public static void main(final String[] args) {
String url, html;
HtmlLangDetector detector = new HtmlLangDetector("conf/lang.properties");
// fileName = "/home/xcyao/CityU/work/welcome.html";
url = "http://news.sina.com.cn";
// url = "http://ngramj.sourceforge.net/use_ngramj.html";
// url = "http://www.let.rug.nl/~vannoord/TextCat/ShortTexts/dutch.txt";
url = "http://alias-i.com/lingpipe/demos/tutorial/langid/read-me.html";
url = "http://www.cas.gov.hk/eng/notice/notice_remove.html";
url = "http://www.wsd.gov.hk/tc/job_opportunities/index_t.html";
url =
"http://www.wsd.gov.hk/sc/customer_services_and_water_bills/application_for_water_supply/water_supply_for_new_buildings/index_t.html";
// html = HtmlLangDetector.fetch(url);
// html = "证券简称,今日开盘价,昨日收盘价,最近成交价,最高成交价,最低成交价,买入价";
// html = "证券简称，今日开盘价，昨日收盘价，最近成交价，最高成交价，最低成交价，买入价";
// html = "今日开盘价，的，好，人，了，是，不";
html = HTML2TEXT.getText(url);
System.out.println(html);
System.out.println(detector.guess(html));
}
public String detectFromRaw(final String raw) {
String text = HTML2TEXT.getTextFromRaw(raw);
return guess(text);
}
public String detect(final String url) {
String text = HTML2TEXT.getText(url);
return guess(text);
}
