/// <summary>
///
/// </summary>
/// <param name="content"></param>
protected override void ValidatePage(string content)
{
try
{
var doc = new HtmlDocument();
doc.LoadHtml(content);
var document = doc.DocumentNode;
var target = HtmlUtil.GetNodeByAttribute(document, "table", "id", "proxylisttable");
PageIsValid = target != null;
}
catch (Exception ex)
{
}
}
/// <summary>
///
/// </summary>
/// <param name="hmaPgDoc"></param>
private int ExtractProxies(string hmaPgDoc)
{
var rowCount = 0;
var doc = new HtmlDocument();
doc.LoadHtml(hmaPgDoc);
var tblNode = HtmlUtil.GetNodeByAttribute(doc.DocumentNode, "table", "id", "listtable");
if (tblNode == null)
{
return(rowCount);
}
var tRowNodes = tblNode.Descendants("tr");
var htmlNodes = tRowNodes as HtmlNode[] ?? tRowNodes.ToArray();
rowCount = htmlNodes.Count();
if (tRowNodes == null || rowCount <= 0)
{
return(rowCount);
}
for (var ctr = 1; ctr < htmlNodes.Count(); ctr++)
{
var protocolOk = true;
var anonymityLevelOk = true;
var speedOk = true;
var connectionTimeOk = true;
FreeProxy freeProxy = null;
try
{
if (ParseRowToProxy(htmlNodes.ToArray()[ctr], ref freeProxy))
{
if (freeProxy != null)
{
if (_protocolOptions != null && _protocolOptions.Count() > 0)
{
protocolOk = _protocolOptions.Any(x => x != freeProxy.Protocol);
}
else if (Protocol.HasValue)
{
protocolOk = (Protocol.Value == freeProxy.Protocol);
}
if (AnonymityLevel.HasValue)
{
if (AnonymityLevel.Value == ProxyAnonymityLevelEnum.High__KA)
{
anonymityLevelOk = freeProxy.AnonymityLevel == ProxyAnonymityLevelEnum.High__KA;
}
else if (AnonymityLevel.Value == ProxyAnonymityLevelEnum.PlanetLab)
{
anonymityLevelOk = freeProxy.AnonymityLevel == ProxyAnonymityLevelEnum.PlanetLab;
}
else
{
if (AnonymityLevel.Value == ProxyAnonymityLevelEnum.High)
{
anonymityLevelOk = freeProxy.AnonymityLevel == ProxyAnonymityLevelEnum.High;
}
else if (AnonymityLevel.Value == ProxyAnonymityLevelEnum.Medium)
{
anonymityLevelOk = freeProxy.AnonymityLevel == ProxyAnonymityLevelEnum.High |
freeProxy.AnonymityLevel == ProxyAnonymityLevelEnum.Medium;
}
else if (AnonymityLevel.Value == ProxyAnonymityLevelEnum.Low)
{
anonymityLevelOk = freeProxy.AnonymityLevel == ProxyAnonymityLevelEnum.High |
freeProxy.AnonymityLevel == ProxyAnonymityLevelEnum.Medium |
freeProxy.AnonymityLevel == ProxyAnonymityLevelEnum.Low;
}
else if (AnonymityLevel.Value == ProxyAnonymityLevelEnum.None)
{
anonymityLevelOk = freeProxy.AnonymityLevel == ProxyAnonymityLevelEnum.High |
freeProxy.AnonymityLevel == ProxyAnonymityLevelEnum.Medium |
freeProxy.AnonymityLevel == ProxyAnonymityLevelEnum.Low |
freeProxy.AnonymityLevel == ProxyAnonymityLevelEnum.None;
}
}
}
if (Speed.HasValue)
{
if (Speed.Value == ProxySpeedEnum.Fast)
{
speedOk = freeProxy.Speed == ProxySpeedEnum.Fast;
}
else if (Speed.Value == ProxySpeedEnum.Medium)
{
speedOk = freeProxy.Speed == ProxySpeedEnum.Medium | freeProxy.Speed == ProxySpeedEnum.Fast;
}
else
{
speedOk = true;
}
}
if (ConnectionTime.HasValue)
{
if (ConnectionTime.Value == ProxyConnectionSpeedEnum.Fast)
{
connectionTimeOk = freeProxy.ConnectionTime == ProxyConnectionSpeedEnum.Fast;
}
else if (ConnectionTime.Value == ProxyConnectionSpeedEnum.Medium)
