protected override TextLineCollection GetLinesCore()
{
var oldLineInfo = _oldText.Lines;
var lineStarts = ArrayBuilder <int> .GetInstance();
lineStarts.Add(0);
// position in the original document
var position = 0;
// delta generated by already processed changes (position in the new document = position + delta)
var delta = 0;
// true if last segment ends with CR and we need to check for CR+LF code below assumes that both CR and LF are also line breaks alone
var endsWithCR = false;
foreach (var change in _changes)
{
// change.Span.Start < position already ruled out by SourceText.WithChanges
// if we've skipped a range, add
if (change.Span.Start > position)
{
if (endsWithCR && _newText[position + delta] == '\n')
{
lineStarts.RemoveLast();
}
var lps = oldLineInfo.GetLinePositionSpan(TextSpan.FromBounds(position, change.Span.Start));
for (int i = lps.Start.Line + 1; i <= lps.End.Line; i++)
{
lineStarts.Add(oldLineInfo[i].Start + delta);
}
endsWithCR = _oldText[change.Span.Start - 1] == '\r';
// in case change is inserted between CR+LF we treat CR as line break alone, but this line break might be retracted and replaced with new one in case LF is inserted
if (endsWithCR && change.Span.Start < _oldText.Length && _oldText[change.Span.Start] == '\n')
{
lineStarts.Add(change.Span.Start + delta);
}
}
if (change.NewLength > 0)
{
var text = GetSubText(new TextSpan(change.Span.Start + delta, change.NewLength));
// optimizations copied from SourceText.LineInfo.ParseLineStarts
var index = 0;
while (index < text.Length)
{
char c = text[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;
}
if (endsWithCR && c == '\n')
{
lineStarts.RemoveLast();
}
else if (c == '\r' && index < text.Length && text[index] == '\n')
{
index++;
}
else if (!TextUtilities.IsAnyLineBreakCharacter(c))
{
continue;
}
lineStarts.Add(change.Span.Start + delta + index);
}
endsWithCR = text[change.NewLength - 1] == '\r';
}
position = change.Span.End;
delta += (change.NewLength - change.Span.Length);
}
if (position < _oldText.Length)
{
if (endsWithCR && _newText[position + delta] == '\n')
{
lineStarts.RemoveLast();
}
var lps = oldLineInfo.GetLinePositionSpan(TextSpan.FromBounds(position, _oldText.Length));
for (int i = lps.Start.Line + 1; i <= lps.End.Line; i++)
{
lineStarts.Add(oldLineInfo[i].Start + delta);
}
}
return(new LineInfo(this, lineStarts.ToArrayAndFree()));
}
private int[] ParseLineStarts()
{
// Corner case check
if (0 == this.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(
(int position, char[] buffer, int 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());
}
private int[] ParseLineStarts()
{
int length = this.Length;
// Corner case check
if (0 == this.Length)
{
return(new int[] { 0 });
}
var position = 0;
var index = 0;
var arrayBuilder = ArrayBuilder <int> .GetInstance();
var lineNumber = 0;
// 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.
while (index < length)
{
char c = this[index];
int lineBreakLength;
// common case - ASCII & not a line break
if (c > '\r' & c <= 127)
{
index++;
continue;
}
else if (c == '\r' && index + 1 < length && this[index + 1] == '\n')
{
lineBreakLength = 2;
}
else if (c == '\n')
{
lineBreakLength = 1;
}
else
{
lineBreakLength = TextUtilities.GetLengthOfLineBreak(this, index);
}
if (0 == lineBreakLength)
{
index++;
}
else
{
arrayBuilder.Add(position);
index += lineBreakLength;
position = index;
lineNumber++;
}
}
// Create a start for the final line.
arrayBuilder.Add(position);
return(arrayBuilder.ToArrayAndFree());
}
