public void Log <TState>(LogLevel logLevel, EventId eventId, TState state,
Exception exception, Func <TState, Exception, string> formatter)
{
if (IsEnabled(logLevel))
{
Trace.TraceEvent(TraceMapping[logLevel], $"{_categoryName}: {formatter(state, exception)}");
}
}
/// <inheritdoc />
public IEnumerable <IDocument> Execute(IReadOnlyList <IDocument> inputs, IExecutionContext context)
{
#pragma warning disable RCS1163 // Unused parameter.
// Handle invalid HTTPS certificates and allow alternate security protocols (see http://stackoverflow.com/a/5670954/807064)
ServicePointManager.ServerCertificateValidationCallback = (s, cert, chain, ssl) => true;
#pragma warning restore RCS1163 // Unused parameter.
// Key = link, Value = source, tag HTML
ConcurrentDictionary <string, ConcurrentBag <(string documentSource, string outerHtml)> > links =
new ConcurrentDictionary <string, ConcurrentBag <(string documentSource, string outerHtml)> >();
// Key = source, Value = tag HTML
ConcurrentDictionary <string, ConcurrentBag <string> > failures = new ConcurrentDictionary <string, ConcurrentBag <string> >();
// Gather all links
HtmlParser parser = new HtmlParser();
context.ParallelForEach(inputs, input => GatherLinks(input, parser, links));
// This policy will limit the number of executing link validations.
// Limit the amount of concurrent link checks to avoid overwhelming servers.
Task[] tasks = links.Select(
async link =>
{
// Attempt to parse the URI
if (!Uri.TryCreate(link.Key, UriKind.RelativeOrAbsolute, out Uri uri))
{
AddOrUpdateFailure(link.Value, failures);
}
// Adjustment for double-slash link prefix which means use http:// or https:// depending on current protocol
// The Uri class treats these as relative, but they're really absolute
if (uri.ToString().StartsWith("//") && !Uri.TryCreate($"http:{link.Key}", UriKind.Absolute, out uri))
{
AddOrUpdateFailure(link.Value, failures);
}
// Relative
if (!uri.IsAbsoluteUri && _validateRelativeLinks && !(await ValidateRelativeLink(uri, context).ConfigureAwait(false)))
{
AddOrUpdateFailure(link.Value, failures);
}
// Absolute
if (uri.IsAbsoluteUri && _validateAbsoluteLinks && !(await ValidateAbsoluteLink(uri, context).ConfigureAwait(false)))
{
AddOrUpdateFailure(link.Value, failures);
}
}).ToArray();
Task.WaitAll(tasks);
// Report failures
if (failures.Count > 0)
{
int failureCount = failures.Sum(x => x.Value.Count);
string failureMessage = string.Join(
Environment.NewLine,
failures.Select(x => $"{x.Key}{Environment.NewLine} - {string.Join(Environment.NewLine + " - ", x.Value)}"));
Trace.TraceEvent(
_asError ? TraceEventType.Error : TraceEventType.Warning,
$"{failureCount} link validation failures:{Environment.NewLine}{failureMessage}");
}
return(inputs);
}
public IEnumerable <IDocument> Execute(IReadOnlyList <IDocument> inputs, IExecutionContext context)
{
// Handle invalid HTTPS certificates and allow alternate security protocols (see http://stackoverflow.com/a/5670954/807064)
ServicePointManager.ServerCertificateValidationCallback = (s, cert, chain, ssl) => true;
// Key = link, Value = source, tag HTML
ConcurrentDictionary <string, ConcurrentBag <Tuple <FilePath, string> > > links =
new ConcurrentDictionary <string, ConcurrentBag <Tuple <FilePath, string> > >();
// Key = source, Value = tag HTML
ConcurrentDictionary <FilePath, ConcurrentBag <string> > failures = new ConcurrentDictionary <FilePath, ConcurrentBag <string> >();
// Gather all links
HtmlParser parser = new HtmlParser();
context.ParallelForEach(inputs, input =>
{
GatherLinks(input, parser, links);
});
// Perform validation
Parallel.ForEach(links, link =>
{
// Attempt to parse the URI
Uri uri;
if (!Uri.TryCreate(link.Key, UriKind.RelativeOrAbsolute, out uri))
{
AddOrUpdateFailure(link.Value, failures);
}
// Adjustment for double-slash link prefix which means use http:// or https:// depending on current protocol
// The Uri class treats these as relative, but they're really absolute
if (uri.ToString().StartsWith("//") && !Uri.TryCreate($"http:{link.Key}", UriKind.Absolute, out uri))
{
AddOrUpdateFailure(link.Value, failures);
}
// Relative
if (!uri.IsAbsoluteUri && _validateRelativeLinks && !ValidateRelativeLink(uri, context))
{
AddOrUpdateFailure(link.Value, failures);
}
// Absolute
if (uri.IsAbsoluteUri && _validateAbsoluteLinks && !ValidateAbsoluteLink(uri, context))
{
AddOrUpdateFailure(link.Value, failures);
}
});
// Report failures
if (failures.Count > 0)
{
int failureCount = failures.Sum(x => x.Value.Count);
string failureMessage = string.Join(Environment.NewLine,
failures.Select(x => $"{x.Key.FullPath}{Environment.NewLine} - {string.Join(Environment.NewLine + " - ", x.Value)}"));
Trace.TraceEvent(
_asError ? TraceEventType.Error : TraceEventType.Warning,
$"{failureCount} link validation failures:{Environment.NewLine}{failureMessage}");
}
return(inputs);
}
