Made changes to scanner and grammar after going over around 3/4 of error...

Made changes to scanner and grammar after going over around 3/4 of error public test cases. (e.g. bitwise operations, integer range checking)

Scanning.py

@@ -85,7 +85,11 @@ wrongJavaKeyWordDict = {
     'transient', 
     'try', 
     'volatile',
-    '_'
+    '_',
+    'this',
+    'super',
+    'Long',
+    'Float'
 }
 
 # a set that contains unsed operators or seperators 
@@ -126,9 +130,6 @@ JoosDFATokens = set([
     'SUB',            # -
     'MULT',           # *
     'DIV',            # /
-    'BITAND',         # &
-    'BITOR',          # |
-    'EXP',            # ^
     'MOD',            # %
 
     # Separators:
@@ -142,9 +143,8 @@ JoosDFATokens = set([
     'COMMA',          # ,
     'PERIOD',         # .
 
-    # Unused Sep and Operators that we need to recognize
+    # Unused Seperator (TODO: refactor this)
     'ELLIPSIS',       # ...
-    'ARROW',          # ->
 
     ])
 
@@ -209,8 +209,6 @@ def JoosTransition(input, state):
             return 'BITAND'
         if (input == '|'):
             return 'BITOR'
-        if (input == '^'):
-            return 'EXP'
         if (input == '%'):
             return 'MOD'
 
@@ -240,6 +238,31 @@ def JoosTransition(input, state):
         if (input == '\''):
             return 'LCHAR'
 
+    # Handling all operators that are not allowed in Joos (some cases are handled elsewhere)
+    elif state == 'ADD':
+        if input == '+':
+            return 'OPDISCARD'
+        if input == '=':
+            return 'OPDISCARD'
+        return None
+    elif state == 'SUB':
+        if (input == '>'):
+            return 'OPDISCARD'
+        if input == '-':
+            return 'OPDISCARD'
+        if input == '=':
+            return 'OPDISCARD'
+        return None
+    elif state == 'MULT':
+        if input == '=':
+            return 'OPDISCARD'
+        return None
+    elif state == 'MOD':
+        if input == '=':
+            return 'OPDISCARD'
+        return None
+
+            
     elif (state == 'ID'):
         # dealing with compound names
         if input == '.':
@@ -267,6 +290,8 @@ def JoosTransition(input, state):
     elif (state == 'NUM'):
         if(input.isdigit()):
             return 'NUM'
+        if input == '.':
+            return 'FLOAT' # not accepted
         return None
 
     # string literal
@@ -277,6 +302,8 @@ def JoosTransition(input, state):
             return 'LITERALSTRING'
         return 'LSTRING'
     elif (state == 'STRINGESC'):
+        if input == 'u':
+            return 'UNICODE' #going to be discarded
         return 'LSTRING'
 
     # char literal
@@ -299,10 +326,14 @@ def JoosTransition(input, state):
     elif (state == 'GT'):
         if (input == '='):
             return 'GE'
+        elif input == '>':
+            return 'OPDISCARD'
         return None
     elif (state == 'LT'):
         if (input == '='):
             return 'LE'
+        elif input == '<':
+            return 'OPDISCARD'
         return None
     elif (state == 'NOT'):
         if (input == '='):
@@ -311,11 +342,16 @@ def JoosTransition(input, state):
     elif (state == 'BITAND'):
         if (input == '&'):
             return 'AND'
+        elif input == '=':
+            return 'OPDISCARD'
         return None
     elif (state == 'BITOR'):
         if (input == '|'):
             return 'OR'
+        elif input == '=':
+            return 'OPDISCARD'
         return None
+
       
     # Comments
     elif(state == 'DIV'):
@@ -323,6 +359,8 @@ def JoosTransition(input, state):
             return 'COMMENT'
         elif (input == '*'):
             return 'LCOMMENT'
+        elif input == '=':
+            return 'OPDISCARD'
         return None
     elif(state == 'LCOMMENT'):
         if (input == '*'):
@@ -354,12 +392,14 @@ def JoosTransition(input, state):
         if(input == '.'):
             return 'ELLIPSIS'
         return None
-    elif (state == 'SUB'):
-        if (input == '>'):
-            return 'ARROW'
-        return None
-
-
+    
+    # Handling hexidecimal literals (error case)
+    elif state == 'ZERO':
+        if input == 'x':
+            return 'HEXLITERAL'
+        elif input == '.':
+            return 'FLOAT'
+    
     else:
         return None
 
@@ -367,7 +407,7 @@ def JoosTransition(input, state):
 #TODO: remove alphabets since it's unecessary in our DFA implementation
 specialChars = set(list(".;:,@{}()[]<>!?+-*/&|^%=''\"\\"))
 JoosAccept = JoosDFATokens.union({'WHITESPACE', 'COMMENT', 'NEWLINE', 'LCOMMENT', 'RCOMMENT', 'NEWLINEC', 'LCOM2', 'LCOM3'})
-JoosStates = JoosAccept.union({'START', 'PERIOD2', 'HALFCOMP'})
+JoosStates = JoosAccept.union({'START', 'PERIOD2', 'HALFCOMP', 'HEXLITERAL', 'OPDISCARD', 'BITAND', 'BITOR', 'FLOAT', 'UNICODE'})
 JoosAlphabet = set(string.ascii_lowercase).union(set(string.ascii_uppercase)).union(set(string.digits)).union(specialChars)
 
 #########################  DFA  #######################################
@@ -431,12 +471,25 @@ def scan(input):
                 if index < indexRange-1:
                     if tokens[index+1].name == 'NUM':
                         return (None, "wrong integer literal: starts with 0")
+            # Checking integer range (does not cover all edge cases)
+            elif token.name == 'NUM' and index > 0 and tokens[index-1].name == 'SUB' and int(token.lex) > 2147483648:
+                return (None, "integer too small")
+            elif token.name == 'NUM' and int(token.lex) > 2147483647:
+                return (None, "interger too large")
+
+
             # dealing with keywords in Java but not in Joos
             elif token.name == 'ID' and token.lex in wrongJavaKeyWordDict:
                 return (None, "keyword in Java but not in Joos")
             # dealing with operators and seperators in Java but not in Joos
             elif token.name in wrongJavaOpDict:
                 return (None, "operator in Java but not in Joos")
+
+            # Checking wrong keywords in compIDs
+            elif token.name == 'COMPID':
+                temp = token.lex.split('.')
+                if temp[0] in wrongJavaKeyWordDict:
+                    return (None, "wrong keyword in comp id")
             
             # Checking if the multi line comment has a closing tag
             if token.name == 'LCOMMENT':

cfg/lr1GenInput.cfg

-64
+63
 EOF
 BOF
 INTERFACE
@@ -24,7 +24,6 @@ PUBLIC
 RETURN
 SHORT
 STATIC
-THIS
 VOID
 WHILE
 ID
@@ -160,7 +159,7 @@ primaryNoArrayAccess
 arrayID
 methodID
 start
-200
+199
 start BOF packageDcl importDcls topDcls EOF
 packageDcl PACKAGE name SEMICO
 packageDcl PACKAGE ID SEMICO
@@ -300,7 +299,6 @@ arrayCreationExpr NEW ID LSQRBRACK RSQRBRACK
 primary arrayAccess
 primary primaryNoArrayAccess 
 primaryNoArrayAccess literal
-primaryNoArrayAccess THIS
 primaryNoArrayAccess LPAREN expr RPAREN
 primaryNoArrayAccess classInstanceCreate
 primaryNoArrayAccess fieldAccess