private void ProcessBookElement(HtmlNode trElement, List<SearchResult> SearchResultUrlList)
{
if (trElement.Name != "tr") return;
var item = new SearchResult();
var tdElements = trElement.DescendantNodes().Where(n => n.Name == "td").ToArray();
if (tdElements.Length == 5)
{
//Parse Book name and index page url
var aElement = tdElements[1].DescendantNodes().Where(n => n.Name == "a").First();
if (aElement != null)
{
item.IndexPageUri = new Uri("http://www.xiaoelang.com" + aElement.Attributes["href"].Value, UriKind.Absolute);
item.BookName = aElement.InnerText.Trim();
item.Book = new Book();
item.Book.Name = item.BookName;
}
//Parse last update chapter name
aElement = tdElements[2].DescendantNodes().Where(n => n.Name == "a").First();
if (aElement != null)
{
item.LastUpdateChapterName = aElement.InnerText.Trim();
}
//Parse last update time
//<td class="xt" align="center" valign="middle" width="78">2011-08-26</td>
}
SearchResultUrlList.Add(item);
}
public static IEnumerable<HtmlNode> GetDirectChildrenByType(HtmlNode node, string type)
{
if (node == null) return null;
var childs =
node.DescendantNodes().Where(
n => n.Name.ToLower(CultureInfo.InvariantCulture) == type.ToLower(CultureInfo.InvariantCulture));
return childs;
}
public static HtmlNode GetSingleDirectChildByTypeAndIndex(HtmlNode node, string type, int index)
{
if (node == null) return null;
var childs =
node.DescendantNodes().Where(
n => n.Name.ToLower(CultureInfo.InvariantCulture) == type.ToLower(CultureInfo.InvariantCulture));
int count = childs.Count();
if (count == 0) return null;
return childs.Skip(index).First();
}
public static HtmlNode FindElementById (HtmlNode node, string id)
{
foreach (var child in node.DescendantNodes())
{
if (child.Id.ToLower() == id.ToLower() )
{
return child;
}
else
{
var result = FindElementById(child, id);
if (result != null)
return result;
}
}
return null;
}
//tags are closed weird, can't just get descendants
public IEnumerable<HtmlForm> GetFormsFromContentNode(HtmlNode content)
{
var forms = new List<HtmlForm>();
foreach (var child in content.DescendantNodes())
{
if (child.Name.Equals("form"))
{
forms.Add(new HtmlForm { FormNode = child, ActionUrl = child.GetAttributeValue("action", "") });
}
else if (child.Name.Equals("input"))
{
var key = child.GetAttributeValue("name", null);
var value = child.GetAttributeValue("value", null);
if (!string.IsNullOrEmpty(key) && !key.Equals("reset"))
{
forms.Last().Inputs[key] = value;
}
}
else if (child.Name.Equals("select"))
{
var key = child.GetAttributeValue("name", null);
var options = child.Descendants("option")
.Select(y => y.GetAttributeValue("value", null))
.Where(y => !string.IsNullOrEmpty(y))
.ToList();
if (!string.IsNullOrEmpty(key))
{
forms.Last().Selects[key] = options;
}
}
}
return forms;
}
/// <summary>
/// <div style='display:none'>
/// </summary>
/// <param name="node"></param>
/// <param name="nodesToBeDeleted"></param>
public static void RemoveNonDisplayNode(HtmlNode node , List<HtmlNode> nodesToBeDeleted)
{
foreach (var child in node.DescendantNodes())
{
if (child.Attributes["style"]!= null && child.Attributes["style"].Value.ToLower().Contains("display:none"))
{
nodesToBeDeleted.Add(child);
}
else
{
RemoveNonDisplayNode(child , nodesToBeDeleted);
}
}
}
public static void RemoveNodeById(HtmlNode node, string Id, List<HtmlNode> nodesToBeDeleted)
{
foreach (var child in node.DescendantNodes())
{
if (child.Id.ToLower() == Id.ToLower())
{
nodesToBeDeleted.Add(child);
return;
}
else
{
RemoveNodeByType(child, Id, nodesToBeDeleted);
}
}
}
public static void RemoveNodeByClassName(HtmlNode node, string className, List<HtmlNode> nodesToBeDeleted)
{
foreach (var child in node.DescendantNodes())
{
if (child.Attributes["class"] == null ) continue;
if (child.Attributes["class"].Value.ToLower() == className)
{
nodesToBeDeleted.Add(child);
}
else
{
RemoveNodeByType(child, className, nodesToBeDeleted);
}
}
}
/// <summary>
/// Returns a node in the immediate decendants of the node which holds the attribute-value pair
/// </summary>
/// <param name="node">an HTML Node</param>
/// <param name="attribute"></param>
/// <param name="value"></param>
/// <returns></returns>
public HtmlNode getChild(HtmlNode node, string attribute, string value)
{
if (node == null || attribute == null || value == null)
return null;
foreach (HtmlNode child in node.DescendantNodes())
