public void CleanContentTrusted(AtomContent content)
{
Logger.Info("Cleaning content to be valid xhtml.");
string text = content.Text;
if (content.Type == "html")
{
text = "<div xmlns=\"" + Atom.XhtmlNs.NamespaceName + "\">" + text + "</div>";
}
Tidy tidy = new Tidy();
/* Set the options you want */
tidy.Options.DocType = DocType.Strict;
//tidy.Options.DropFontTags = true;
tidy.Options.LogicalEmphasis = true;
tidy.Options.Xhtml = true;
tidy.Options.XmlOut = true;
tidy.Options.MakeClean = true;
tidy.Options.TidyMark = false;
tidy.Options.QuoteNbsp = false;
tidy.Options.NumEntities = true;
tidy.Options.CharEncoding = CharEncoding.UTF8;
tidy.Options.FixBackslash = true;
tidy.Options.FixComments = true;
TidyMessageCollection tmc = new TidyMessageCollection();
using (MemoryStream input = new MemoryStream())
using (MemoryStream output = new MemoryStream())
{
byte[] bytes = Encoding.UTF8.GetBytes(text);
input.Write(bytes, 0, bytes.Length);
input.Position = 0;
tidy.Parse(input, output, tmc);
text = Encoding.UTF8.GetString(output.ToArray());
if (string.IsNullOrEmpty(text)) throw new FailedToCleanContentException(
string.Format("{0} HTML Tidy Error(s)" + Environment.NewLine, tmc.Errors)
+ string.Join(Environment.NewLine,
tmc.Cast<TidyMessage>()
.Where(m => m.Level == MessageLevel.Error)
.Select(m => m.ToString()).ToArray()));
}
//remove html/body that gets added by tidy
//int start = text.IndexOf("<div xmlns");
//text = text.Substring(start, text.IndexOf("</body>") - start);
XElement div = XElement.Parse(text).Element(Atom.XhtmlNs + "body").Element(Atom.XhtmlNs + "div");
//remove decendent xmlns that gets added by tidy
div.Descendants().Where(d => d.Attribute("xmlns") != null &&
d.Attribute("xmlns").Value == Atom.XhtmlNs.NamespaceName)
.Select(d => d.Attribute("xmlns")).Remove();
//text = text.Replace(" xmlns=\"http://www.w3.org/1999/xhtml\"", string.Empty);
//text = "<div xmlns=\"http://www.w3.org/1999/xhtml\">" + text.Substring("<div>".Length);
//set as xhtml
content.Type = "xhtml";
content.Text = div.ToString(SaveOptions.None);
}
public override List<SearchResult> Search(ChapterInfo chapterInfo)
{
string result = string.Empty;
using (WebClient wc = new WebClient())
{
//NameValueCollection vars = new NameValueCollection();
//vars.Add("txtTitle", chapterInfo.Title);
//vars.Add("btnSearch", "Search");
//wc.UploadValues(uri, "POST", vars);
wc.Headers["Content-Type"] = "application/x-www-form-urlencoded";
Uri uri = new Uri("http://www.e-home.no/metaservices/search.aspx");
result = wc.UploadString(uri, "POST",
//__VIEWSTATE=%2FwEPDwUKLTM3MTkwMDA5NQ9kFgICAQ9kFgICDQ88KwALAGRkg%2BhH%2F3tiaQDjnQncD1sYDdeni%2BA%3D&txtTitle=batman&btnSearch=Search&__EVENTVALIDATION=%2FwEWAwLXiqPdDAL55JyzBAKln%2FPuCgMJnDvHIVAx2tPEYdjNUbwqrR67
string.Format("__VIEWSTATE=%2FwEPDwUKLTM3MTkwMDA5NQ9kFgICAQ9kFgICDQ88KwALAGRkg%2BhH%2F3tiaQDjnQncD1sYDdeni%2BA%3D&txtTitle={0}&btnSearch=Search&__EVENTVALIDATION=%2FwEWAwLXiqPdDAL55JyzBAKln%2FPuCgMJnDvHIVAx2tPEYdjNUbwqrR67", HttpUtility.UrlEncode(chapterInfo.Title)));
}
//__VIEWSTATE=%2FwEPDwUKLTM3MTkwMDA5NQ9kFgICAQ9kFgICDQ88KwALAGRkg%2BhH%2F3tiaQDjnQncD1sYDdeni%2BA%3D&txtTitle=batman&btnSearch=Search&__EVENTVALIDATION=%2FwEWAwLXiqPdDAL55JyzBAKln%2FPuCgMJnDvHIVAx2tPEYdjNUbwqrR67
Tidy tidy = new Tidy();
/* Set the options you want */
tidy.Options.DocType = DocType.Strict;
//tidy.Options.DropFontTags = true;
tidy.Options.LogicalEmphasis = true;
tidy.Options.Xhtml = true;
tidy.Options.XmlOut = true;
tidy.Options.MakeClean = true;
tidy.Options.TidyMark = false;
tidy.Options.QuoteNbsp = false;
tidy.Options.NumEntities = true;
tidy.Options.CharEncoding = CharEncoding.UTF8;
tidy.Options.FixBackslash = true;
tidy.Options.FixComments = true;
TidyMessageCollection tmc = new TidyMessageCollection();
using (MemoryStream input = new MemoryStream())
using (MemoryStream output = new MemoryStream())
{
byte[] bytes = Encoding.UTF8.GetBytes(result);
input.Write(bytes, 0, bytes.Length);
input.Position = 0;
tidy.Parse(input, output, tmc);
result = Encoding.UTF8.GetString(output.ToArray());
if (tmc.Errors > 0) throw new Exception(
string.Format("{0} HTML Tidy Error(s)" + Environment.NewLine, tmc.Errors)
+ string.Join(Environment.NewLine,
tmc.Cast<TidyMessage>()
.Where(m => m.Level == MessageLevel.Error)
.Select(m => m.ToString()).ToArray()));
XNamespace ns = "http://www.w3.org/1999/xhtml";
//parse titles
XDocument searchXhtml = XDocument.Parse(result);
Debug.Write(searchXhtml.Descendants(ns + "tr")
.Where(tr => (tr.Attribute("id") != null && tr.Attribute("id").Value.Length == 17)).Count());
var titles = searchXhtml.Descendants(ns + "tr")
.Where(tr => (tr.Attribute("id") != null && tr.Attribute("id").Value.Length == 17))
.Select(tr => new SearchResult()
{
Id = (string)tr.Attribute("id"),
Name = (string)tr.Elements(ns + "td").First()
});
OnSearchComplete();
return titles.ToList();
}
}