public static ILineBreaks CreateLineBreaks(string source)
{
ILineBreaksEditor lineBreaks = null;
int index = 0;
while (index < source.Length)
{
int breakLength = TextUtilities.LengthOfLineBreak(source, index, source.Length);
if (breakLength == 0)
{
++index;
}
else
{
if (lineBreaks == null)
{
lineBreaks = LineBreakManager.CreateLineBreakEditor(source.Length);
}
lineBreaks.Add(index, breakLength);
index += breakLength;
}
}
return(lineBreaks ?? Empty);
}
/// <summary>
/// Consolidate two string rebuilders, taking advantage of the fact that they have already extracted the line breaks.
/// </summary>
public static StringRebuilder Consolidate(StringRebuilder left, StringRebuilder right)
{
Debug.Assert(left.Length > 0);
Debug.Assert(right.Length > 0);
int length = left.Length + right.Length;
char[] result = new char[length];
left.CopyTo(0, result, 0, left.Length);
right.CopyTo(0, result, left.Length, right.Length);
ILineBreaks lineBreaks;
if ((left.LineBreakCount == 0) && (right.LineBreakCount == 0))
{
lineBreaks = LineBreakManager.Empty;
//_lineBreakSpan defaults to 0, 0 which is what we want
}
else
{
ILineBreaksEditor breaks = LineBreakManager.CreateLineBreakEditor(length, left.LineBreakCount + right.LineBreakCount);
int offset = 0;
if ((result[left.Length] == '\n') && (result[left.Length - 1] == '\r'))
{
//We have a \r\n spanning the seam ... add that as a special linebreak later.
offset = 1;
}
int leftLines = left.LineBreakCount - offset;
for (int i = 0; (i < leftLines); ++i)
{
Span extent;
int lineBreakLength;
left.GetLineFromLineNumber(i, out extent, out lineBreakLength);
breaks.Add(extent.End, lineBreakLength);
}
if (offset == 1)
{
breaks.Add(left.Length - 1, 2);
}
for (int i = offset; (i < right.LineBreakCount); ++i)
{
Span extent;
int lineBreakLength;
right.GetLineFromLineNumber(i, out extent, out lineBreakLength);
breaks.Add(extent.End + left.Length, lineBreakLength);
}
lineBreaks = breaks;
}
return(StringRebuilderForChars.Create(result, length, lineBreaks));
}
public static StringRebuilder Create(string text)
{
if (text == null)
{
throw new ArgumentNullException("text");
}
#if DEBUG
Interlocked.Add(ref _totalCharactersScanned, text.Length);
#endif
return((text.Length == 0)
? StringRebuilder.Empty
: StringRebuilderForString.Create(text, text.Length, LineBreakManager.CreateLineBreaks(text)));
}
// Evil performance hack (but we are on a hot path here):
// thresholdForInvalidCharacters should be '\u0001' if we are throwing on invalid characters.
// should be '\0' if we are not.
// (otherwise we need to check both a throwOnInvalidCharacters boolean and that c == 0).
private static ILineBreaks ParseBlock(char[] buffer, int length, char thresholdForInvalidCharacters,
ref NewlineState newlineState,
ref LeadingWhitespaceState leadingWhitespaceState,
ref int currentLineLength,
ref int longestLineLength,
ref bool nextCharIsStartOfLine)
{
// Note that the lineBreaks created here will (internally) use the pooled list of line breaks.
IPooledLineBreaksEditor lineBreaks = LineBreakManager.CreatePooledLineBreakEditor(length);
int index = 0;
while (index < length)
{
int breakLength = TextUtilities.LengthOfLineBreak(buffer, index, length);
if (breakLength == 0)
{
char c = buffer[index];
// If we are checking for invalid characters, throw if we encounter a \0
if (c < thresholdForInvalidCharacters)
{
throw new FileFormatException("File contains NUL characters");
}
++currentLineLength;
++index;
if (nextCharIsStartOfLine)
{
switch (c)
{
case ' ':
leadingWhitespaceState.Increment(LeadingWhitespaceState.LineLeadingCharacter.Space, 1);
break;
case '\t':
leadingWhitespaceState.Increment(LeadingWhitespaceState.LineLeadingCharacter.Tab, 1);
break;
default:
leadingWhitespaceState.Increment(LeadingWhitespaceState.LineLeadingCharacter.Printable, 1);
break;
}
nextCharIsStartOfLine = false;
}
}
else
{
lineBreaks.Add(index, breakLength);
longestLineLength = Math.Max(longestLineLength, currentLineLength);
currentLineLength = 0;
if (breakLength == 2)
{
newlineState.Increment(NewlineState.LineEnding.CRLF, 1);
}
else
{
switch (buffer[index])
{
// This code needs to be kep consistent with TextUtilities.LengthOfLineBreak()
case '\r': newlineState.Increment(NewlineState.LineEnding.CR, 1); break;
case '\n': newlineState.Increment(NewlineState.LineEnding.LF, 1); break;
