source: gs3-extensions/mat/trunk/src/org/greenstone/mat/PrintHTML.java@ 22171

Last change on this file since 22171 was 22171, checked in by sjm84, 14 years ago

Updated the MAT code so that it now uses the new util classes
File size: 100.5 KB
Line 
1package org.greenstone.mat;
2
3import java.io.*;
4import java.text.SimpleDateFormat;
5import java.util.ArrayList;
6import java.util.Calendar;
7import java.util.Collection;
8import java.util.HashMap;
9import java.util.Iterator;
10import java.util.Map;
11import java.util.Set;
12
13import javax.xml.parsers.DocumentBuilder;
14import javax.xml.parsers.DocumentBuilderFactory;
15
16import org.w3c.dom.Document;
17import org.w3c.dom.Element;
18import org.w3c.dom.Node;
19import org.w3c.dom.NodeList;
20
21import org.greenstone.util.GlobalProperties;
22
23public class PrintHTML {
24
25 final String NORMAL = "normal";
26 final String WORST = "worst";
27 final String BEST = "best";
28
29 final String title1 ="Hide Empty Metadata ,";
30 final String title2 ="Hide Completed Metadata ,";
31 final String title3 ="Hide Documents with Empty Metadata ,";
32 final String title4 ="Hide Documents with Completed Metadata ,";
33 final String title5 ="No Available Graph";
34
35 final String metadataSet1 = "Dublin Core";
36 final String metadataSet2 = "Extracted";
37
38 final String spaceLeft = "<I>&laquo;";
39 final String spaceRight = "&raquo;</I>";
40 final String htmlSpace = "&nbsp;";
41 final String oddChar ="odd Character";
42
43 String linkIdentifier = "";
44 //String cssString = "<link rel=\"stylesheet\" href=\"http://www.nzdl.org/greenstone3/mat/script/mat.css\" type=\"text/css\" >";
45
46 final String space ="space";
47 private HashMap url;
48
49 String destination ="";
50 MetadataStats mds;
51 DataMaker dm;
52 Calendar cl;
53 SimpleDateFormat sdf;
54
55 public PrintHTML(MetadataStats ms){
56
57 mds = ms;
58
59 File current_dir = new File (".");
60 try{
61 destination = mds.HTMLDirectory;
62 }catch(Exception ex){
63 ex.printStackTrace();
64 }
65
66 dm = new DataMaker(mds);
67 cl=Calendar.getInstance();
68 linkIdentifier = "dc.Identifier";
69
70 sdf = new SimpleDateFormat(" dd MMM yyyy 'at' HH:mm:ss z 'GMT'Z");
71 setupMetadataLink("metadataElementURL");
72 }
73
74 private ArrayList getHideEmptyMetadataDetail(ArrayList dataset, String[] id, String[] metadataName){
75
76 if(id.length==0 || metadataName.length==0){
77 return new ArrayList();
78 }
79
80 ArrayList detailList = new ArrayList();
81 ArrayList alist = new ArrayList();
82 ArrayList nameList = new ArrayList();
83
84 for(int i = 0; i< metadataName.length; i++){
85 if(!dm.IsElementEmpty(metadataName[i])){
86 alist.add(dm.getMetadataRows(metadataName[i]));
87 nameList.add(metadataName[i]);
88 }
89 }
90
91 String[] metaDataElementName = new String[nameList.size()];
92
93 for(int i = 0; i< nameList.size(); i++){
94 metaDataElementName[i] = nameList.get(i).toString();
95 }
96
97 if(id.length==0 || metaDataElementName.length==0){
98 return new ArrayList();
99 }
100
101 detailList.add(alist);
102 detailList.add(id);
103 detailList.add(metaDataElementName);
104 return detailList;
105 }
106
107 private ArrayList getHideFullMetadataDetail(ArrayList dataset, String[] id, String[] metadataName){
108
109 if(id.length==0 || metadataName.length==0){
110 return new ArrayList();
111 }
112
113 ArrayList detailList = new ArrayList();
114 ArrayList alist = new ArrayList();
115 ArrayList nameList = new ArrayList();
116
117 for(int i = 0; i< metadataName.length; i++){
118 if(!dm.IsElementFull(metadataName[i])){
119 alist.add(dm.getMetadataRows(metadataName[i]));
120 nameList.add(metadataName[i]);
121 }
122 }
123
124 String[] metaDataElementName = new String[nameList.size()];
125
126 for(int i = 0; i< nameList.size(); i++){
127 metaDataElementName[i] = nameList.get(i).toString();
128 }
129
130 if(id.length==0 || metaDataElementName.length==0){
131 return new ArrayList();
132 }
133
134 detailList.add(alist);
135 detailList.add(id);
136 detailList.add(metaDataElementName);
137
138 return detailList;
139 }
140
141 private ArrayList getHideEmptyDocumentDetail(ArrayList dataset, String[] id, String[] metadataName){
142
143 if(id.length==0 || metadataName.length==0){
144 return new ArrayList();
145 }
146
147 ArrayList detailList = new ArrayList();
148 ArrayList alist = dm.removeDocument(dataset, id, 0);
149 ArrayList idList = dm.getRemovedID();
150 ArrayList newIDList = new ArrayList();
151
152 for(int i = 0; i< id.length; i++){
153 if(!idList.contains(id[i])){
154 newIDList.add(id[i]);
155 }
156 }
157
158 id = new String[newIDList.size()];
159
160 for(int i = 0; i<newIDList.size(); i++){
161 id[i] = newIDList.get(i).toString();
162 }
163
164 if(id.length==0 || metadataName.length==0){
165 return new ArrayList();
166 }
167
168 detailList.add(alist);
169 detailList.add(id);
170 detailList.add(metadataName);
171
172 return detailList;
173 }
174
175 private ArrayList getHideFullDocumentDetail(ArrayList dataset, String[] id, String[] metadataName){
176
177 if(id.length==0 || metadataName.length==0){
178 return new ArrayList();
179 }
180
181 ArrayList detailList = new ArrayList();
182 ArrayList alist = dm.removeDocument(dataset, id ,1);
183 ArrayList idList = dm.getRemovedID();
184 ArrayList newIDList = new ArrayList();
185
186 for(int i = 0; i< id.length; i++){
187 if(!idList.contains(id[i])){
188 newIDList.add(id[i]);
189 }
190 }
191
192 id = new String[newIDList.size()];
193
194 for(int i = 0; i<newIDList.size(); i++){
195 id[i] = newIDList.get(i).toString();
196 }
197
198 if(id.length==0 || metadataName.length==0){
199 return new ArrayList();
200 }
201
202 detailList.add(alist);
203 detailList.add(id);
204 detailList.add(metadataName);
205
206 return detailList;
207 }
208
209
210 public void generateAllPossibleGraph(ArrayList Namelist,String[]ids,String[] names,String condition){
211
212 ArrayList dataList= Namelist;
213 String[] id = ids;
214 String[] metadataName = names;
215
216 ArrayList alist = new ArrayList();
217 ArrayList detailList = new ArrayList();
218 ArrayList new_dataList = new ArrayList();
219 String[] idList;
220 String[] metadataNameList;
221
222 //-------------------SS
223 //-----generate SSSS
224 if(id.length==0 || metadataName.length==0){
225 generateEmptyGraph("SSSS",WORST,condition);
226 generateEmptyGraph("SSSS",BEST,condition);
227 }
228 else{
229 new_dataList = transformDataList(dataList,id);
230
231 WriteHTML("SSSS",new_dataList,id,metadataName,WORST,"Show completed graph",condition);
232 WriteHTML("SSSS",new_dataList,id,metadataName,BEST,"Show completed graph",condition);
233 }
234
235 //-----SSSH
236 detailList = getHideFullDocumentDetail(dataList,id,metadataName);
237
238 if(detailList.size()==0){
239 generateEmptyGraph("SSSH",WORST,condition);
240 generateEmptyGraph("SSSH",BEST,condition);
241 }
242 else{
243 alist = (ArrayList)detailList.get(0);
244 idList = (String[])detailList.get(1);
245 metadataNameList = (String[])detailList.get(2);
246 new_dataList = transformDataList(alist,idList);
247
248 WriteHTML("SSSH",new_dataList,idList,metadataNameList,WORST,title4,condition);
249 WriteHTML("SSSH",new_dataList,idList,metadataNameList,BEST,title4,condition);
250 }
251
252 //-----SSHS
253 detailList = getHideEmptyDocumentDetail(dataList,id,metadataName);
254 if(detailList.size()==0){
255 generateEmptyGraph("SSHS",WORST,condition);
256 generateEmptyGraph("SSHS",BEST,condition);
257 generateEmptyGraph("SSHH",WORST,condition);
258 generateEmptyGraph("SSHH",BEST,condition);
259 }
260 else{
261 alist = (ArrayList)detailList.get(0);
262 idList = (String[])detailList.get(1);
263 metadataNameList = (String[])detailList.get(2);
264 new_dataList = transformDataList(alist,idList);
265
266 WriteHTML("SSHS",new_dataList,idList,metadataNameList,WORST,title3,condition);
267 WriteHTML("SSHS",new_dataList,idList,metadataNameList,BEST,title3,condition);
268
269 //-----SSHH
270 detailList = getHideFullDocumentDetail(alist,idList,metadataNameList);
271 if(detailList.size()==0){
272 generateEmptyGraph("SSHH",WORST,condition);
273 generateEmptyGraph("SSHH",BEST,condition);
274 }
275 else{
276 alist = (ArrayList)detailList.get(0);
277 idList = (String[])detailList.get(1);
278 metadataNameList = (String[])detailList.get(2);
279 new_dataList = transformDataList(alist,idList);
280
281 WriteHTML("SSHH",new_dataList,idList,metadataNameList,WORST,title3+title4,condition);
282 WriteHTML("SSHH",new_dataList,idList,metadataNameList,BEST,title3+title4,condition);
283 }
284 }
285
286
287 //-------------------SH
288 ArrayList xList = new ArrayList();
289 String[] idListCopy;
290 String[] metadataNameListCopy;
291
292 //-----SHSS
293 detailList = getHideFullMetadataDetail(dataList,id,metadataName);
294
295 if(detailList.size()==0){
296 generateEmptyGraph("SHSS",WORST,condition);
297 generateEmptyGraph("SHSS",BEST,condition);
298
299 generateEmptyGraph("SHSH",WORST,condition);
300 generateEmptyGraph("SHSH",BEST,condition);
301
302 generateEmptyGraph("SHHS",WORST,condition);
303 generateEmptyGraph("SHHS",BEST,condition);
304
305 generateEmptyGraph("SHHH",WORST,condition);
306 generateEmptyGraph("SHHH",BEST,condition);
307 }
308 else{
309 alist = (ArrayList)detailList.get(0);
310 idList = (String[])detailList.get(1);
311 metadataNameList = (String[])detailList.get(2);
312
313 xList = (ArrayList)alist.clone();
314 idListCopy = (String[]) idList.clone();
315 metadataNameListCopy = (String[])metadataNameList.clone();
316 new_dataList = transformDataList(alist,idList);
317
318 WriteHTML("SHSS",new_dataList,idList,metadataNameList,WORST,title2,condition);
319 WriteHTML("SHSS",new_dataList,idList,metadataNameList,BEST,title2,condition);
320
321 //-----SHHS
322 detailList = getHideEmptyDocumentDetail((ArrayList)xList.clone(),(String[])idListCopy.clone(),(String[])metadataNameListCopy.clone());
323
324 if(detailList.size()==0){
325 generateEmptyGraph("SHHS",WORST,condition);
326 generateEmptyGraph("SHHS",BEST,condition);
327
328 generateEmptyGraph("SHHH",WORST,condition);
329 generateEmptyGraph("SHHH",BEST,condition);
330 }
331 else{
332 alist = (ArrayList)detailList.get(0);
333 idList = (String[])detailList.get(1);
334 metadataNameList = (String[])detailList.get(2);
335 new_dataList = transformDataList(alist,idList);
336
337 WriteHTML("SHHS",new_dataList,idList,metadataNameList,WORST,title2+title3,condition);
338 WriteHTML("SHHS",new_dataList,idList,metadataNameList,BEST,title2+title3,condition);
339
340 //-----SHHH
341 detailList = getHideFullDocumentDetail(alist,idList,metadataNameList);
342
343 if(detailList.size()==0){
344 generateEmptyGraph("SHHH",WORST,condition);
345 generateEmptyGraph("SHHH",BEST,condition);
346 }
347 else{
348 alist = (ArrayList)detailList.get(0);
349 idList = (String[])detailList.get(1);
350 metadataNameList = (String[])detailList.get(2);
351 new_dataList = transformDataList(alist,idList);
352
353 WriteHTML("SHHH",new_dataList,idList,metadataNameList,WORST,title2+title3+title4,condition);
354 WriteHTML("SHHH",new_dataList,idList,metadataNameList,BEST,title2+title3+title4,condition);
355 }
356 }
357
358 //-----SHSH
359 detailList = getHideFullDocumentDetail((ArrayList)xList.clone(),(String[])idListCopy.clone(),(String[])metadataNameListCopy.clone());
360
361 if(detailList.size()==0){
362 generateEmptyGraph("SHSH",WORST,condition);
363 generateEmptyGraph("SHSH",BEST,condition);
364 }
365 else{
366 alist = (ArrayList)detailList.get(0);
367 idList = (String[])detailList.get(1);
368 metadataNameList = (String[])detailList.get(2);
369 new_dataList = transformDataList(alist,idList);
370
371 WriteHTML("SHSH",new_dataList,idList,metadataNameList,WORST,title2+title4,condition);
372 WriteHTML("SHSH",new_dataList,idList,metadataNameList,BEST,title2+title4,condition);
373 }
374 }
375
376 //-------------------HS
377 ArrayList xList1 = new ArrayList();
378 String[] idListCopy1;
379 String[] metadataNameListCopy1;
380
381 //-----HSSS
382 detailList = getHideEmptyMetadataDetail(dataList,id,metadataName);
