Enable parsing lists

AST.py

@@ -22,13 +22,17 @@ class Node:
     def pop(self, index):
         return self.children.pop(index)
     
-    def _print(self, level):
-        print(f"{pad(level)}{self.__class__.__name__}({self.parent.__class__.__name__}):")
+    def _print(self, level):        
+        string = f"{pad(level)}{self.__class__.__name__}({self.parent.__class__.__name__}):\n"
         for child in self.children:
             if isinstance(child, str) or isinstance(child, int) or isinstance(child, Note):
-                print(pad(level+1) + f"'{child}'")           
+                string += pad(level+1) + f"'{child}'\n"
             else:
-                child._print(level+1)
+                string += child._print(level+1)
+        return string
+    
+    def __str__(self):
+        return self._print(0)
     
 def pad(level):
     return ("   " * level)
@@ -37,26 +41,26 @@ class Program(Node):
     def __init__(self):
         Node.__init__(self, None, (-1, -1))        
     
-    def __str__(self):
-        return "Program:\n" + "\n".join([str(e) for e in self.children])
+    #def __str__(self):
+        #return "Program:\n" + "\n".join([str(e) for e in self.children])
     
     def print(self):
-        self._print(0)
+        print(self._print(0))
 
 class BlockNode(Node):
     def __init__(self, parent, pos):      
         Node.__init__(self, parent, pos)              
     
-    def __str__(self):
-        return "B{\n" + "\n".join([str(e) for e in self.children]) + "\n}"
+    #def __str__(self):
+        #return "B{\n" + "\n".join([str(e) for e in self.children]) + "\n}"
 
 
 class ListNode(Node):
     def __init__(self, parent, pos):
         Node.__init__(self, parent, pos)              
     
-    def __str__(self):
-        return "@(" + ", ".join([str(e) for e in self.children]) + ")"      
+    #def __str__(self):
+        #return "@(" + ", ".join([str(e) for e in self.children]) + ")"      
         
 class IdentifierNode(Node):
     def __init__(self, identifier, parent, pos): 
@@ -65,8 +69,8 @@ class IdentifierNode(Node):
         
         self.identifier = self.children[0]
     
-    def __str__(self):
-        return f"L'{self.identifier}'"
+    #def __str__(self):
+        #return f"L'{self.identifier}'"
         
 class AssignExpression(Node):
     def __init__(self, target, value, parent, pos):
@@ -76,8 +80,8 @@ class AssignExpression(Node):
         self.target = self.children[0]
         self.value = self.children[1]
         
-    def __str__(self):
-        return f"A[{self.target} = {self.value}]"
+    #def __str__(self):
+        #return f"A[{self.target} = {self.value}]"
    
 class AsteriskStatementNode(Node):
     def __init__(self, iterator, statement, parent, pos):
@@ -87,8 +91,8 @@ class AsteriskStatementNode(Node):
         self.iterator = self.children[0]
         self.statement = self.children[1]
         
-    def __str__(self):
-        return f"*({self.iterator}: {self.statement})"
+    #def __str__(self):
+        #return f"*({self.iterator}: {self.statement})"
    
 class ColonNode(Node):
     def __init__(self, a, b, parent, pos):
@@ -98,16 +102,15 @@ class ColonNode(Node):
         self.a = self.children[0]
         self.b = self.children[1]
         
-    def __str__(self):
-        return f":({self.a}, {self.b})"
+    #def __str__(self):
+        #return f":({self.a}, {self.b})"
    
 class ExpressionNode(Node):
     def __init__(self, parent, pos):
         Node.__init__(self, parent, pos)   
         
-    def __str__(self):
-        return f"{self.__class__.__name__}('{self.value}')"
-
+    #def __str__(self):
+        #return f"{self.__class__.__name__}('{self.value}')"
 
 class IntegerLiteralNode(ExpressionNode):
     def __init__(self, value, parent, pos):
@@ -116,8 +119,8 @@ class IntegerLiteralNode(ExpressionNode):
         
         self.value = self.children[0]
     
