added grammar class ; added first set

lab/testGrNul.grm

@@ -0,0 +1,8 @@
+$S -> $A$B
+$A -> $C/a
+$A -> /~
+$B -> /c$Bpr
+$Bpr -> /a$A$C$Bpr
+$Bpr -> /~
+$C -> /b
+$C -> /~

src/grammar.py

@@ -0,0 +1,80 @@
+from terminal import Terminal
+from nonTerminal import NonTerminal
+
+class Grammar:
+    __atomKeyMapping = []
+    __productions = []
+    __first = {}
+
+    __nonTerminals = {}
+    __terminals = {}
+    
+    def addProduction(self, nonTerminal, atoms):
+
+
+        self.__addAtom(nonTerminal);
+
+        for atom in atoms:
+            self.__addAtom(atom)
+
+        key = nonTerminal.toString()
+        self.__nonTerminals[key].addProduction(atoms)
+
+    def __addAtom(self, atom):
+        if isinstance(atom, Terminal):
+            self.__addAtomToSet(self.__terminals, atom)
+        elif isinstance(atom, NonTerminal):
+            self.__addAtomToSet(self.__nonTerminals, atom)
+
+    def __addAtomToSet(self, atomSet, atom):
+        key = atom.toString()
+
+        if atomSet.get(key) == None:
+            atomSet[key] = atom
+
+
+    def toString(self):
+        retStr = ""
+
+        for key in self.__nonTerminals:
+            nonTerminal = self.__nonTerminals[key]
+
+            retStr += nonTerminal.productionsToString() + "\n";
+
+        return retStr
+
+    def generateMetrics(self):
+        self.__generateFirstSets()
+
+
+    def __generateFirstSets(self):
+        for key in self.__nonTerminals:
+            nonTerminal = self.__nonTerminals[key]
+            nonTerminal.generateFirstSet();
+
+
+    def printNullableSet(self):
+        output = "Nullable = { "
+
+        for key in self.__nonTerminals:
+            nonTerminal = self.__nonTerminals[key]
+
+            if nonTerminal.isNullable():
+                output += nonTerminal.toString() + " "
+
+        output += "}"
+
+        print(output)
+
+    def printFirstSets(self):
+        print("First Sets:")
+
+        for key in self.__nonTerminals:
+            nonTerminal = self.__nonTerminals[key]
+
+            nonTerminal.printFirstSet();
+
+        for key in self.__terminals:
+            terminal = self.__terminals[key]
+
+            terminal.printFirstSet();

src/grammarParser.py

@@ -2,6 +2,9 @@
 
 from grammarLexer import GrammarLexer
 from grammarLexer import T_TOKEN
+from grammar import Grammar
+from terminal import Terminal
+from nonTerminal import NonTerminal
 
 class GrammarParser:
     __lxr = GrammarLexer()
@@ -10,13 +13,14 @@ class GrammarParser:
 
     def parse(self, grammarFile):
         lines = grammarFile.readlines()
-        
+
         for line in lines:
 
             self.__lxr.processLine(line)
 
             tokenLine = self.__lxr.getParsedLine()
-            
+
+
             if self.__lineIsValid(tokenLine):
                 leftSide = tokenLine.popleft().getLexeme()
                 tokenLine.popleft()
@@ -30,8 +34,38 @@ class GrammarParser:
 
                 self.__grammar[leftSide].append(rightSide)
 
-        print(self.__grammar)
-    
+    def parseGrm(self, grammarFile):
+        grm = Grammar()
+        lines = grammarFile.readlines()
+
+        for line in lines:
+            self.__lxr.processLine(line)
+            tokenLine = self.__lxr.getParsedLine()
+
+            if self.__lineIsValid(tokenLine):
+                leftSide = NonTerminal(tokenLine.popleft().getLexeme())
+                tokenLine.popleft()
+
+                rightSide = []
+                tmpToken = None
+
+                while len(tokenLine) > 0:
+                    tmpToken = tokenLine.popleft()
+
+                    if tmpToken.getClass() == T_TOKEN.NON_TERMINAL:
+                        rightSide.append(NonTerminal(tmpToken.getLexeme()))
+                    elif tmpToken.getClass() == T_TOKEN.TERMINAL:
+                        lexeme = tmpToken.getLexeme();
+
+                        if lexeme == "~":
+                            lexeme = None;
+
+                        rightSide.append(Terminal(lexeme))
+
+                grm.addProduction(leftSide, rightSide)
+
+        return grm
+
 
     def __lineIsValid(self, tokenLine):
         if len(tokenLine) < 3:
@@ -42,7 +76,7 @@ class GrammarParser:
 
         if tokenLine[1].getClass() != T_TOKEN.ARROW:
             return False;
-        
+
         return True;
 
     def __isNewNonTerm(self, token):
@@ -54,7 +88,8 @@ class GrammarParser:
     def printGrammar(self):
         for left in self.__grammar:
             self.__printRow(left, self.__grammar[left]);
-    
+
+
     def __printRow(self, left, right):
         line = left + " -> "
 