383
384 if(detailList.size()==0){
385 generateEmptyGraph("HSSS",WORST,condition);
386 generateEmptyGraph("HSSS",BEST,condition);
387
388 generateEmptyGraph("HSHS",WORST,condition);
389 generateEmptyGraph("HSHS",BEST,condition);
390
391 generateEmptyGraph("HSHH",WORST,condition);
392 generateEmptyGraph("HSHH",BEST,condition);
393
394 generateEmptyGraph("HSSH",WORST,condition);
395 generateEmptyGraph("HSSH",BEST,condition);
396 }
397 else{
398 alist = (ArrayList)detailList.get(0);
399 idList = (String[])detailList.get(1);
400 metadataNameList = (String[])detailList.get(2);
401
402 xList1 = (ArrayList)alist.clone();
403 idListCopy1 = (String[]) idList.clone();
404 metadataNameListCopy1 = (String[])metadataNameList.clone();
405 new_dataList = transformDataList(alist,idList);
406
407 WriteHTML("HSSS",new_dataList,idList,metadataNameList,WORST,title1,condition);
408 WriteHTML("HSSS",new_dataList,idList,metadataNameList,BEST,title1,condition);
409
410 //-----HSHS
411 detailList = getHideEmptyDocumentDetail((ArrayList)xList1.clone(),(String[])idListCopy1.clone(),(String[])metadataNameListCopy1.clone());
412
413 if(detailList.size()==0){
414 generateEmptyGraph("HSHS",WORST,condition);
415 generateEmptyGraph("HSHS",BEST,condition);
416
417 generateEmptyGraph("HSHH",WORST,condition);
418 generateEmptyGraph("HSHH",BEST,condition);
419 }
420 else{
421 alist = (ArrayList)detailList.get(0);
422 idList = (String[])detailList.get(1);
423 metadataNameList = (String[])detailList.get(2);
424 new_dataList = transformDataList(alist,idList);
425
426 WriteHTML("HSHS",new_dataList,idList,metadataNameList,WORST,title1+title3,condition);
427 WriteHTML("HSHS",new_dataList,idList,metadataNameList,BEST,title1+title3,condition);
428
429 //-----HSHH
430 detailList = getHideFullDocumentDetail(alist,idList,metadataNameList);
431 if(detailList.size()==0){
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 new_dataList = transformDataList(alist,idList);
440
441 WriteHTML("HSHH",new_dataList,idList,metadataNameList,WORST,title1+title3+title4,condition);
442 WriteHTML("HSHH",new_dataList,idList,metadataNameList,BEST,title1+title3+title4,condition);
443 }
444 }
445
446 //-----HSSH
447 detailList = getHideFullDocumentDetail((ArrayList)xList1.clone(),(String[])idListCopy1.clone(),(String[])metadataNameListCopy1.clone());
448
449 if(detailList.size()==0){
450 generateEmptyGraph("HSSH",WORST,condition);
451 generateEmptyGraph("HSSH",BEST,condition);
452 }
453 else{
454 alist = (ArrayList)detailList.get(0);
455 idList = (String[])detailList.get(1);
456 metadataNameList = (String[])detailList.get(2);
457 new_dataList = transformDataList(alist,idList);
458
459 WriteHTML("HSSH",new_dataList,idList,metadataNameList,WORST,title1+title4,condition);
460 WriteHTML("HSSH",new_dataList,idList,metadataNameList,BEST,title1+title4,condition);
461 }
462 }
463
464 //-------------------HH
465 ArrayList xList2 = new ArrayList();
466 String[] idListCopy2;
467 String[] metadataNameListCopy2;
468
469 //-----HHSS
470 detailList = getHideEmptyMetadataDetail(dataList,id,metadataName);
471
472 if(detailList.size()==0){
473 generateEmptyGraph("HHSS",WORST,condition);
474 generateEmptyGraph("HHSS",BEST,condition);
475
476 generateEmptyGraph("HHHS",WORST,condition);
477 generateEmptyGraph("HHHS",BEST,condition);
478
479 generateEmptyGraph("HHHH",WORST,condition);
480 generateEmptyGraph("HHHH",BEST,condition);
481
482 generateEmptyGraph("HHSH",WORST,condition);
483 generateEmptyGraph("HHSH",BEST,condition);
484 }
485 else{
486 alist = (ArrayList)detailList.get(0);
487 idList = (String[])detailList.get(1);
488 metadataNameList = (String[])detailList.get(2);
489 detailList = getHideFullMetadataDetail(alist,idList,metadataNameList);
490
491 if(detailList.size()==0){
492 generateEmptyGraph("HHSS",WORST,condition);
493 generateEmptyGraph("HHSS",BEST,condition);
494
495 generateEmptyGraph("HHHS",WORST,condition);
496 generateEmptyGraph("HHHS",BEST,condition);
497
498 generateEmptyGraph("HHHH",WORST,condition);
499 generateEmptyGraph("HHHH",BEST,condition);
500
501 generateEmptyGraph("HHSH",WORST,condition);
502 generateEmptyGraph("HHSH",BEST,condition);
503 }
504 else{
505 alist = (ArrayList)detailList.get(0);
506 idList = (String[])detailList.get(1);
507 metadataNameList = (String[])detailList.get(2);
508
509 xList2 = (ArrayList)alist.clone();
510 idListCopy2 = (String[]) idList.clone();
511 metadataNameListCopy2 = (String[])metadataNameList.clone();
512 new_dataList = transformDataList(alist,idList);
513
514 WriteHTML("HHSS",new_dataList,idList,metadataNameList,WORST,title1+title2,condition);
515 WriteHTML("HHSS",new_dataList,idList,metadataNameList,BEST,title1+title2,condition);
516
517 //----- HHSH
518 detailList = getHideFullDocumentDetail((ArrayList)xList2.clone(),(String[])idListCopy2.clone(),(String[])metadataNameListCopy2.clone());
519
520 if(detailList.size()==0){
521 generateEmptyGraph("HHSH",WORST,condition);
522 generateEmptyGraph("HHSH",BEST,condition);
523 }
524 else{
525 alist = (ArrayList)detailList.get(0);
526 idList = (String[])detailList.get(1);
527 metadataNameList = (String[])detailList.get(2);
528 new_dataList = transformDataList(alist,idList);
529
530 WriteHTML("HHSH",new_dataList,idList,metadataNameList,WORST,title1+title2+title4,condition);
531 WriteHTML("HHSH",new_dataList,idList,metadataNameList,BEST,title1+title2+title4,condition);
532 }
533 //-----HHHS
534 detailList = getHideEmptyDocumentDetail((ArrayList)xList2.clone(),(String[])idListCopy2.clone(),(String[])metadataNameListCopy2.clone());
535
536 if(detailList.size()==0){
537 generateEmptyGraph("HHHS",WORST,condition);
538 generateEmptyGraph("HHHS",BEST,condition);
539 generateEmptyGraph("HHHH",WORST,condition);
540 generateEmptyGraph("HHHH",BEST,condition);
541 }
542 else{
543 alist = (ArrayList)detailList.get(0);
544 idList = (String[])detailList.get(1);
545 metadataNameList = (String[])detailList.get(2);
546 new_dataList = transformDataList(alist,idList);
547
548 WriteHTML("HHHS",new_dataList,idList,metadataNameList,WORST,title1+title2+title3,condition);
549 WriteHTML("HHHS",new_dataList,idList,metadataNameList,BEST,title1+title2+title3,condition);
550
551 //-----HHHH
552 detailList = getHideFullDocumentDetail((ArrayList)alist.clone(),(String[])idList.clone(),(String[])metadataNameList.clone());
553
554 if(detailList.size()==0){
555 generateEmptyGraph("HHHH",WORST,condition);
556 generateEmptyGraph("HHHH",BEST,condition);
557 }
558 else{
559 alist = (ArrayList)detailList.get(0);
560 idList = (String[])detailList.get(1);
561 metadataNameList = (String[])detailList.get(2);
562 new_dataList = transformDataList(alist,idList);
563
564 WriteHTML("HHHH",new_dataList,idList,metadataNameList,WORST,title1+title2+title3+title4,condition);
565 WriteHTML("HHHH",new_dataList,idList,metadataNameList,BEST,title1+title2+title3+title4,condition);
566 }
567 }
568 }
569 }
570 }
571
572 private ArrayList transformDataList(ArrayList list, String[] ids){
573
574 ArrayList wholeList = new ArrayList();
575
576 for(int i = 0; i< ids.length; i++){
577 ArrayList idList = new ArrayList();
578 for(int j = 0; j<list.size();j++){
579 int[] datarows = (int[]) list.get(j);
580 idList.add(new Integer (datarows[i]));
581 }
582 wholeList.add(idList);
583 }
584
585 for(int i = 0; i< wholeList.size(); i++){
586 ArrayList alist = (ArrayList)wholeList.get(i);
587 int[] rows = new int[alist.size()];
588 for(int j = 0; j< alist.size(); j++){
589 rows[j] = ((Integer)alist.get(j)).intValue();
590 }
591 wholeList.remove(i);
592 wholeList.add(i,rows);
593 }
594 return wholeList;
595 }
596
597 private void generateEmptyGraph(String fileName,String condition,String suffix){
598
599 String suf = suffix;
600
601 if(!suffix.equalsIgnoreCase("dublin")){
602 suf = "other";
603 }
604
605 String cases = condition;
606 String collectionFolder = mds.getCollectionName();
607
608 try{
609 FileWriter fstream = new FileWriter(destination+suf+"_"+fileName+"_"+cases+".html");
610 BufferedWriter out = new BufferedWriter(fstream);
611
612 out.write("<!-- This comment keeps IE6/7 in the reliable quirks mode -->\r\n");
613 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">\r\n");
614 out.write("<html>\r\n");
615 out.write("<head>\r\n<title> No Available Chart</title>\r\n");
616 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
617 out.write("<link rel=\"stylesheet\" href=\"../script/doby2.css\" type=\"text/css\"/>\r\n");
618 out.write("</head>\r\n");
619 out.write("<p><a href=\" Overall.html \">Summary</a></p>");
620 out.write("<body>\r\n");
621 out.write("<p>No data available to render chart.</p>\r\n");
622 out.write("<p>Reason: Document number is zero or Metadata element number is zero </p>\r\n");
623 out.write("</body></html>\r\n");
624 out.close();
625 fstream.close();
626
627 }catch(Exception ex){
628 ex.printStackTrace();
629 }
630 }
631
632 private void WriteHTML(String fileName, ArrayList dataset, String[] ids, String[] metadataName, String condition, String title, String suffix){
633
634 int blueDot = 0;
635 String cases = condition;
636 String suf = suffix;
637
638 if(!suffix.equalsIgnoreCase("dublin")){
639 suf = "other";
640 }
641
642 try{
643 ArrayList tempList = mds.getMetadataNameList();
644 ArrayList urlIDList;
645
646 if(tempList.contains("dc.Identifier") && linkIdentifier.equals("dc.Identifier")){
647 urlIDList = dm.getURLMap("dc.Identifier");
648 }
649 else if(tempList.contains("nzir_internal.Link") && linkIdentifier.equals("nzir_internal.Link")){
650 urlIDList = dm.getURLMap("nzir_internal.Link");
651 }
652 else{
653 urlIDList = new ArrayList();
654 }
655
656 String collectionFolder = mds.getCollectionName();
657 FileWriter fstream = new FileWriter(destination+suf+"_"+fileName+"_"+cases+".html");
658 BufferedWriter out = new BufferedWriter(fstream);
659
660 out.write("<!-- This comment keeps IE6/7 in the reliable quirks mode -->\r\n");
661 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">\r\n");
662 out.write("<html>\r\n");
663 out.write("<head>\r\n<title>"+title+"</title>\r\n");
664 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
665 out.write("<link rel=\"stylesheet\" href=\"../script/doby2.css\" type=\"text/css\"/>\r\n");
666 out.write("<script type=\"text/javascript\" src=\"../script/getInfomation.js\"></script>\r\n");
667 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/yahoo/yahoo-min.js\"></script>\r\n");
668 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/event/event-min.js\"></script>\r\n");
669 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/connection/connection-min.js\"></script>\r\n");
670 out.write("</head>\r\n");
671 out.write ("<p><a href=\" Overall.html \">Summary</a></p>");
672 out.write("<body id=\""+mds.getCollectionName()+"\" onLoad=\"reconfig()\">\r\n");
673 out.write("<div id=\"container\">\r\n");
674 out.write("<div class=\"tableContainer\">\r\n");
675 out.write ("<table cellspacing=\"0\">\r\n");
676 out.write ("<thead>\r\n");
677 out.write ("<tr>\r\n");
678 out.write("<td class=\"qh\">Info\r\n");
679 out.write("<td class=\"qh\">URL\r\n");
680
681 for(int a = 0; a< metadataName.length; a++){
682 out.write("<td>"+ metadataName[a]+"\r\n");
683 }
684
685 out.write ("</thead>\r\n<tfoot>\r\n<tr>\r\n");
686 out.write("<td class=\"qh\">&nbsp;\r\n");
687 out.write("<td class=\"qh\">&nbsp;\r\n");
688
689 for(int a = 0; a< metadataName.length; a++){
690 out.write("<td>"+ dm.Mean(metadataName[a])+"%\r\n");
691 }
692
693 out.write ("</tfoot>\r\n<tbody>\r\n");
694
695 if(cases.equals("normal")){
696 for(int i = ids.length; i>0; i--){
697 out.write ("<tr>");
698 int[] datarows = (int[])dataset.get(i-1);
699
700 for(int j = 0; j<datarows.length; j++){
701 if(datarows[j]==1){
702 out.write ("<td class=\"b\">");
703 blueDot++;
704 }
705 else {
706 out.write ("<td>&nbsp;");
707 }
708 }
709 }
710 }
711 else{
712 ArrayList idList = new ArrayList();
713
714 for(int i = 0; i<ids.length;i++){
715 idList.add(ids[i]);