case '\u0085': newlineState.Increment(NewlineState.LineEnding.NEL, 1); break;
case '\u2028': newlineState.Increment(NewlineState.LineEnding.LS, 1); break;
case '\u2029': newlineState.Increment(NewlineState.LineEnding.PS, 1); break;
default: throw new InvalidOperationException("Unexpected line ending");
}
}
if (nextCharIsStartOfLine)
{
leadingWhitespaceState.Increment(LeadingWhitespaceState.LineLeadingCharacter.Empty, 1);
}
nextCharIsStartOfLine = true;
}
index += breakLength;
}
lineBreaks.ReleasePooledLineBreaks();
return(lineBreaks);
}
internal static StringRebuilder Load(TextReader reader, long fileSize, string id,
out bool hasConsistentLineEndings, out int longestLineLength,
int blockSize = 0,
int minCompressedBlockSize = TextImageLoader.BlockSize) // Exposed for unit tests
{
LineEndingState lineEnding = LineEndingState.Unknown;
int currentLineLength = 0;
longestLineLength = 0;
bool useCompressedStringRebuilders = (fileSize >= TextModelOptions.CompressedStorageFileSizeThreshold);
if (blockSize == 0)
{
blockSize = useCompressedStringRebuilders ? TextModelOptions.CompressedStoragePageSize : TextImageLoader.BlockSize;
}
PageManager pageManager = null;
char[] buffer;
if (useCompressedStringRebuilders)
{
pageManager = new PageManager();
buffer = new char[blockSize];
}
else
{
buffer = TextImageLoader.AcquireBuffer(blockSize);
}
StringRebuilder content = StringRebuilderForChars.Empty;
try
{
while (true)
{
int read = TextImageLoader.LoadNextBlock(reader, buffer);
if (read == 0)
{
break;
}
var lineBreaks = LineBreakManager.CreateLineBreakEditor(read);
TextImageLoader.ParseBlock(buffer, read, lineBreaks, ref lineEnding, ref currentLineLength, ref longestLineLength);
char[] bufferForStringBuilder = buffer;
if (read < (buffer.Length / 2))
{
// We read far less characters than buffer so copy the contents to a new buffer and reuse the original buffer.
bufferForStringBuilder = new char[read];
Array.Copy(buffer, bufferForStringBuilder, read);
}
else
{
// We're using most of bufferForStringRebuilder so allocate a new block for the next chunk.
buffer = new char[blockSize];
}
var newContent = (useCompressedStringRebuilders && (read > minCompressedBlockSize))
? StringRebuilderForCompressedChars.Create(new Page(pageManager, bufferForStringBuilder, read), lineBreaks)
: StringRebuilderForChars.Create(bufferForStringBuilder, read, lineBreaks);
content = content.Insert(content.Length, newContent);
}
longestLineLength = Math.Max(longestLineLength, currentLineLength);
hasConsistentLineEndings = lineEnding != LineEndingState.Inconsistent;
}
finally
{
if (!useCompressedStringRebuilders)
{
TextImageLoader.ReleaseBuffer(buffer);
}
}
return(content);
}
private static ILineBreaks ParseBlock(char[] buffer, int length,
ref LineEndingState lineEnding, ref int currentLineLength, ref int longestLineLength)
{
// Note that the lineBreaks created here will (internally) use the pooled list of line breaks.
IPooledLineBreaksEditor lineBreaks = LineBreakManager.CreatePooledLineBreakEditor(length);
int index = 0;
while (index < length)
{
int breakLength = TextUtilities.LengthOfLineBreak(buffer, index, length);
if (breakLength == 0)
{
++currentLineLength;
++index;
}
else
{
lineBreaks.Add(index, breakLength);
longestLineLength = Math.Max(longestLineLength, currentLineLength);
currentLineLength = 0;
if (lineEnding != LineEndingState.Inconsistent)
{
if (breakLength == 2)
{
if (lineEnding == LineEndingState.Unknown)
{
lineEnding = LineEndingState.CRLF;
}
else if (lineEnding != LineEndingState.CRLF)
{
lineEnding = LineEndingState.Inconsistent;
}
}
else
{
LineEndingState newLineEndingState;
switch (buffer[index])
{
// This code needs to be kep consistent with TextUtilities.LengthOfLineBreak()
case '\r': newLineEndingState = LineEndingState.CR; break;
case '\n': newLineEndingState = LineEndingState.LF; break;
case '\u0085': newLineEndingState = LineEndingState.NEL; break;
case '\u2028': newLineEndingState = LineEndingState.LS; break;
case '\u2029': newLineEndingState = LineEndingState.PS; break;
default: throw new InvalidOperationException("Unexpected line ending");
}
if (lineEnding == LineEndingState.Unknown)
{
lineEnding = newLineEndingState;
}
else if (lineEnding != newLineEndingState)
{
lineEnding = LineEndingState.Inconsistent;
}
}
}
index += breakLength;
}
}
lineBreaks.ReleasePooledLineBreaks();
return(lineBreaks);
}
