internal HtmlPage(WebBytes wb, WebString prelimContent) : base(wb.Location)
{
HeadersCharset = wb.CharacterSet;
EncHeaders = prelimContent.UsedEncoding;
DetermineEncoding(wb.Data, prelimContent.Document);
}
FetcherResult Load(Uri uri, bool useCookies, int redirects, string previousPage)
{
var req = SetupRequest(uri, useCookies);
var wb = WebBytes.Create(req, MaxSizeHtml, MaxSizeNonHtml);
if (wb.ContentIsHtml)
{
var page = WebString.Create(wb, EncHelp.Windows1252);
Uri refreshUrl;
if (redirects < MaxRefreshes &&
TryGetMetaRefresh(page, out refreshUrl) &&
VerifyRefresh(page.Location, refreshUrl))
{
// If during the redirects we get a different page/HTML string, refrain from following more
// redirects. It probably means we've arrived, but only more real world testing will tell us
// if that's true.
// Otherwise keep trying to follow the redirects.
if (redirects == 0 || page.Document == previousPage)
{
return(Load(refreshUrl, useCookies, (redirects + 1), page.Document));
}
}
return(new FetcherResult(wb, page));
}
// Silently handle cookie exceptions, Mono/.NET can be very strict with which cookies it accepts.
if (!wb.Success && wb.Exception.InnerException is CookieException)
{
return(Load(uri, false, redirects, previousPage));
}
// If either not HTML or there was an error in getting the resource, return as-is.
return(new FetcherResult(wb));
}
public static HtmlPage Create(System.Net.WebRequest request)
{
if (request == null)
{
throw new ArgumentNullException("request");
}
return(Create(WebBytes.ReadOnlyHtml(request)));
}
public static WebString Create(WebRequest request, Encoding fallbackEnc)
{
if (request == null)
{
throw new ArgumentNullException("request");
}
return(Create(WebBytes.Create(request), fallbackEnc));
}
public FetcherResult(WebBytes bytes)
{
if (bytes == null)
{
throw new ArgumentNullException(nameof(bytes));
}
Bytes = bytes;
Page = HtmlPage.Create(bytes);
}
public static HtmlPage Create(WebBytes wb)
{
if (wb == null)
{
throw new ArgumentNullException("wb");
}
if (wb.ContentIsHtml)
{
return(new HtmlPage(wb.Location, wb.Data, wb.CharacterSet));
}
else if (wb.Success)
{
var ex = new NotHtmlException("Content isn't (X)HTML. Content-Type: " + wb.ContentType);
return(new HtmlPage(wb.Location, ex));
}
else
{
return(new HtmlPage(wb));
}
}
public static WebString Create(WebBytes wb, Encoding fallbackEnc)
{
if (wb == null)
{
throw new ArgumentNullException("wb");
}
else if (fallbackEnc == null)
{
throw new ArgumentNullException("fallbackEnc");
}
if (wb.Success)
{
var enc = EncHelp.GetEncoding(wb.CharacterSet) ?? fallbackEnc;
return(new WebString(wb.Location, wb.Data, enc));
}
else
{
return(new WebString(wb));
}
}
internal FetcherResult(WebBytes bytes, WebString page)
{
Bytes = bytes;
Page = new HtmlPage(bytes, page);
}