-    def __str__(self):
-        return f"i'{self.value}'"
+    #def __str__(self):
+        #return f"i'{self.value}'"
 
 class StringLiteralNode(ExpressionNode):
     def __init__(self, value, parent, pos):
@@ -126,8 +129,8 @@ class StringLiteralNode(ExpressionNode):
         
         self.value = self.children[0]
 
-    def __str__(self):
-        return f"s'{self.value}'"
+    #def __str__(self):
+        #return f"s'{self.value}'"
     
 class NoteLiteralNode(ExpressionNode):
     def __init__(self, value, parent, pos):
@@ -136,8 +139,8 @@ class NoteLiteralNode(ExpressionNode):
         
         self.value = self.children[0]
         
-    def __str__(self):
-        return f"n'{self.value.note}[{self.value.octave}, {self.value.duration}]'"
+    #def __str__(self):
+        #return f"n'{self.value.note}[{self.value.octave}, {self.value.duration}]'"
 
 class FunctionCallNode(Node):
     def __init__(self, identifier, arguments, parent, pos):
@@ -147,15 +150,15 @@ class FunctionCallNode(Node):
         self.identifier = self.children[0]
         self.arguments = self.children[1]
         
-    def __str__(self):
-        return f"F({self.identifier}: {self.arguments})"
+    #def __str__(self):
+        #return f"F({self.identifier}: {self.arguments})"
 
 class CommaNode(Node):    
     def __init__(self, parent, pos):
         Node.__init__(self, parent, pos)   
         
-    def __str__(self):
-        return "[,]"
+    #def __str__(self):
+        #return "[,]"
 
 class PercentNode(Node):
     def __init__(self, value, parent, pos):        
@@ -164,8 +167,8 @@ class PercentNode(Node):
         
         self.value = self.children[0]
         
-    def __str__(self):
-        return f"%'{self.value}'"
+    #def __str__(self):
+        #return f"%'{self.value}'"
 
 class FunctionDefinitionNode(Node):
     def __init__(self, name, parameters, body, parent, pos):
@@ -176,8 +179,8 @@ class FunctionDefinitionNode(Node):
         self.parameters = self.children[1]
         self.body = self.children[2]
     
-    def __str__(self):
-        return f"$F'{self.name}{self.parameters}{self.body}"
+    #def __str__(self):
+        #return f"$F'{self.name}{self.parameters}{self.body}"
     
 class ReturnNode(Node):
     def __init__(self, value, parent, pos):
@@ -186,5 +189,15 @@ class ReturnNode(Node):
         
         self.value = self.children[0]
     
-    def __str__(self):
-        return f"Ret({self.value})"
+    #def __str__(self):
+        #return f"Ret({self.value})"
+    
+class ListItemNode(Node):
+    def __init__(self, value, parent, pos):
+        Node.__init__(self, parent, pos)
+        self.children.append(value)
+        
+        self.value = self.children[0]
+        
+class CloseListNode(Node):
+    pass

Parser.py

@@ -1,230 +1,117 @@
-from Tokenizer import *
-from Note import *
 from AST import *
+from Tokenizer import TokenType
 from Error import SyntaxException
 
-def expectedFound(expected, found):
-    raise SyntaxException(None, f"Expected: {expected}, found: {found}")
+def assertToken(expected, input):
+    if expected != input.current().type:
+        raise SyntaxException(input.current().pos, f"Expected '{expected}', found '{input.current().value}'")
 
-def assertType(expected, found):
-    if expected != found:
-        raise SyntaxException(None, f"Expected: {expected}, found: {found}")
+def runParsers(input, parent, parsers):
+    for parser in parsers:
+        value = parser(input, parent)
+        if value is not None:
+            return value
+    return None
 
