public NFA maybe()
{
NFA ret = new NFA(this);
ret.endStates.Add(ret.startState);
return(ret);
}
private NFA parseAlternates(String str, ref int loc)
{
//str[loc] == '['
loc++;
bool exclusion = false;
if (str[loc] == '^')
{
exclusion = true;
loc++;
}
NFA builtNFA = null;
while (str[loc] != ']')
{
if (loc >= str.Length)
{
throw new RegexException("Regex has a [ without a matching ]");
}
NFA nextNFA;
if (str[loc] == '\\')
{
nextNFA = parseSpecial(str, ref loc);
}
else if (loc < str.Length - 1 && str[loc + 1] == '-')
{
char start = str[loc];
loc += 2;
char end = str[loc];
HashSet <char> chars = new HashSet <char>();
for (char ch = start; ch <= end; ch++)
{
chars.Add(ch);
}
nextNFA = new NFA(chars);
loc++;
}
else
{
nextNFA = new NFA(str.Substring(loc, 1));
loc++;
}
if (builtNFA == null)
{
builtNFA = nextNFA;
}
else
{
builtNFA = builtNFA.or(nextNFA);
}
}
if (exclusion)
{
builtNFA = builtNFA.complement();
}
builtNFA = builtNFA.complement().or(new NFA(1).complement()).complement();
loc++;
//loc is first character after ]
return(builtNFA);
}
public NFA(NFA second)
{
//we want to create a deep copy instead of modifying the original
connections = second.connections.ConvertAll(c => c.copy());
endStates = new HashSet <int>(second.endStates);
startState = second.startState;
numStates = second.numStates;
hasEpsilons = second.hasEpsilons;
calculateConnectionPerState();
}
private NFA parseRegex(String str, ref int loc)
{
NFA builtNFA = new NFA("");
while (loc < str.Length)
{
NFA nextNFA = parseOne(str, ref loc);
builtNFA = builtNFA.conc(nextNFA);
}
return(builtNFA);
}
public NFA not()
{
NFA ret = new NFA(this);
HashSet <int> newEndStates = new HashSet <int>();
for (int i = 0; i < ret.numStates; i++)
{
if (!ret.endStates.Contains(i))
{
newEndStates.Add(i);
}
}
ret.endStates = newEndStates;
return(ret);
}
private NFA parseParenthesis(string str, ref int loc)
{
//str[loc] == '('
loc++;
NFA totalNFA = null;
NFA builtNFA = null;
while (str[loc] != ')')
{
if (loc >= str.Length)
{
throw new RegexException("Regex has a ( without a matching )");
}
if (str[loc] == '|')
{
if (totalNFA == null)
{
totalNFA = builtNFA;
}
else
{
totalNFA = totalNFA.or(builtNFA);
}
builtNFA = null;
loc++;
}
NFA nextNFA = parseOne(str, ref loc);
if (builtNFA == null)
{
builtNFA = nextNFA;
}
else
{
builtNFA = builtNFA.conc(nextNFA);
}
}
if (totalNFA == null)
{
totalNFA = builtNFA;
}
else
{
totalNFA = totalNFA.or(builtNFA);
}
loc++;
//loc is first character after )
return(totalNFA);
}
public NFA closure()
{
NFA ret = new NFA(this);
//add new start state
int oldStartState = ret.startState;
int newStartState = ret.addState(false);
ret.startState = newStartState;
ret.connect(newStartState, oldStartState, new EpsilonConnection(0, 0));
//add connections from old end states to old start state
ret.addConnectionToEndStates(oldStartState, new EpsilonConnection(0, 0));
//old start state is a valid end state
ret.endStates.Add(newStartState);
ret.hasEpsilons = true;
return(ret);
}
public NFA conc(NFA second)
{
NFA ret = new NFA(this);
//add new states
ret.incrementIndices(second.numStates);
foreach (Connection c in second.connections)
{
ret.connections.Add(c.copy());
}
ret.calculateConnectionPerState();
//connect old end states to second's start state
int secondStartState = second.startState;
ret.addConnectionToEndStates(secondStartState, new EpsilonConnection(0, 0));
//fix end states as second's end states
ret.endStates = new HashSet <int>(second.endStates);
ret.hasEpsilons = true;
return(ret);
}
private NFA parseOne(String str, ref int loc)
{
NFA retNFA;
if (str[loc] == '(')
{
retNFA = parseParenthesis(str, ref loc);
}
else if (str[loc] == '[')
{
retNFA = parseAlternates(str, ref loc);
}
else if (str[loc] == '\\')
{
retNFA = parseSpecial(str, ref loc);
}
else if (str[loc] == '.')
{
retNFA = new NFA(1);
loc++;
}
else
{
retNFA = new NFA(str.Substring(loc, 1));
loc++;
}
if (loc < str.Length)
{
if (str[loc] == '*')
{
retNFA = retNFA.closure();
loc++;
}
else if (str[loc] == '?')
