Package org.htmlparser.util

Examples of org.htmlparser.util.NodeList


          new CssClassFilter("list-mpaa"),
          new AndFilter(new NodeClassFilter(ImageTag.class),
              new CssClassFilter("star"))
          });
 
      NodeList discs = parser.extractAllNodesThatMatch(discFilter);
 
      Pattern ratingPattern = Pattern.compile(RATING_EXPR);
      Pattern movieIdPattern = Pattern.compile(MOVIE_ID_EXPR);
     
      Node disc;
      NodeList info = new NodeList();
      SimpleNodeIterator iter = discs.elements();
      while (iter.hasMoreNodes()) {
        info.removeAll();
        infoCollector.clear();
       
        disc = iter.nextNode();
       
        //collect all the nodes for the movie items
View Full Code Here


              new CssClassFilter("star"))
          });
 
      parser.extractAllNodesThatMatch(searchCollector);
      Node activeTab = searchCollector.getNode(0);
      NodeList discs = searchCollector.getNodeList(1);
     
      if (totalResultsOut != null) {
        String totalStr = NodeUtils.getTextData(activeTab);
        Pattern totalResultsPattern = Pattern.compile(SEARCH_RESULT_COUNT_EXPR);
        Matcher matcher = totalResultsPattern.matcher(totalStr);
        if (matcher.matches()) {
          totalResultsOut[0] = Integer.parseInt(matcher.group(1));
        }
      }
 
      Pattern ratingPattern = Pattern.compile(RATING_EXPR);
      Pattern movieIdPattern = Pattern.compile(MOVIE_ID_EXPR);
     
      Node disc;
      NodeList info = new NodeList();
      SimpleNodeIterator iter = discs.elements();
      while (iter.hasMoreNodes()) {
        info.removeAll();
        infoCollector.clear();
       
        disc = iter.nextNode();
       
        //collect all the nodes for the movie items
        disc.collectInto(info, infoCollector);
 
        int i=0;
 
        //parse out all the information
        NodeList linkNodes = infoCollector.getNodeList(i++);
        LinkTag titleLink = (LinkTag)linkNodes.elementAt(0);
        String movieId = null;
        Matcher matcher = movieIdPattern.matcher(titleLink.getLink());
        if (matcher.matches()) {
          movieId = matcher.group(1);
        } else {
          //not a valid movie
          continue;
        }
       
        String title = titleLink.getLinkText();
        String mpaa = NodeUtils.getTextData(linkNodes.elementAt(1));
 
        String year = NodeUtils.getTextData(infoCollector.getNode(i++));
        year = year.replaceAll("\\((\\d*)\\)", "$1");
 
        int rating = -1;
View Full Code Here

              new HasParentFilter(new CssClassFilter("searchSmallText"))),
          new AndFilter(new NodeClassFilter(ImageTag.class),
              new CssClassFilter("star"))
          });

      NodeList discs = parser.extractAllNodesThatMatch(discFilter);

      Pattern ratingPattern = Pattern.compile(RATING_EXPR);
      Pattern movieIdPattern = Pattern.compile(MOVIE_ID_EXPR);
     
      Node disc;
      NodeList info = new NodeList();
      SimpleNodeIterator iter = discs.elements();
      while (iter.hasMoreNodes()) {
        info.removeAll();
        infoCollector.clear();
       
        disc = iter.nextNode();
       
        //collect all the nodes for the movie items
        disc.collectInto(info, infoCollector);

        int i=0;

        //parse out all the information
        NodeList links = infoCollector.getNodeList(i++);
        LinkTag titleLink = (LinkTag)links.elementAt(0);
        LinkTag mpaaLink = (LinkTag)links.elementAt(1);
        String movieId = null;
        Matcher matcher = movieIdPattern.matcher(titleLink.getLink());
        if (matcher.matches()) {
          movieId = matcher.group(1);
        } else {
View Full Code Here

      NodeFilter removeFilter = new AndFilter(
          new NodeClassFilter(LinkTag.class),
          new CssClassFilter("bvr-qremove")
      );

      NodeList discNodes = parser.extractAllNodesThatMatch(discFilter);

      if (discNodes.size() == 1) {
        Node disc = discNodes.elementAt(0);
        NodeList list = new NodeList();
        disc.collectInto(list, removeFilter);
                String queueId = null;
                boolean discSet = false;
                if (list.size() == 1) {
                  LinkTag removeNode = (LinkTag)list.elementAt(0);
                    queueId = removeNode.getLink();
                    discSet = queueId.indexOf("setId") != -1;
                    int index = queueId.lastIndexOf('=');
                    if (index != -1) {
                        queueId = queueId.substring(index+1);
View Full Code Here