716 }
717
718 HashMap hp = new HashMap();
719
720 for(int i = ids.length; i>0; i--){
721 String idValue = ids[i-1];
722 int dots = 0;
723 int[] datarows = (int[])dataset.get(i-1);
724 for(int j = 0; j<datarows.length; j++){
725 if(datarows[j]==1){
726 dots++;
727 }
728 }
729 hp.put(idValue,new Integer(dots));
730 }
731
732 ArrayList alist = dm.sortMap(hp);
733 String[] idIntValue = new String[alist.size()];
734
735 for(int i = 0; i< alist.size(); i++){
736 Map.Entry entry = (Map.Entry) alist.get(i);
737 String idElement = ((String) entry.getKey());
738 idIntValue[i] = idElement;
739 }
740
741 if(cases.equals("worst")){
742
743 for(int i = 0; i<idIntValue.length; i++){
744 int value = idList.indexOf(idIntValue[i]);
745 int[] datarows = (int[])dataset.get(value);
746
747 out.write ("<tr id=\""+idIntValue[i].substring(4)+"\">\r\n");
748 out.write("<td class=\"E\" onclick=\"GD(this)\">&nbsp;\r\n");
749
750 if(urlIDList.contains(idIntValue[i])){
751 out.write("<td class=\"qh\" onclick=\"GX(this)\"><span title=\"open URL in new window\">open</span>\r\n");
752 }
753 else{
754 out.write("<td>\r\n");
755 }
756
757 for(int j = 0; j<datarows.length; j++){
758 if(datarows[j]==1){
759 out.write ("<td class=\"b\">\r\n");
760 blueDot++;
761 }
762 else {
763 out.write ("<td class=\"w\">\r\n");
764 }
765 }
766 }
767 }
768
769 if(cases.equals("best")){
770
771 for(int i = idIntValue.length; i>0; i--){
772 int value = idList.indexOf(idIntValue[i-1]);
773 int[] datarows = (int[])dataset.get(value);
774
775 out.write ("<tr id=\""+idIntValue[i-1].substring(4)+"\">\r\n");
776 out.write("<td class=\"E\" onclick=\"GD(this)\">&nbsp;\r\n");
777
778 if(urlIDList.contains(idIntValue[i-1])){
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 if(datarows[j]==1){
787 out.write ("<td class=\"b\">\r\n");
788 blueDot++;
789 }
790 else {
791 out.write ("<td class=\"w\">\r\n");
792 }
793 }
794 }
795 }
796 }
797
798 out.write ("</tbody>\r\n");
799 out.write ("</table>\r\n</div></div>\r\n");
800
801 int t1 = ids.length;
802 int t3 = metadataName.length;
803 int t4 = t1*t3;
804 HashMap tempMap = mds.getMetadataSetMap();
805 MetadataSet ms = (MetadataSet)tempMap.get(suffix);
806
807 out.write("<table>\r\n<tbody class=\"table1\">");
808 out.write("<tr>\r\n");
809 out.write("<td class=\"bfont\">This subset shows "+t1+" out of "+mds.getDocNum()+" documents");
810 out.write("<td class=\"bfont\">"+blueDot+" out of "+(mds.getDocNum()*t3)+" metadata items are defined");
811 out.write("<tr>\r\n");
812 out.write("<td class=\"bfont\">This subset shows "+t3+" out of "+(ms.getIndexsList().size())+" metadata elements");
813 out.write("<td class=\"bfont\">Subset completeness: "+dm.round((double)(blueDot*100/t4),5)+"%");
814 out.write("</table>\r\n");
815 out.write ("<p> <a href=\"Overall.html \">Summary</a></p>");
816 out.write ("</body></html>");
817 out.close();
818 fstream.close();
819
820 }catch (Exception e){//Catch exception if any
821 e.printStackTrace();
822 }
823 }
824
825 public void generateOverallStatisticsPage(HashMap MetadataSetMap){
826
827 String fileName = "Overall";
828 ArrayList wholeList = new ArrayList();
829 HashMap hp = mds.getMetadataSetMap();
830 Collection c = hp.values();
831 Iterator i = c.iterator();
832 int counter = 0;
833 String[][] MetadataData = new String[c.size()][2];
834
835 while(i.hasNext()){
836 MetadataSet mds = (MetadataSet)i.next();
837 wholeList.add(mds);
838 ArrayList newMDS = new ArrayList();
839 newMDS.add(mds);
840 MetadataData[counter][0] = mds.getName();
841 MetadataData[counter][1] = dm.getSingleMetadataSetCompleteness(newMDS)+"%";
842 counter++;
843 }
844
845 try{
846 String str = sdf.format(cl.getTime());
847 String collectionFolder = mds.getCollectionName();
848 FileWriter fstream = new FileWriter(destination+fileName+".html");
849 BufferedWriter out = new BufferedWriter(fstream);
850
851 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
852 out.write("<html>\r\n");
853 out.write("<head><title>Summary</title>\r\n");
854 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
855 out.write("<style type=\"text/css\">\r\n");
856 out.write(".tam{height:20; text-align:right}\r\n");
857 out.write("th{height:20; text-align:left}\r\n");
858 out.write("body{font-family: Arial;}\r\n");
859 out.write("</style>\r\n");
860 out.write("<script type=\"text/javascript\" src=\"../script/status3.js\"></script>\r\n");
861 out.write("</head>\r\n");
862 out.write("<body>\r\n");
863 out.write ("<td> <div style=\"float:left;\"> <a href=\"http://www.nzdl.org/greenstone3/mat\">Mat Home</a></div>");
864 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>");
865 out.write("<h3 align=\"center\">Summary</h3>\r\n");
866 out.write("<table border=1 align=\"center\">\r\n");
867 out.write("<colgroup width=\"300\" span=\"2\">\r\n");
868 out.write("<tr>\r\n");
869 out.write("<th>OAI URL: </th>\r\n");
870
871 if(mds.getOAIURL().length()>50){
872 out.write("<td class=\"tam\"><font size=\"2px\"><a href=\""+mds.getOAIURL()+"\">"+mds.getOAIURL().substring(0,47)+"....</a></font>\r\n");
873 }
874 else{
875 out.write("<td class=\"tam\"><font size=\"2px\"><a href=\""+mds.getOAIURL()+"\">"+mds.getOAIURL()+"</a></font>\r\n");
876 }
877 out.write("</tr>\r\n");
878 out.write("<tr>\r\n");
879 out.write("<th>Number of Records:\r\n");
880 out.write("<td class=\"tam\">"+mds.getDocNum()+"\r\n");
881 out.write("</tr>\r\n");
882 out.write("<table border=1 align=\"center\">\r\n");
883 out.write("<colgroup width=\"300\" span=\"2\">\r\n");
884 out.write("<tr>\r\n");
885 out.write("<th>Metadata:\r\n");
886 out.write("<td class=\"tam\"><b>Completeness</b>\r\n");
887 out.write("</tr><br>\r\n");
888
889 for(int a = 0; a<MetadataData.length; a++){
890 out.write ("<tr>\r\n");
891
892 if(MetadataData[a][0].equalsIgnoreCase("dublin")){
893 out.write ("<th><a href =\""+MetadataData[a][0]+".html\">"+metadataSet1+"</a>");
894 }
895
896 else if (MetadataData[a][0].equalsIgnoreCase("extracted")){
897 out.write ("<th><a href =\""+MetadataData[a][0]+".html\">"+metadataSet2+"</a>");
898 }
899 else{
900 out.write ("<th><a href =\""+MetadataData[a][0]+".html\">"+MetadataData[a][0]+"</a>");
901 }
902
903 out.write ("<td class=\"tam\"> "+MetadataData[a][1]);
904 out.write ("</tr>\r\n");
905 }
906
907 out.write ("</table>");
908 out.write("<br>\r\n");
909 out.write("<FORM name=\"test\" onsubmit=\"checkStatus()\" action=\"\">");
910 out.write("<table border=1 align=\"center\">\r\n");
911 out.write ("<colgroup width=\"606\" span=\"1\">\r\n");
912 out.write("<tr>\r\n<th>Customize Visualization");
913 out.write("<tr>\r\n<td><LABEL FOR=\"H1\"><INPUT align=\"left\" TYPE=\"checkbox\" ID=\"H1\">Hide Empty Metadata Elements</LABEL>");
914 out.write("<tr>\r\n<td><LABEL FOR=\"H2\"><INPUT align=\"left\" TYPE=\"checkbox\" ID=\"H2\">Hide Completed Metadata Elements</LABEL>");
915 out.write("<tr>\r\n<td><LABEL FOR=\"H3\"><INPUT align=\"left\" TYPE=\"checkbox\" ID=\"H3\">Hide Documents with Empty Metadata Elements</LABEL>");
916 out.write("<tr>\r\n<td><LABEL FOR=\"H4\"><INPUT align=\"left\" TYPE=\"checkbox\" ID=\"H4\">Hide Documents with Completed Metadata Elements</LABEL>");
917 out.write("<tr>\r\n<th>Metadata: ");
918
919 if(MetadataData.length==1){
920 if(MetadataData[0][0].equalsIgnoreCase("dublin")){
921 out.write("<tr>\r\n<td><LABEL FOR=\"C1\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C1\" checked> Dublin Core </LABEL>");
922 }
923 else {
924 out.write("<tr>\r\n<td><LABEL FOR=\"C4\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C4\" checked>"+MetadataData[0][0]+"</LABEL>");
925 }
926 }
927 else{
928 for(int a = 0; a<MetadataData.length; a++){
929 if(a==0){
930 if(MetadataData[a][0].equalsIgnoreCase("dublin")){
931 out.write("<tr>\r\n<td><LABEL FOR=\"C1\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C1\" checked>Dublin Core</LABEL>");
932 }
933 else {
934 out.write("<tr>\r\n<td><LABEL FOR=\"C4\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C4\" checked>"+MetadataData[a][0]+"</LABEL>");
935 }
936 }
937 else{
938 if(MetadataData[a][0].equalsIgnoreCase("dublin")){
939 out.write("<tr>\r\n<td><LABEL FOR=\"C1\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C1\">"+MetadataData[a][0]+"</LABEL>");
940 }
941 else{
942 out.write("<tr>\r\n<td><LABEL FOR=\"C4\"><INPUT align=\"left\" TYPE=\"radio\" name = \"col\" ID=\"C4\">"+MetadataData[a][0]+"</LABEL>");
943 }
944 }
945 }
946 }
947
948 out.write("<tr>\r\n<th>Order By Completeness : ");
949 out.write("<tr>\r\n<td><LABEL FOR=\"R1\"><INPUT align=\"left\" TYPE=\"radio\" name = \"order\" ID=\"R1\">Best Case to Worst Case</LABEL>");
950 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>");
951 out.write("</table><p align=\"center\"><INPUT TYPE=\"button\" VALUE=\"Show Visualization\" onClick=\"checkStatus()\"> </FORM>");
952 out.write("<p align=\"center\">"+str+"</p>");
953 out.write("</body></html>");
954
955 out.close();
956 fstream.close();
957 }catch (Exception e){//Catch exception if any
958 e.printStackTrace();
959 }
960 }
961
962
963 public void WriteMetadataSetDetailHTML(MetadataSet mdset){
964
965 String fileName = mdset.getName();
966
967 try{
968 String str = sdf.format(cl.getTime());
969 String collectionFolder = mds.getCollectionName();
970 FileWriter fstream = new FileWriter(destination+fileName+".html");
971 BufferedWriter out = new BufferedWriter(fstream);
972
973 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
974 out.write("<html>\r\n");
975 out.write("<head>\r\n<title>Metadata Detail</title>\r\n");
976 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
977 out.write("<style type=\"text/css\">\r\n");
978 out.write("td{height:20;text-align:right;}\r\n");
979 out.write("th{height:20;text-align:left;}\r\n");
980 out.write("body{font-family: Arial;}\r\n");
981 out.write("</style>\r\n");
982 out.write( "</head>\r\n");
983 out.write ("<body>\r\n");
984 out.write ("<td> <div style=\"float:left;\"> <a href=\" Overall.html \">Summary</a></div>");
985 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>");
986
987 if(fileName.equalsIgnoreCase("dublin")){
988 out.write ("<h3 align=\"center\">Metadata Detail: "+metadataSet1+"</h3>\r\n");
989 }
990 else if (fileName.equalsIgnoreCase("extracted")){
991 out.write ("<h3 align=\"center\">Metadata Detail: "+metadataSet2+"</h3>\r\n");
992 }
993 else{
994 out.write ("<h3 align=\"center\">Metadata Detail: "+fileName+"</h3>\r\n");
995 }
996
997 out.write ("<table border=\"1\" align=\"center\">\r\n");
998 out.write ("<colgroup width=\"200\" span=\"2\">\r\n");
999 out.write ("<tr>\r\n");
1000 out.write ("<th>Elements:\r\n");
1001 out.write ("<td><b>Completeness</b>\r\n");
1002 out.write ("</tr>\r\n");
1003
1004 ArrayList elementList = mdset.getIndexsList();
1005 HashMap hp = new HashMap();
1006 int num = elementList.size();
1007
1008 for(int i = 0; i<num; i++){
1009 String elementName = (String)elementList.get(i);
1010 Double elementValue = new Double(dm.Mean(elementName));
1011 hp.put(elementName, elementValue);
1012 }
1013
1014 elementList = new ArrayList();
1015 elementList = dm.sortMap(hp);
1016 num = elementList.size();
1017
1018 for(int i = 0; i<num; i++){
1019 Map.Entry entry = (Map.Entry) elementList.get(i);
1020 String elementName = (String) entry.getKey();
1021
1022 out.write ("<tr>\r\n");
1023 out.write ("<th><a href=\""+ elementName +".html\">"+elementName+"</a>\r\n");
1024 out.write ("<td>"+ dm.Mean(elementName)+"%\r\n");
1025 out.write ("</tr>\r\n");
1026 }
1027
1028 out.write("</table>\r\n");
1029 out.write ("<p align=\"center\"> <a href=\" Overall.html \">Summary</a></p>");
1030 out.write("<p align=\"center\">"+str+"</p>");
1031 out.write ("</body></html>\r\n");
1032
1033 out.close();
1034 fstream.close();
1035 }catch (Exception e){//Catch exception if any
1036 e.printStackTrace();
1037 }
1038 }
1039
1040 public void generateMetadataElementDetailPage(MetadataSet mds){
1041
1042 ArrayList nameList = mds.getIndexsList();
1043
1044 for(int i = 0; i<nameList.size();i++){
