public async Task <PingRequest> ExamineSourceAsync(string sourceUrl, string targetUrl, int timeoutSeconds = 30)
{
try
{
var regexHtml = new Regex(
@"</?\w+((\s+\w+(\s*=\s*(?:"".*?""|'.*?'|[^'"">\s]+))?)+\s*|\s*)/?>",
RegexOptions.Singleline | RegexOptions.Compiled);
var regexTitle = new Regex(
@"(?<=<title.*>)([\s\S]*)(?=</title>)", RegexOptions.IgnoreCase | RegexOptions.Compiled);
using var httpClient = new HttpClient { Timeout = TimeSpan.FromSeconds(timeoutSeconds) };
var html = await httpClient.GetStringAsync(sourceUrl);
var title = regexTitle.Match(html).Value.Trim();
var containsHtml = regexHtml.IsMatch(title);
var sourceHasLink = html.ToUpperInvariant().Contains(targetUrl.ToUpperInvariant());
var pingRequest = new PingRequest
{
Title = title,
ContainsHtml = containsHtml,
SourceHasLink = sourceHasLink,
TargetUrl = targetUrl,
SourceUrl = sourceUrl
};
return(pingRequest);
}
catch (WebException ex)
{
_logger.LogError(ex, nameof(ExamineSourceAsync));
return(null);
}
}
private async Task <PingRequest> ExamineSourceAsync()
{
try
{
var regexHtml = new Regex(
@"</?\w+((\s+\w+(\s*=\s*(?:"".*?""|'.*?'|[^'"">\s]+))?)+\s*|\s*)/?>",
RegexOptions.Singleline | RegexOptions.Compiled);
var regexTitle = new Regex(
@"(?<=<title.*>)([\s\S]*)(?=</title>)", RegexOptions.IgnoreCase | RegexOptions.Compiled);
using (var httpClient = new HttpClient {
Timeout = TimeSpan.FromSeconds(RemoteTimeout)
})
{
var html = await httpClient.GetStringAsync(_sourceUrl);
var title = regexTitle.Match(html).Value.Trim();
Logger.LogInformation($"ExamineSourceAsync:title: {title}");
var containsHtml = regexHtml.IsMatch(title);
Logger.LogInformation($"ExamineSourceAsync:containsHtml: {containsHtml}");
var sourceHasLink = html.ToUpperInvariant().Contains(_targetUrl.ToUpperInvariant());
Logger.LogInformation($"ExamineSourceAsync:sourceHasLink: {sourceHasLink}");
var pingRequest = new PingRequest
{
SourceDocumentInfo = new SourceDocumentInfo
{
Title = title,
ContainsHtml = containsHtml,
SourceHasLink = sourceHasLink
},
TargetUrl = _targetUrl,
SourceUrl = _sourceUrl,
SourceIpAddress = _remoteIpAddress
};
return(pingRequest);
}
}
catch (WebException ex)
{
Logger.LogError(ex, nameof(ExamineSourceAsync));
return(new PingRequest
{
SourceDocumentInfo = new SourceDocumentInfo
{
SourceHasLink = false
},
SourceUrl = _sourceUrl,
SourceIpAddress = _remoteIpAddress
});
}
}
public PingbackServiceResponse ProcessReceivedPingback(PingRequest req, Func <bool> ifTargetResourceExists, Func <bool> ifAlreadyBeenPinged)
{
try
{
if (null == req)
{
return(PingbackServiceResponse.InvalidPingRequest);
}
var ti = ifTargetResourceExists();
if (!ti)
{
return(PingbackServiceResponse.Error32TargetUriNotExist);
}
var pd = ifAlreadyBeenPinged();
if (pd)
{
return(PingbackServiceResponse.Error48PingbackAlreadyRegistered);
}
if (req.SourceDocumentInfo.SourceHasLink && !req.SourceDocumentInfo.ContainsHtml)
{
Logger.LogInformation("Adding received pingback...");
var domain = GetDomain(_sourceUrl);
OnPingSuccess?.Invoke(this, new PingSuccessEventArgs(domain, req));
return(PingbackServiceResponse.Success);
}
if (!req.SourceDocumentInfo.SourceHasLink)
{
Logger.LogError("Pingback error: The source URI does not contain a link to the target URI, and so cannot be used as a source.");
return(PingbackServiceResponse.Error17SourceNotContainTargetUri);
}
Logger.LogWarning("Spam detected on current Pingback...");
return(PingbackServiceResponse.SpamDetectedFakeNotFound);
}
catch (Exception ex)
{
Logger.LogError(ex, nameof(ProcessReceivedPingback));
return(PingbackServiceResponse.GenericError);
}
}
public PingSuccessEventArgs(string domain, PingRequest pingRequest)
{
Domain = domain;
PingRequest = pingRequest;
}