/// <summary>
///
/// </summary>
/// <param name="htmlDocument"></param>
/// <param name="actionAdExtracted"></param>
/// <returns></returns>
public Models.Advertisement.Advertisements ExtractAds(string htmlDocument, Action <Handlers.EventHandlers.AdExtractedEventArgs> actionAdExtracted)
{
Models.Advertisement.Advertisements _ads = null;
try
{
HtmlAgilityPack.HtmlDocument _doc = new HtmlAgilityPack.HtmlDocument();
_doc.LoadHtml(htmlDocument);
var _nodes = HtmlUtil.GetNodeCollection(_doc.DocumentNode, "div", "class", "ypgListing clearfix");
if (_nodes != null && _nodes.Count() > 0)
{
_ads = new Models.Advertisement.Advertisements();
int _pgItemIdx = 0;
foreach (HtmlAgilityPack.HtmlNode _n in _nodes)
{
Models.Advertisement _ad = new Models.Advertisement();
_pgItemIdx++;
Stopwatch _stopwatch = new Stopwatch();
_stopwatch.Reset();
_stopwatch.Start();
_ad = this.EscrapeAdInfo(_n);
if (_ad != null)
{
_ad = this.EscrapeAdInfoExtend(_ad);
}
_stopwatch.Stop();
_ads.Add(_ad);
if (actionAdExtracted != null)
{
actionAdExtracted(new Handlers.EventHandlers.AdExtractedEventArgs(_ad, _stopwatch.Elapsed, _pgItemIdx));
}
}
}
}
catch (Exception ex)
{
if (this.frameworkExceptionInvoke != null)
{
Exception _ex = new Exception(string.Format("Exception in {0}.{1}(?)", this.directoryProviderSetting.ServicedCountry.ToString(), "ExtractAds"), ex);
this.frameworkExceptionInvoke(new Handlers.EventHandlers.FrameworkExceptionEventArgs(_ex));
}
}
return(_ads);
}
/// <summary>
///
/// </summary>
/// <param name="htmlDocument"></param>
/// <returns></returns>
public int GetResultsPerPage(string htmlDocument)
{
HtmlAgilityPack.HtmlDocument _doc = new HtmlAgilityPack.HtmlDocument();
_doc.LoadHtml(htmlDocument);
var _nodes = HtmlUtil.GetNodeCollection(_doc.DocumentNode, "li", "class", "gold mappableListing listingContainer omnitureListing");
if (_nodes != null && _nodes.Count() > 0)
{
return(_nodes.Count());
}
return(0);
}
/// <summary>
///
/// </summary>
/// <param name="htmlDocument"></param>
/// <returns></returns>
public int GetResultsPerPage(string htmlDocument)
{
HtmlAgilityPack.HtmlDocument _doc = new HtmlAgilityPack.HtmlDocument();
_doc.LoadHtml(htmlDocument);
var _nodes = HtmlUtil.GetNodeCollection(_doc.DocumentNode, "div", "class", "ypgListing clearfix");
if (_nodes != null && _nodes.Count() > 0)
{
return(_nodes.Count());
}
return(0);
}
/// <summary>
///
/// </summary>
/// <param name="pageContentDocument"></param>
/// <param name="freeProxyFetched"></param>
/// <returns></returns>
public override Queue <FreeProxy> LoadUpIPProxies(string pageContentDocument, Action <FreeProxy> freeProxyFetched)
{
Queue <FreeProxy> freeProxies = null;
try
{
var doc = new HtmlDocument();
doc.LoadHtml(pageContentDocument);
var document = doc.DocumentNode;
var target = HtmlUtil.GetNodeByAttribute(document, "table", "id", "proxylist");
if (target == null)
{
return(null);
}
var lines = HtmlUtil.GetNodeCollection(target, "tr");
lines.ToList().ForEach(e =>
{
if (freeProxies == null)
{
freeProxies = new Queue <FreeProxy>();
}
if (e.Descendants("td").Count() <= 1)
{
return;
}
try
{
var proxy = new FreeProxy();
var cells = e.Descendants("td").ToArray();
var unwanted = cells[0].Descendants("script");
var enumerable = unwanted as HtmlNode[] ?? unwanted.ToArray();
if (unwanted != null && enumerable.ToArray().Any())
{
enumerable.ToArray()[0].Remove();
}
var address = cells[0];
var addressParts = address.InnerText.Split(new char[] { ':' }, StringSplitOptions.RemoveEmptyEntries);
proxy.IPAddress = addressParts[0];
proxy.PortNo = int.Parse(addressParts[1]);
// get anonymity level
//proxy.AnonymityLevel = cells[1].InnerText.Contains("high")
// ? ProxyAnonymityLevelEnum.High
// : ProxyAnonymityLevelEnum.Medium;
proxy.AnonymityLevel = cells[1].InnerText;
// get last checked time
var checkdate = cells[2].InnerText;
//todo:
//proxy.LastValidationCheck = DateTime.Parse(checkdate);
