/// <summary>
/// Marks characters from the specified languages as safe.
/// </summary>
/// <param name="lowerCodeCharts">The combination of lower code charts to use.</param>
/// <param name="lowerMidCodeCharts">The combination of lower mid code charts to use.</param>
/// <param name="midCodeCharts">The combination of mid code charts to use.</param>
/// <param name="upperMidCodeCharts">The combination of upper mid code charts to use.</param>
/// <param name="upperCodeCharts">The combination of upper code charts to use.</param>
/// <remarks>The safe list affects all HTML and XML encoding functions.</remarks>
public static void MarkAsSafe(
LowerCodeCharts lowerCodeCharts,
LowerMidCodeCharts lowerMidCodeCharts,
MidCodeCharts midCodeCharts,
UpperMidCodeCharts upperMidCodeCharts,
UpperCodeCharts upperCodeCharts)
{
if (lowerCodeCharts == currentLowerCodeChartSettings &&
lowerMidCodeCharts == currentLowerMidCodeChartSettings &&
midCodeCharts == currentMidCodeChartSettings &&
upperMidCodeCharts == currentUpperMidCodeChartSettings &&
upperCodeCharts == currentUpperCodeChartSettings)
{
return;
}
AcquireWriteLock();
try
{
// Reset back to everything hashed.
characterValues = SafeList.Generate(65536, SafeList.HashThenValueGenerator);
SafeList.PunchUnicodeThrough(
ref characterValues,
lowerCodeCharts,
lowerMidCodeCharts,
midCodeCharts,
upperMidCodeCharts,
upperCodeCharts);
ApplyHtmlSpecificValues();
currentLowerCodeChartSettings = lowerCodeCharts;
currentLowerMidCodeChartSettings = lowerMidCodeCharts;
currentMidCodeChartSettings = midCodeCharts;
currentUpperMidCodeChartSettings = upperMidCodeCharts;
currentUpperCodeChartSettings = upperCodeCharts;
}
finally
{
ReleaseWriteLock();
}
}
/// <summary>
/// Initializes the HTML safe list.
/// </summary>
private static void InitialiseSafeList()
{
AcquireWriteLock();
try
{
if (characterValues == null)
{
// We use decimal encoding to support some older Japanese mobile browsers which don't support hex encoding.
characterValues = SafeList.Generate(0xFFFF, SafeList.HashThenValueGenerator);
SafeList.PunchUnicodeThrough(
ref characterValues,
currentLowerCodeChartSettings,
currentLowerMidCodeChartSettings,
currentMidCodeChartSettings,
currentUpperMidCodeChartSettings,
currentUpperCodeChartSettings);
ApplyHtmlSpecificValues();
}
}
finally
{
ReleaseWriteLock();
}
}
/// <summary>
/// Encodes input strings for use in HTML.
/// </summary>
/// <param name="input">String to be encoded</param>
/// <param name="useNamedEntities">Value indicating if the HTML 4.0 named entities should be used.</param>
/// <param name="encoderTweak">A <see cref="MethodSpecificEncoder"/> function, if needed.</param>
/// <returns>
/// Encoded string for use in HTML.
/// </returns>
private static string HtmlEncode(string input, bool useNamedEntities, MethodSpecificEncoder encoderTweak)
{
if (string.IsNullOrEmpty(input))
{
return(input);
}
if (characterValues == null)
{
InitialiseSafeList();
}
char[][] namedEntities = null;
if (useNamedEntities)
{
namedEntities = namedEntitiesLazy.Value;
}
// Setup a new StringBuilder for output.
// Worse case scenario - the longest entity name, thetasym is 10 characters, including the & and ;.
StringBuilder builder = EncoderUtil.GetOutputStringBuilder(input.Length, 10);
AcquireReadLock();
try
{
Utf16StringReader stringReader = new Utf16StringReader(input);
while (true)
{
int currentCodePoint = stringReader.ReadNextScalarValue();
if (currentCodePoint < 0)
{
break; // EOF
}
if (currentCodePoint > char.MaxValue)
{
// We don't have a pre-generated mapping of characters beyond the Basic Multilingual
// Plane (BMP), so we need to generate these encodings on-the-fly. We should encode
// the code point rather than the surrogate code units that make up this code point.
// See: http://www.w3.org/International/questions/qa-escapes#bytheway
char[] encodedCharacter = SafeList.HashThenValueGenerator(currentCodePoint);
builder.Append('&');
builder.Append(encodedCharacter);
builder.Append(';');
}
else
{
// If we reached this point, the code point is within the BMP.
char currentCharacter = (char)currentCodePoint;
char[] tweekedValue;
if (encoderTweak != null && encoderTweak(currentCharacter, out tweekedValue))
{
builder.Append(tweekedValue);
}
else if (useNamedEntities && namedEntities[currentCodePoint] != null)
{
char[] encodedCharacter = namedEntities[currentCodePoint];
builder.Append('&');
builder.Append(encodedCharacter);
builder.Append(';');
}
else if (characterValues[currentCodePoint] != null)
{
// character needs to be encoded
char[] encodedCharacter = characterValues[currentCodePoint];
builder.Append('&');
builder.Append(encodedCharacter);
builder.Append(';');
}
else
{
// character does not need encoding
builder.Append(currentCharacter);
}
}
}
}
finally
{
ReleaseReadLock();
}
return(builder.ToString());
}
