public Slug Scrape(Uri programUri)
{
if (programUri == null)
{
throw new ArgumentNullException("programUri");
}
if (!programUri.IsAbsoluteUri)
{
throw new ArgumentException("Only absolute URIs are supported.", "programUri");
}
var request = _webRequestWrapper.CreateHttp(programUri);
var response = request.GetResponse();
string html;
using (var streamReader = new StreamReader(response.GetResponseStream()))
{
html = streamReader.ReadToEnd();
}
var regex = new Regex("programSerieSlug:\\s*\"(?<slug>.*)\"");
var match = regex.Match(html);
if (!match.Success)
{
throw new ScraperException("Unable to find programSerieSlug.");
}
return(match.Groups["slug"].Value);
}
public IEnumerable <Uri> Scrape(Slug slug)
{
if (slug == null)
{
throw new ArgumentNullException("slug");
}
var episodesUri = new Uri(string.Format("http://www.dr.dk/TV/play/AllEpisodes?slug={0}&episodesperpage=100&pagenumber=1", slug));
var request = _webRequestWrapper.CreateHttp(episodesUri);
var response = request.GetResponse();
var htmlDocument = new HtmlDocument();
htmlDocument.Load(response.GetResponseStream());
var liElements = htmlDocument.DocumentNode.SelectNodes("//li");
if (liElements == null)
{
throw new ScraperException("No li elements found.");
}
var uris = new List <Uri>();
foreach (var liElement in liElements)
{
var aElement = (from ae in liElement.ChildNodes
where ae.Name == "a"
select ae).FirstOrDefault();
if (aElement == null || aElement.Attributes["href"] == null)
{
continue;
}
uris.Add(new Uri(string.Format("http://www.dr.dk{0}", aElement.Attributes["href"].Value)));
}
if (!uris.Any())
{
throw new ScraperException("No program URIs found.");
}
return(uris);
}
public Resource Scrape(Uri resourceUri)
{
if (resourceUri == null)
{
throw new ArgumentNullException("resourceUri");
}
if (!resourceUri.IsAbsoluteUri)
{
throw new ArgumentException("Only absolute URIs are supported.", "resourceUri");
}
var request = _webRequestWrapper.CreateHttp(resourceUri);
var response = request.GetResponse();
string json;
using (var streamReader = new StreamReader(response.GetResponseStream()))
{
json = streamReader.ReadToEnd();
}
return(_jsonConvertWrapper.DeserializeObject <Resource>(json));
}
public Uri Scrape(Uri programUri)
{
if (programUri == null)
{
throw new ArgumentNullException("programUri");
}
if (!programUri.IsAbsoluteUri)
{
throw new ArgumentException("Only absolute URIs are supported.", "programUri");
}
try
{
var request = _webRequestWrapper.CreateHttp(programUri);
var response = request.GetResponse();
string html;
using (var streamReader = new StreamReader(response.GetResponseStream()))
{
html = streamReader.ReadToEnd();
}
var regex = new Regex("resource:\\s*\"(?<uri>.*)\"");
var match = regex.Match(html);
if (!match.Success)
{
throw new ScraperException("Unable to find resource.");
}
return(new Uri(match.Groups["uri"].Value));
}
catch (WebException exception)
{
throw new ScraperException("Unable to find resource.", exception);
}
}
