public async Task <PageCheckResult> GetPageResult(string pageUrl)
{
await _throttler.WaitAsync();
try
{
string rawHtml = await _client.GetRawResultOfBasicGetRequestAsync(pageUrl);
var page = new Page(rawHtml, pageUrl);
var pageResult = new PageCheckResult(pageUrl, page.Title);
CheckForOutdated(page, pageResult);
CheckForProblematicWords(page, pageResult);
await CheckForBadLinks(page.InPostExternalLinks, pageResult);
return(pageResult);
}
catch (Exception ex)
{
return(new PageCheckResult(pageUrl, string.Empty, "This page generated an exception on parsing."));
}
finally
{
_throttler.Release();
}
}
private void CheckForOutdated(Page page, PageCheckResult pageResult)
{
if (ContainsNonCurrentYear(page.Title) || page.H1s.Any(h1 => ContainsNonCurrentYear(h1)))
{
pageResult.AddIssue("Is possibly outdated");
}
}
public async Task <PageCheckResult> GetPageResultAsync(string url)
{
var rawHtml = await _client.GetRawResultOfBasicGetRequestAsync(url);
var page = new Page(rawHtml, url);
var pageResult = new PageCheckResult(url, page.Title);
CheckForOutdated(page, pageResult);
return(pageResult);
}
private async Task CheckForBadLinks(IEnumerable <Link> inPostExternalLinks, PageCheckResult pageResult)
{
var nonSslLinks = inPostExternalLinks.Where(ipel => ipel.Url.StartsWith("http://"));
pageResult.AddIssues(nonSslLinks.Select(nssl => $"Non-ssl link {nssl.Url} with anchor text {nssl.AnchorText}."));
var linkCheckingTasks = new List <Task <string> >(inPostExternalLinks.Select(ipl => GetIssuesWithLink(ipl)));
await Task.WhenAll(linkCheckingTasks);
var badResponseCodeLinkProblems = linkCheckingTasks.Select(lct => lct.Result).Where(lcr => !string.IsNullOrEmpty(lcr));
pageResult.AddIssues(badResponseCodeLinkProblems);
}
private void CheckForProblematicWords(Page page, PageCheckResult pageResult)
{
var problemWords = GetProblematicWords(page).ToList();
pageResult.AddIssues(problemWords.Select(pw => $"Contains term \"{pw}\""));
}
public void AddPageCheckResult(PageCheckResult result)
{
_pageCheckResults.Add(result);
}