Package org.vietspider.html

Examples of org.vietspider.html.NodeIterator


    }
    return false;
  }
 
  private boolean isBlockLink(HTMLNode node) {
    NodeIterator iterator = node.iterator();
    int counter = 0;
    while(iterator.hasNext()) {
      HTMLNode n = iterator.next();
      if(n.isNode(Name.A)) counter++;
    }
    return counter > 3;
  }
View Full Code Here


* Feb 10, 2009 
*/
public class FormChecker {
 
  boolean hasForm(HTMLNode node) {
    NodeIterator iterator = node.iterator();
    while(iterator.hasNext()) {
      HTMLNode n = iterator.next();
      if(n.isNode(Name.FORM)) return true;
      if(isFormElement(n)) return true;
    }
    return false;
  }
View Full Code Here

* Feb 10, 2009 
*/
public class FormNodeRemover {
 
  boolean hasForm(HTMLNode node) {
    NodeIterator iterator = node.iterator();
    while(iterator.hasNext()) {
      HTMLNode n = iterator.next();
      if(n.isNode(Name.FORM)) return true;
//      if(isFormElement(n)) return true;
    }
    return false;
  }
View Full Code Here

    if(isTextBlock(node, checkLink, 1, 1)) return false;
    return true;
  }
 
  public boolean isTextBlock(HTMLNode node, boolean checkLink, int w_size, int s_size) {
    NodeIterator nodeIterator = node.iterator();
    while(nodeIterator.hasNext()) {
      HTMLNode iterNode = nodeIterator.next();
      if(checkLink) {
        if(RenderNodeUtils.getAncestor(iterNode, Name.A, 0, 5) != null) continue;
      }
      if(iterNode.isNode(Name.CONTENT)) {
        String text = iterNode.getTextValue();
View Full Code Here

    if(rate >= 85) return true;
    return (onlyLinkCounter*100)/list.size() > 85;
  }
 
  private boolean hasLink(HTMLNode node) {
    NodeIterator iterator = node.iterator();
    while(iterator.hasNext()) {
      HTMLNode n = iterator.next();
      if(n.isNode(Name.A)) return true;
    }
    return false;
  }
View Full Code Here

  }
 
  @Deprecated()
  public void searchTextNode(HTMLNode node, List<char[]> list){
    HTMLText.EmptyVerify verify = new HTMLText.EmptyVerify()
    NodeIterator iterator = node.iterator();
    while(iterator.hasNext()) {
      HTMLNode n = iterator.next();
      if(!n.isNode(Name.CONTENT)) return;
      if(verify.isValid(n.getValue())) list.add(n.getValue());
    }
  }
View Full Code Here

    String path2 = getIndexPath(element2);
    return getCommonIndexPath(path1, path2);
 
 
  public void searchAnchors(HTMLNode node, List<HTMLNode> anchors) {
    NodeIterator iterator = node.iterator();
    while(iterator.hasNext()) {
      HTMLNode n = iterator.next();
      if(n.isNode(Name.A)) anchors.add(n);
    }
  }
View Full Code Here

    }
  }
 
  public List<HTMLNode> search(HTMLNode root, Name name) {
    final List<HTMLNode> values = new ArrayList<HTMLNode>();
    NodeIterator iterator = root.iterator();
    while(iterator.hasNext()) {
      HTMLNode n = iterator.next();
      if(n.isNode(name)) values.add(n);
    }
    return values;
  }
View Full Code Here

    }
    return values;
  }

  public void search(HTMLNode root, final List<HTMLNode> values, Name name) {
    NodeIterator iterator = root.iterator();
    while(iterator.hasNext()) {
      HTMLNode n = iterator.next();
      if(n.isNode(name)) values.add(n);
    }
  }
View Full Code Here

      buildText(builder, node);
      return;
    }
   
    if(node == null) return;
    NodeIterator iterator = node.iterator();
   
    while(iterator.hasNext()) {
      HTMLNode n = iterator.next();
      if(!n.isNode(Name.CONTENT)) continue;
      char [] chars = n.getValue();
      if(!verify.isValid(chars)) continue;
      if(builder.length() > 0) builder.append(' ');
      builder.append(chars);
View Full Code Here

TOP

Related Classes of org.vietspider.html.NodeIterator

Copyright © 2018 www.massapicom. All rights reserved.
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.