//this is wicked fast //do not refactor extract methods from this if you want to keep the speed public MatchResult Match(string text, int startIndex) { if (StringUtils.IsNullOrEmpty(text)) { throw new ArgumentException("text may not be empty", "text"); // do not localize } MatchResult lastMatch = new MatchResult(); lastMatch.Text = text; int textLength = text.Length; for (int currentIndex = startIndex; currentIndex < textLength; currentIndex++) { //call any prefixless patternmatchers #region HasExpressions if (root.FirstExpression != null) { //begin with the first expression of the _root node_ PatternMatchReference patternMatcherReference = root.FirstExpression; while (patternMatcherReference != null) { int expressionMatchIndex = patternMatcherReference.Matcher.Match(text, currentIndex); if (expressionMatchIndex > 0 && expressionMatchIndex > lastMatch.Length) { lastMatch.Index = currentIndex; lastMatch.Length = expressionMatchIndex; lastMatch.Found = true; lastMatch.Tag = patternMatcherReference.Tag; } patternMatcherReference = patternMatcherReference.NextSibling; } } #endregion //lookup the first token tree node TokenTreeNode node = nodes[(int)text[currentIndex]]; if (node == null) { if (lastMatch.Found) { break; } else { continue; } } for (int matchIndex = currentIndex + 1; matchIndex <= textLength; matchIndex++) { //call patternmatchers for the current prefix #region HasExpressions if (node.FirstExpression != null) { //begin with the first expression of the _current node_ PatternMatchReference patternMatcherReference = node.FirstExpression; while (patternMatcherReference != null) { int expressionMatchIndex = patternMatcherReference.Matcher.Match(text, matchIndex); if (expressionMatchIndex > 0 && expressionMatchIndex > lastMatch.Length) { lastMatch.Index = currentIndex; lastMatch.Length = expressionMatchIndex + matchIndex - currentIndex; lastMatch.Found = true; lastMatch.Tag = patternMatcherReference.Tag; } patternMatcherReference = patternMatcherReference.NextSibling; } } #endregion #region IsEndNode if (node.IsEnd && matchIndex - currentIndex >= lastMatch.Length) { bool leftIsSeparator = currentIndex == 0 ? true : separatorCharLookup[text[currentIndex - 1]]; bool rightIsSeparator = matchIndex == textLength ? true : separatorCharLookup[text[matchIndex]]; if (!node.NeedSeparators || (leftIsSeparator && rightIsSeparator)) { //this node does not require separators on the sides lastMatch.Index = currentIndex; lastMatch.Tag = node.Tag; lastMatch.Found = true; lastMatch.Length = matchIndex - currentIndex; //TODO:perform case test here , case sensitive words might be matched even if they have incorrect case if (currentIndex + lastMatch.Length == textLength) { break; } } } #endregion if (matchIndex >= textLength) { break; } //try fetch a node at this index node = node.ChildNodes[ node.ContainsCaseInsensitiveData ? (int)CharUtils.ToLower(textLookup[(int)text[matchIndex]]) & 0xff : (int)textLookup[(int)text[matchIndex]] & 0xff]; //if node is not null then do: if caseinsensitive then do: insensitivelookup else do: casesesnsitivelookup while (node != null ? (node.ContainsCaseInsensitiveData ? (node.Char != CharUtils.ToLower(textLookup[(int)text[matchIndex]])) : (node.Char != textLookup[(int)text[matchIndex]])) : false) { node = node.NextSibling; } //we found no node on the lookupindex or none of the siblingnodes at that index matched the current char if (node == null) { break; // continue with the next character } } //return last match if (lastMatch.Found) { return(lastMatch); } } if (lastMatch.Found) { return(lastMatch); } else { //no match was found return(MatchResult.NoMatch); } }
//this is wicked fast //do not refactor extract methods from this if you want to keep the speed public MatchResult Match(string text, int startIndex) { if (StringUtils.IsNullOrEmpty(text)) throw new ArgumentException("text may not be empty", "text"); // do not localize MatchResult lastMatch = new MatchResult(); lastMatch.Text = text; int textLength = text.Length; for (int currentIndex = startIndex; currentIndex < textLength; currentIndex ++) { //call any prefixless patternmatchers #region HasExpressions if (root.FirstExpression != null) { //begin with the first expression of the _root node_ PatternMatchReference patternMatcherReference = root.FirstExpression; while (patternMatcherReference != null) { int expressionMatchIndex = patternMatcherReference.Matcher.Match(text, currentIndex); if (expressionMatchIndex > 0 && expressionMatchIndex > lastMatch.Length) { lastMatch.Index = currentIndex; lastMatch.Length = expressionMatchIndex; lastMatch.Found = true; lastMatch.Tag = patternMatcherReference.Tag; } patternMatcherReference = patternMatcherReference.NextSibling; } } #endregion //lookup the first token tree node TokenTreeNode node = nodes[(int) text[currentIndex]]; if (node == null) { if (lastMatch.Found) break; else continue; } for (int matchIndex = currentIndex + 1; matchIndex <= textLength; matchIndex++) { //call patternmatchers for the current prefix #region HasExpressions if (node.FirstExpression != null) { //begin with the first expression of the _current node_ PatternMatchReference patternMatcherReference = node.FirstExpression; while (patternMatcherReference != null) { int expressionMatchIndex = patternMatcherReference.Matcher.Match(text, matchIndex); if (expressionMatchIndex > 0 && expressionMatchIndex > lastMatch.Length) { lastMatch.Index = currentIndex; lastMatch.Length = expressionMatchIndex + matchIndex - currentIndex; lastMatch.Found = true; lastMatch.Tag = patternMatcherReference.Tag; } patternMatcherReference = patternMatcherReference.NextSibling; } } #endregion #region IsEndNode if (node.IsEnd && matchIndex - currentIndex >= lastMatch.Length) { bool leftIsSeparator = currentIndex == 0 ? true : separatorCharLookup[text[currentIndex - 1]]; bool rightIsSeparator = matchIndex == textLength ? true : separatorCharLookup[text[matchIndex]]; if (!node.NeedSeparators || (leftIsSeparator && rightIsSeparator)) { //this node does not require separators on the sides lastMatch.Index = currentIndex; lastMatch.Tag = node.Tag; lastMatch.Found = true; lastMatch.Length = matchIndex - currentIndex; //TODO:perform case test here , case sensitive words might be matched even if they have incorrect case if (currentIndex + lastMatch.Length == textLength) break; } } #endregion if (matchIndex >= textLength) break; //try fetch a node at this index node = node.ChildNodes[ node.ContainsCaseInsensitiveData ? (int) CharUtils.ToLower(textLookup[(int) text[matchIndex]]) & 0xff : (int) textLookup[(int) text[matchIndex]] & 0xff]; //if node is not null then do: if caseinsensitive then do: insensitivelookup else do: casesesnsitivelookup while (node != null ? (node.ContainsCaseInsensitiveData ? (node.Char != CharUtils.ToLower(textLookup[(int) text[matchIndex]])) : (node.Char != textLookup[(int) text[matchIndex]])) : false) node = node.NextSibling; //we found no node on the lookupindex or none of the siblingnodes at that index matched the current char if (node == null) break; // continue with the next character } //return last match if (lastMatch.Found) return lastMatch; } if (lastMatch.Found) { return lastMatch; } else { //no match was found return MatchResult.NoMatch; } }