-def parseInteger(input, parent):
-    token = input.pop(0)
-    return IntegerLiteralNode(int(token.value), parent, token.pos)
-    
-def parseString(input, parent):   
-    token = input.pop(0)
-    return StringLiteralNode(token.value[1:-1], parent, token.pos)
-    
-def parseNote(input, parent): 
-    token = input.pop(0)
-    value = token.value
-    consumedChars = 1
-    notePitch = value[consumedChars]
-    consumedChars += 1
-    octave = 4
-    duration = 4
-    dot = False
-    if consumedChars < len(value) and value[consumedChars] in ('b', '#'):
-        notePitch += value[consumedChars]
-        consumedChars += 1
-    if consumedChars < len(value) and re.match(r'\d', value[consumedChars]):
-        octave = int(value[consumedChars])
-        consumedChars += 1
-    if consumedChars < len(value) and value[consumedChars] == '.':
-        consumedChars += 1
-        durationString = ''            
-        while consumedChars < len(value) and re.match(r'\d', value[consumedChars]):
-            durationString += value[consumedChars]      
-            consumedChars += 1  
-            duration = int(durationString)
-        if consumedChars < len(value) and value[consumedChars] == '.':
-            dot = True
-            consumedChars += 1
-    
-    return NoteLiteralNode(Note(notePitch, octave, duration, dot), parent, token.pos)
+def returnAndGoAhead(input, getValue):
+    value = getValue(input.current())
+    input.ahead()
+    return value
 
-def parseComma(input, parent):
-    token = input.pop(0)
-    return CommaNode(parent, token.pos)
-
-def parseList(input, parent):       
-    token = input.pop(0)
-    
-    node = ListNode(parent, token.pos)
-    
-    while input[0].type != TokenType.CLOSE_PAREN:  
-        element = parseArrayElement(input, node)
-        if element is None:
-            raise SyntaxException(input[0].pos, "Invalid element '{input[0].value}'")
-        node.append(element)
-    
-    if input[0].type != TokenType.CLOSE_PAREN:
-        expectedFound(TokenType.CLOSE_PAREN, input[0].type)
-    input.pop(0)
-    
-    return node
-            
-def parseBlock(input, parent):
-    token = input.pop(0)
-    
-    block = BlockNode(parent, token.pos)
-    
-    while input[0].type != TokenType.CLOSE_BRACKET:
-        block.append(parseToken(input, block))
-    
-    if input[0].type != TokenType.CLOSE_BRACKET:
-        expectedFound(TokenType.CLOSE_BRACKET, input[0].type)
-    input.pop(0)
-    
-    return block
-
-
-def parseAsterisk(input, parent):
-    token = input.pop(0)
-    
-    iterator = parent.pop(-1)
-    value = parseStatement(input, parent)
-    
-    asterisk = AsteriskStatementNode(iterator, value, parent, token.pos)        
-    iterator.parent = asterisk
-    value.parent = asterisk
-    return asterisk
-   
-def parseNoteOrColon(input, parent):
-    note = parseNote(input, parent)    
-    if len(input) > 1 and input[0].type == TokenType.COLON:
-        token = input.pop(0)                
-        b = parseNote(input, parent)
-        if b is None:
-            raise SyntaxException(input[0].pos, f"Invalid colon argument '{input[0].value}'")
-        colon = ColonNode(note, b, parent, token.pos)
-        note.parent = colon
-        b.parent = colon
-        return colon
-    
-    return note
-   
-def parseIntegerOrColonOrPercent(input, parent):
-    integer = parseInteger(input, parent)    
-    if len(input) > 1 and input[0].type == TokenType.COLON:
-        token = input.pop(0)                
-        b = parseInteger(input, parent) 
-        if b is None:
-            raise SyntaxException(input[0].pos, f"Invalid colon argument '{input[0].value}'")
-        colon = ColonNode(integer, b, parent, token.pos)
-        integer.parent = colon
-        b.parent = colon
-        return colon
-    
-    if len(input) > 0 and input[0].type == TokenType.PERCENT:
-        input.pop(0)
-        percent = PercentNode(integer, parent, integer.pos)
-        integer.parent = percent
-        return percent
-    
-    return integer   
-   
-def parseFunctionCallOrAssignOrIdentifier(input, parent):   
-    token = input.pop(0)
-    identifier = IdentifierNode(token.value, parent, token.pos)
-    # Function call
-    if len(input) > 0 and input[0].type == TokenType.OPEN_PAREN:            
-        arguments = parseList(input, parent)        
-        func = FunctionCallNode(identifier, arguments, parent, token.pos)
-        identifier.parent = func
-        arguments.parent = func
-        return func
-    # Assign
-    if len(input) > 1 and input[0].type == TokenType.ASSIGN:          
-        token = input.pop(0)            
-        value = parseExpression(input, parent) #
-        assign = AssignExpression(identifier, value, parent, token.pos)
-        identifier.parent = assign
-        value.parent = assign
-        return assign
+# int -> INTEGER
+def parseInteger(input, parent):    
+    if input.current().type == TokenType.INTEGER:
+        integer = IntegerLiteralNode(input.current().value, parent, input.current().pos)
+        input.ahead()
         
