private int SkipWhitespaces(LexerStringReader reader)
{
var whitespaceRuntimeInfo = new LexerRuntimeInfo(reader);
var whitespaceResult = _whitespaceTokenizer.Tokenize(ref whitespaceRuntimeInfo);
return(whitespaceResult.Token?.Length ?? 0);
}
public bool TryFind(LexerStringReader reader, out T value, out int length)
{
if (TryFindNode(ref reader, out var node, out length))
{
value = node.Value;
return(true);
}
value = null;
return(false);
}
public static LexerStringPositionProvider Create(LexerStringReader reader)
{
reader.MoveTo(new LexerStringReader.Position(0));
// TODO: heap alloc :(
var linesToPositionMap = new List <int>(10)
{
0
};
while (reader.MoveToNextLine())
{
linesToPositionMap.Add(reader.CurrentPosition);
}
return(new LexerStringPositionProvider(linesToPositionMap));
}
public LexerResult Tokenize(string query)
{
if (string.IsNullOrWhiteSpace(query))
{
throw new ArgumentException();
}
var finalResult = new LexerResult();
var reader = new LexerStringReader(query, 0);
while (reader.IsEndOfQuery() == false)
{
reader.MoveBy(SkipWhitespaces(reader));
var tokenFound = false;
foreach (var tokenizer in _languageConfig.TokenizersSource.OrderedTokenizers)
{
var runtimeInfo = new LexerRuntimeInfo(reader);
var result = tokenizer.Tokenize(ref runtimeInfo);
if (result.Success || result.IsPartiallySuccessful)
{
finalResult.Tokens.Add(result.Token);
reader.ForceMoveBy(result.Token.Length);
tokenFound = true;
break;
}
}
if (tokenFound == false)
{
var startPosition = reader.CurrentPosition;
var word = reader.ReadTillEndOfWord();
var unknownToken = new UnknownToken(startPosition, word.Length);
finalResult.Tokens.Add(unknownToken);
}
}
return(finalResult);
}
public LexerRuntimeInfo(LexerStringReader reader)
{
Reader = reader;
}
private bool TryFindNode(ref LexerStringReader reader, out Node node, out int length)
{
var initialPosition = reader.CurrentPosition;
var currentNode = _root;
Node lastSuccessfulNode = null;
var lastSuccessfulLength = 0;
var endOfQuery = false;
while (currentNode != null)
{
var progressed = false;
var currentChar = reader.CurrentChar;
// PERF NOTE: bisection could be implemented here, but since number of children should remain small
// then overhead could be larger than actual gains
foreach (var child in currentNode.Children)
{
var isWhitespace = false;
if (char.IsWhiteSpace(child.Key) && char.IsWhiteSpace(currentChar))
{
isWhitespace = true;
}
else if (currentChar != child.Key &&
(_caseSensitive || char.ToLower(currentChar) != char.ToLower(child.Key)))
{
// TODO: PERF: use zero cost extension point trick to implement case sensivity flag (see Performance.md)
continue;
}
// check if not the end
if (reader.IsEndOfQuery())
{
endOfQuery = true;
}
// read characters
if (isWhitespace)
{
reader.ReadTillEndOfWhitespace(); // TODO: multi whitespace read opt in (perhaps use extension point trick?)
}
else
{
reader.MoveBy(1);
}
// assign results
currentNode = child;
if (currentNode.Value != null)
{
lastSuccessfulNode = currentNode;
lastSuccessfulLength = reader.CurrentPosition - initialPosition;
}
// break if nothing more to be processed
if (endOfQuery)
{
break;
}
progressed = true;
break;
}
if (endOfQuery)
{
break;
}
if (!progressed)
{
break;
}
}
if (lastSuccessfulNode != null)
{
length = lastSuccessfulLength;
if (endOfQuery)
{
length++;
}
node = lastSuccessfulNode;
return(true);
}
node = null;
length = 0;
return(false);
}
public void RestorePosition(ref LexerStringReader reader)
{
reader.CurrentPosition = _capturedPosition;
}
public void CaptureCurrentPosition(ref LexerStringReader reader)
{
_capturedPosition = reader.CurrentPosition;
}