|
public CreateNavigator ( ) : |
||
| return | ||
private void RemoveXPaths(ConfigSection config, HtmlAgilityPack.HtmlNode parentNode)
{
if (parentNode != null && config != null && config.RemoveXPathRules != null && config.RemoveXPathRules.Count > 0)
{
foreach (var removeXPathRule in config.RemoveXPathRules)
{
var navigator = parentNode.CreateNavigator();
var nodes = navigator.Select(removeXPathRule);
foreach (HtmlNodeNavigator node in nodes)
{
node.CurrentNode.Remove();
}
}
}
}
private void FillRow(HtmlAgilityPack.HtmlNode node)
{
nav = node.CreateNavigator();
bool letolt = false;
string nyelv = "";
string magyar = "";
string eredeti = "";
string link = "";
DateTime datum = DateTime.Now;
//expr = nav.Compile(@"./td/small");
//expr = nav.Compile(@"./td[small='Magyar']");
//expr = nav.Compile(@"./td/div[@class='eredeti']");
// Nyelv:
expr = nav.Compile(@"./td/small");
iterator = nav.Select(expr);
while (iterator.MoveNext())
{
XPathNavigator nav2 = iterator.Current.Clone();
//Console.WriteLine("nyelv:" + nav2.Value);
nyelv = nav2.Value;
}
// magyar cím:
expr = nav.Compile(@"./td/div[@class='magyar']");
iterator = nav.Select(expr);
while (iterator.MoveNext())
{
XPathNavigator nav2 = iterator.Current.Clone();
//Console.WriteLine("magyar cím:" + nav2.Value);
magyar = nav2.Value;
}
// eredeti cím:
expr = nav.Compile(@"./td/div[@class='eredeti']");
iterator = nav.Select(expr);
while (iterator.MoveNext())
{
XPathNavigator nav2 = iterator.Current.Clone();
//Console.WriteLine("eredeti cím:" + nav2.Value);
eredeti = nav2.Value;
}
// Link:
expr = nav.Compile(@"./td/a[@href]");
iterator = nav.Select(expr);
while (iterator.MoveNext())
{
XPathNavigator nav2 = iterator.Current.Clone();
//Console.WriteLine("link:" + website + nav2.GetAttribute("href", ""));
link = website + nav2.GetAttribute("href", "");
}
// dátum:
expr = nav.Compile(@"./td");
iterator = nav.Select(expr);
while (iterator.MoveNext())
{
XPathNavigator nav2 = iterator.Current.Clone();
string datestring = nav2.Value.Trim();
string format = "yyyy-MM-dd";
DateTime dt;
if (DateTime.TryParseExact(datestring, format, CultureInfo.InvariantCulture, DateTimeStyles.None, out dt))
{
datum = dt.Date;
//DataGridViewCellStyle style = new DataGridViewCellStyle();
//style.Format = "yyyy.MM.dd";
//newrow.Cells[5].Style = style;
}
}
backgroundWorker.ReportProgress(1, new NewRowObject()
{
datum = datum,
eredeti = eredeti,
letolt = letolt,
link = link,
magyar = magyar,
nyelv = nyelv
});
}
private object Extract(string name, ConfigSection config, HtmlAgilityPack.HtmlNode parentNode, List <HtmlAgilityPack.HtmlNode> logicalParents)
{
this.RemoveUnwantedTags(config, parentNode);
// We will try to extract text for this item because it does not have children
var containers = new JArray();
if (config.XPathRules != null && config.XPathRules.Count > 0)
{
var navigator = parentNode.CreateNavigator();
foreach (var xpath in config.XPathRules)
{
// TODO: Add try catch Exception
var nodes = navigator.Select(xpath);
if (nodes != null && nodes.Count > 0)
{
var newLogicalParents = logicalParents.GetRange(0, logicalParents.Count);
newLogicalParents.Add(parentNode);
foreach (HtmlNodeNavigator node in nodes)
{
if (config.Children != null && config.Children.Count > 0)
{
var container = new JObject();
this.ExtractChildren(config: config, parentNode: node.CurrentNode, container: container, logicalParents: newLogicalParents);
containers.Add(container);
}
else if (config.Transformations != null && config.Transformations.Count > 0)
{
var obj = this.RunTransformations(config.Transformations, node.CurrentNode, newLogicalParents);
if (obj != null)
{
containers.Add(obj);
}
}
else if (node.Value != null)
{
containers.Add(HtmlEntity.DeEntitize(node.Value).Trim());
}
}
}
}
}
else
{
var container = new JObject();
this.ExtractChildren(config: config, parentNode: parentNode, container: container, logicalParents: logicalParents);
containers.Add(container);
}
if (!config.ForceArray && containers.Count == 0)
{
return(new JObject());
}
else if (!config.ForceArray && containers.Count == 1)
{
return(containers.First);
}
else
{
return(containers);
}
}