-    return identifier
+        return integer
+    return None
 
-def parseMinus(input, parent):
-    token = input.pop(0)
-    
-    value = parseInteger(input, parent)
-    
-    return IntegerLiteralNode(-value.value, parent, token.pos)
+# list -> CLOSE_PAREN | expr listTail
+def parseList(input, parent):      
+    if input.current().type == TokenType.OPEN_PAREN:
+        node = ListNode(parent, input.current().pos)                
+        input.ahead()
+        
+        # list -> CLOSE_PAREN (end of list)
+        if input.current().type == TokenType.CLOSE_PAREN:
+            close = CloseListNode(node, input.current().pos)
+            node.append(close)
+            input.ahead()
+            return node                
+        
+        # list -> expr listTail
+        token = input.current()
+        expr = parseExpression(input, node)
+        item = ListItemNode(expr, node, token.pos)
+        node.append(item)
+        listTail = parseListTail(input, item)
+        item.append(listTail)
+        
+        #while input.current().type != TokenType.CLOSE_PAREN:                        
+            #elem = parseListTail(input, node)            
+            #if elem is None:
+                #raise SyntaxException(input.current().pos, f"Invalid element '{input.current().value}'")
+            #node.append(elem)
+        return node
+    return None
+            
 
-def parseFunctionDefinition(input, parent):
-    input.pop(0)
+# listTail -> COMMA expr listTail | CLOSE_PAREN
+def parseListTail(input, parent):        
+    # listTail -> CLOSE_PAREN
+    if input.current().type == TokenType.CLOSE_PAREN:   
+        close = CloseListNode(parent, input.current().pos)
+        input.ahead()
+        return close
     
-    assertType(TokenType.IDENTIFIER, input[0].type)
-    token = input.pop(0)
-    name = IdentifierNode(token.value, parent, token.pos)
+    assertToken(TokenType.COMMA, input)
+    input.ahead()
+        
+    expr = parseExpression(input, parent)        
+    if expr is not None: 
+        item = ListItemNode(expr, parent, expr.pos)            
+        expr.parent = item           
+        listTail = parseListTail(input, item)
+        item.append(listTail)
+        listTail.parent = item
+        input.ahead()
+        return item
     
-    assertType(TokenType.OPEN_PAREN, input[0].type)
-    parameters = parseList(input, parent)
-    
-    assertType(TokenType.OPEN_BRACKET, input[0].type)
-    body = parseBlock(input, parent)
-    
-    func = FunctionDefinitionNode(name, parameters, body, parent, token.pos)
-    name.parent = func
-    parameters.parent = func
-    body.parent = func
-    return func
+    return None
 
-def parseReturn(input, parent):
-    token = input.pop(0)
+def parseExpression(input, parent):
+    value = runParsers(input, parent, [
+        parseInteger,
+        parseList        
+    ])
     