1045 WriteMetadataElementDetailHTML((String)nameList.get(i),mds.getName());
1046 }
1047 }
1048
1049 public void WriteMetadataElementDetailHTML(String name, String linkName){
1050
1051 String fileName = name;
1052
1053 try{
1054 String collectionFolder = mds.getCollectionName();
1055
1056 FileWriter fstream = new FileWriter(destination+fileName+".html");
1057 BufferedWriter out = new BufferedWriter(fstream);
1058
1059 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
1060 out.write ("<html>\r\n");
1061 out.write("<head>\r\n<title> "+ name +" </title>\r\n");
1062 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
1063 out.write("<style type=\"text/css\">\r\n");
1064 out.write("td{height:20; text-align:left;}\r\n");
1065 out.write(".tam{height:20; text-align:center}\r\n");
1066 out.write("th{height:20; text-align:left;}\r\n");
1067 out.write("body{font-family: Arial;}\r\n");
1068 out.write("</style>\r\n");
1069 out.write("<script type=\"text/javascript\" src=\"http://www.nzdl.org/greenstone3/mat/script/status3.js\"></script>");
1070 out.write("</head>\r\n");
1071 out.write("<body>\r\n");
1072 out.write ("<td> <div style=\"float:left;\"> <a href=\" Overall.html \">Summary</a></div>");
1073 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>");
1074 out.write("&raquo;");
1075
1076 if(linkName.equalsIgnoreCase("dublin")){
1077 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+metadataSet1+")</a>");
1078 }
1079 else if (linkName.equalsIgnoreCase("extracted")){
1080 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+metadataSet2+")</a>");
1081 }
1082 else{
1083 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+linkName+")</a>");
1084 }
1085
1086 if(name.startsWith("dc.")){
1087 int dot = name.lastIndexOf('.');
1088 dot++;
1089 String nameReplace = name.substring(dot);
1090
1091 if(url.containsKey(nameReplace)){
1092 out.write ("<h3 align=\"center\">Metadata Element Detail:<a href=\""+ url.get(nameReplace) +"\">"+ name +"</a></h3>\r\n");
1093 }
1094 else{
1095 out.write ("<h3 align=\"center\">Metadata Element Detail: "+ name +"</a></h3>\r\n");
1096 }
1097 }
1098
1099 else{
1100 out.write ("<h3 align=\"center\">Metadata Element Detail: "+ name +"</h3>\r\n");
1101 }
1102
1103 out.write ("<table border=\"1\" align=\"center\" width=1000>\r\n");
1104 out.write ("<colgroup width=\"500\" span=\"2\">\r\n");
1105 out.write ("<tr>\r\n");
1106 out.write ("<td><b>Total Number of Records</b>\r\n");
1107 out.write ("<td class=\"tam\">"+mds.getDocNum()+"\r\n");
1108 out.write ("</tr>\r\n");
1109 out.write ("<tr>\r\n");
1110 out.write ("<td><b>Unique Values</b>\r\n");
1111 out.write ("<td class=\"tam\">"+dm.getDistinctNumber(name)+"\r\n");
1112 out.write ("</tr>\r\n");
1113 out.write ("<tr>\r\n");
1114 out.write ("<td><b>Total times element used</b>\r\n");
1115 out.write ("<td class=\"tam\">"+dm.getFrequency(name) +"\r\n");
1116 out.write ("</tr>\r\n");
1117 out.write ("<tr>\r\n");
1118 out.write ("<td><b>No. of records containing element</b>\r\n");
1119 out.write ("<td class=\"tam\"> "+dm.getDocumentUsedElement(name)+"\r\n");
1120 out.write ("</tr>\r\n");
1121 out.write ("<tr>\r\n");
1122
1123 double percentage = dm.Mean(name);
1124
1125 out.write ("<th title = \"The completeness means the arithmetic average \"> <div style=\"float:left;\"> Completeness</div>");
1126 out.write ("<td class=\"tam\"> "+dm.Mean(name) +"%\r\n");
1127 out.write ("</tr>\r\n");
1128 out.write ("<tr>\r\n");
1129 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");
1130 out.write ("<td class=\"tam\"> "+dm.getMinRange(name) +"\r\n");
1131 out.write ("</tr>\r\n");
1132 out.write ("<tr>\r\n");
1133 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");
1134 out.write ("<td class=\"tam\"> "+dm.getMaxRange(name) +"\r\n");
1135 out.write ("</tr>\r\n");
1136 out.write ("<tr>\r\n");
1137 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");
1138 out.write ("<td class=\"tam\"> "+dm.Average(name) +"\r\n");
1139 out.write ("</tr>\r\n");
1140 out.write ("<tr>\r\n");
1141 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");
1142 out.write ("<td class=\"tam\"> "+dm.getMode(name) +"\r\n");
1143 out.write ("</tr>\r\n");
1144 out.write ("<tr>\r\n");
1145 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");
1146 out.write ("<td class=\"tam\">"+dm.ModeFrequency(name) +"%\r\n");
1147 out.write ("</tr>\r\n");
1148
1149 if(mds.getOaiPrefix().equalsIgnoreCase("oai_dc")){
1150
1151 HashMap suggestionMap = generateMetadataElementSortList(fileName,"ASCII",linkName);
1152
1153 if(suggestionMap.size()>0){
1154 boolean status = compareElement(fileName,collectionFolder,suggestionMap,linkName);
1155 if(status){
1156 out.write("<tr><td class=\"tam\"><a href =\""+fileName+"_Suggestion.html\">View Potential Duplicate List</a>");
1157 if(percentage<100){
1158 out.write ("<td class=\"tam\"><a href =\""+fileName+"_IncompletedList.html\">Records missing "+fileName+"</a></td>");
1159 createIncompletedList(fileName,linkName,collectionFolder);
1160 }
1161 else{
1162 out.write("<td class=\"tam\">No Records Missing "+ fileName);
1163 }
1164 }
1165 else{
1166 out.write("<tr><td class=\"tam\">No Potential Duplicates");
1167 if(percentage<100){
1168 out.write ("<td class=\"tam\"><a href =\""+fileName+"_IncompletedList.html\">Records missing "+fileName+"</a></td>");
1169 createIncompletedList(fileName,linkName,collectionFolder);
1170 }
1171 else{
1172 out.write("<td class=\"tam\">No Records Missing "+ fileName);
1173 }
1174 }
1175 }
1176 else{
1177 out.write("<tr><td class=\"tam\">No Potential Duplicates");
1178 if(percentage<100){
1179 out.write ("<td class=\"tam\"><a href =\""+fileName+"_IncompletedList.html\">Records missing "+fileName+"</a></td>");
1180 createIncompletedList(fileName,linkName,collectionFolder);
1181 }
1182 else{
1183 out.write("<td class=\"tam\">No Records Missing "+ fileName);
1184 }
1185 }
1186 generateMetadataElementSortList(fileName,"Frequency-based",linkName);
1187 }
1188 else{
1189
1190 HashMap suggestionMap = generateMetadataElementSortList(fileName,"ASCII",linkName);
1191
1192 if(suggestionMap.size()>0){
1193 boolean status = compareElement(fileName,collectionFolder,suggestionMap,linkName);
1194 if(status){
1195 out.write("<tr><td class=\"tam\"><a href =\""+fileName+"_Suggestion.html\">View Potential Duplicate List</a>");
1196 }
1197 else{
1198 out.write("<tr><td class=\"tam\">No Potential Duplicates");
1199 }
1200 }
1201 else{
1202 out.write("<tr><td class=\"tam\">No Potential Duplicates");
1203 }
1204
1205 if(percentage<100){
1206 out.write ("<td class=\"tam\"><a href =\""+fileName+"_IncompletedList.html\">Records missing "+fileName+"</a></td>");
1207 createIncompletedList(fileName,linkName,collectionFolder);
1208 }
1209 else{
1210 out.write("<td class=\"tam\">No Records Missing "+ fileName);
1211 }
1212
1213 generateMetadataElementSortList(fileName,"Frequency-based",linkName);
1214 }
1215
1216 out.write ("<tr>\r\n");
1217 out.write ("<td class=\"tam\"><a href =\""+fileName+"_Frequency-based.html\">"+"View Full Frequency Sorted list</a>" +
1218 " <td class=\"tam\"><a href =\""+fileName+"_ASCII.html\">"+"View Full ASCII Sorted list</a></td>");
1219 out.write ("</tr>\r\n");
1220 out.write ("</table><br>\r\n");
1221 out.write ("<table border=\"1\" align=\"center\" width=1000>\r\n");
1222 out.write ("<colgroup width=\"500\" span=\"2\">\r\n");
1223 out.write ("<tr>\r\n");
1224 out.write ("<th>ASCII-Based\r\n");
1225 out.write ("<th>First Five\r\n");
1226 out.write ("</tr>\r\n");
1227
1228 String[] temp = dm.getSortList(name,"ASCII");
1229 String[] temp2 = {"&nbsp;","&nbsp;","&nbsp;","&nbsp;","&nbsp;"};
1230 int length = 0;
1231
1232 if(temp.length>=5){
1233 length=5;
1234 }
1235 else if(temp.length<5){
1236 length = temp.length;
1237 }
1238
1239 for(int i =0; i<length; i++){
1240 temp2[i] = temp[i];
1241 }
1242
1243 int x = temp2.length;
1244 for(int a = 0; a<x; a++){
1245 out.write ("<tr>\r\n");
1246 if(!temp2[a].equals("&nbsp;")){
1247 out.write ("<th>"+(a+1)+"\r\n");
1248 }
1249 else{
1250 out.write ("<th>&nbsp;\r\n");
1251 }
1252
1253 if(temp2[a].startsWith("http://")){
1254 if(temp2[a].length()>60){
1255 out.write ("<td><a href = \""+temp2[a]+"\">"+temp2[a].substring(0, 60)+"...</a>");
1256 }
1257 else {
1258 out.write ("<td><a href = \""+temp2[a]+"\">"+temp2[a]+"</a>");
1259 }
1260 }
1261 else{
1262 char singleChar = temp2[a].charAt(0);
1263 if(temp2[a].length()>61 ){
1264 if(temp2[a].startsWith(" ") && ((int)singleChar!=65279)){
1265 temp2[a] = temp2[a].substring(1);
1266 out.write ("<td>"+spaceLeft+space+spaceRight+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#topFive\"> ... </a>\r\n");
1267 }
1268 else if (((int)singleChar==65279)){
1269 temp2[a] = temp2[a].substring(1);
1270 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#topFive\"> ... </a>\r\n");
1271 }
1272 else{
1273 out.write ("<td>"+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#topFive\"> ... </a>\r\n");
1274 }
1275 }
1276 else {
1277 if(temp2[a].startsWith(" ")&& ((int)singleChar!=65279)){
1278 temp2[a] = temp2[a].substring(1);
1279 out.write ("<td>"+spaceLeft+space+spaceRight+temp2[a]+"\r\n");
1280 }
1281 else if (((int)singleChar==65279)){
1282 temp2[a] = temp2[a].substring(1);
1283 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp2[a]+"\r\n");
1284 }
1285 else{
1286 out.write ("<td>"+temp2[a]+"\r\n");
1287 }
1288 }
1289 }
1290 out.write ("</tr>\r\n");
1291 }
1292
1293 String[] temp3 ={"&nbsp;","&nbsp;","&nbsp;","&nbsp;","&nbsp;"};
1294 length = 0;
1295 int start = temp.length;
1296
1297 if(temp.length>=5){
1298 length= 5;
1299 }
1300 else if(temp.length<5){
1301 length = temp.length;
1302 }
1303
1304 for(int i = length; i>0; i--){
1305 temp3[i-1] = temp[start-1];
1306 start--;
1307 }
1308
1309 out.write ("<tr>\r\n");
1310 out.write ("<th>......\r\n");
1311 out.write ("<th>Last Five\r\n");
1312 out.write ("</tr>\r\n");
1313
1314 int counter = temp.length;
1315 start = temp.length;
1316 x = temp3.length;
1317
1318 for(int a = 0; a<x; a++){
1319 out.write ("<tr>\r\n");
1320 if(!temp3[a].equals("&nbsp;")){
1321 out.write ("<th>"+(start-length+1+a));
1322 }
1323 else{
1324 out.write ("<th>&nbsp;");
1325 }
1326 if(temp3[a].startsWith("http://")){
1327 if(temp3[a].length()>60){
1328 out.write ("<td><a href = \""+temp3[a]+"\">"+temp3[a].substring(0, 60)+"...</a>");
1329 }
1330 else {
1331 out.write ("<td><a href = \""+temp3[a]+"\">"+temp3[a]+"</a>");
1332 }
1333 }
1334 else{
1335 char singleChar = temp3[a].charAt(0);
1336 if(temp3[a].length()>61){
1337 if(temp3[a].startsWith(" ") && (int)singleChar!=65279){
1338 temp3[a] = temp3[a].substring(1);
1339 out.write ("<td>"+spaceLeft+space+spaceRight+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#lastFive\"> ... </a>\r\n");
1340 }
1341 else if((int)singleChar==65279){
1342 temp3[a] = temp3[a].substring(1);
1343 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#lastFive\"> ... </a>\r\n");
1344 }
1345 else{
1346 out.write ("<td>"+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_ASCII.html#lastFive\"> ... </a>\r\n");
1347 }
1348 }
1349 else{
1350 if(temp3[a].startsWith(" ")&& (int)singleChar!=65279){
1351 temp3[a] = temp3[a].substring(1);
1352 out.write ("<td>"+spaceLeft+space+spaceRight+temp3[a]+"\r\n");
1353 }
1354 else if((int)singleChar==65279){
1355 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp3[a]+"\r\n");
1356 }
1357 else{
1358 out.write ("<td>"+temp3[a]+"\r\n");
1359 }
1360 }
1361 out.write ("</tr>\r\n");
1362 }
1363 }
1364 out.write ("</table><br>\r\n");
1365 out.write ("<table border=\"1\" align=\"center\" width=1000>\r\n");
1366 out.write ("<colgroup width=\"500\" span=\"2\">\r\n");
1367 out.write ("<tr>\r\n");
1368 out.write ("<th>Frequency-Based:\r\n");
1369 out.write ("<th>First Five\r\n");
1370 out.write ("</tr>\r\n");
1371
1372 HashMap xMap = dm.getDistinctValueMap(name);
1373 temp = dm.getSortList(name,"Frequency-based");
1374 temp2 = new String[] {"&nbsp;","&nbsp;","&nbsp;","&nbsp;","&nbsp;"};
