Examples of org.htmlcleaner.TagNode

org.htmlcleaner.TagNode

XML node tag - basic node of the cleaned HTML tree. At the same time, it represents start tag token after HTML parsing phase and before cleaning phase. After cleaning process, tree structure remains containing tag nodes (TagNode class), content (text nodes - ContentNode), comments (CommentNode) and optionally doctype node (DoctypeToken).

    props.setAllowHtmlInsideAttributes( false );
    props.setPruneTags( "style, script" );


    HtmlCleaner cleaner = new HtmlCleaner( props );
    try {
      TagNode node = cleaner.clean( new StringReader( content ) );


      TagNode[] nodes = node.getElementsByName( "a", true );
      for (TagNode tagnode : nodes) {
        tagnode.removeAttribute( "target" );
        tagnode.addAttribute( "target", "_blank" );
      }
      nodes = node.getElementsByName( "img", true );
      for (TagNode tagnode : nodes) {
        tagnode.removeAttribute( "src" );
        tagnode.addAttribute( "src", "NO_IMAGE" );
      }
      System.out.println( node.getText() );


      return cleaner.getInnerHtml( node );


    } catch (IOException e) {

View Full Code Here


    HtmlCleaner cleaner = new HtmlCleaner( CLEANER_PROPERTIES );
    String result = "";


    try {
      TagNode rootNode = cleaner.clean( new StringReader( messageText ) );


      TagNode[] nodes = rootNode.getElementsByName( "a", true );
      if ( nodes != null && nodes.length > 0 ) {
        for (TagNode tagnode : nodes) {
          tagnode.removeAttribute( "target" );
          tagnode.addAttribute( "target", "_blank" );
        }
      }


      nodes = rootNode.getElementsByName( "img", true );
      if ( nodes != null && nodes.length > 0 ) {
        hasImages[0] = true;
        if ( !loadImages ) {
          for (TagNode tagnode : nodes) {
            tagnode.removeAttribute( "src" );

View Full Code Here

  public static String convertHtml2PlainText( String htmlText ) {


    HtmlCleaner cleaner = new HtmlCleaner( CLEANER_PROPERTIES );


    try {
      TagNode rootNode = cleaner.clean( new StringReader( htmlText ) );
      return rootNode.getText().toString();
    }
    catch (IOException e) {
      log.error( e.getMessage(), e );
    }

View Full Code Here


    HtmlCleaner cleaner = new HtmlCleaner( CLEANER_PROPERTIES );
    String result = "";


    try {
      TagNode rootNode = cleaner.clean( new StringReader( messageText ) );


      TagNode[] nodes = rootNode.getElementsByName( "a", true );
      if ( nodes != null && nodes.length > 0 ) {
        for (TagNode tagnode : nodes) {
          tagnode.removeAttribute( "target" );
          tagnode.addAttribute( "target", "_blank" );
        }
      }


      nodes = rootNode.getElementsByName( "img", true );
      if ( nodes != null && nodes.length > 0 ) {
        hasImages[0] = true;
        if ( !loadImages ) {
          for (TagNode tagnode : nodes) {
            tagnode.removeAttribute( "src" );

View Full Code Here

  public static String convertHtml2PlainText( String htmlText ) {


    HtmlCleaner cleaner = new HtmlCleaner( CLEANER_PROPERTIES );


    try {
      TagNode rootNode = cleaner.clean( new StringReader( htmlText ) );
      return rootNode.getText().toString();
    }
    catch (IOException e) {
      log.error( e.getMessage(), e );
    }

View Full Code Here

  
  /* (non-Javadoc)
   * @see org.apache.wookie.util.html.IHtmlProcessor#injectScript(java.lang.String)
   */
  public void injectScript(String script) {
    TagNode js = new TagNode(SCRIPT_TAG);
    js.addAttribute(TYPE_ATTRIBUTE, TYPE_ATTRIBUTE_VALUE);
    js.addAttribute(SRC_ATTRIBUTE, script);
    headNode.addChild(js);
  }

View Full Code Here


  /* (non-Javadoc)
   * @see org.apache.wookie.util.html.IHtmlProcessor#injectStylesheet(java.lang.String)
   */
  public void injectStylesheet(String stylesheet) {
    TagNode js = new TagNode(LINK_TAG);
    js.addAttribute(TYPE_ATTRIBUTE, CSS_TYPE_ATTRIBUTE_VALUE);
    js.addAttribute(REL_ATTRIBUTE, CSS_REL_ATTRIBUTE_VALUE);
    js.addAttribute(HREF_ATTRIBUTE, stylesheet);
    headNode.addChild(js);
  }

View Full Code Here

    
    //
    // Check if the page already has a META http-equiv=content-type tag,
    // if it doesn't create one and add it to the head node
    //
    TagNode meta = headNode.findElementByAttValue("http-equiv", "content-type", true, false);
    if (meta == null) {
      meta = new TagNode(META_TAG);
      meta.addAttribute("http-equiv", "Content-Type");
      headNode.getChildren().add(0, meta);
    }
    //
    // Force UTF into lowercase
    //
    if (charset.equals("UTF-8")) charset = "utf-8";
    
    //
    // Override the charset and content-type values for the 
    // META http-equiv=content-type tag
    //
    meta.addAttribute("content", type + ";charset=" + charset);
  }

View Full Code Here

      //
      // Check for charset overrides in the HTML start page
      //
      HtmlCleaner cleaner = new HtmlCleaner();
 
      TagNode httpEquivNode = cleaner.clean(get.getResponseBodyAsStream()).findElementByAttValue("http-equiv", "content-type", true, false);
      if (httpEquivNode != null && httpEquivNode.hasAttribute("content")){
        String value = httpEquivNode.getAttributeByName("content");
        int offset = value.indexOf("charset=");
        if (offset >= -1){
            charset = value.substring(offset+8).toUpperCase();
        }
      }

View Full Code Here

     
    props.setTranslateSpecialEntities(true);
    props.setTransResCharsToNCR(true);
    props.setOmitComments(true);
     
    final TagNode tagNode = new HtmlCleaner(props).clean(xml);
    final String cleansData = new CompactXmlSerializer(props).getAsString(tagNode);
     
    final SAXParserFactory spf = SAXParserFactory.newInstance();
      spf.setNamespaceAware(true);

View Full Code Here

0 1 2 3 4 5 6

TOP

Related Classes of org.htmlcleaner.TagNode

at.newmedialab.ldpath.model.functions.CleanHtmlFunction

ch.entwine.weblounge.preview.xhtmlrenderer.XhtmlRendererPagePreviewGenerator

com.atlantbh.nutch.filter.xpath.XPathHtmlParserFilter

com.cubusmail.mail.text.MessageTextUtil

com.cubusmail.mail.text.test.HtmlParserTest

com.cubusmail.server.mail.text.MessageTextUtil

com.ikanow.infinit.e.harvest.enrichment.custom.UnstructuredAnalysisHarvester

com.jeck.microblogging.utils.HtmlUtils

com.netfever.site.dynovisz.tools.utils.XmlUtils

com.skrul.greasefire.DownloadScripts

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.