private string HtmlText(string sourceHtml)
{
hParser.Parser parser = hParser.Parser.CreateParser(sourceHtml.Replace(System.Environment.NewLine, ""), "utf-8");
StringBuilder builderHead = new StringBuilder();
StringBuilder builderBody = new StringBuilder();
hParser.NodeFilter html = new TagNameFilter("TR");
hParser.INode nodes = parser.Parse(html)[0];
builderHead.Append(nodes.Children[0].ToHtml());
hParser.INode body = nodes.Children[1];
hParser.INode div = body.Children[0];
for (int i = 0; i < div.Children.Count; i++)
{
if (div.Children[i] is hParser.ITag)
{
builderBody.Append(div.Children[i].ToHtml());
}
}
StringBuilder builder = new StringBuilder();
builder.Append("<html>");
builder.Append(builderHead.ToString());
builder.Append("<body>");
builder.Append(string.Format("<{0}>", div.GetText()));
builder.Append(builderBody.ToString());
builder.Append("</div>");
builder.Append("</body>");
builder.Append("</html>");
return(builder.ToString());
}
private void paserData(hParser.INode node)
{
hParser.ITag tag = getTag(node);
if (tag != null && !tag.IsEndTag() && !start.Contains(tag.StartPosition))
{
object oId = tag.GetAttribute("ID");
object oName = tag.GetAttribute("name");
object oClass = tag.GetAttribute("class");
parseResult += tag.TagName + ":\r\nID:" + oId + " Name:" + oName
+ " Class:" + oClass + " StartPosition:" + tag.StartPosition.ToString() + "\r\n";
start.Add(tag.StartPosition);
}
//子节点
if (node.Children != null && node.Children.Count > 0)
{
paserData(node.FirstChild);
}
//兄弟节点
hParser.INode siblingNode = node.NextSibling;
while (siblingNode != null)
{
paserData(siblingNode);
siblingNode = siblingNode.NextSibling;
}
}
private void parserTR(hParser.INode node)
{
hParser.Tags.TableRow tagTR = getTagRow(node);
//TD在子节点
if (tagTR.Headers != null && tagTR.Headers.Count() > 0)
{
for (int i = 0; i < tagTR.Headers.Count(); i++)
{
var header = tagTR.Headers[i] as hParser.Tags.TableHeader; // th
if (header.TagName == "TH" && !string.IsNullOrEmpty(header.StringText))
{
parseResult += header.TagName + ":\r\nStringText:" + header.StringText + " ChildrenHTML:" + header.ChildrenHTML
+ " StartPosition:" + header.StartPosition.ToString() + " EndPosition:" + header.EndPosition.ToString() + "\r\n";
}
}
}
if ((tagTR.Headers == null || tagTR.Headers.Count() == 0) && tagTR.ChildrenAsNodeArray != null && tagTR.ChildrenAsNodeArray.Count() > 0)
{
for (int i = 0; i < tagTR.ChildrenAsNodeArray.Count(); i++)
{
var colum = tagTR.ChildrenAsNodeArray[i] as hParser.Tags.TableColumn; //td
if (colum != null && colum.TagName == "TD" && !string.IsNullOrEmpty(colum.StringText) && colum.StringText != "\n")
{
parseResult += colum.TagName + ":\r\nStringText:" + colum.StringText + " ChildrenHTML:" + colum.ChildrenHTML
+ " StartPosition:" + colum.StartPosition.ToString() + " EndPosition:" + colum.EndPosition.ToString() + "\r\n";
}
}
}
}
private hParser.ITag getTag(hParser.INode node)
{
if (node == null)
{
return(null);
}
return(node is hParser.ITag ? node as hParser.ITag : null);
}
private hParser.Tags.TableRow getTagRow(hParser.INode node)
{
if (node == null)
{
return(null);
}
return(node is hParser.Tags.TableRow ? node as hParser.Tags.TableRow : null);
}
private hParser.Tags.TableRow parserTR(hParser.INode node)
{
hParser.Tags.TableRow tagTR = getTagRow(node);
bool isValid = false;
//TD在子节点
//抓取th
if (tagTR.Headers != null && tagTR.Headers.Count() > 0)
{
for (int i = 0; i < tagTR.Headers.Count(); i++)
{
var header = tagTR.Headers[i] as hParser.Tags.TableHeader; // th
if (header.TagName == "TH" && !string.IsNullOrEmpty(header.StringText))
{
// parseResult += header.TagName + ":\r\nStringText:" + header.StringText + " ChildrenHTML:" + header.ChildrenHTML
//+ " StartPosition:" + header.StartPosition.ToString() + " EndPosition:" + header.EndPosition.ToString() + "\r\n";
}
}
}
//抓取td
if ((tagTR.Headers == null || tagTR.Headers.Count() == 0) && tagTR.ChildrenAsNodeArray != null && tagTR.ChildrenAsNodeArray.Count() > 0
)
{
for (int i = 0; i < tagTR.ChildrenAsNodeArray.Count(); i++)
{
var colum = tagTR.ChildrenAsNodeArray[i] as hParser.Tags.TableColumn; //td
if (colum != null && colum.TagName == "TD" && !string.IsNullOrEmpty(colum.StringText) && colum.StringText != "\n"
//&&
//(0 == string.Compare(colum.StringText.Trim(), "底账", StringComparison.InvariantCultureIgnoreCase)
//|| 0 == string.Compare(colum.StringText.Trim(), "认证", StringComparison.InvariantCultureIgnoreCase)
//|| 0 == string.Compare(colum.StringText.Trim(), "全部", StringComparison.InvariantCultureIgnoreCase)
//)
)
{
isValid = true;
}
if (isValid)
{
break;
}
}
}
return(isValid ? tagTR : null);
}