1375 length = 0;
1376
1377 if(temp.length>=5){
1378 length=5;
1379 }
1380 else if(temp.length<5){
1381 length = temp.length;
1382 }
1383
1384 for(int i =0; i<length; i++){
1385 temp2[i] = temp[i];
1386 }
1387
1388 ArrayList tempList = new ArrayList();
1389 for(int i =0; i<length; i++){
1390 if(((Integer)xMap.get(temp2[i])).intValue()==1){
1391 tempList.add(temp2[i]);
1392 }
1393 }
1394
1395 x = temp2.length;
1396 for(int a = 0; a<x; a++){
1397 out.write ("<tr>\r\n");
1398 if(!temp2[a].equals("&nbsp;")){
1399 char singleChar = temp2[a].charAt(0);
1400 out.write ("<th>"+(a+1)+". (No. of occurrences: "+((Integer)xMap.get(temp2[a])).toString()+")\r\n");
1401 if(temp2[a].startsWith("http://") && (int)singleChar != 65279){
1402 if(temp2[a].length()>61){
1403 out.write ("<td><a href = \""+temp2[a]+"\">"+temp2[a].substring(0, 60)+"...</a>");
1404 }
1405 else {
1406 out.write ("<td><a href = \""+temp2[a]+"\">"+temp2[a]+"</a>");
1407 }
1408 }
1409
1410 else{
1411 if(temp2[a].length()>61){
1412 if(temp2[a].startsWith(" ")){
1413 temp2[a] = temp2[a].substring(1);
1414 out.write ("<td>"+spaceLeft+space+spaceRight+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#topFive\"> ... </a>\r\n");
1415 }
1416 else if((int)singleChar == 65279){
1417 temp2[a] = temp2[a].substring(1);
1418 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#topFive\"> ... </a>\r\n");
1419 }
1420 else{
1421 out.write ("<td>"+temp2[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#topFive\"> ... </a>\r\n");
1422 }
1423 }
1424 else{
1425 if(temp2[a].startsWith(" ")){
1426 temp2[a] = temp2[a].substring(1);
1427 out.write ("<td>"+spaceLeft+space+spaceRight+temp2[a]+"\r\n");
1428 }
1429 else if((int)singleChar == 65279){
1430 temp2[a] = temp2[a].substring(1);
1431 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp2[a]+"\r\n");
1432 }
1433 else{
1434 out.write ("<td>"+temp2[a]+"\r\n");
1435 }
1436 }
1437 }
1438 }
1439 else{
1440 out.write ("<th>&nbsp;\r\n");
1441 out.write ("<td>\r\n");
1442 }
1443 out.write ("</tr>\r\n");
1444 }
1445
1446 temp3 = new String[]{"&nbsp;","&nbsp;","&nbsp;","&nbsp;","&nbsp;"};
1447 length = 0;
1448 start = temp.length;
1449
1450 if(temp.length>=5){
1451 length= 5;
1452 }
1453 else if(temp.length<5){length = temp.length;}
1454
1455 for(int i = length; i>0; i--){
1456 temp3[i-1] = temp[start-1];
1457 start--;
1458 }
1459
1460 out.write ("<tr>\r\n");
1461 out.write ("<th>......\r\n");
1462 out.write ("<th>Last Five\r\n");
1463 out.write ("</tr>\r\n");
1464
1465 x = temp3.length;
1466 start = temp.length;
1467 for(int a = 0; a<x; a++){
1468 out.write ("<tr>\r\n");
1469 if(!temp3[a].equals("&nbsp;")){
1470 out.write ("<th>"+(start-length+1+a)+". (No. of occurrences: "+((Integer)xMap.get(temp3[a])).toString()+")\r\n");
1471 char singleChar = temp3[a].charAt(0);
1472 if(temp3[a].startsWith("http://")){
1473 if(temp3[a].length()>60){
1474 out.write ("<td><a href = \""+temp3[a]+"\">"+temp3[a].substring(0, 60)+"...</a>");
1475 }
1476 else {
1477 out.write ("<td><a href = \""+temp3[a]+"\">"+temp3[a]+"</a>");
1478 }
1479 }
1480 else{
1481 if(temp3[a].length()>61){
1482 if(temp3[a].startsWith(" ")){
1483 temp3[a] = temp3[a].substring(1);
1484 out.write ("<td>"+spaceLeft+space+spaceRight+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#lastFive\"> ... </a>\r\n");
1485 }
1486 else if((int)singleChar == 65279){
1487 temp3[a] = temp3[a].substring(1);
1488 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#lastFive\"> ... </a>\r\n");
1489 }
1490 else{
1491 out.write ("<td>"+temp3[a].substring(0, 60)+"<a href=\""+fileName+"_Frequency-based.html#lastFive\"> ... </a>\r\n");
1492 }
1493 }
1494 else{
1495 if(temp3[a].startsWith(" ")){
1496 out.write ("<td>"+spaceLeft+space+spaceRight+temp3[a]+"\r\n");
1497 }
1498 else if((int)singleChar == 65279){
1499 temp3[a] = temp3[a].substring(1);
1500 out.write ("<td>"+spaceLeft+oddChar+spaceRight+temp3[a]+"\r\n");
1501 }
1502 else{
1503 out.write ("<td>"+temp3[a]+"\r\n");
1504 }
1505 }
1506 }
1507 }
1508 else{
1509 out.write ("<th>&nbsp;\r\n");
1510 out.write ("<th>\r\n");
1511 }
1512 out.write ("</tr>\r\n");
1513 }
1514
1515 out.write ("</table>\r\n");
1516 out.write ("<p> <a href=\" Overall.html \">Summary</a>");
1517 out.write ("&raquo;");
1518
1519 if(linkName.equalsIgnoreCase("dublin")){
1520 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+metadataSet1+")</a>");
1521 }
1522 else if (linkName.equalsIgnoreCase("extracted")){
1523 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+metadataSet2+")</a>");
1524 }
1525 else{
1526 out.write ("<a href=\""+linkName+".html\">Metadata Detail ("+linkName+")</a>");
1527 }
1528
1529 out.write ("</body></html>\r\n");
1530 out.close();
1531 fstream.close();
1532 }catch (Exception e){//Catch exception if any
1533 e.printStackTrace();
1534 }
1535 }
1536
1537 private HashMap generateMetadataElementSortList(String title, String sort, String metadataSetName){
1538
1539 String fileName = title+"_"+sort;
1540 String collectionFolder = mds.getCollectionName();
1541 String IDENTIFIER = linkIdentifier;
1542 SearchLink sl = new SearchLink(mds.StatsDirectory);
1543 HashMap suggestionMap = new HashMap();
1544 HashMap valueMap = sl.createValueMap(title,collectionFolder);
1545 HashMap linkMap = sl.createLinkMap(IDENTIFIER,collectionFolder);
1546 HashMap internalIDMap = dm.getInternalIdentifier(title);
1547 boolean status = false;
1548
1549 try{
1550 FileWriter fstream = new FileWriter(destination+fileName+".html");
1551 BufferedWriter out = new BufferedWriter(fstream);
1552
1553 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
1554 out.write ("<html>\r\n");
1555 out.write("<head>\r\n<title>Metadata Element Sort List</title>\r\n");
1556 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
1557 out.write("<script type=\"text/javascript\" src=\"../script/getInfomation.js\"></script>\r\n");
1558 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/yahoo/yahoo-min.js\"></script>\r\n");
1559 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/event/event-min.js\"></script>\r\n");
1560 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/connection/connection-min.js\"></script>\r\n");
1561 out.write("<style type=\"text/css\">\r\n");
1562 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
1563 out.write(".tam{height:20; text-align:center}\r\n");
1564 out.write("th{height:20; text-align:center;}\r\n");
1565 out.write("body{font-family: Arial;}\r\n");
1566 out.write("</style>\r\n");
1567 out.write("</head><body id=\""+collectionFolder+"\">\r\n");
1568 out.write("<td><div style=\"float:left;\"><a href=\" Overall.html \">Summary</a>");
1569 out.write("&raquo;");
1570
1571 if(metadataSetName.equalsIgnoreCase("dublin")){
1572 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
1573 }
1574 else if (metadataSetName.equalsIgnoreCase("extracted")){
1575 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
1576 }
1577 else{
1578 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
1579 }
1580
1581 out.write("&raquo;");
1582 out.write("<a href=\""+title+".html\">"+title+"</a></div>");
1583 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>");
1584
1585 String[] list = dm.getSortList(title, sort);
1586
1587 if(list.length>=1){
1588 if(sort.equals("ASCII")){
1589 out.write("<table border=\"1\" align=\"center\" width=\"800\">\r\n");
1590 out.write ("<h2 align=\"center\">"+ title+"</h2>\r\n");
1591 out.write ("<th class=\"tam\">ASCII Sort\r\n");
1592 out.write ("<th class=\"tam\">Element Values\r\n");
1593 out.write ("<th class=\"tam\">Source Documents\r\n");
1594 out.write ("<th class=\"tam\">Internal Link\r\n");
1595 out.write ("<a name='topFive'>\r\n");
1596
1597 int counter = 0;
1598
1599 for(int i = 0; i<list.length; i++){
1600 if(list.length<=5 && i==0){
1601 out.write ("<a name='lastFive'>\r\n");
1602 }
1603 else if((list.length>5) && (list.length-5==i)){
1604 out.write ("<a name='lastFive'>\r\n");
1605 }
1606
1607 InternalLink il= (InternalLink)internalIDMap.get(list[i]);
1608 ArrayList alist2 = il.retrieveList();
1609 String id = (String)alist2.get(0);
1610 id = id.substring(4);
1611 out.write("<tr id=\""+id+"\" >\r\n");
1612
1613 if(list[i].length()>=201){
1614 if(list[i].startsWith("http://")){
1615 if(title.equals(IDENTIFIER)){
1616 if(list[i].length()>=100){
1617 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1618 }
1619 else{
1620 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1621 }
1622 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
1623 }
1624 else{
1625 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1626 if(list[i].length()>=100){
1627 String url = (String)alist.get(0);
1628 if(alist.size()==1){
1629 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1630 if(url.startsWith("http://")){
1631 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1632 }else{
1633 out.write("<td>Source\r\n");
1634 }
1635 }
1636 else if(alist.size()>1){
1637 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1638 if(url.startsWith("http://")){
1639 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1640 }else{
1641 out.write("<td>Source\r\n");
1642 }
1643 }
1644 else{
1645 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1646 out.write("<td>No Source Available\r\n");
1647 }
1648 suggestionMap.put(list[i], url);
1649 }
1650 else{
1651 String url = (String)alist.get(0);
1652 if(alist.size()==1){
1653 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1654 if(url.startsWith("http://")){
1655 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1656 }else{
1657 out.write("<td>Source\r\n");
1658 }
1659 }
1660 else if(alist.size()>1){
1661 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1662 if(url.startsWith("http://")){
1663 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1664 }else{
1665 out.write("<td>Source\r\n");
1666 }
1667 }
1668 else{
1669 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1670 out.write("<td>No Source Available\r\n");
1671 }
1672 suggestionMap.put(list[i], url);
1673 }
1674 }
1675 }
1676 else{
1677 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1678
1679
1680 if(list[i].startsWith(" ")){
1681 String elements = list[i];
1682 list[i] = list[i].substring(1);
1683
1684 if(alist.size()==1){
1685 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i].substring(0, 200)+"...<td>"+
1686 "<a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1687 suggestionMap.put(elements, (String)alist.get(0));
1688 }
1689 else if(alist.size()>1){
1690 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i].substring(0, 200)+"...<td>"+
1691 "<a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1692 suggestionMap.put(elements, (String)alist.get(0));
1693 }
1694 else{
1695 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i].substring(0, 200)+"...<td>No Source Available \r\n");
1696 suggestionMap.put(elements, "No Source Available");
1697 }
1698 }
1699 else{
1700 if(alist.size()==1){
1701 out.write("<td>"+(counter+1)+"<td>"+list[i].substring(0, 200)+"...<td>" +
1702 "<a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1703 suggestionMap.put(list[i], (String)alist.get(0));
1704 }
1705 else if(alist.size()>1){
1706 out.write("<td>"+(counter+1)+"<td>"+list[i].substring(0, 200)+"...<td>" +
1707 "<a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1708 suggestionMap.put(list[i], (String)alist.get(0));
1709 }
1710 else{
1711 out.write("<td>"+(counter+1)+"<td>"+list[i].substring(0, 200)+"... <td>No Source Available\r\n");
1712 suggestionMap.put(list[i],"No Source Available");
1713 }
1714 }
1715 }
1716
1717 }
1718 else{
1719 if(list[i].startsWith("http://")){
1720 if(title.equals(IDENTIFIER)){
1721 if(list[i].length()>=100){
1722 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1723 }
1724 else{
1725 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1726 }
1727 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
1728 }
1729 else{
1730 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1731
1732 if(list[i].length()>=100){
