HTML parser选型测试

内容管理(cms)常常需要将网站频道的摘要(summery)合并到父频道的封面，引入HTML parser，
 可以结构化方式操作HTML内容，使网页内容的提取、重构变得容易。
 以下链接列出了相关的java opensource项目
 http://www.open-open.com/30.htm
 根据网友的评论，将htmlcleaner、htmlparser、nekohtml列入候选。
 以附件html作为测试用例，按照常见的getElementsByTagName提取Body,
 以getElementById获取id为'content6'的script
 测试编码如下：

java 代码

static public String Neko(String path) throws SAXException, IOException{
    DOMParser parser = new DOMParser();
    //InputSource in = new InputSource(new Reader());
    parser.parse(TPPath+path);
    Document doc=parser.getDocument();
    org.w3c.dom.NodeList nl = doc.getElementsByTagName("body");
    System.out.println(printNode(nl.item(0)));
    System.out.println("----------------------------------------");
    org.w3c.dom.Element n = doc.getElementById("content6");
    System.out.println(printNode(n));
    return "";
}
static public String htmlparser(String path) throws ParserException{
    // one of several constructors
    Parser p = new Parser(TPPath+path);
    NodeList nl=p.parse(new TagNameFilter("body"));
    System.out.println(nl.elementAt(0).toHtml());
    System.out.println("----------------------------------------");
    Parser p2 = new Parser(TPPath+path);
    NodeList nl2=p2.parse(new HasAttributeFilter("id","content6"));
    System.out.println(nl2.elementAt(0).toHtml());
    return "";
}
static public String htmlcleaner(String path) throws Exception{
    // one of several constructors
    HtmlCleaner cleaner = new HtmlCleaner(new File(TPPath+path));
    org.w3c.dom.Document doc = cleaner.createDOM();
    org.w3c.dom.NodeList nl = doc.getElementsByTagName("body");
    System.out.println(printNode(nl.item(0)));
    System.out.println("----------------------------------------");
    org.w3c.dom.Element n=doc.getElementById("content6");
    System.out.println(printNode(n));
    return "";
}

一个打印dom节点的辅助方法如下：

java 代码

public static String printNode(Node node) {
    StringBuffer sbuf=new StringBuffer();
    String nn=node.getNodeName();
    boolean btag=true;
    if(nn.equals("#text")) btag=false;
    if(btag){
      if(node.hasAttributes()){
      NamedNodeMap attrs=node.getAttributes();
      StringBuffer abuf= new StringBuffer();
      for(int i=0,len=attrs.getLength(); i<len; i++){
          Node attr=attrs.item(i);
          abuf.append(" "+attr.getNodeName()+"=\""+attr.getNodeValue()+"\"");
      }
      sbuf.append("<"+nn+abuf.toString()+">");
      }else  sbuf.append("<"+nn+">");
    }
    if(node.hasChildNodes()){
     Node child = node.getFirstChild();
     sbuf.append(child.getNodeValue());
     while (child != null) {
         sbuf.append(printNode(child));
         child = child.getNextSibling();
     }
    }
    if(btag)
      sbuf.append("</"+nn+">");
    return sbuf.toString();
}

 测试结果如下：
                   getElementsByTagName        getElementById
htmlcleaner         抛出异常java.lang.NoSuchFieldError: fRecognizedFeatures
htmlparser         在分析到script中的字符串包含"</b>"出现逻辑错误，将该</b>误判为script结束
nekohtml            pass                        pass

nekohtml入选。