                            new NodeClassFilter(TextNode.class)),
                    new AndFilter(
                            new HasAncestorFilter(new CssClassFilter("detailSpecs")),
                            new CssClassFilter("movieInfo"))
            });
            NodeList all = parser.extractAllNodesThatMatch(infoCollector);

            if (all.size() >= 3) {
                int i=0;
                //get the icon
                ImageTag icon = (ImageTag)infoCollector.getNode(i++);
                item.setIconUrl(icon.getImageURL().replace("&&", "&"));

                //get the description
                NodeList list = infoCollector.getNodeList(i++);
                item.setSummary(NodeUtils.getTextData(list.elementAt(list.size()-1)));
               
                //strip off the query params
                int index = item.getIconUrl().indexOf('?');
                if (index != -1) {
                    item.setIconUrl(item.getIconUrl().substring(0, index)+ "?wid=130&hei=182&cvt=jpeg");
View Full Code Here

            }

            Node infoNode = infoCollector.getNode(i++);
           
            //info
            NodeList list = new NodeList();
            NodeFilter groupsFilter = new CssClassFilter("movieInfo");
            infoNode.collectInto(list, groupsFilter);
            setListValues(item, list);
           
            //summary
            list.removeAll();
            NodeFilter summaryFilter = new CssClassFilter("summary");
            infoNode.collectInto(list, summaryFilter);
            item.setSummary(NodeUtils.getTextData(list));
           
            item.addDetailFlag(MovieItem.DETAIL_MEDIUM);
View Full Code Here

        });
   
    Pattern ratingPattern = Pattern.compile(RATING_IMG_EXPR);
        SimpleDateFormat dateFormat = new SimpleDateFormat("yyyy-MM-dd");
   
    NodeList discs = new NodeList();
    queue.collectInto(discs, discFilter);

        int index;
    NodeList all = new NodeList();
    SimpleNodeIterator iter = discs.elements();
    while (iter.hasMoreNodes()) {
            all.removeAll();
            infoCollector.clear();
      Node disc = iter.nextNode();
     
      //collect all the nodes for the movie items
      disc.collectInto(all, infoCollector);
View Full Code Here

                new CssClassFilter("summary")
        });

    Pattern ratingPattern = Pattern.compile(RATING_IMG_EXPR);

    NodeList all = new NodeList();
    movie.collectInto(all, infoCollector);

        String movieId = ((Tag)movie).getAttribute("id");
    MovieItem item = null;
    if (all.size() > 0) {
            item = createMovieItem(movieId);
     
      int i=0;
           
            //save off the title and image
View Full Code Here

          new CssClassFilter("movie"),
      });
       
        parser.extractAllNodesThatMatch(searchCollector);
        Node moviesTab = searchCollector.getNode(0);
        NodeList searchNodes = searchCollector.getNodeList(1);
       
            if (searchNodes != null) {
            if (totalResultsOut != null) {
              //parse out the total results
            String totalStr = NodeUtils.getTextData(moviesTab);
            Pattern totalResultsPattern = Pattern.compile(SEARCH_RESULT_COUNT_EXPR);
            Matcher matcher = totalResultsPattern.matcher(totalStr);
            if (matcher.matches()) {
              totalResultsOut[0] = Integer.parseInt(matcher.group(1));
            }
            }
       
            if (searchNodes.size() > 0) {
              NodeIterator iter = searchNodes.elements();
              while (iter.hasMoreNodes()) {
              MovieItem item = parseMovieSearch(iter.nextNode());
              items.add(item);
              }
            }
View Full Code Here

                                new OrFilter(new CssClassFilter("column33"),
                                        new CssClassFilter("column34"))),
            new CssClassFilter("listMain")
        );
 
        NodeList categoryNodes = parser.extractAllNodesThatMatch(categoryFilter);

                List<Category> categories = new ArrayList<Category>();
        if (categoryNodes.size() > 0) {
          NodeIterator iter = categoryNodes.elements();
          while (iter.hasMoreNodes()) {
            Node node = iter.nextNode();
            parseCategory(node, categories);
          }
          this.categories = categories;
View Full Code Here

TOP

Related Classes of org.htmlparser.util.NodeList

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.