1733 if(alist.size()==1){
1734 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1735 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1736 }
1737 else if(alist.size()>1){
1738 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
1739 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1740 }
1741 else{
1742 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1743 out.write("<td>No Source Available\r\n");
1744 }
1745 }
1746 else{
1747 if(alist.size()==1){
1748 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1749 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1750 suggestionMap.put(list[i], (String)alist.get(0));
1751 }
1752 else if(alist.size()>1){
1753 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1754 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1755 suggestionMap.put(list[i], (String)alist.get(0));
1756 }
1757 else{
1758 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1759 out.write("<td>No Source Available\r\n");
1760 suggestionMap.put(list[i],"No Source Available");
1761 }
1762 }
1763 }
1764 }
1765 else{
1766 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1767
1768 if(list[i].startsWith(" ")){
1769 String elements = list[i];
1770 list[i] = list[i].substring(1);
1771
1772 if(alist.size()==1){
1773 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i]+
1774 "<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1775 suggestionMap.put(list[i], (String)alist.get(0));
1776 }
1777 else if(alist.size()>1){
1778 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i]+
1779 "<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1780 suggestionMap.put(list[i], (String)alist.get(0));
1781 }
1782 else{
1783 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i]+"<td>No Source Available\r\n");
1784 suggestionMap.put(elements,"No Source Available");
1785 }
1786 }
1787 else{
1788 if(alist.size()==1){
1789 out.write("<td>"+(counter+1)+"<td>"+list[i]+"<td><a href=\""+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]+"<td><a href=\""+alist.get(0)+"\">Source</a>...\r\n");
1794 suggestionMap.put(list[i],(String)alist.get(0));
1795 }
1796 else{
1797 out.write("<td>"+(counter+1)+"<td>"+list[i]+"<td>No Source Available\r\n");
1798 suggestionMap.put(list[i],"No Source Available");
1799 }
1800 }
1801 }
1802 }
1803 counter++;
1804 out.write("<td onclick=\"GD(this)\">View");
1805 }
1806
1807 if(suggestionMap.size()>1 && !title.equals(IDENTIFIER)){
1808 status = true;
1809 }
1810 }
1811 else{
1812 out.write("<table border=\"1\" align=\"center\" width=\"800\">\r\n");
1813 out.write ("<h2 align=\"center\">"+ title+"</h2>\r\n");
1814 out.write ("<th class=\"tam\">&nbsp;\r\n");
1815 out.write ("<th class=\"tam\">Frequency\r\n");
1816 out.write ("<th class=\"tam\">Element Values\r\n");
1817 out.write ("<th class=\"tam\">Source Documents\r\n");
1818 out.write ("<th class=\"tam\">Internal Link\r\n");
1819 out.write ("<a name='topFive'>\r\n");
1820
1821 HashMap xMap = dm.getDistinctValueMap(title);
1822 int counter = 0;
1823
1824 for(int i = 0; i<list.length; i++){
1825 if(list.length<=5 && i==0){
1826 out.write ("<a name='lastFive'>\r\n");
1827 }
1828 else if((list.length>5) && (list.length-5==i)){
1829 out.write ("<a name='lastFive'>\r\n");
1830 }
1831
1832 InternalLink il= (InternalLink)internalIDMap.get(list[i]);
1833 ArrayList alist2 = il.retrieveList();
1834 String id = (String)alist2.get(0);
1835 id = id.substring(4);
1836
1837 out.write("<tr id=\""+id+"\" >\r\n");
1838 out.write("<td>"+(counter+1)+"<th> "+((Integer)xMap.get(list[i])).toString()+"\r\n");
1839
1840 if(list[i].length()>=201){
1841 if(list[i].startsWith("http://")){
1842 if(title.equals(IDENTIFIER)){
1843 if(list[i].length()>=100){
1844 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1845 }
1846 else{
1847 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1848 }
1849 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
1850 }
1851 else{
1852 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1853 if(list[i].length()>=100){
1854 if(alist.size()==1){
1855 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1856 out.write("<td><a href=\""+alist.get(0)+"\">Source</a>\r\n");
1857 }
1858 else if (alist.size()>1){
1859 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1860 out.write("<td><a href=\""+alist.get(0)+"\">Source</a>...\r\n");
1861 }
1862 else{
1863 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1864 out.write("<td>No Source Available\r\n");
1865 }
1866 }
1867 else{
1868 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1869 }
1870 }
1871 }
1872 else{
1873 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1874
1875 if(list[i].startsWith(" ")){
1876 list[i]=list[i].substring(1);
1877 if(alist.size()==1){
1878 out.write("<td>"+spaceLeft+space+spaceRight+list[i].substring(0,200)+"...<td>"+
1879 "<a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1880 }
1881 else if(alist.size()>1){
1882 out.write("<td>"+spaceLeft+space+spaceRight+list[i].substring(0,200)+"...<td>"+
1883 "<a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1884 }
1885 else{
1886 out.write("<td>"+spaceLeft+space+spaceRight+list[i].substring(0,200)+"...<td>No Source Available\r\n");
1887 }
1888 }
1889 else{
1890 if(alist.size()==1){
1891 out.write("<td>"+list[i].substring(0,200)+"...<td>" +
1892 "<a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1893 }
1894 else if(alist.size()>1){
1895 out.write("<td>"+list[i].substring(0,200)+"...<td>"+
1896 "<a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1897 }
1898 else{
1899 out.write("<td>"+list[i].substring(0,200)+"...<td>No Source Available\r\n");
1900 }
1901 }
1902 }
1903 }
1904 else{
1905 if(list[i].startsWith("http://")){
1906 if(title.equals(IDENTIFIER)){
1907 if(list[i].length()>=100){
1908 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1909 }
1910 else{
1911 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1912 }
1913 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
1914 }
1915 else{
1916 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1917
1918 if(list[i].length()>=100){
1919 if(alist.size()==1){
1920 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1921 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1922 }
1923 else if(alist.size()>1){
1924 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1925 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1926 }
1927 else{
1928 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
1929 out.write("<td>No Source Available\r\n");
1930 }
1931 }
1932 else{
1933 if(alist.size()==1){
1934 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1935 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1936 }
1937 else if(alist.size()>1){
1938 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1939 out.write("<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1940 }
1941 else{
1942 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
1943 out.write("<td>No Source Available\r\n");
1944 }
1945 }
1946 }
1947 }
1948 else{
1949
1950 ArrayList alist= sl.CreateIndentifierLinkPage(title, list[i],IDENTIFIER, collectionFolder,valueMap,linkMap);
1951
1952 if(list[i].startsWith(" ")){
1953 list[i]=list[i].substring(1);
1954 if(alist.size()==1){
1955 out.write("<td>"+spaceLeft+space+spaceRight+list[i]+
1956 "<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1957 }
1958 else if(alist.size()>1){
1959 out.write("<td>"+spaceLeft+space+spaceRight+list[i]+
1960 "<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1961 }
1962 else{
1963 out.write("<td>"+spaceLeft+space+spaceRight+list[i]+"<td>No Source Available\r\n");
1964 }
1965 }
1966 else{
1967 if(alist.size()==1){
1968 out.write("<td>"+list[i]+"<td><a href=\""+(String)alist.get(0)+"\">Source</a>\r\n");
1969 }
1970 else if(alist.size()>1){
1971 out.write("<td>"+list[i]+"<td><a href=\""+(String)alist.get(0)+"\">Source</a>...\r\n");
1972 }
1973 else{
1974 out.write("<td>"+list[i]+"<td>No Source Available\r\n");
1975 }
1976 }
1977 }
1978 }
1979 counter++;
1980 out.write("<td onclick=\"GD(this)\">View");
1981 }
1982 }
1983 out.write("</table>\r\n");
1984 }
1985 else{
1986 out.write("<p>Sorry! The list is empty</p>");
1987 }
1988 out.write("<tr>\r\n");
1989 out.write("<p><a href=\" Overall.html \">Summary</a>");
1990 out.write("&raquo;");
1991
1992 if(metadataSetName.equalsIgnoreCase("dublin")){
1993 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
1994 }
1995 else if (metadataSetName.equalsIgnoreCase("extracted")){
1996 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
1997 }
1998 else{
1999 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2000 }
2001
2002 out.write("&raquo;");
2003 out.write("<a href=\""+title+".html\">"+title+"</a></p>");
2004 out.write("</body></html>\r\n");
2005 out.close();
2006
2007 fstream.close();
2008 }catch(Exception e){
2009 e.printStackTrace();
2010 }
2011
2012 if(status){
2013 return suggestionMap;
2014 }
2015 else{
2016 return new HashMap();
2017 }
2018 }
2019
2020 private void setupMetadataLink(String fileName){
2021
2022 try{
2023
2024 DocumentBuilderFactory docBuilderFactory = DocumentBuilderFactory.newInstance();
2025 DocumentBuilder docBuilder = docBuilderFactory.newDocumentBuilder();
2026 Document doc = docBuilder.newDocument();
2027 //doc = docBuilder.parse (new File("/research/cc108/greenstone3/web/mat/script/"+fileName+".xml"));
2028 String gsdl3Home = GlobalProperties.getGSDL3Home();
2029 String fileSeparator = File.separator;
2030 doc = docBuilder.parse (new File(gsdl3Home+ fileSeparator + "mat"+ fileSeparator +"script"+ fileSeparator + fileName+".xml"));
2031 Element rootNode = doc.getDocumentElement();
2032 NodeList listOfName = rootNode.getElementsByTagName("metadataElement");
2033 url = new HashMap();
2034
2035 for(int i = 0; i<listOfName.getLength(); i++){
2036 Node NameNode = listOfName.item(i);
2037 Element docElement = (Element)NameNode;
2038 NodeList valueList = docElement.getElementsByTagName("URL");
2039 Node urlParentNode = valueList.item(0);
2040 String urlText = urlParentNode.getChildNodes().item(0).getNodeValue();
2041
2042 valueList = docElement.getElementsByTagName("name");
2043 Node urlNameParentNode = valueList.item(0);
2044 String urlNameText = urlNameParentNode.getChildNodes().item(0).getNodeValue();
2045
2046 url.put(urlNameText, urlText);
2047 }
2048 }catch(Exception e){
2049 e.printStackTrace();
2050 }
2051 }
2052
2053 private void createIncompletedList(String fileName, String metadataSetName, String collectionFolder){
2054
2055 HashMap hp = dm.getIdentifierLink(linkIdentifier);
2056 String[] ids = dm.getDocumentIDList(fileName);
2057
2058 for(int i = 0; i<ids.length; i++){
2059 hp.remove(ids[i]);
2060 }
2061
2062 try{
2063 FileWriter fstream = new FileWriter(destination+fileName+"_IncompletedList.html");
2064 BufferedWriter out = new BufferedWriter(fstream);
2065
2066 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
2067 out.write ("<html>\r\n");
2068 out.write("<head>\r\n<title>Incompleted Document List</title>\r\n");
2069 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
2070 out.write("<style type=\"text/css\">\r\n");
2071 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
2072 out.write(".tam{height:20; text-align:center}\r\n");
2073 out.write("th{height:20; text-align:center;}\r\n");
2074 out.write("body{font-family: Arial;}\r\n");
2075 out.write("</style>\r\n");
2076 out.write("</head><body>\r\n");
2077 out.write("<p><a href=\" Overall.html \">Summary</a>");
2078 out.write("&raquo;");
2079
2080 if(metadataSetName.equalsIgnoreCase("dublin")){
2081 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2082 }
2083 else if (metadataSetName.equalsIgnoreCase("extracted")){
2084 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2085 }
2086 else{
2087 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2088 }
2089
2090 out.write("&raquo;");
2091 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
2092 out.write("<h3 align=center>"+fileName+" does not appear in the following documents</h3>");
