public ENFA_Regex_MatchPath(ENFA_Controller controller)
{
if (controller.ParserType == ParserType.Grammar)
{
throw new ENFA_RegexRuntime_Exception(ErrorText.ControllerGrammarTypeInRegex);
}
_controller = controller;
_patternLocation = _controller.PatternStart;
_node = new LinkedListNode <ENFA_Regex_MatchPath>(this);
}
internal virtual ENFA_Base NewState(string stateName, StateType stateType)
{
foreach (ENFA_Transition transition in GetTransitions)
{
ENFA_Base state = transition.Transition();
if (state.StateName.Equals(StateName) && state.StateType == stateType)
{
return(state);
}
}
return(Controller.Factory.CreateState(this, stateName, stateType));
}
internal virtual ENFA_Transition NewRegexTransition(RegexTransitionType transitionType, ENFA_Base nextState)
{
if (Controller.ParserType == ParserType.Grammar)
{
throw new ENFA_GrammarBuild_Exception(ErrorText.TryingToCreateNewRegexTransitionInGrammar);
}
foreach (ENFA_Transition transition in GetTransitions)
{
ENFA_Base state = transition.Transition();
if ((transition as ENFA_Regex_Transition).TransitionType == transitionType && state.Equals(nextState))
{
return(transition);
}
}
return((Controller.Factory as ENFA_Regex_Factory).CreateRegexTransition(transitionType, nextState));
}
private void Kill()
{
if (Node.List.Count > 1)
{
Node.List.Remove(this);
// TODO remove matches from matchTree until split
_controller = null;
_node = null;
_patternLocation = null;
}
else
{
// TODO Insert Error token for one char and go back
}
}
public ENFA_Regex_Transition(RegexTransitionType transitionType, ENFA_Base nextState) : base(nextState)
{
_transitionType = transitionType;
_literal = new List <char>();
}
private void SetTransitionCount(ENFA_Base state, ENFA_Transition transition, int minRepetitions, int maxRepetitions, MatchingType matchingType)
{
state.MinRepetitions = minRepetitions;
state.MaxRepetitions = maxRepetitions;
transition.MatchingType = matchingType;
}
public override bool Tokenize(string terminalName, StreamReader reader)
{
_matchedCharInRegexBuild = new List <char>();
_currectTerminalName = terminalName;
char?nextChar = NextCharInStream(reader);
bool escaped = false;
bool success = false; // Error until proven correct
bool exit = false;
ENFA_GroupingStart _parentStart = Controller.PatternStart;
ENFA_PatternEnd _patternEnd = Controller.Factory.CreatePatternEnd(_parentStart as ENFA_PatternStart, terminalName, StateType.Accepting);
ENFA_GroupingEnd _parentEnd = _patternEnd;
ENFA_Base lastState = _parentStart;
ENFA_Base nextState;
ENFA_Transition activeTransition = null;
while (nextChar.HasValue && !exit)
{
char? tempNextChar = PeekNextChar(reader);
MatchingType matchingType = MatchingType.NotSet;
if (!escaped)
{
switch (nextChar.Value)
{
case Constants.Backslash:
escaped = true;
break;
case Constants.ExitContext:
activeTransition = lastState.NewRegexTransition(RegexTransitionType.GroupingEnd, _parentEnd);
lastState.AddTransition(activeTransition);
exit = true;
if (_parentEnd is ENFA_PatternEnd)
{
success = true;
}
else
{
ThrowBuildException(ErrorText.ExitContextBeforePatternEnd);
}
break;
case Constants.StartOfLine:
nextState = lastState.NewState(RegexTransitionType.StartOfLine.ToString(), StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.StartOfLine, nextState);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case Constants.AllButNewLine:
nextState = lastState.NewState(RegexTransitionType.NegateNewLine.ToString(), StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.NegateNewLine, nextState);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case Constants.EndOfLine:
if (lastState is ENFA_PatternStart)
{
ThrowBuildException(ErrorText.EndOfLineAsFirstCharInPattern);
}
nextState = lastState.NewState(RegexTransitionType.EndOfLine.ToString(), StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.EndOfLine, nextState);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case Constants.RightCurlyBracket:
