3febafb429d2a6b0b6d0bb228059fb59e7bb3c11,edu.usc.cssl.tacit.crawlers.frontier/src/edu/usc/cssl/tacit/crawlers/frontier/services/FrontierCrawl.java,FrontierCrawl,crawl,#String#String#number#IProgressMonitor#,28
Before Change
if(i != -1){
String journalBody = abstractBody.substring(0, i);
jsonGenerator.writeObjectField("journal_body", Jsoup.parse(journalBody).text());
jsonGenerator.writeObjectField("references", Jsoup.parse(abstractBody.substring(i+10)).text());
}else{
jsonGenerator.writeObjectField("journal_body", Jsoup.parse(abstractBody).text());
}
After Change
Elements title = docJournalAbstract.select("h1");
Elements abs = docJournalAbstract.select("p");
jsonGenerator.writeStartObject();
if(jsonFilter[0])
jsonGenerator.writeObjectField("title", Jsoup.parse(title.toString()).text());
if(jsonFilter[1])
jsonGenerator.writeObjectField("abstract_body", Jsoup.parse(abs.toString()).text());
String abstractBody = d.body().child(2).child(4).child(0).child(1).child(1).child(0).child(0).child(0).child(2).toString();
int i = abstractBody.indexOf("References");
if(i != -1){
String journalBody = abstractBody.substring(0, i);
if(jsonFilter[2])
jsonGenerator.writeObjectField("journal_body", Jsoup.parse(journalBody).text());
if(jsonFilter[3])
jsonGenerator.writeObjectField("references", Jsoup.parse(abstractBody.substring(i+10)).text());
}else{
if(jsonFilter[2])
jsonGenerator.writeObjectField("journal_body", Jsoup.parse(abstractBody).text());
}
jsonGenerator.writeEndObject();
count++;