AbstractTokenizer tokenizer = NLPGetter.getTokenizer("en");
Map<String,String> redirects = Maps.newHashMap();
// Pattern paren = Pattern.compile("\\(.+?\\)$");
Set<String> titles = Sets.newHashSet();
String line, title, redirect;
PTHtml html = new PTHtml();
int count = 0;
String[] t;
while ((line = reader.readLine()) != null)
{