public void TestDecodeNamedEntities ()
{
var path = Path.Combine ("..", "..", "TestData", "html", "HtmlEntities.json");
var decoder = new HtmlEntityDecoder ();
using (var json = new JsonTextReader (new StreamReader (path))) {
while (json.Read ()) {
string name, value;
if (json.TokenType == JsonToken.StartObject)
continue;
if (json.TokenType != JsonToken.PropertyName)
break;
name = (string) json.Value;
if (!json.Read () || json.TokenType != JsonToken.StartObject)
break;
// read to the "codepoints" property
if (!json.Read () || json.TokenType != JsonToken.PropertyName)
break;
// skip the array of integers...
if (!json.Read () || json.TokenType != JsonToken.StartArray)
break;
while (json.Read ()) {
if (json.TokenType == JsonToken.EndArray)
break;
}
// the property should be "characters" - this is what we want
if (!json.Read () || json.TokenType != JsonToken.PropertyName)
break;
value = json.ReadAsString ();
if (!json.Read () || json.TokenType != JsonToken.EndObject)
break;
for (int i = 0; i < name.Length && name[i] != ';'; i++)
Assert.IsTrue (decoder.Push (name[i]), "Failed to push char #{0} of \"{1}\".", i, name);
Assert.AreEqual (value, decoder.GetValue (), "Decoded entity did not match for \"{0}\".", name);
decoder.Reset ();
}
}
}
/// <summary>
/// Decode HTML character data.
/// </summary>
/// <remarks>
/// Decodes HTML character data.
/// </remarks>
/// <param name="output">The <see cref="System.IO.TextWriter"/> to output the result.</param>
/// <param name="data">The character data to decode.</param>
/// <param name="startIndex">The starting index of the character data.</param>
/// <param name="count">The number of characters in the data.</param>
/// <exception cref="System.ArgumentNullException">
/// <para><paramref name="output"/> is <c>null</c>.</para>
/// <para>-or-</para>
/// <para><paramref name="data"/> is <c>null</c>.</para>
/// </exception>
/// <exception cref="System.ArgumentOutOfRangeException">
/// <para><paramref name="startIndex"/> and <paramref name="count"/> do not specify
/// a valid range in the data.</para>
/// </exception>
public static void HtmlDecode(TextWriter output, string data, int startIndex, int count)
{
if (output == null)
{
throw new ArgumentNullException("output");
}
if (data == null)
{
throw new ArgumentNullException("data");
}
if (startIndex < 0 || startIndex >= data.Length)
{
throw new ArgumentOutOfRangeException("startIndex");
}
if (count < 0 || count > (data.Length - startIndex))
{
throw new ArgumentOutOfRangeException("count");
}
var entity = new HtmlEntityDecoder();
int endIndex = startIndex + count;
int index = startIndex;
while (index < endIndex)
{
if (data[index] == '&')
{
while (index < endIndex && entity.Push(data[index]))
{
index++;
}
output.Write(entity.GetValue());
entity.Reset();
if (index < endIndex && data[index] == ';')
{
index++;
}
}
else
{
output.Write(data[index++]);
}
}
}
/// <summary>
/// Decode HTML character data.
/// </summary>
/// <remarks>
/// Decodes HTML character data.
/// </remarks>
/// <param name="output">The <see cref="System.IO.TextWriter"/> to output the result.</param>
/// <param name="data">The character data to decode.</param>
/// <param name="startIndex">The starting index of the character data.</param>
/// <param name="count">The number of characters in the data.</param>
/// <exception cref="System.ArgumentNullException">
/// <para><paramref name="output"/> is <c>null</c>.</para>
/// <para>-or-</para>
/// <para><paramref name="data"/> is <c>null</c>.</para>
/// </exception>
/// <exception cref="System.ArgumentOutOfRangeException">
/// <para><paramref name="startIndex"/> and <paramref name="count"/> do not specify
/// a valid range in the data.</para>
/// </exception>
public static void HtmlDecode(TextWriter output, string data, int startIndex, int count)
{
if (output == null)
throw new ArgumentNullException("output");
if (data == null)
throw new ArgumentNullException("data");
if (startIndex < 0 || startIndex >= data.Length)
throw new ArgumentOutOfRangeException("startIndex");
if (count < 0 || count > (data.Length - startIndex))
throw new ArgumentOutOfRangeException("count");
var entity = new HtmlEntityDecoder();
int endIndex = startIndex + count;
int index = startIndex;
while (index < endIndex)
{
if (data[index] == '&')
{
while (index < endIndex && entity.Push(data[index]))
index++;
output.Write(entity.GetValue());
entity.Reset();
if (index < endIndex && data[index] == ';')
index++;
}
else
{
output.Write(data[index++]);
}
}
}
static void TestDecodeNumericEntity (string text, string expected)
{
var decoder = new HtmlEntityDecoder ();
for (int i = 0; i < text.Length; i++)
Assert.IsTrue (decoder.Push (text[i]), "Failed to push char #{0} of \"{1}\".", i, text);
Assert.AreEqual (expected, decoder.GetValue (), "Decoded entity did not match for \"{0}\".", text);
}