String extractContentFromDocument(
LesothoCourtDocument scd, String pdfFile, int startpage, int endpage) {
Document document = null;
File input = new File(pdfFile);
String textContent = null;
String extractedCaseId = "";
ArrayList<String> participants = new ArrayList<String>();
String extractedDecisionDate = "";
String participantsString = "";
if (pdfFile == null) {
System.out.println("File name is not valid");
} else {
try {
document = Jsoup.parse(input, "UTF-8");
String textToBeExtracted = document.select("h1").text();
String[] extractedElements = textToBeExtracted.toUpperCase().split("V");
if (extractedElements != null) {
participants.add(extractedElements[0]);
String[] extractedEle = null;
extractedEle = (extractedElements[extractedElements.length - 1]).split("\\(");
participants.add(extractedEle[0]);
for (String s : participants) {
participantsString += s + "\t";
}
System.out.println(participantsString);
extractedCaseId = extractedEle[1];
extractedCaseId = extractedCaseId.replaceAll("\\)", "");
textContent = document.body().toString();
String temp = "Judgment Date:";
/**
* int startIndex = textContent.indexOf(temp)+temp.length(); String subContent =
* textContent.substring(startIndex); int endIndex = subContent.indexOf("\n"); *
*/
Element decDate = document.select("span.date-display-single").first();
extractedDecisionDate = decDate.text();
System.out.println("Decision Date: " + extractedDecisionDate);
System.out.println("Case Id: " + extractedCaseId);
}
scd.setCaseId(extractedCaseId);
scd.setDecisionDate(extractedDecisionDate);
scd.setHeardDate(null);
scd.setParticipantsName(participantsString);
} catch (Exception e) {
System.out.println("Error in parsing html : " + e.getMessage());
}
}
return textContent;
}
CourtDocument processCaseDetails(String caseFile, String sourceFileName) {
int startPage = 1;
int endPage = 1;
LesothoCourtDocument scd =
new LesothoCourtDocument(
this.getCountryName(),
this.getCourtName(),
this.getProcessedUserName(),
sourceFileName);
String htmlFileContent = extractContentFromDocument(scd, caseFile, startPage, endPage);
int filelength = caseFile.length();
this.setFileLength(filelength);
List<String> pageContentList = new ArrayList<String>();
pageContentList.add(htmlFileContent);
scd.extractCitations(this.getCourtName(), pageContentList);
return scd;
}