ThrowBuildException(ErrorText.RightCurlyBracketWithoutMatchingLeftCurlyBracket);
break;
case Constants.LeftCurlyBracket:
if (lastState is ENFA_PatternStart)
{
ThrowBuildException(ErrorText.LeftCurlyBracketAsFirstCharInPattern);
}
int minRepetitions;
int maxRepetitions;
CheckQuantifiers(reader, out minRepetitions, out maxRepetitions, out matchingType);
if (lastState is ENFA_GroupEnd)
{
SetTransitionCount((lastState as ENFA_GroupEnd).GroupStart, activeTransition, minRepetitions, maxRepetitions, matchingType);
}
else
{
SetTransitionCount(lastState, activeTransition, minRepetitions, maxRepetitions, matchingType);
}
break;
case Constants.RightSquareBracket:
ThrowBuildException(ErrorText.RightSquareBracketWithoutMatchingLeftSquareBracket);
break;
case Constants.LeftSquareBracket:
if (tempNextChar.HasValue && tempNextChar.Value == Constants.CircumflexAccent)
{
nextState = lastState.NewState("Negate Character Group", StateType.Transition);
/* Remove CircumfelxAccent */
ConsumeNextChar(reader);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.NegateLiteral, nextState);
}
else
{
nextState = lastState.NewState("Character Group", StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.Literal, nextState);
}
AddCharacterGroup((ENFA_Regex_Transition)activeTransition, reader);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case Constants.GroupingStart:
bool recording = Controller.DefaultGroupingRecording;
string groupName = null;
if (tempNextChar.HasValue && tempNextChar.Value == Constants.QuestionMark)
{
/* Consume QuetionMark */
ConsumeNextChar(reader);
char?tempNextChar2 = PeekNextChar(reader);
switch (tempNextChar2.Value)
{
case Constants.Colon:
/* Consume Colon */
ConsumeNextChar(reader);
/* non-recording group */
recording = false;
_parentStart = Controller.Factory.CreateGroupStart(_parentStart);
_parentEnd = Controller.Factory.CreateGroupEnd(_parentStart as ENFA_GroupStart, recording, groupName, _parentEnd);
break;
case Constants.GreaterThanSign:
/* Consume Greater Than Sign */
ConsumeNextChar(reader);
/* recording group */
recording = true;
_parentStart = Controller.Factory.CreateGroupStart(_parentStart);
_parentEnd = Controller.Factory.CreateGroupEnd(_parentStart as ENFA_GroupStart, recording, groupName, _parentEnd);
break;
case Constants.EqualsSign:
/* Consume EqualSign */
ConsumeNextChar(reader);
/* positive lookahead */
_parentStart = Controller.Factory.CreateLookaheadStart(AssertionType.Positive, _parentStart);
_parentEnd = Controller.Factory.CreateLookaheadEnd(_parentStart as ENFA_LookaheadStart, _parentEnd);
break;
case Constants.ExclamationMark:
/* Consume ExclamationMark */
ConsumeNextChar(reader);
/* negative lookahead */
_parentStart = Controller.Factory.CreateLookaheadStart(AssertionType.Negative, _parentStart);
_parentEnd = Controller.Factory.CreateLookaheadEnd(_parentStart as ENFA_LookaheadStart, _parentEnd);
break;
case Constants.LessThanSign:
/* Consume Less Than Sign */
ConsumeNextChar(reader);
char?tempNextChar3 = PeekNextChar(reader);
if (tempNextChar3.Value == Constants.EqualsSign)
{
/* Consume Equals Sign */
ConsumeNextChar(reader);
/* positive lookbehind */
_parentStart = Controller.Factory.CreateLookbehindStart(AssertionType.Positive, _parentStart);
_parentEnd = Controller.Factory.CreateLookbehindEnd(_parentStart as ENFA_LookbehindStart, _parentEnd);
}
else if (tempNextChar3.Value == Constants.ExclamationMark)
{
/* Consume Exclamation Mark */
ConsumeNextChar(reader);
/* negative lookbehind */
_parentStart = Controller.Factory.CreateLookbehindStart(AssertionType.Negative, _parentStart);
_parentEnd = Controller.Factory.CreateLookbehindEnd(_parentStart as ENFA_LookbehindStart, _parentEnd);
}
else
{
/* named group */
recording = true;
groupName = GetGroupName(reader);
_parentStart = Controller.Factory.CreateGroupStart(_parentStart);
_parentEnd = Controller.Factory.CreateGroupEnd(_parentStart as ENFA_GroupStart, recording, groupName, _parentEnd);