2093 out.write("<table border=1 align=center><tr><th>Document ID<th>Source Link\r\n");
2094
2095 int counter = 0;
2096 Set s = hp.keySet();
2097 Iterator i = s.iterator();
2098
2099 while(i.hasNext()){
2100 counter++;
2101 String keys = (String)i.next();
2102 InternalLink il = (InternalLink) hp.get(keys);
2103 ArrayList alist = il.retrieveList();
2104 String url = (String)alist.get(0);
2105 out.write("<tr><td>"+counter+"<td><a href=\""+url+"\">"+url+"</a>\r\n");
2106 }
2107
2108 out.write("</table></body></html>\r\n");
2109 out.write("<p><a href=\" Overall.html \">Summary</a>");
2110 out.write("&raquo;");
2111
2112 if(metadataSetName.equalsIgnoreCase("dublin")){
2113 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2114 }
2115 else if (metadataSetName.equalsIgnoreCase("extracted")){
2116 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2117 }
2118 else{
2119 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2120 }
2121
2122 out.write("&raquo;");
2123 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
2124
2125 out.close();
2126 fstream.close();
2127 }catch(IOException ex){
2128 ex.printStackTrace();
2129 }
2130 }
2131
2132 public boolean compareElement(String fileName,String collectionFolder,HashMap suggestionMap, String metadataSetName){
2133
2134 boolean status = false;
2135 Set kset = suggestionMap.keySet();
2136 ArrayList arrayList = new ArrayList();
2137 Iterator is = kset.iterator();
2138
2139 while(is.hasNext()){
2140 arrayList.add((String)is.next());
2141 }
2142
2143 HashMap distanceMap = new HashMap();
2144 int totalLength = 0;
2145 int arrayListLength = arrayList.size();
2146 int counter = 0;
2147 double distance ;
2148
2149 try{
2150 for(int i = 0; i<arrayListLength; i++){
2151 String keyword = (String)arrayList.get(i);
2152 int spaceCounterFront1 = 0;
2153 int spaceCounterEnd1 = 0;
2154 String keywordClone = keyword;
2155 String _keywordClone = keyword;
2156 String bkKeyWord = keyword;
2157 keywordClone = removeUnusedCharacter(keywordClone);
2158
2159 while(true){
2160 if(_keywordClone.length()>1){
2161 if(_keywordClone.charAt(0)==' '){
2162 _keywordClone = _keywordClone.substring(1);
2163 spaceCounterFront1++;
2164 }
2165 else{
2166 if(_keywordClone.charAt(_keywordClone.length()-1)==' '){
2167 _keywordClone = _keywordClone.substring(0,_keywordClone.length()-1);
2168 spaceCounterEnd1++;
2169 }
2170 else{
2171 break;
2172 }
2173 }
2174 }
2175 else{
2176 break;
2177 }
2178 }
2179
2180 int keywordLength = keywordClone.length();
2181 totalLength = totalLength + keywordLength;
2182
2183 for(int j = i+1; j<arrayListLength; j++){
2184 String keyword2 = (String)arrayList.get(j);
2185 counter ++;
2186 double pre_cost = 0;
2187 String keywordClone2 = keyword2;
2188 String _keywordClone2 = keyword2;
2189 String bkKeyWord2 = keyword2;
2190 CostModel cm = new CostModel();
2191 cm = removeUnusedCharacter(keywordClone2,pre_cost);
2192 keywordClone2 = cm.getString();
2193 pre_cost = cm.getCost();
2194
2195 int spaceCounterFront2 = 0;
2196 int spaceCounterEnd2 = 0;
2197
2198 while(true){
2199 if(_keywordClone2.length()>1){
2200 if(_keywordClone2.charAt(0)==' '){
2201 _keywordClone2 = _keywordClone2.substring(1);
2202 spaceCounterFront2++;
2203 }
2204 else{
2205 if(_keywordClone2.charAt(_keywordClone2.length()-1)==' '){
2206 _keywordClone2 = _keywordClone2.substring(0,_keywordClone2.length()-1);
2207 spaceCounterEnd2++;
2208 }
2209 else{
2210 break;
2211 }
2212 }
2213 }
2214 else{
2215 break;
2216 }
2217 }
2218
2219 if(counter == 50000){
2220 counter = 0;
2221 }
2222
2223 int keyword2Length = keywordClone2.length();
2224
2225 if(keywordLength>(keyword2Length+2) || (keywordLength+2)<keyword2Length ){}
2226 else{
2227 distance = calculateEditDistance(keywordClone.toLowerCase().toCharArray(),keywordClone2.toLowerCase().toCharArray());
2228 distance = distance + pre_cost;
2229
2230 if(distance<=2){
2231 if(distanceMap.containsKey(_keywordClone)){
2232 InternalLink il = (InternalLink)distanceMap.get(_keywordClone);
2233 String keywordHolder = _keywordClone2;
2234
2235 while(true){
2236 if(keywordHolder.indexOf(" ")!=-1){
2237 keywordHolder = keywordHolder.replaceFirst(" ",spaceLeft+space+spaceRight+" ");
2238 }
2239 else{
2240 break;
2241 }
2242 }
2243
2244 for(int a = 0; a<spaceCounterFront2; a++){
2245 keywordHolder = spaceLeft+space+spaceRight+keywordHolder;
2246 }
2247
2248 for(int a = 0; a<spaceCounterEnd2; a++){
2249 keywordHolder = keywordHolder+spaceLeft+space+spaceRight;
2250 }
2251
2252 InternalLink il2 = new InternalLink();
2253 il2.setValue(keywordHolder);
2254
2255 if(suggestionMap.containsKey(bkKeyWord2)){
2256 il2.increaseElement((String)suggestionMap.get(bkKeyWord2));
2257 }
2258 else{
2259 il2.increaseElement("&nbsp;");
2260 }
2261
2262 il.increaseNode(il2);
2263 distanceMap.put(_keywordClone, il);
2264 }
2265
2266 else{
2267 InternalLink il = new InternalLink();
2268 String keywordHolder = _keywordClone;
2269
2270 while(true){
2271 if(keywordHolder.indexOf(" ")!=-1){
2272 keywordHolder = keywordHolder.replaceFirst(" ",spaceLeft+space+spaceRight+" ");
2273 }
2274 else{
2275 break;
2276 }
2277 }
2278
2279 for(int a = 0; a<spaceCounterFront1; a++){
2280 keywordHolder = spaceLeft+space+spaceRight+keywordHolder;
2281 }
2282
2283 for(int a = 0; a<spaceCounterEnd1; a++){
2284 keywordHolder = keywordHolder+spaceLeft+space+spaceRight;
2285 }
2286
2287 InternalLink il2 = new InternalLink();
2288 il2.setValue(keywordHolder);
2289 if(suggestionMap.containsKey(bkKeyWord)){
2290 il2.increaseElement((String)suggestionMap.get(bkKeyWord));
2291 }
2292 else{
2293 il2.increaseElement("&nbsp;");
2294 }
2295
2296 il.increaseNode(il2);
2297 keywordHolder = _keywordClone2;
2298
2299 while(true){
2300 if(keywordHolder.indexOf(" ")!=-1){
2301 keywordHolder = keywordHolder.replaceFirst(" ",spaceLeft+space+spaceRight+" ");
2302 }
2303 else{
2304 break;
2305 }
2306 }
2307
2308 for(int a = 0; a<spaceCounterFront2; a++){
2309 keywordHolder = spaceLeft+space+spaceRight+keywordHolder;
2310 }
2311
2312 for(int a = 0; a<spaceCounterEnd2; a++){
2313 keywordHolder = keywordHolder+spaceLeft+space+spaceRight;
2314 }
2315
2316 InternalLink il3 = new InternalLink();
2317 il3.setValue(keywordHolder);
2318 if(suggestionMap.containsKey(bkKeyWord2)){
2319 il3.increaseElement((String)suggestionMap.get(bkKeyWord2));
2320 }
2321 else{
2322 il3.increaseElement("&nbsp;");
2323 }
2324
2325 il.increaseNode(il3);
2326 distanceMap.put(_keywordClone, il);
2327 }
2328 }
2329 }
2330 }
2331 }
2332
2333 if(distanceMap.size()!=0){
2334 generateHTML(distanceMap,fileName,metadataSetName);
2335 status = true;
2336 }
2337 }catch(Exception ex){
2338 ex.printStackTrace();
2339 }
2340 return status;
2341 }
2342
2343 private void generateHTML(HashMap distanceMap, String fileName,String metadataSetName){
2344
2345 String collectionFolder = mds.getCollectionName();
2346
2347 try{
2348 FileWriter fstream = new FileWriter(destination+fileName+"_Suggestion.html");
2349 BufferedWriter out = new BufferedWriter(fstream);
2350
2351 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
2352 out.write ("<html>\r\n");
2353 out.write("<head>\r\n<title>Potential Duplicate List</title>\r\n");
2354 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
2355 out.write("<style type=\"text/css\">\r\n");
2356 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
2357 out.write(".tam{height:20; text-align:center}\r\n");
2358 out.write("th{height:20; text-align:left;}\r\n");
2359 out.write("body{font-family: Arial;}\r\n");
2360 out.write("</style>\r\n");
2361 out.write("</head><body>\r\n");
2362 out.write("<p><a href=\" Overall.html \">Summary</a>");
2363 out.write("&raquo;");
2364
2365 if(metadataSetName.equalsIgnoreCase("dublin")){
2366 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2367 }
2368 else if (metadataSetName.equalsIgnoreCase("extracted")){
2369 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2370 }
2371 else{
2372 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2373 }
2374
2375 out.write("&raquo;");
2376 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
2377 out.write("<h3 align=center>Potential Duplicate List</h3>");
2378 out.write("<table border=1 align=center>\r\n");
2379
2380 Set s = distanceMap.keySet();
2381 Iterator i = s.iterator();
2382
2383 while(i.hasNext()){
2384 String keyword = (String)i.next();
2385 InternalLink il = (InternalLink)distanceMap.get(keyword);
2386 ArrayList alist = il.retrieveNodeList();
2387
2388 out.write("<tr><th>Original Text<th>Source Link\r\n");
2389
2390 for(int a = 0; a<alist.size(); a++){
2391 InternalLink il2 = new InternalLink();
2392 il2 = (InternalLink)alist.get(a);
2393 String url = (String)il2.retrieveList().get(0);
2394 out.write("<tr><td>"+il2.getValue()+"<td><a href=\""+url+"\">"+url+"</a>");
2395 }
2396 out.write("</tr>");
2397 }
2398 out.write("</table>\r\n");
2399 out.write("<p><a href=\" Overall.html \">Summary</a>");
2400 out.write("&raquo;");
2401
2402 if(metadataSetName.equalsIgnoreCase("dublin")){
2403 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2404 }
2405 else if (metadataSetName.equalsIgnoreCase("extracted")){
2406 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2407 }
2408 else{
2409 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2410 }
2411
2412 out.write("&raquo;");
2413 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
2414 out.write("</body></html>\r\n");
2415 out.close();
2416
2417 fstream.close();
2418 }catch(IOException ex){
2419 ex.printStackTrace();
2420 }
2421 }
2422
2423 public int calculateEditDistance(char[] args1, char[] args2){
2424
2425 int n = args1.length;
2426 int m = args2.length;
2427
2428 if (n == 0) {
2429 return m;
2430 }
2431 else if (m == 0) {
2432 return n;
2433 }
2434
2435 int[] p = new int[n + 1];
2436 int[] d = new int[n + 1];
2437 int[] _d;
2438 int i;
2439 int j;
2440 int cost; // cost
2441
2442 for (i = 0; i <= n; i++) {
2443 p[i] = i;
2444 }
2445
2446 for (j = 1; j <= m; j++) {
2447 d[0] = j;
2448 for (i = 1; i <= n; i++) {
2449 cost = (args1[i-1] == args2[j-1]) ? 0 : 1;
2450 d[i] = Math.min(Math.min(d[i - 1] + 1, p[i] + 1),
2451 p[i - 1] + cost);
2452 }
2453
2454 // copy current distance counts to 'previous row' distance counts
2455 _d = p;
2456 p = d;
2457 d = _d;
2458 }
2459 // our last action in the above loop was to switch d and p, so p now
2460 // actually has the most recent cost counts
2461 return p[n];
2462 }
2463
2464 private String removeUnusedCharacter(String target){
2465
2466 while(true){
2467 if(target.indexOf("\\n")!=-1){
2468 target = target.replaceFirst("\\\\n","");
2469 }
2470 else{
2471 break;
2472 }
2473 }
2474
2475 //remove spaces at the end of string
2476 if(target.length()>1){
2477 while(true){
2478 if(target.length()>1){
2479 if(target.charAt(target.length()-1)==' '){
2480 target = target.substring(0,target.length()-1);
2481 }
2482 else{
2483 break;
2484 }
2485 }
2486 else{
2487 break;
2488 }
2489 }
2490 }
2491
2492 //remove leading spaces
2493 if(target.length()>1){
2494 while(true){
2495 if(target.length()>1){
2496 if(target.charAt(0)==' '){
2497 target = target.substring(1,target.length());
2498 }
2499 else{
2500 break;
2501 }
2502 }
2503 else{
2504 break;
2505 }
2506 }
2507 }
2508
2509 //remove multiple spaces between words
2510 while(true){
2511 if(target.indexOf(" ")!=-1){
2512 target = target.replaceFirst(" "," ");
2513 }
2514 else{
2515 break;
2516 }
2517 }
2518 return target;
2519 }
2520
2521 private CostModel removeUnusedCharacter(String target, double cost){
2522
2523 CostModel cm = new CostModel();
2524
2525 while(true){
2526 if(target.indexOf("\\n")!=-1){
2527 target = target.replaceFirst("\\\\n","");
2528 cost = cost + 0.2;
2529 }
2530 else{
2531 break;
2532 }
2533 }
2534
2535 //remove spaces at the end of string
2536 if(target.length()>1){
2537 while(true){
2538 if(target.length()>1){
2539 if(target.charAt(target.length()-1)==' '){
2540 target = target.substring(0,target.length()-1);
