internal Processor(ReadOnlySpan <T> buffer, DecodeFirstRune <T> decoder)
{
_buffer = buffer;
_decoder = decoder;
_codeUnitLengthOfCurrentScalar = 0;
CurrentType = GraphemeClusterBreakType.Other;
CurrentCodeUnitOffset = 0;
}
private static int GetLengthOfFirstExtendedGraphemeCluster <T>(ReadOnlySpan <T> input, DecodeFirstRune <T> decoder)
{
// Algorithm given at https://www.unicode.org/reports/tr29/#Grapheme_Cluster_Boundary_Rules.
Processor <T> processor = new Processor <T>(input, decoder);
processor.MoveNext();
// First, consume as many Prepend scalars as we can (rule GB9b).
while (processor.CurrentType == GraphemeClusterBreakType.Prepend)
{
processor.MoveNext();
}
// Next, make sure we're not about to violate control character restrictions.
// Essentially, if we saw Prepend data, we can't have Control | CR | LF data afterward (rule GB5).
if (processor.CurrentCodeUnitOffset > 0)
{
if (processor.CurrentType == GraphemeClusterBreakType.Control ||
processor.CurrentType == GraphemeClusterBreakType.CR ||
processor.CurrentType == GraphemeClusterBreakType.LF)
{
goto Return;
}
}
// Now begin the main state machine.
GraphemeClusterBreakType previousClusterBreakType = processor.CurrentType;
processor.MoveNext();
switch (previousClusterBreakType)
{
case GraphemeClusterBreakType.CR:
if (processor.CurrentType != GraphemeClusterBreakType.LF)
{
goto Return; // rules GB3 & GB4 (only <LF> can follow <CR>)
}
processor.MoveNext();
goto case GraphemeClusterBreakType.LF;
case GraphemeClusterBreakType.Control:
case GraphemeClusterBreakType.LF:
goto Return; // rule GB4 (no data after Control | LF)
case GraphemeClusterBreakType.L:
if (processor.CurrentType == GraphemeClusterBreakType.L)
{
processor.MoveNext(); // rule GB6 (L x L)
goto case GraphemeClusterBreakType.L;
}
else if (processor.CurrentType == GraphemeClusterBreakType.V)
{
processor.MoveNext(); // rule GB6 (L x V)
goto case GraphemeClusterBreakType.V;
}
else if (processor.CurrentType == GraphemeClusterBreakType.LV)
{
processor.MoveNext(); // rule GB6 (L x LV)
goto case GraphemeClusterBreakType.LV;
}
else if (processor.CurrentType == GraphemeClusterBreakType.LVT)
{
processor.MoveNext(); // rule GB6 (L x LVT)
goto case GraphemeClusterBreakType.LVT;
}
else
{
break;
}
case GraphemeClusterBreakType.LV:
case GraphemeClusterBreakType.V:
if (processor.CurrentType == GraphemeClusterBreakType.V)
{
processor.MoveNext(); // rule GB7 (LV | V x V)
goto case GraphemeClusterBreakType.V;
}
else if (processor.CurrentType == GraphemeClusterBreakType.T)
{
processor.MoveNext(); // rule GB7 (LV | V x T)
goto case GraphemeClusterBreakType.T;
}
else
{
break;
}
case GraphemeClusterBreakType.LVT:
case GraphemeClusterBreakType.T:
if (processor.CurrentType == GraphemeClusterBreakType.T)
{
processor.MoveNext(); // rule GB8 (LVT | T x T)
goto case GraphemeClusterBreakType.T;
}
else
{
break;
}
case GraphemeClusterBreakType.Extended_Pictograph:
// Attempt processing extended pictographic (rules GB11, GB9).
// First, drain any Extend scalars that might exist
while (processor.CurrentType == GraphemeClusterBreakType.Extend)
{
processor.MoveNext();
}
// Now see if there's a ZWJ + extended pictograph again.
if (processor.CurrentType != GraphemeClusterBreakType.ZWJ)
{
break;
}
processor.MoveNext();
if (processor.CurrentType != GraphemeClusterBreakType.Extended_Pictograph)
{
break;
}
processor.MoveNext();
goto case GraphemeClusterBreakType.Extended_Pictograph;
case GraphemeClusterBreakType.Regional_Indicator:
// We've consumed a single RI scalar. Try to consume another (to make it a pair).
if (processor.CurrentType == GraphemeClusterBreakType.Regional_Indicator)
{
processor.MoveNext();
}
// Standlone RI scalars (or a single pair of RI scalars) can only be followed by trailers.
break; // nothing but trailers after the final RI
default:
break;
}
// rules GB9, GB9a
while (processor.CurrentType == GraphemeClusterBreakType.Extend ||
processor.CurrentType == GraphemeClusterBreakType.ZWJ ||
processor.CurrentType == GraphemeClusterBreakType.SpacingMark)
{
processor.MoveNext();
}
Return:
return(processor.CurrentCodeUnitOffset); // rules GB2, GB999
}