public async Task <WebCrawlerOutput> PerformCrawlingAsync(Uri uri, int currentDepth, int parentId)
{
if (IsUriUnique(uri))
{
// Load web resource
var pageLoader = new WebPageLoader();
var loadResult = await pageLoader.LoadAsync(uri);
// Register resource
int uriId = AddUri(uri);
// Generate crawler output
WebCrawlerOutput output = new WebCrawlerOutput(
uriId, uri, loadResult.Response, loadResult.Content);
NotifyResourceLoadingFinished(parentId, output);
currentDepth++;
if (loadResult.Content != null && currentDepth < MaxDepth)
{
ExtractPageLinks(loadResult.Content, uri, output, currentDepth);
}
return(output);
}
return(null);
}
private void NotifyResourceLoadingFinished(int parentId, WebCrawlerOutput output)
{
if (LoadingFinished != null)
{
LoadingFinished(parentId, output);
}
}
private async void ExtractPageLinks(byte[] content, Uri parentUri, WebCrawlerOutput output, int depth)
{
LinkExtractor extractor = new LinkExtractor();
Encoding responseEncoding = GetEncodingFromResponse(output.Response);
List <Uri> childUris = extractor.ExtractLinksFromPage(parentUri, content, responseEncoding);
foreach (var link in childUris)
{
var childOutput = await PerformCrawlingAsync(link, depth, GetUriId(parentUri));
if (childOutput != null)
{
output.AddChild(childOutput);
}
}
}
public void AddChild(WebCrawlerOutput child)
{
Children.Add(child);
}