2541 cost = cost + 0.2;
2542 }
2543 else{
2544 break;
2545 }
2546 }
2547 else{
2548 break;
2549 }
2550 }
2551 }
2552
2553 //remove leading spaces
2554 if(target.length()>1){
2555 while(true){
2556 if(target.length()>1){
2557 if(target.charAt(0)==' '){
2558 target = target.substring(1,target.length());
2559 cost = cost + 0.2;
2560 }
2561 else{
2562 break;
2563 }
2564 }
2565 else{
2566 break;
2567 }
2568 }
2569 }
2570
2571 //remove multiple spaces between words
2572 while(true){
2573 if(target.indexOf(" ")!=-1){
2574 target = target.replaceFirst(" "," ");
2575 cost = cost + 0.2;
2576 }
2577 else{
2578 break;
2579 }
2580 }
2581
2582 cm.setCost(cost);
2583 cm.setString(target);
2584 return cm;
2585 }
2586
2587 private HashMap generateMetadataElementSortListNoIdentifier(String title, String sort, String metadataSetName){
2588
2589 String fileName = title+"_"+sort;
2590 String collectionFolder = mds.getCollectionName();
2591 SearchLink sl = new SearchLink(mds.StatsDirectory);
2592 HashMap suggestionMap = new HashMap();
2593
2594 HashMap internalIDMap = dm.getInternalIdentifier(title);
2595 boolean status = false;
2596
2597 try{
2598
2599 FileWriter fstream = new FileWriter(destination+fileName+".html");
2600 BufferedWriter out = new BufferedWriter(fstream);
2601
2602 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
2603 out.write ("<html>\r\n");
2604 out.write("<head>\r\n<title>Metadata Element Sort List</title>\r\n");
2605 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
2606 out.write("<script type=\"text/javascript\" src=\"../script/getInfomation.js\"></script>\r\n");
2607 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/yahoo/yahoo-min.js\"></script>\r\n");
2608 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/event/event-min.js\"></script>\r\n");
2609 out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/connection/connection-min.js\"></script>\r\n");
2610 out.write("<style type=\"text/css\">\r\n");
2611 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
2612 out.write(".tam{height:20; text-align:center}\r\n");
2613 out.write("th{height:20; text-align:center;}\r\n");
2614 out.write("body{font-family: Arial;}\r\n");
2615 out.write("</style>\r\n");
2616 out.write("</head><body id=\""+collectionFolder+"\">\r\n");
2617 out.write("<td><div style=\"float:left;\"><a href=\" Overall.html \">Summary</a>");
2618 out.write("&raquo;");
2619
2620 if(metadataSetName.equalsIgnoreCase("dublin")){
2621 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2622 }
2623 else if (metadataSetName.equalsIgnoreCase("extracted")){
2624 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2625 }
2626 else{
2627 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2628 }
2629 out.write("&raquo;");
2630 out.write("<a href=\""+title+".html\">"+title+"</a></div>");
2631 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>");
2632
2633 String[] list = dm.getSortList(title, sort);
2634
2635 if(list.length>=1){
2636 if(sort.equals("ASCII")){
2637
2638 out.write("<table border=\"1\" align=\"center\" width=\"800\">\r\n");
2639 out.write ("<h2 align=\"center\">"+ title+"</h2>\r\n");
2640 out.write ("<th class=\"tam\">ASCII Sort\r\n");
2641 out.write ("<th class=\"tam\">Element Values\r\n");
2642 out.write ("<th class=\"tam\">Source Documents\r\n");
2643 out.write ("<th class=\"tam\">Internal Link\r\n");
2644 out.write ("<a name='topFive'>\r\n");
2645
2646 int counter = 0;
2647
2648 for(int i = 0; i<list.length; i++){
2649 if(list.length<=5 && i==0){
2650 out.write ("<a name='lastFive'>\r\n");
2651 }
2652 else if((list.length>5) && (list.length-5==i)){
2653 out.write ("<a name='lastFive'>\r\n");
2654 }
2655
2656
2657 InternalLink il= (InternalLink)internalIDMap.get(list[i]);
2658 ArrayList alist2 = il.retrieveList();
2659 String id = (String)alist2.get(0);
2660 id = id.substring(4);
2661
2662 out.write("<tr id=\""+id+"\" >\r\n");
2663
2664 if(list[i].length()>=201){
2665 if(list[i].startsWith("http://")){
2666 if(list[i].length()>=100){
2667 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2668 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2669 }
2670 else{
2671 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2672 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2673 }
2674 }
2675 else{
2676
2677 if(list[i].startsWith(" ")){
2678 String elements = list[i];
2679 list[i] = list[i].substring(1);
2680 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i].substring(0, 200)+"...<td>No Source Available \r\n");
2681 }
2682 else{
2683 out.write("<td>"+(counter+1)+"<td>"+list[i].substring(0, 200)+"... <td>No Source Available\r\n");
2684 }
2685 }
2686 }
2687 else{
2688 if(list[i].startsWith("http://")){
2689 if(list[i].length()>=100){
2690 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"...</a>");
2691 }
2692 else{
2693 out.write("<td>"+(counter+1)+"<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2694 }
2695 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2696 }
2697 else{
2698 if(list[i].startsWith(" ")){
2699 String elements = list[i];
2700 list[i] = list[i].substring(1);
2701 out.write("<td>"+(counter+1)+"<td>"+spaceLeft+space+spaceRight+list[i]+"<td>No Source Available\r\n");
2702 suggestionMap.put(elements,"No Source Available");
2703 }
2704 else{
2705 out.write("<td>"+(counter+1)+"<td>"+list[i]+"<td>No Source Available\r\n");
2706 }
2707 }
2708 }
2709 counter++;
2710 out.write("<td onclick=\"GD(this)\">View");
2711 }
2712 }
2713 else{
2714 out.write("<table border=\"1\" align=\"center\" width=\"800\">\r\n");
2715 out.write ("<h2 align=\"center\">"+ title+"</h2>\r\n");
2716 out.write ("<th class=\"tam\">&nbsp;\r\n");
2717 out.write ("<th class=\"tam\">Frequency\r\n");
2718 out.write ("<th class=\"tam\">Element Values\r\n");
2719 out.write ("<th class=\"tam\">Source Documents\r\n");
2720 out.write ("<th class=\"tam\">Internal Link\r\n");
2721 out.write ("<a name='topFive'>\r\n");
2722
2723 HashMap xMap = dm.getDistinctValueMap(title);
2724 int counter = 0;
2725
2726 for(int i = 0; i<list.length; i++){
2727 if(list.length<=5 && i==0){
2728 out.write ("<a name='lastFive'>\r\n");
2729 }
2730 else if((list.length>5) && (list.length-5==i)){
2731 out.write ("<a name='lastFive'>\r\n");
2732 }
2733
2734 InternalLink il= (InternalLink)internalIDMap.get(list[i]);
2735 ArrayList alist2 = il.retrieveList();
2736 String id = (String)alist2.get(0);
2737 id = id.substring(4);
2738
2739 out.write("<tr id=\""+id+"\" >\r\n");
2740 out.write("<td>"+(counter+1)+"<th> "+((Integer)xMap.get(list[i])).toString()+"\r\n");
2741
2742 if(list[i].length()>=201){
2743 if(list[i].startsWith("http://")){
2744 if(list[i].length()>=100){
2745 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2746 }
2747 else{
2748 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2749 }
2750 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2751 }
2752 else{
2753 if(list[i].startsWith(" ")){
2754 list[i]=list[i].substring(1);
2755 out.write("<td>"+spaceLeft+space+spaceRight+list[i].substring(0,200)+"...<td>No Source Available\r\n");
2756 }
2757 else{
2758 out.write("<td>"+list[i].substring(0,200)+"...<td>No Source Available\r\n");
2759 }
2760 }
2761 }
2762 else{
2763 if(list[i].startsWith("http://")){
2764 if(list[i].length()>=100){
2765 out.write("<td><a href=\""+list[i]+"\">"+list[i].substring(0,100)+"</a>");
2766 }
2767 else{
2768 out.write("<td><a href=\""+list[i]+"\">"+list[i]+"</a>");
2769 }
2770 out.write("<td><a href=\""+list[i]+"\">Source</a>\r\n");
2771 }
2772 else{
2773 if(list[i].startsWith(" ")){
2774 list[i]=list[i].substring(1);
2775 out.write("<td>"+spaceLeft+space+spaceRight+list[i]+"<td>No Source Available\r\n");
2776 }
2777 else{
2778 out.write("<td>"+list[i]+"<td>No Source Available\r\n");
2779 }
2780 }
2781 }
2782 counter++;
2783 out.write("<td onclick=\"GD(this)\">View");
2784 }
2785 }
2786 out.write("</table>\r\n");
2787 }
2788 else{
2789 out.write("<p>Sorry! The list is empty</p>");
2790 }
2791
2792 out.write("<tr>\r\n");
2793 out.write("<p><a href=\" Overall.html \">Summary</a>");
2794 out.write("&raquo;");
2795
2796 if(metadataSetName.equalsIgnoreCase("dublin")){
2797 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2798 }
2799 else if (metadataSetName.equalsIgnoreCase("extracted")){
2800 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2801 }
2802 else{
2803 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2804 }
2805
2806 out.write("&raquo;");
2807 out.write("<a href=\""+title+".html\">"+title+"</a></p>");
2808 out.write("</body></html>\r\n");
2809 out.close();
2810
2811 fstream.close();
2812 }catch(Exception e){
2813 e.printStackTrace();
2814 }
2815 return new HashMap();
2816 }
2817
2818 private void createIncompletedListNoIdentifier(String fileName, String metadataSetName, String collectionFolder){
2819
2820 HashMap hp = dm.getIdentifierLinkNoIdentifier();
2821 String[] ids = dm.getDocumentIDList(fileName);
2822
2823 Set st = hp.keySet();
2824 Iterator it = st.iterator();
2825
2826 while(it.hasNext()){
2827 System.out.println(it.next());
2828 }
2829 for(int i = 0; i<ids.length; i++){
2830 System.out.println("IDS:" +ids[i]);
2831 hp.remove(ids[i]);
2832 }
2833
2834 //System.out.println("Length: "+ids.length+" "+fileName+ " "+hp.size());
2835
2836 try{
2837 FileWriter fstream = new FileWriter(destination+fileName+"_IncompletedList.html");
2838 BufferedWriter out = new BufferedWriter(fstream);
2839
2840 out.write("<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.01 Transitional//EN\" \"http://www.w3.org/TR/html4/loose.dtd\">");
2841 out.write ("<html>\r\n");
2842 out.write("<head>\r\n<title>Incompleted Document List</title>\r\n");
2843 out.write("<meta http-equiv=\"content-type\" content=\"text/html; charset=utf-8\">\r\n");
2844 out.write("<style type=\"text/css\">\r\n");
2845 out.write("td{height:20; text-align:left;padding-left:5px;}\r\n");
2846 out.write(".tam{height:20; text-align:center}\r\n");
2847 out.write("th{height:20; text-align:center;}\r\n");
2848 out.write("body{font-family: Arial;}\r\n");
2849 out.write("</style>\r\n");
2850 out.write("</head><body>\r\n");
2851 out.write("<p><a href=\" Overall.html \">Summary</a>");
2852 out.write("&raquo;");
2853
2854 if(metadataSetName.equalsIgnoreCase("dublin")){
2855 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2856 }
2857 else if (metadataSetName.equalsIgnoreCase("extracted")){
2858 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2859 }
2860 else{
2861 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2862 }
2863
2864 out.write("&raquo;");
2865 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
2866 out.write("<h3 align=center>"+fileName+" does not appear in the following documents</h3>");
2867 int counter = 0;
2868 Set s = hp.keySet();
2869 Iterator i = s.iterator();
2870 out.write("<table border=1 align=center><tr><th>Document ID<th>Source Link\r\n");
2871
2872 while(i.hasNext()){
2873 counter++;
2874 String keys = (String)i.next();
2875 InternalLink il = (InternalLink) hp.get(keys);
2876 ArrayList alist = il.retrieveList();
2877 String url = (String)alist.get(0);
2878 out.write("<tr><td>"+counter+"<td><a href=\""+url+"\">"+url+"</a>\r\n");
2879 }
2880 out.write("</table></body></html>\r\n");
2881 out.write("<p><a href=\" Overall.html \">Summary</a>");
2882 out.write("&raquo;");
2883
2884 if(metadataSetName.equalsIgnoreCase("dublin")){
2885 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet1+")</a>");
2886 }
2887 else if (metadataSetName.equalsIgnoreCase("extracted")){
2888 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSet2+")</a>");
2889 }
2890 else{
2891 out.write ("<a href=\""+metadataSetName+".html\">Metadata Detail ("+metadataSetName+")</a>");
2892 }
2893 out.write("&raquo;");
2894 out.write("<a href=\""+fileName+".html \">"+fileName+"</a>");
2895
2896 out.close();
2897 fstream.close();
2898 }catch(IOException ex){
2899 ex.printStackTrace();
2900 }
2901 }
2902}
Note: See TracBrowser for help on using the repository browser.