2967
2968
2969 out.write("<p><a href=\" Overall.html \">Summary</a>");
2970 out.write("&raquo;");
2971
2972 if(metadataSetName.equalsIgnoreCase("dublin")){
2973 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2974 }
2975 else if (metadataSetName.equalsIgnoreCase("extracted")){
2976 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2977 }
2978 else{
2979 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2980 }
2981 out.write("&raquo;");
2982 out.write("<a href=\""+title+".html\">"+title+"</a></p>");
2983 out.write("</body></html>\r\n");
2984 out.close();
2985 fstream.close();
2986 }catch(Exception e){
2987
2988 e.printStackTrace(outx);
2989 }
2990
2991 return new HashMap();
2992 }
2993
2994
2995
2996
2997
2998
2999
3000
3001
3002
3003
3004
3005
3006
3007
3008
3009
3010
3011
3012
3013
3014
3015
3016
3017
3018
3019
3020
3021
3022
3023
3024
3025 private void createIncompletedListNoIdentifier(String fileName, String metadataSetName, String collectionFolder,PrintWriter outx){
3026 //HashMap hp = dm.getIdentifierLink("dc.Identifier");
3027 HashMap hp = dm.getIdentifierLinkNoIdentifier();
3028 String[] ids = dm.getDocumentIDList(fileName);
3029 System.out.println("Length: "+ids.length+" "+fileName+ " "+hp.size());
3030
3031 Set st = hp.keySet();
3032 Iterator it = st.iterator();
3033 while(it.hasNext()){
3034 System.out.println(it.next());
3035 }
3036 for(int i = 0; i<ids.length; i++){
3037 System.out.println("IDS:" +ids[i]);
3038 hp.remove(ids[i]);
3039 }
3040 System.out.println("Length: "+ids.length+" "+fileName+ " "+hp.size());
3041 //String collectionFolder = mds.getCollectionName();
3042 //System.out.println(hp.size());
3043 try{
3044 FileWriter fstream = new FileWriter(destination+fileName+"_IncompletedList.html");
3045 BufferedWriter out = new BufferedWriter(fstream);
3046
3047 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
3048 out.write ("<html>\r\n");
3049 out.write("<head>\r\n<title>Incompleted Document List</title>\r\n");
3050 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
3051 out.write("<style type=\"text/css\">\r\n");
3052 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
3053 out.write(".tam{height:20; text-align:center}\r\n");
3054 out.write("th{height:20; text-align:center;}\r\n");
3055 out.write("body{font-family: Arial;}\r\n");
3056 out.write("</style>\r\n");
3057 out.write("</head><body>\r\n");
3058
3059 out.write("<p><a href=\" Overall.html \">Summary</a>");
3060 out.write("&raquo;");
3061 if(metadataSetName.equalsIgnoreCase("dublin")){
3062 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
3063 }
3064 else if (metadataSetName.equalsIgnoreCase("extracted")){
3065 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
3066 }
3067 else{
3068 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
3069 }
3070 out.write("&raquo;");
3071 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
3072
3073 out.write("<h3 align=center>"+fileName+" does not appear in the following documents</h3>");
3074 int counter = 0;
3075 System.out.println("link.....");
3076 Set s = hp.keySet();
3077 Iterator i = s.iterator();
3078 System.out.println("link2.....");
3079 out.write("<table border=1 align=center><tr><th>Document ID<th>Source Link\r\n");
3080 while(i.hasNext()){
3081 //printWriter.write("<!-- generating all possible page-->");
3082 System.out.println("link.....3");
3083 counter++;
3084 String keys = (String)i.next();
3085 System.out.println("link.....4"+keys);
3086 InternalLink il = (InternalLink) hp.get(keys);
3087
3088 ArrayList alist = il.retrieveList();
3089 System.out.println("link.....5"+alist.size());
3090 String url = (String)alist.get(0);
3091 out.write("<tr><td>"+counter+"<td><a href=\""+url+"\">"+url+"</a>\r\n");
3092 }
3093 out.write("</table></body></html>\r\n");
3094
3095
3096 out.write("<p><a href=\" Overall.html \">Summary</a>");
3097 out.write("&raquo;");
3098 if(metadataSetName.equalsIgnoreCase("dublin")){
3099 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
3100 }
3101 else if (metadataSetName.equalsIgnoreCase("extracted")){
3102 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
3103 }
3104 else{
3105 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
3106 }
3107 out.write("&raquo;");
3108 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
3109
3110 out.close();
3111 fstream.close();
3112 }catch(IOException ex){ex.printStackTrace();}
3113
3114
3115 }
3116 ///////////////////////////////////////////////////////////////////////////////////////
3117}
3118
Note: See TracBrowser for help on using the repository browser.