}
break;
default:
ThrowBuildException(ErrorText.GroupingExpectedSpecifierAfterQuestionMark);
break;
}
}
else
{
_parentStart = Controller.Factory.CreateGroupStart(_parentStart);
_parentEnd = Controller.Factory.CreateGroupEnd(_parentStart as ENFA_GroupStart, recording, groupName, _parentEnd);
}
if (_parentEnd is ENFA_LookbehindEnd)
{
nextState = _parentEnd;
}
else
{
nextState = _parentStart;
}
activeTransition = lastState.NewRegexTransition(RegexTransitionType.GroupingStart, nextState);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case Constants.GroupingEnd:
if (!(_parentStart is ENFA_GroupingStart))
{
ThrowBuildException(ErrorText.GroupingEndWithoutGroupingStart);
}
if (_parentEnd is ENFA_LookbehindEnd)
{
nextState = _parentStart;
}
else
{
nextState = _parentEnd;
}
activeTransition = lastState.NewRegexTransition(RegexTransitionType.GroupingEnd, nextState);
lastState.AddTransition(activeTransition);
lastState = nextState;
_parentStart = _parentStart.Parent;
_parentEnd = _parentEnd.Parent;
break;
case Constants.VerticalLine:
if (_parentEnd is ENFA_LookbehindEnd)
{
nextState = _parentEnd;
}
else
{
nextState = _parentStart;
}
activeTransition = lastState.NewRegexTransition(RegexTransitionType.GroupingEnd, _parentEnd);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case Constants.PlusSign:
if (lastState is ENFA_PatternStart)
{
ThrowBuildException(ErrorText.PlusSignAsFirstCharInPattern);
}
if (tempNextChar == Constants.QuestionMark)
{
/* Consume Quention Mark */
ConsumeNextChar(reader);
/* Lazy matching overwriting default */
matchingType = MatchingType.LazyMatching;
}
else if (tempNextChar == Constants.GreaterThanSign)
{
/* Consume Greater Than Sign */
ConsumeNextChar(reader);
/* Greedy matching overwriting default */
matchingType = MatchingType.GreedyMatching;
}
else
{
/* Use default matching */
matchingType = Controller.DefaultMatchType;
}
minRepetitions = 1;
maxRepetitions = -1;
if (lastState is ENFA_GroupEnd)
{
SetTransitionCount((lastState as ENFA_GroupEnd).GroupStart, activeTransition, minRepetitions, maxRepetitions, matchingType);
}
else
{
SetTransitionCount(lastState, activeTransition, minRepetitions, maxRepetitions, matchingType);
}
break;
case Constants.Asterisk:
if (lastState is ENFA_PatternStart)
{
ThrowBuildException(ErrorText.AsterixAsFirstCharInPattern);
}
if (tempNextChar == Constants.QuestionMark)
{
/* Consume Quention Mark */
ConsumeNextChar(reader);
/* Lazy matching overwriting default */
matchingType = MatchingType.LazyMatching;
}
else if (tempNextChar == Constants.GreaterThanSign)
{
/* Consume Greater Than Sign */
ConsumeNextChar(reader);
/* Greedy matching overwriting default */
matchingType = MatchingType.GreedyMatching;
}
else
{
/* Use default matching */
matchingType = Controller.DefaultMatchType;
}
minRepetitions = 0;
maxRepetitions = -1;
if (lastState is ENFA_GroupEnd)
{
SetTransitionCount((lastState as ENFA_GroupEnd).GroupStart, activeTransition, minRepetitions, maxRepetitions, matchingType);
}
else
{
SetTransitionCount(lastState, activeTransition, minRepetitions, maxRepetitions, matchingType);
}
break;
case Constants.QuestionMark:
if (lastState is ENFA_PatternStart)
{
ThrowBuildException(ErrorText.QuestionMarkAsFirstCharInPattern);
}
if (tempNextChar == Constants.QuestionMark)
{
/* Consume Quention Mark */
ConsumeNextChar(reader);
/* Lazy matching overwriting default */
matchingType = MatchingType.LazyMatching;
}
else if (tempNextChar == Constants.GreaterThanSign)
{
/* Consume Greater Than Sign */
ConsumeNextChar(reader);
/* Greedy matching overwriting default */
matchingType = MatchingType.GreedyMatching;
}
else
{
/* Use default matching */
matchingType = Controller.DefaultMatchType;
}
minRepetitions = 0;
maxRepetitions = 1;
if (lastState is ENFA_GroupEnd)
{
SetTransitionCount((lastState as ENFA_GroupEnd).GroupStart, activeTransition, minRepetitions, maxRepetitions, matchingType);
}
else
{
