private static object DomObjectsToNodeObjectsConvernet(HtmlNode node)
{
if (node.NodeType == HtmlNodeType.Text)
{
if (node.InnerText.Length == 0)
{
return(null);
}
return(WebUtility.HtmlDecode(node.InnerText));
}
if (node.NodeType != HtmlNodeType.Element)
{
return(null);
}
var telegrahpNode = new TelegraphNode();
telegrahpNode.tag = node.OriginalName.ToLower();
var hrefValue = node.GetAttributeValue("href", null);
var srcValue = node.GetAttributeValue("src", null);
if ((hrefValue != null || srcValue != null) && telegrahpNode.attrs == null)
{
telegrahpNode.attrs = new Dictionary <string, string>();
}
if (hrefValue != null)
{
telegrahpNode.attrs["href"] = hrefValue;
}
if (srcValue != null)
{
telegrahpNode.attrs["src"] = srcValue;
}
if (node.ChildNodes.Count > 0)
{
telegrahpNode.children = new List <object>();
foreach (var childNode in node.ChildNodes)
{
telegrahpNode.children.Add(DomObjectsToNodeObjectsConvernet(childNode));
}
}
return(telegrahpNode);
}
private string TelegraphParserMethod(List <string> imageUrls, HtmlNode htmlNode)
{
var listOfNodes = new List <Object>();
TelegraphNode urlPNode = new TelegraphNode()
{
tag = "p"
};
urlPNode.children = new List <object>();
TelegraphNode urlNode = new TelegraphNode()
{
tag = "a"
};
urlNode.attrs = new Dictionary <string, string>();
urlNode.attrs["href"] = url;
urlNode.children = new List <object>();
urlNode.children.Add("Джерело");
urlPNode.children.Add(urlNode);
listOfNodes.Add(urlPNode);
foreach (var node in htmlNode.ChildNodes)
{
listOfNodes.Add(DomObjectsToNodeObjectsConvernet(node));
}
foreach (var item in imageUrls)
{
TelegraphNode imgNode = new TelegraphNode()
{
tag = "img"
};
imgNode.attrs = new Dictionary <string, string>();
imgNode.attrs["src"] = item;
listOfNodes.Add(imgNode);
}
string output = JsonConvert.SerializeObject(listOfNodes, Formatting.None, new JsonSerializerSettings {
NullValueHandling = NullValueHandling.Ignore
});
return(output);
}
