public static string HtmlDecode(string value)
{
if (value == null)
{
throw new ArgumentNullException("value");
}
var sb = new StringBuilder();
for (int i = 0; i < value.Length; i++)
{
if (value[i] == '&' && value.Length > i + 2)
{
// Scan for the ;.
int maxSearch = Math.Min(value.Length, i + _longestHtmlEntity + 2);
int endPosition = -1;
for (int j = i + 1; j < maxSearch; j++)
{
if (value[j] == ';')
{
endPosition = j;
break;
}
}
// If we did not find an end separator, just skip over this
// entity and treat is at text.
if (endPosition == -1)
{
sb.Append(value[i]);
continue;
}
// Are we in a numeric separator?
if (value[i + 1] == '#')
{
int offset = 2;
bool isHexNumeric = false;
if (value[i + 2] == 'x' || value[i + 2] == 'X')
{
isHexNumeric = true;
offset++;
}
// All parts of the numeric separator must be digits.
bool isNumeric = true;
for (int j = i + offset; j < endPosition; j++)
{
if (!(
Char.IsDigit(value[j]) ||
(isHexNumeric && HttpUtil.IsHex(value[j]))
))
{
isNumeric = false;
break;
}
}
// If not all numeric, just skip over this
// entity and treat is at text.
if (!isNumeric)
{
sb.Append(value[i]);
continue;
}
// Convert the numeric entity to unicode.
string numericEntity = value.Substring(i + offset, endPosition - (i + offset));
sb.Append((char)int.Parse(numericEntity, isHexNumeric ? NumberStyles.HexNumber : NumberStyles.Integer));
i = endPosition;
}
else
{
string entity = value.Substring(i + 1, endPosition - (i + 1));
int codePoint;
if (_htmlEntitiesByEntity.TryGetValue(entity, out codePoint))
{
sb.Append((char)codePoint);
i = endPosition;
}
else
{
// If we don't know the entity, just skip over this
// entity and treat is at text.
sb.Append(value[i]);
}
}
}
else
{
sb.Append(value[i]);
}
}
return(sb.ToString());
}