@@ -67,4 +102,3 @@ class GrammarParser:
             line += str(token.getClass()) + "(" + token.getLexeme() + ")"
 
         print(line)
-         

src/nonTerminal.py

@@ -0,0 +1,117 @@
+from terminal import Terminal
+
+class NonTerminal:
+    def __init__(self, string):
+        self.__string = None
+        self.__isNullable = False
+        self.__productions = []
+        self.__first = []
+        self.__firstSetIsGenerated = False;
+
+        self.__string = string
+
+    def addProduction(self, rhs):
+        self.__firstSetIsGenrated = False;
+
+        if (len(rhs) == 1 and
+            isinstance(rhs[0], Terminal) and
+            rhs[0].isEmptyString()):
+
+            self.__isNullable = True
+
+        self.__productions.append(rhs)
+
+    def productionsToString(self):
+        if len(self.__productions) == 0:
+            return "N/A"
+
+        output = self.toString() + " -> "
+
+        for index in range(len(self.__productions) - 1):
+            production = self.__productions[index]
+
+            for atom in production:
+                output += atom.toString()
+
+            output += " | "
+
+        for atom in self.__productions[len(self.__productions) - 1]:
+            output += atom.toString()
+
+        return output
+
+    def printProductions(self):
+        print(self.productionsToString())
+
+    def generateFirstSet(self):
+        for production in self.__productions:
+            self.__addFirsSetForProduction(production)
+
+        self.__firstSetIsGenrated = True;
+
+    def __addFirsSetForProduction(self, production):
+        stillNullable = True
+        lenProd = len(production)
+        index = 0;
+
+        while stillNullable and index < lenProd:
+            atom = production[index]
+
+            if isinstance(atom, Terminal):
+                self.__addAtomToFirstSet(atom)
+                stillNullable = False
+            elif atom != self:
+                if atom.isNullable():
+                    if index < lenProd - 1:
+                        self.__copyFirstSetFrom(atom, False)
+                    else:
+                        self.__copyFirstSetFrom(atom, True)
+                else:
+                    self.__copyFirstSetFrom(atom, False)
+
+
+            index += 1;
+
+    def __addAtomToFirstSet(self, newAtom):
+        key = newAtom.toString()
+
+        for atom in self.__first:
+            if atom.toString() == key:
+                return
+
+        self.__first.append(newAtom)
+
+    def __copyFirstSetFrom(self, atom, includeEmptyString):
+
+        for innerAtom in atom.getFirstSet():
+            if isinstance(innerAtom, Terminal):
+                if innerAtom.isEmptyString():
+                    if includeEmptyString:
+                        self.__addAtomToFirstSet(innerAtom)
+                else:
+                    self.__addAtomToFirstSet(innerAtom)
+
+    def getFirstSet(self):
+        if self.__firstSetIsGenrated == False:
+            self.generateFirstSet();
+
+        return self.__first
+
+    def isNullable(self):
+        return self.__isNullable
+
+    def toString(self):
+        return "N(" + self.__string + ")"
+
+    def getString(self):
+        return self.__string
+
+    def printFirstSet(self):
+        output = "First("+ self.toString() +") = { "
+
+        for atom in self.__first:
+            output += atom.toString() + " ";
+
+        output += "}"
+
+        print(output)

src/terminal.py

@@ -0,0 +1,24 @@
+class Terminal:
+
+    def __init__(self, string):
+        self.__string = None
+        self.__isEmptyString = True
+
+        if string == "":
+            self.__string = ""
+        else:
+            self.__string = string
+            self.__isEmptyString = False
+
+
+    def getFirstSet(self):
+        return [self]
+
+    def isEmptyString(self):
+        return self.__isEmptyString
+
+    def toString(self):
+        return "T(" + self.__string + ")"
+
+    def printFirstSet(self):
+        print("First("+ self.toString() +") = { " + self.toString() + " }")

src/testParser.py

@@ -7,7 +7,16 @@ prs = GrammarParser()
 
 inp = input("grammar file: ");
 
-prs.parse(open(inp, "r"));
+grm = prs.parseGrm(open(inp, "r"));
 
-prs.printGrammar();
+print(grm.toString())
 
+grm.generateMetrics()
+
+print("----------------")
+grm.printNullableSet()
+
+print("----------------")
+grm.printFirstSets()
+
+#prs.printGrammar();