SetTransitionCount(lastState, activeTransition, minRepetitions, maxRepetitions, matchingType);
}
break;
default:
nextState = lastState.NewState(nextChar.Value, StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.Literal, nextState);
(activeTransition as ENFA_Regex_Transition).AddLiteral(nextChar.Value);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
}
}
else
{
/* Escaped characters */
switch (nextChar.Value)
{
case '0':
nextState = lastState.NewState("Null Char", StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.Literal, nextState);
(activeTransition as ENFA_Regex_Transition).AddLiteral(Constants.NullChar);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 'a':
nextState = lastState.NewState("Alert", StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.Literal, nextState);
(activeTransition as ENFA_Regex_Transition).AddLiteral(Constants.Alert);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 'e':
nextState = lastState.NewState("Escape", StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.Literal, nextState);
(activeTransition as ENFA_Regex_Transition).AddLiteral(Constants.Escape);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 'y':
nextState = lastState.NewState("Backspace", StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.Literal, nextState);
(activeTransition as ENFA_Regex_Transition).AddLiteral(Constants.Backspace);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 'f':
nextState = lastState.NewState("Form Feed", StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.Literal, nextState);
(activeTransition as ENFA_Regex_Transition).AddLiteral(Constants.FormFeed);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 'r':
nextState = lastState.NewState("Carriage Return", StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.Literal, nextState);
(activeTransition as ENFA_Regex_Transition).AddLiteral(Constants.CarriageReturn);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 't':
nextState = lastState.NewState("Horizontal Tab", StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.Literal, nextState);
(activeTransition as ENFA_Regex_Transition).AddLiteral(Constants.HorizontalTab);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 'v':
nextState = lastState.NewState("Vertical Tab", StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.Literal, nextState);
(activeTransition as ENFA_Regex_Transition).AddLiteral(Constants.VerticalTab);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 'n':
nextState = lastState.NewState(RegexTransitionType.NewLine.ToString(), StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.NewLine, nextState);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 'w':
nextState = lastState.NewState(RegexTransitionType.Word.ToString(), StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.Word, nextState);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 'W':
nextState = lastState.NewState(RegexTransitionType.NegateWord.ToString(), StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.NegateWord, nextState);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 'd':
nextState = lastState.NewState(RegexTransitionType.Digit.ToString(), StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.Digit, nextState);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 'D':
nextState = lastState.NewState(RegexTransitionType.NegateDigit.ToString(), StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.NegateDigit, nextState);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 's':
nextState = lastState.NewState(RegexTransitionType.Whitespace.ToString(), StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.Whitespace, nextState);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 'S':
nextState = lastState.NewState(RegexTransitionType.NegateWhitespace.ToString(), StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.NegateWhitespace, nextState);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 'l':
