void ProcessMetaTag(HtmlReader reader, WorkspaceResult result)
{
var propertyName = reader.GetAttribute("property").ToLower();
if (!string.IsNullOrWhiteSpace(propertyName))
{
propertyName = SanitiseProperty(propertyName);
var content = SanitiseAttribute(reader.GetAttribute("content"));
if (!result.PropertyCache.ContainsKey(propertyName))
{
result.PropertyCache.Add(propertyName, reader.GetAttribute("content"));
}
switch (propertyName)
{
case "og_title": // og:title
if (!result.PropertyCache.ContainsKey("title"))
{
result.PropertyCache.Add("title", content);
}
result.PropertyCache["title"] = content;
result.Title = content;
break;
case "og_description": // og:description
result.Description = content;
if (!result.PropertyCache.ContainsKey("description"))
{
result.PropertyCache.Add("description", content);
}
result.PropertyCache["description"] = content;
break;
default:
break;
}
}
else
{
// other random historical meta tags here
var metaname = reader.GetAttribute("name").ToLower();
switch (metaname)
{
case "description":
var value = SanitiseAttribute(reader.GetAttribute("content"));
if (!result.PropertyCache.ContainsKey("description"))
{
result.PropertyCache.Add("description", value);
}
break;
}
}
}
public void AttributeValueNamedCharacterReference()
{
reader = HtmlReaderFactory.FromString("<a title=\"<\">", parseErrors);
Assert.IsTrue(reader.Read());
Assert.AreEqual("<", reader.GetAttribute("title"));
Assert.IsFalse(reader.Read());
Assert.AreEqual(0, parseErrors.Count);
}
public void TagMissingAttribute()
{
reader = HtmlReaderFactory.FromString("<a>", parseErrors);
Assert.IsTrue(reader.Read());
Assert.AreEqual(HtmlTokenKind.Tag, reader.TokenKind);
Assert.AreEqual("a", reader.Name);
Assert.IsNull(reader.GetAttribute("href"));
Assert.IsFalse(reader.SelfClosingElement);
Assert.IsFalse(reader.Read());
Assert.AreEqual(0, parseErrors.Count);
}
public void TagUnquotedAttributeValue()
{
reader = HtmlReaderFactory.FromString("<a href=javascript:;>", parseErrors);
Assert.IsTrue(reader.Read());
Assert.AreEqual(HtmlTokenKind.Tag, reader.TokenKind);
Assert.AreEqual("a", reader.Name);
Assert.AreEqual("javascript:;", reader.GetAttribute("href"));
Assert.IsFalse(reader.SelfClosingElement);
Assert.IsFalse(reader.Read());
Assert.AreEqual(0, parseErrors.Count);
}
public void GetAttributeReturnsFirstAttributeValue()
{
reader = HtmlReaderFactory.FromString("<img src=\"a\" src=\"b\" />", parseErrors);
Assert.IsTrue(reader.Read());
Assert.AreEqual(HtmlTokenKind.Tag, reader.TokenKind);
Assert.AreEqual("img", reader.Name);
Assert.AreEqual(2, reader.AttributeCount);
Assert.AreEqual("a", reader.GetAttribute("src"));
Assert.IsFalse(reader.Read());
Assert.AreEqual(0, parseErrors.Count);
}
public List <string> ExtractLinks()
{
stream.Seek(0, SeekOrigin.Begin);
var htmlReader = new HtmlReader(new StreamReader(stream));
var links = new List <string>();
while (htmlReader.Read())
{
if (htmlReader.TokenKind == HtmlTokenKind.Tag && htmlReader.Name == "a")
{
var hrefAttributeValue = htmlReader.GetAttribute("href");
if (hrefAttributeValue != null)
{
links.Add(hrefAttributeValue);
}
}
}
return(links);
}
public override async Task ProcessResponseStream(Stream dataStream)
{
if (dataStream == null)
{
return;
}
var ms = new MemoryStream();
dataStream.CopyTo(ms);
ms.Seek(0, SeekOrigin.Begin);
if (ContentType.IsUnknownOrNull(ContentType))
{
ContentType = ContentType.Guess(ms);
}
ms.Seek(0, SeekOrigin.Begin);
if (ContentType.IsUnknownOrNull(ContentType) || ContentType.MediaType == "text")
{
using (var p = new HtmlReader(ms))
{
while (!p.EOF)
{
var node = p.NextNode();
if (node.Type == HtmlTokenType.StartTag)
{
if (node.Value == "script")
{
string src = p.GetAttribute("src");
if (!string.IsNullOrWhiteSpace(src))
{
Console.WriteLine(src);
}
}
}
}
}
}
}
public Uri NextUri()
{
Uri uri = null;
while (!reader.EOF && uri == null)
{
var node = reader.NextNode();
if (node.Value == "a")
{
var href = reader.GetAttribute("href");
if (!string.IsNullOrWhiteSpace(href))
{
var links = Utility.GetLinks(CurrentSourceUri, href);
if (links.Any())
{
uri = links.First();
}
}
}
else if (node.Value == "script")
{
var href = reader.GetAttribute("src");
if (!string.IsNullOrWhiteSpace(href))
{
var links = Utility.GetLinks(CurrentSourceUri, href);
if (links.Any())
{
uri = links.First();
}
}
}
else if (node.Value == "link")
{
var href = reader.GetAttribute("href");
if (!string.IsNullOrWhiteSpace(href))
{
var links = Utility.GetLinks(CurrentSourceUri, href);
if (links.Any())
{
uri = links.First();
}
}
}
else if (node.Value == "base")
{
var href = reader.GetAttribute("href");
if (!string.IsNullOrWhiteSpace(href))
{
// sometimes the links are bogus?!
var l = Utility.GetLinks(null, href);
if (l.Any())
{
CurrentSourceUri = l.First();
}
}
}
}
return(uri);
}