{
connectionTimeOk = freeProxy.ConnectionTime == ProxyConnectionSpeedEnum.Medium | freeProxy.ConnectionTime == ProxyConnectionSpeedEnum.Fast;
}
else
{
connectionTimeOk = true;
}
}
}
}
if (protocolOk && anonymityLevelOk && speedOk && connectionTimeOk)
{
var exFreeProxy = FreeProxies
.ToArray().FirstOrDefault(p => p.HostIP == freeProxy.HostIP && p.PortNo == freeProxy.PortNo);
if (exFreeProxy == null)
{
lock (FreeProxies)
{
FreeProxies.Add(freeProxy);
}
InvokeEventFreeIPProxyFetched(new EventHandlers.FreeIPProxyFetchedEventArgs(freeProxy));
}
}
}
catch (Exception ex)
{
throw;
}
}
return(rowCount);
}
/// <summary>
///
/// </summary>
/// <param name="rowNode"></param>
/// <param name="freeProxy"></param>
/// <returns></returns>
private bool ParseRowToProxy(HtmlNode rowNode, ref FreeProxy freeProxy)
{
try
{
var prxCells = rowNode.Descendants("td");
var htmlNodes = prxCells as HtmlNode[] ?? prxCells.ToArray();
if (prxCells != null && htmlNodes.Any())
{
freeProxy = new FreeProxy();
freeProxy.HostIP = SnatchIPAddress(htmlNodes.ToArray()[1].InnerHtml);
freeProxy.PortNo = int.Parse(htmlNodes.ToArray()[2].InnerText);
var country = htmlNodes.ToArray()[3].InnerText.Trim().ToUpper();
country = country.Replace("; ", "___")
.Replace(", ", "__")
.Replace(" ", "_")
.Replace("'", "____");
freeProxy.Country = (ProxyCountry)Enum.Parse(typeof(ProxyCountry), country);
var connspeedNode = htmlNodes.ToArray()[5];
connspeedNode = HtmlUtil.GetNodeByAttribute(connspeedNode, "div", "class", "speedbar connection_time");
connspeedNode = connspeedNode.Descendants("div").ToArray()[0];
var rate = connspeedNode.Attributes["style"].Value.Replace("width:", "").Replace("%", "");
freeProxy.ResponseRate = int.Parse(rate);
if (freeProxy.ResponseRate <= 35)
{
freeProxy.ConnectionTime = ProxyConnectionSpeedEnum.Slow;
}
else if (freeProxy.ResponseRate > 35 && freeProxy.ResponseRate <= 65)
{
freeProxy.ConnectionTime = ProxyConnectionSpeedEnum.Medium;
}
else if (freeProxy.ResponseRate > 65)
{
freeProxy.ConnectionTime = ProxyConnectionSpeedEnum.Medium;
}
var speedNode = htmlNodes.ToArray()[4];
speedNode = HtmlUtil.GetNodeByAttribute(speedNode, "div", "class", "speedbar response_time");
speedNode = speedNode.Descendants("div").ToArray()[0];
rate = speedNode.Attributes["style"].Value.Replace("width:", "").Replace("%", "");
freeProxy.SpeedRate = int.Parse(rate);
if (freeProxy.SpeedRate <= 35)
{
freeProxy.Speed = ProxySpeedEnum.Slow;
}
else if (freeProxy.SpeedRate > 35 && freeProxy.SpeedRate <= 65)
{
freeProxy.Speed = ProxySpeedEnum.Medium;
}
else if (freeProxy.SpeedRate > 65)
{
freeProxy.Speed = ProxySpeedEnum.Medium;
}
var protocol = htmlNodes.ToArray()[6].InnerText.Replace("/", "_");
freeProxy.Protocol = (ProxyProtocolEnum)Enum.Parse(typeof(ProxyProtocolEnum), protocol);
var anonymity = htmlNodes.ToArray()[7].InnerText.Replace(" +", "__");
freeProxy.AnonymityLevel = (ProxyAnonymityLevelEnum)Enum.Parse(typeof(ProxyAnonymityLevelEnum), anonymity);
freeProxy.Valid = true;
return(true);
}
}
catch (Exception ex)
{
throw;
}
return(false);
}
/// <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);
}