private static string DetectCharacterSetFromContent(byte[] bytes, string httpContentType)
{
string text = UnknownEncodingToString(bytes, 4096);
HTMLParser htmlParser = new HTMLParser(text);
string mimeType = GetMIMETypeFromContentType(httpContentType) ?? String.Empty;
string charSet;
if (mimeType.Equals("application/xhtml+xml", StringComparison.OrdinalIgnoreCase) ||
mimeType.Equals("application/xml", StringComparison.OrdinalIgnoreCase) ||
mimeType.Equals("text/xml", StringComparison.OrdinalIgnoreCase))
{
if (text.StartsWith("<?xml", StringComparison.OrdinalIgnoreCase))
{
// XML declaration
HTMLParser xmlParser = new HTMLParser("<" + text.Substring(2));
HTMLTag xmlTag = xmlParser.Tags.Count >= 1 ? xmlParser.Tags[0] : null;
if (xmlTag != null && xmlTag.NameEquals("xml") && xmlTag.Offset == 0)
{
charSet = xmlTag.GetAttributeValue("encoding");
if (!String.IsNullOrEmpty(charSet))
{
return(charSet);
}
}
}
// Default
return("UTF-8");
}
foreach (HTMLTag tag in htmlParser.FindStartTags("meta"))
{
// charset attribute
charSet = tag.GetAttributeValue("charset");
if (!String.IsNullOrEmpty(charSet))
{
return(charSet);
}
// http-equiv and content attributes
if (tag.GetAttributeValueOrEmpty("http-equiv").Trim().Equals("Content-Type", StringComparison.OrdinalIgnoreCase))
{
charSet = GetCharSetFromContentType(tag.GetAttributeValue("content"));
if (!String.IsNullOrEmpty(charSet))
{
return(charSet);
}
}
}
return(null);
}
public static bool ClassAttributeValueHas(HTMLTag tag, string targetClassName)
{
string attributeValue = tag.GetAttributeValue("class");
return(attributeValue != null && ClassAttributeValueHas(attributeValue, targetClassName));
}
public static bool ClassAttributeValueHas(HTMLTag tag, string targetClassName)
{
string attributeValue = tag.GetAttributeValue("class");
return attributeValue != null && ClassAttributeValueHas(attributeValue, targetClassName);
}