nextState = lastState.NewState(RegexTransitionType.Letter.ToString(), StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.Letter, nextState);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 'L':
nextState = lastState.NewState(RegexTransitionType.NegateLetter.ToString(), StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.NegateLetter, nextState);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 'b':
nextState = lastState.NewState(RegexTransitionType.WordBoundary.ToString(), StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.WordBoundary, nextState);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 'B':
nextState = lastState.NewState(RegexTransitionType.NegateWordBoundary.ToString(), StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.NegateWordBoundary, nextState);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
case '8':
case '9':
/* Back reference */
int groupNumber = int.Parse(nextChar.Value.ToString());
nextState = lastState.NewPlaceHolder(_patternEnd.LookupGroupNameFromNumber(groupNumber));
activeTransition = lastState.NewRegexTransition(RegexTransitionType.BackReference, nextState);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case 'k':
/* Named back reference like \k<Bartho> */
string groupName = null;
if (tempNextChar.HasValue && tempNextChar.Value == Constants.LessThanSign)
{
/* Consume LessThanSign */
ConsumeNextChar(reader);
groupName = GetGroupName(reader);
if (!_patternEnd.GroupNameExists(groupName))
{
ThrowBuildException(ErrorText.SpecifiedGroupNameDoesNotExist);
}
}
else
{
ThrowBuildException(ErrorText.NamedBackreferenceMissingStartGroupName);
}
nextState = lastState.NewPlaceHolder(groupName);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.BackReference, nextState);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
case Constants.DoubleQuote:
case Constants.LeftCurlyBracket:
case Constants.LeftSquareBracket:
case Constants.LeftParanthesis:
case Constants.RightParanthesis:
case Constants.VerticalLine:
case Constants.Backslash:
case Constants.FullStop:
case Constants.DollarSign:
case Constants.QuestionMark:
case Constants.PlusSign:
case Constants.Asterisk:
case Constants.CircumflexAccent:
nextState = lastState.NewState(nextChar.Value, StateType.Transition);
activeTransition = lastState.NewRegexTransition(RegexTransitionType.Literal, nextState);
(activeTransition as ENFA_Regex_Transition).AddLiteral(nextChar.Value);
lastState.AddTransition(activeTransition);
lastState = nextState;
break;
default:
ThrowBuildException(ErrorText.CharacterEscapedWithoutBeingExpectedTo);
break;
}
escaped = false;
}
if (!exit)
{
nextChar = NextCharInStream(reader);
}
}
return(success);
}
public ENFA_Grammar_Transition(GrammarTransitionType transitionType, ENFA_Base nextState) : base(nextState)
{
_transitionType = transitionType;
}
public ENFA_Grammar_Transition CreateGrammarTransition(GrammarTransitionType transitionType, ENFA_Base nextState)
{
return(new ENFA_Grammar_Transition(transitionType, nextState));
}
public override ENFA_Base CreateState(ENFA_Base prevoiusState, string stateName, StateType stateType)
{
return(new ENFA_State(Controller, prevoiusState, stateName, stateType));
}
public ENFA_Transition(ENFA_Base nextState)
{
_nextState = nextState;
_matchingType = MatchingType.NotSet;
}
public ENFA_State(ENFA_Controller controller, ENFA_Base prevoiusState, string stateName, StateType stateType) : base(controller, stateType)
{
_stateName = stateName;
_prevoiusState = prevoiusState;
}
public abstract ENFA_Base CreateState(ENFA_Base prevoiusState, string stateName, StateType stateType);
public ENFA_Regex_MatchPath(ENFA_Regex_MatchPath cloneThis)
{
_controller = cloneThis.Controller;
_node = cloneThis.Node;
_patternLocation = cloneThis.PatternLocation;
}
public ENFA_Regex_Transition CreateRegexTransition(RegexTransitionType transitionType, ENFA_Base nextState)
{
return(new ENFA_Regex_Transition(transitionType, nextState));
}