// get the country
var countryPartial = cells[3].InnerText.ToLower().Replace(" ", "_");
countryPartial = countryPartial.Split(new[] { ',' }, StringSplitOptions.RemoveEmptyEntries)[0];
var pxycountry = FindProxyCountryFromPartial(countryPartial);
proxy.Country = pxycountry;
//if (!ProxyTestHelper.CanPing(string.Format("{0}://{1}:{2}", proxy.Protocol == ProxyProtocolEnum.HTTP ? "http" : "https", proxy.HostIP, proxy.PortNo)))
//if (!ProxyTestHelper.ProxyIsGood(proxy.HostIP, proxy.PortNo)) return;
freeProxies.Enqueue(proxy);
freeProxyFetched(proxy);
}
catch (Exception ex)
{
throw;
}
});
}
catch (Exception ex)
{
throw;
}
return(freeProxies);
}
/// <summary>
///
/// </summary>
/// <param name="pageContentDocument"></param>
/// <param name="freeProxyFetched"></param>
/// <returns></returns>
public override Queue <FreeProxy> LoadUpIPProxies(string pageContentDocument, Action <FreeProxy> freeProxyFetched)
{
Queue <FreeProxy> freeProxies = null;
try
{
var doc = new HtmlDocument();
doc.LoadHtml(pageContentDocument);
var document = doc.DocumentNode;
var target = HtmlUtil.GetNodeByAttribute(document, "table", "class", "proxytbl");
if (target == null)
{
return(null);
}
var lines = HtmlUtil.GetNodeCollection(target, "tr");
lines.ToList().ForEach(e =>
{
if (freeProxies == null)
{
freeProxies = new Queue <FreeProxy>();
}
if (e.Descendants("td").Count() <= 1)
{
return;
}
var proxy = new FreeProxy();
var cells = e.Descendants("td").ToArray();
//ip
var ip = cells[0].InnerText;
//port
var scriptPart = cells[1].Descendants("script");
var htmlNodes = scriptPart as HtmlNode[] ?? scriptPart.ToArray();
if (htmlNodes.Any())
{
htmlNodes.ToArray()[0].Remove();
}
var port = HtmlUtil.Resolve(cells[1].InnerText);
// country
var country = HtmlUtil.Resolve(cells[2].InnerText);
// anon
var anon = HtmlUtil.Resolve(cells[3].InnerText);
//https
var http = HtmlUtil.Resolve(cells[4].Attributes["class"].Value);
//last check
var lastChecked = HtmlUtil.Resolve(cells[5].InnerText);
freeProxies.Enqueue(proxy);
freeProxyFetched(proxy);
});
}
catch (Exception ex)
{
throw;
}
return(freeProxies);
}
/// <summary>
///
/// </summary>
/// <param name="content"></param>
protected override void ParseProxyPage(string content)
{
var doc = new HtmlDocument();
doc.LoadHtml(content);
var document = doc.DocumentNode;
var target = HtmlUtil.GetNodeByAttribute(document, "table", "class", "proxytbl");
if (target == null)
{
return;
}
var lines = HtmlUtil.GetNodeCollection(target, "tr");
lines.ToList().ForEach(e =>
{
if (e.Descendants("td").Count() <= 1)
{
return;
}
var proxy = new IPProxy()
{
ProviderId = GetType().Name.Replace("Cartridge", ""),
};
var cells = e.Descendants("td").ToArray();
var scriptPart = cells[1].Descendants("script");
var htmlNodes = scriptPart as HtmlNode[] ?? scriptPart.ToArray();
if (htmlNodes.Any())
{
htmlNodes.ToArray()[0].Remove();
}
//ip
proxy.IPAddress = cells[0].InnerText.Trim();
//port
proxy.PortNo = int.Parse(HtmlUtil.Resolve(cells[1].InnerText.Trim()).Replace("\r\n", "").Trim());
// country
var country = HtmlUtil.Resolve(cells[2].InnerText.Trim());
var countryPrts = country.Split(new[] { ',' }, StringSplitOptions.RemoveEmptyEntries);
proxy.Country = Helper.FindProxyCountryFromPartial(countryPrts[0].Replace(" ", "_"));
// anon
proxy.AnonymityLevel = cells[3].InnerText.Trim().Replace("\r\n", "").Trim().Contains("anonymous")
? ProxyAnonymityLevelsEnum.Anonymous
: ProxyAnonymityLevelsEnum.Elite;
//protocol
var https = HtmlUtil.Resolve(cells[4].Attributes["class"].Value);
proxy.Protocol = https.ToLower().Contains("https")
? ProxyProtocolsEnum.HTTPS
: ProxyProtocolsEnum.HTTP;
//last check
var lastChecked = (new DateTime(DateTime.Now.Year, DateTime.Now.Month, DateTime.Now.Day));
proxy.LastChecked = lastChecked.Add(TimeSpan.Parse(HtmlUtil.Resolve(cells[5].InnerText.Trim())));
RegisterProxy(proxy);
});
base.ParseProxyPage(content);
}