private int[] ParseLineStarts()
{
// Corner case check
if (0 == Length)
{
return(new[] { 0 });
}
var lineStarts = ArrayBuilder <int> .GetInstance();
lineStarts.Add(0); // there is always the first line
var lastWasCR = false;
// The following loop goes through every character in the text. It is highly
// performance critical, and thus inlines knowledge about common line breaks
// and non-line breaks.
EnumerateChars((position, buffer, length) =>
{
var index = 0;
if (lastWasCR)
{
if (length > 0 && buffer[0] == '\n')
{
index++;
}
lineStarts.Add(position + index);
lastWasCR = false;
}
while (index < length)
{
char c = buffer[index];
index++;
// Common case - ASCII & not a line break
// if (c > '\r' && c <= 127)
// if (c >= ('\r'+1) && c <= 127)
const uint bias = '\r' + 1;
if (unchecked (c - bias) <= 127 - bias)
{
continue;
}
// Assumes that the only 2-char line break sequence is CR+LF
if (c == '\r')
{
if (index < length && buffer[index] == '\n')
{
index++;
}
else if (index >= length)
{
lastWasCR = true;
continue;
}
}
else if (!TextUtilities.IsAnyLineBreakCharacter(c))
{
continue;
}
// next line starts at index
lineStarts.Add(position + index);
}
});
return(lineStarts.ToArrayAndFree());
}
