// CharClass : "[" ["^"] {code | code "-" code | FilteredClass}+ "]" ;
internal Leaf CharClass() {
// Assert chr == '['
// Need to build a new string taking into account char escapes
Leaf leaf = new Leaf( RegOp.charClass );
bool invert = false;
scan(); // read past '['
if (!esc && chr == '^') {
invert = true;
scan(); // read past '^'
}
leaf.rangeLit = new RangeLiteral( invert );
// Special case of '-' at start, taken as ordinary class member.
// This is correct for LEX specification, but is undocumented
// behavior for FLEX. GPLEX gives a friendly warning, just in
// case this is actually a typographical error.
if (!esc && chr == '-') {
Warn( 113, index - 1, 1, "-" );
leaf.rangeLit.list.Add( new CharRange( '-' ) );
scan(); // read past -'
}
while (chr != NUL && (esc || chr != ']')) {
int lhCodePoint;
int startIx = index - 1; // save starting index for error reporting
lhCodePoint = (esc ? EscapedChar() : CodePoint());
if (!esc && lhCodePoint == (int)'-')
Error( 82, startIx, index - startIx, null );
//
// There are three possible elements here:
// * a singleton character
// * a character range
// * a filtered class like [:IsLetter:]
//
if (chr == '[' && !esc && peek() == ':') // character category
{
Leaf rslt = FilteredClass();
leaf.Merge( rslt );
}
else {
scan();
if (!esc && chr == '-') // character range
{
scan();
if (!esc && chr == ']') {
// Special case of '-' at end, taken as ordinary class member.
// This is correct for LEX specification, but is undocumented
// behavior for FLEX. GPLEX gives a friendly warning, just in
// case this is actually a typographical error.
leaf.rangeLit.list.Add( new CharRange( lhCodePoint ) );
leaf.rangeLit.list.Add( new CharRange( '-' ) );
//Error(81, idx, index - idx - 1);
Warn( 114, startIx, index - startIx - 1, String.Format(
CultureInfo.InvariantCulture,
"'{0}','{1}'",
CharacterUtilities.Map( lhCodePoint ),
'-' ) );
}
else {
int rhCodePoint = (esc ? EscapedChar() : CodePoint());
if (rhCodePoint < lhCodePoint)
Error( 54, startIx, index - startIx, null );
scan();
leaf.rangeLit.list.Add( new CharRange( lhCodePoint, rhCodePoint ) );
}
}
else // character singleton
{
leaf.rangeLit.list.Add( new CharRange( lhCodePoint ) );
}
}
}
checkAndScan( ']' );
leaf.rangeLit.list.Canonicalize();
return leaf;
}
