support-meta-refresh.java

/**
 * jsoup 工具包
 * founder
 * @author XueLiang
 * @date 2016年9月23日 下午3:16:36
 * @version 1.0
 */
public class JsoupUtils {
	
	/**
	 * 支持 meta 标签跳转
	 * @param document
	 * @return
	 * @throws IOException
	 * @throws URISyntaxException
	 */
	public Document supportMetaRefresh(Document document) throws IOException, URISyntaxException {
		String refresh = getMetaValue(document, "refresh");
		if (refresh == null) {
			return document;
		}
		String baseUri = document.baseUri();
		URI uri = new URI(baseUri);
		Matcher m = Pattern.compile("(?si)\\d+;\\s+url=(.+)|\\d+").matcher(refresh);
		// find the first one that is valid
        if (m.matches()) {
        	String refreshUrl = m.group(1);
            if (refreshUrl != null) {
            	Document doc = Jsoup.connect(uri.resolve(refreshUrl).toString()).userAgent("Mozilla")
            			.followRedirects(true).get();
            	return supportMetaRefresh(doc);
            }
        }
        return document;
	}
}