{
retNFA = retNFA.maybe();
loc++;
}
}
return(retNFA);
}
public NFA or(NFA second)
{
NFA ret = new NFA(this);
//add states
ret.incrementIndices(second.numStates);
foreach (int i in second.endStates)
{
ret.endStates.Add(i);
}
//add connections between newly added states
foreach (Connection c in second.connections)
{
ret.connections.Add(c.copy());
}
//connect new start states with old start state
ret.connect(ret.startState, second.startState, new EpsilonConnection(-1, -1));
ret.hasEpsilons = true;
ret.calculateConnectionPerState();
return(ret);
}
private NFA parseSpecial(String str, ref int loc)
{
//str[loc] == '\\'
loc++;
if (loc >= str.Length)
{
throw new RegexException("Regex ends with a special character marker");
}
NFA ret;
switch (str[loc])
{
case '|':
ret = new NFA("|");
break;
case '\\':
ret = new NFA("\\");
break;
case 'n':
ret = new NFA("\n");
break;
case 't':
ret = new NFA("\t");
break;
case 'r':
ret = new NFA("\r");
break;
case '\'':
ret = new NFA("\'");
break;
case '\"':
ret = new NFA("\"");
break;
case '*':
ret = new NFA("*");
break;
case '(':
ret = new NFA("(");
break;
case ')':
ret = new NFA(")");
break;
case '[':
ret = new NFA("[");
break;
case ']':
ret = new NFA("]");
break;
case '.':
ret = new NFA(".");
break;
case '?':
ret = new NFA("?");
break;
case '^':
ret = new NFA("^");
break;
case '-':
ret = new NFA("-");
break;
default:
throw new RegexException("Unknown escape character in regex");
}
loc++;
//loc is first character after special
return(ret);
}
public Regex(String str)
{
int loc = 0;
nfa = parseRegex(str, ref loc);
}
public NFA toDFA()
{
NFA temp = new NFA(this);
NFA ret = new NFA();
temp.deEpsilonate();
List <HashSet <int> > newStates = new List <HashSet <int> >();
newStates.Add(new HashSet <int>());
Queue <HashSet <int> > statesThatNeedProcessing = new Queue <HashSet <int> >();
HashSet <int> startState = new HashSet <int>();
startState.Add(temp.startState);
ret.startState = ret.addState(temp.endStates.Contains(temp.startState));
newStates.Add(startState);
statesThatNeedProcessing.Enqueue(startState);
while (statesThatNeedProcessing.Count > 0)
{
HashSet <int> newState = statesThatNeedProcessing.Dequeue();
int hasState = newStates.FindIndex(h => newState.SetEquals(h));
List <Connection> allConnections = new List <Connection>();
List <Connection> uninterestingConnections = new List <Connection>();
HashSet <int> anyReachable = new HashSet <int>();
foreach (int i in newState)
{
allConnections.AddRange(temp.connectionPerState[i]);
}
HashSet <char> interestingChars = new HashSet <char>();
foreach (Connection c in allConnections)
{
if (c is AnythingConnection)
{
anyReachable.Add(c.end);
}
if (c is AnythingConnection || c is ConnectionAnythingBut)
{
uninterestingConnections.Add(c);
}
interestingChars.UnionWith(c.interestingCharacters());
}
foreach (char ch in interestingChars)
{
HashSet <int> reachable = new HashSet <int>();
foreach (Connection c in allConnections)
{
if (c.accepts(ch))
{
reachable.Add(c.end);
}
}
int newAddIndex = newStates.FindIndex(h => reachable.SetEquals(h));
if (newAddIndex == -1)
{
bool endState = false;
foreach (int i in reachable)
{
if (temp.endStates.Contains(i))
{
endState = true;
break;
}
}
newAddIndex = ret.addState(endState);
newStates.Add(reachable);
statesThatNeedProcessing.Enqueue(reachable);
}
ret.connect(hasState, newAddIndex, new ConnectionExactCharacter(hasState, newAddIndex, ch));
}
if (uninterestingConnections.Count > 0)
{
HashSet <int> reachable = new HashSet <int>();
foreach (Connection c in uninterestingConnections)
{
reachable.Add(c.end);
}
int newAddIndex = newStates.FindIndex(h => reachable.SetEquals(h));
if (newAddIndex == -1)
{
bool endState = false;
foreach (int i in reachable)
{
if (temp.endStates.Contains(i))
{
endState = true;
break;
}
}
newAddIndex = ret.addState(endState);
newStates.Add(reachable);
statesThatNeedProcessing.Enqueue(reachable);
}
if (interestingChars.Count > 0)
{
ret.connect(hasState, newAddIndex, new ConnectionAnythingBut(hasState, newAddIndex, interestingChars));
}
else
{
ret.connect(hasState, newAddIndex, new AnythingConnection(hasState, newAddIndex));
}
}
}
int garbageState = ret.addState(false);
ret.calculateConnectionPerState();
for (int i = 1; i < ret.numStates; i++)
{
HashSet <char> unUsableConnections = new HashSet <char>();
foreach (Connection c in ret.connectionPerState[i])
{
unUsableConnections.UnionWith(c.acceptSet());
}
bool canUseAll = true;
bool canUseAny = true;
HashSet <char> usableConnections = new HashSet <char>();
foreach (Connection c in ret.connectionPerState[i])
{
if (c is AnythingConnection)
{
canUseAny = false;
break;
}
if (c is ConnectionAnythingBut)
{
if (canUseAll)
{
usableConnections = new HashSet <char>(((ConnectionAnythingBut)c).forbidden);
canUseAll = false;
}
else
{
usableConnections.IntersectWith(((ConnectionAnythingBut)c).forbidden);
}
}
}
if (canUseAny)
{
if (!canUseAll)
{
usableConnections.ExceptWith(unUsableConnections);
if (usableConnections.Count > 0)
{
ret.connect(i, garbageState, new ConnectionAnyOf(i, garbageState, usableConnections));
}
}
else
{
if (unUsableConnections.Count > 0)
{
ret.connect(i, garbageState, new ConnectionAnythingBut(i, garbageState, unUsableConnections));
}
else
{
ret.connect(i, garbageState, new AnythingConnection(i, garbageState));
}
}
}
}
ret.connect(garbageState, garbageState, new AnythingConnection(garbageState, garbageState));
ret.calculateConnectionPerState();
return(ret);
}
