/// <summary>
/// Returns a TextReader that detects the underlying stream's endoding. Allows clients to stream the
/// retured content using a TextReader. This method is similar in purpose to GetStreamAsync, however, GetStreamAsync
/// doesn't detect the Stream's encoding as GetStringAsync does.
/// </summary>
/// <param name="httpClient"></param>
public static async Task <HtmlTextReader> GetHtmlTextReaderAsync(string url, ClientOptions options)
{
HtmlTextReader reader;
ClientOptions optionsToUse = options == null ? HtmlClient.Options : options;
Uri uri = new Uri(url);
// See if the url pointed to a file. If so, return a reader with a file stream
// under the hood.
if (uri.IsFile)
{
FileStream fs = File.OpenRead(uri.AbsolutePath);
HtmlStream stream = new HtmlStream(fs);
reader = new HtmlTextReader(stream, options.DefaultEncoding, EncodingConfidence.Tentative);
reader.OriginatingUrl = url;
return(reader);
}
// Set a user agent if one was specified
if (!string.IsNullOrEmpty(optionsToUse.UserAgent))
{
HttpClient.DefaultRequestHeaders.Remove("User-Agent");
HttpClient.DefaultRequestHeaders.Add("User-Agent", optionsToUse.UserAgent);
}
// Get the Http response (only read the headers at this point) and ensure succes
HttpResponseMessage responseMessage = await HttpClient.GetAsync(uri, HttpCompletionOption.ResponseHeadersRead).ConfigureAwait(false);
responseMessage.EnsureSuccessStatusCode();
// If there is no content to return, return an empty HtmlTextReader
HttpContent content = responseMessage.Content;
if (content == null)
{
reader = new HtmlTextReader(String.Empty);
}
else
{
reader = await content.GetHtmlTextReaderAsync(optionsToUse.DefaultEncoding, optionsToUse.DetectEncoding);
}
// Store some metadata on the reader. Could be used by a parser.
reader.OriginatingUrl = url;
foreach (var header in content.Headers)
{
reader.OriginatingHttpHeaders.Add(new KeyValuePair <string, string>(header.Key, string.Join(";", header.Value)));
}
return(reader);
}