public override string ToString()
{
StringBuilder rslt = new StringBuilder();
rslt.Append('[');
if (invert)
{
rslt.Append('^');
}
if (!isCanonical)
{
Canonicalize();
}
foreach (CharRange range in ranges)
{
if (range.minChr == range.maxChr)
{
rslt.Append(CharacterUtilities.MapForCharSet(range.minChr));
}
else
{
rslt.Append(CharacterUtilities.MapForCharSet(range.minChr));
rslt.Append('-');
rslt.Append(CharacterUtilities.MapForCharSet(range.maxChr));
}
}
rslt.Append(']');
return(rslt.ToString());
}
public override string ToString()
{
if (minChr == maxChr)
{
return(String.Format(CultureInfo.InvariantCulture, "singleton char {0}", CharacterUtilities.Map(minChr)));
}
else
{
return(String.Format(CultureInfo.InvariantCulture, "char range {0} .. {1}, {2} chars",
CharacterUtilities.Map(minChr), CharacterUtilities.Map(maxChr), maxChr - minChr + 1));
}
}
internal override void Op(RegExTree tree)
{
Leaf leaf = tree as Leaf;
if (leaf != null)
{
switch (leaf.op)
{
case RegOp.primitive:
DoSingleton(leaf.chVal);
break;
case RegOp.litStr:
for (int index = 0; ;)
{
// Use CodePoint in case string has surrogate pairs.
int code = CharacterUtilities.CodePoint(leaf.str, ref index);
if (code == -1)
{
break;
}
else
{
DoSingleton(code);
}
}
break;
case RegOp.charClass:
DoLiteral(leaf.rangeLit);
break;
case RegOp.eof: // no action required
break;
default:
throw new GplexInternalException("Unknown RegOp");
}
}
else if (tree.op == RegOp.rightAnchor)
{
DoLiteral(RangeLiteral.RightAnchors);
}
}
/// <summary>
/// This static method expands characters in a literal
/// string, returning a modified string with character
/// escapes replaced by the character which they denote.
/// This includes characters outside the BMP which
/// return a pair of surrogate characters.
/// </summary>
/// <param name="str">the input string</param>
/// <returns>interpreted version of the string</returns>
public static string InterpretCharacterEscapes(string source)
{
int sLen = source.Length;
if (sLen == 0)
{
return(source);
}
char[] arr = new char[sLen];
int sNxt = 0;
int aNxt = 0;
char chr = source[sNxt++];
for (; ; chr = source[sNxt++])
{
if (chr != '\\')
{
arr[aNxt++] = chr;
}
else
{
int codePoint = EscapedChar(source, ref sNxt);
if (codePoint > 0xFFFF)
{
arr[aNxt++] = CharacterUtilities.HiSurrogate(codePoint);
arr[aNxt++] = CharacterUtilities.LoSurrogate(codePoint);
}
else
{
arr[aNxt++] = (char)codePoint;
}
}
if (sNxt == sLen)
{
return(new String(arr, 0, aNxt));
}
}
}