/**
* 去除html标记（类似DOM的element.innerText）
*
* @param htmlcontent
* @return content
*/
private static String getTextByRemoveHTMLTag(String htmlcontent) {
if (StringUtils.isEmpty(htmlcontent)) {
return "";
}
// 不能替换导致多个空格被改成一个空格.
Source source = new Source(htmlcontent);
Renderer renderer = source.getRenderer();
renderer.setMaxLineLength(99999); // 设置一行最长个数，默认76字符
String txt = renderer.toString();
return txt;
}
/**
* Renders the HTML content into text. This provides a human readable version of the content that
* is modeled on the way Mozilla Thunderbird® and other email clients provide an automatic
* conversion of HTML content to text in their alternative MIME encoding of emails.
*
* <p>Using the default settings, the output complies with the <code>
* Text/Plain; Format=Flowed (DelSp=No)</code> protocol described in <a
* href="http://tools.ietf.org/html/rfc3676">RFC-3676</a>.
*
* @param html the HTML text
* @return the rendered HTML text, or <code>null</code> if the HTML text is <code>null</code>
*/
@Override
public String render(String html) {
if (html == null) {
return null;
}
Source source = new Source(html);
Renderer renderer = source.getRenderer();
return renderer.toString();
}