{
if (child.Attributes[attribute] != null)
{
if (child.Attributes[attribute].Value == value)
return child;
}
}
throw new InvalidDataException();
//return null;
}
public static void GetAllHyperlinkElementWithFilter(HtmlNode node, List<HtmlNode> hyperLinkNode)
{
foreach (var child in node.DescendantNodes())
{
if (child.Name.ToLower() == "a"
&& child.Attributes["target"] == null
&& child.Attributes["onclick"] == null
&& child.InnerText.Trim().Length > 1
&& !child.InnerText.Contains("最新章节")
)
{
var href = child.Attributes["href"];
var hrefValue = href == null ? String.Empty : href.Value.ToLower();
if ( href != null
&& !String.IsNullOrEmpty(hrefValue)
&& !hrefValue.EndsWith("/")
&& !hrefValue.Contains("xiazai")
&& !hrefValue.Contains("shuye")
&& !hrefValue.Contains("javascript")
&& !hrefValue.Contains("php")
&& !hrefValue.Contains("list")
&& !hrefValue.Contains("index")
&& !hrefValue.Contains("mailto")
)
{
if (!hyperLinkNode.Contains(child))
hyperLinkNode.Add(child);
}
continue;
}
GetAllHyperlinkElementWithFilter(child, hyperLinkNode);
}
}
public static void GetAllTextElement (HtmlNode node, List<HtmlNode> textNodeParents )
{
foreach (var child in node.DescendantNodes())
{
if (child.NodeType == HtmlNodeType.Text)
{
if ( !String.IsNullOrWhiteSpace(child.InnerText) && child.InnerText.Trim().Length > 1)
{
if (!textNodeParents.Contains(child))
textNodeParents.Add(child);
}
}
GetAllTextElement(child, textNodeParents);
}
}
public static void RemoveNodeByType(HtmlNode node, string type, List<HtmlNode> nodesToBeDeleted)
{
foreach (var child in node.DescendantNodes())
{
if (child.Name.ToLower() == type)
{
nodesToBeDeleted.Add(child);
}
else
{
RemoveNodeByType(child,type, nodesToBeDeleted);
}
}
}
private String getTitleFromURLNode(HtmlNode n)
{
String title = "";
if (n.Name.ToLower() == "a")
{
title = n.InnerText;
}else
{
foreach (HtmlNode i in n.DescendantNodes().ToList())
{
String u = getTitleFromURLNode(i);
if (u != "")
{
title = u;
break;
}
}
}
return title;
}
public String getURLFromNode(HtmlNode n)
{
String url = "";
if (n.Name.ToLower() == "a")
{
foreach (HtmlAttribute att in n.Attributes)
{
if (att.Name.ToLower() == "href")
{
url = att.Value;
break;
}
}
}
else
{
foreach (HtmlNode i in n.DescendantNodes().ToList())
{
String u = getURLFromNode(i);
if (u != "")
{
url = u;
break;
}
}
}
return url;
}
public void helpItem(HtmlNode item, Episode ep)
{
ep.ShowName = item.DescendantNodes().ToArray<HtmlNode>()[0].InnerText;
foreach (HtmlNode titleofzo in item.DescendantNodes())
if (titleofzo.Attributes["class"] != null)
{
switch (titleofzo.Attributes["class"].Value)
{
case "title":
ep.ShowName = titleofzo.DescendantNodes().ToArray<HtmlNode>()[0].InnerText;
break;
case "infos":
string[] build = titleofzo.DescendantNodes().ToArray<HtmlNode>()[0].InnerHtml.Split(new char[] { ' ' });
ep.ISeason = int.Parse(build[1]);
ep.IEpisode = int.Parse(build[3].Replace(":", ""));
ep.EpisodePos = "S" + build[1] + "E" + build[3].Replace(":", "");
string name = "";
for (int i = 4; i < build.Length; i++)
{
name += build[i] + " ";
}
ep.EpisodeName = name;
foreach (HtmlNode searchforSum in titleofzo.DescendantNodes())
{
if (searchforSum.Attributes["class"] != null && searchforSum.Attributes["class"].Value == "summary hidden-xs")
{
ep.summary = searchforSum.InnerText.Replace(@"... (more)", "");
}
}
break;
case "btn btn-default watch-episode-button":
ep.id = titleofzo.Attributes["episodeid"].Value;
break;
}
}
}
public static HtmlNode FindElementByClass(HtmlNode node, string className)
{
foreach (var child in node.DescendantNodes())
{
if (child.Attributes["class"] != null
&& child.Attributes["class"].Value.ToLower() == className.ToLower())
{
return child;
}
else
{
var result = FindElementById(child, className);
if (result != null)
return result;
}
}
return null;
}
protected void RemoveWhitespace(HtmlNode root)
{
root.DescendantNodes().ForEach(x => x.InnerHtml = Regex.Replace(x.InnerHtml, "\\s+", " "));
root.DescendantNodes().ForEach(x => x.InnerHtml = x.InnerHtml.Trim());
}
public static void GetAllImageElementWithFilter(HtmlNode node, List<HtmlNode> imageNodes)
{
foreach (var child in node.DescendantNodes())
{
if (child.Name.ToLower() == "img")
{
if (child.Attributes["src"] != null)
{
if (!imageNodes.Contains(child))
imageNodes.Add(child);
}
continue;
}
GetAllImageElementWithFilter(child, imageNodes);
}
}
