public void testUrlModificationWithVisitor() throws Exception {
Parser parser = Parser.createParser(HTML_WITH_LINK, null);
UrlModifyingVisitor visitor = new UrlModifyingVisitor("localhost://");
parser.visitAllNodesWith(visitor);
String result = visitor.getModifiedResult();
assertStringEquals("Expected HTML", MODIFIED_HTML, result);
}
/**
* 获取新闻的内容
*
* @param newsContentFilter
* @param parser
* @return content 新闻内容
*/
public String getNewsContent(NodeFilter newsContentFilter, Parser parser) {
String content = null;
StringBuilder builder = new StringBuilder();
try {
NodeList newsContentList = parser.parse(newsContentFilter);
for (int i = 0; i < newsContentList.size(); i++) {
Div newsContenTag = (Div) newsContentList.elementAt(i);
builder = builder.append(newsContenTag.getStringText());
}
content = builder.toString(); // 转换为String 类型。
if (content != null) {
parser.reset();
parser = Parser.createParser(content, "utf8");
StringBean sb = new StringBean();
sb.setCollapse(true);
parser.visitAllNodesWith(sb);
content = sb.getStrings();
// String s = "\";} else{ document.getElementById('TurnAD444').innerHTML = \"\";} }
// showTurnAD444(intTurnAD444); }catch(e){}";
content = content.replaceAll("\\\".*[a-z].*\\}", "");
content = content.replace("[我来说两句]", "");
} else {
System.out.println("没有得到新闻内容！");
}
} catch (ParserException ex) {
Logger.getLogger(AreaTest.class.getName()).log(Level.SEVERE, null, ex);
}
return content;
}
public static void setInnerHTML(Element root, String html) {
// remove old root childs
OverrideNodeList<Node> list = (OverrideNodeList<Node>) root.getChildNodes();
list.getList().clear();
if (html != null) {
Parser parser = Parser.createParser(html, "UTF-8");
try {
parser.visitAllNodesWith(new GwtNodeVisitor(root));
} catch (ParserException e) {
throw new RuntimeException(
"error while parsing <" + root.getTagName() + "> element's innerHTML : " + html, e);
}
}
}