-    value = parseExpression(input, parent)
+    if value is None:
+        raise SyntaxException(input.current().pos, f"Expression expected")
+    
+    return value
     
-    returnNode = ReturnNode(value, parent, token.pos)
-    value.parent = returnNode
-    return returnNode
 
-def parseExpression(input, parent):    
-    type = input[0].type
-    if type == TokenType.FUNCTION:
-        return parseFunctionDefinition(input, parent)
-    if type == TokenType.RETURN:
-        return parseReturn(input, parent)
-    if type == TokenType.MINUS:
-        return parseMinus(input, parent)
-    if type == TokenType.INTEGER:
-        return parseIntegerOrColonOrPercent(input, parent)
-    if type == TokenType.STRING:
-        return parseString(input, parent)    
-    if type == TokenType.NOTE:
-        return parseNoteOrColon(input, parent)    
-    if type == TokenType.IDENTIFIER:
-        return parseFunctionCallOrAssignOrIdentifier(input, parent)    
-    if type == TokenType.OPEN_PAREN:
-        return parseList(input, parent)     
-    raise SyntaxException(input[0].pos, f"Unexpected character '{input[0].value}'")
- 
-def parseArrayElement(input, parent):
-    type = input[0].type
-    if type == TokenType.COMMA:
-        return parseComma(input, parent)
-    return parseExpression(input, parent)
- 
-def parseStatement(input, parent):
-    type = input[0].type
-    if type == TokenType.OPEN_BRACKET:
-        return parseBlock(input, parent)
-    if type == TokenType.ASTERISK:
-        return parseAsterisk(input, parent)
+def parseToken(input, parent):
+    value = runParsers(input, parent, [
+        parseExpression
+    ])
     
-    return parseExpression(input, parent)
+    if value is None:
+        raise SyntaxException(input.current().pos, "Unknown statement")
     
-def parseToken(input, parent):      
-    #import pdb; pdb.set_trace()
-    return parseStatement(input, parent)    
-    
-  
-def parse(input):    
+    return value
+
+def parse(input):
     root = Program()
-    while len(input) > 0:
-        root.append(parseToken(input, root))
+    while input.notParsedTokensRemain():
+        root.append(parseToken(input, root))    
     return root
+
+
+
+
+
+
+def parseNote(input, parent):
+    pass

Tokenizer.py

@@ -4,6 +4,51 @@ import re
 import sys
 from Error import SyntaxException
 
+class Tokens:
+    def __init__(self, tokens = []):
+        self.tokens = tokens
+        self.cursor = 0
+        self.snap = 0
+        
+    def append(self, token):
+        self.tokens.append(token)
+        
+    def __getitem__(self, index):
+        return self.tokens[index]
+    
+    def current(self):
+        if self.cursor >= len(self.tokens):
+            raise RuntimeError(f"Cursor points to not existing token! Cursor = {self.cursor}, len = {len(self.tokens)}")
+        return self.tokens[self.cursor]
+    
+    def next(self):
+        return self.tokens[self.cursor+1]
+    
+    def prev(self):
+        return self.tokens[self.cursor-1]
+    
+    def hasMore(self, count):
+        return self.cursor + count < len(self.tokens)
+    
+    def ahead(self):
+        self.cursor += 1        
+    
+    def snapshot(self):
+        self.snapshot = self.cursor
+        
+    def reset(self):
+        self.cursor = self.snapshot
+        return self.tokens[self.cursor]
+    
+    def notParsedTokensRemain(self):
+        return self.cursor < len(self.tokens)
+    
+    def __str__(self):
+        return f"[Cursor: {self.cursor}\n{', '.join([str(token) for token in self.tokens])}]"
+    
+    def __repr__(self):
+        return self.__str__()
+
 class TokenType(Enum):
     OPEN_PAREN = 1
     CLOSE_PAREN = 2
@@ -189,7 +234,7 @@ tokenizers = (
 )
 
 def doTokenize(lines):    
-    tokens = []         
+    tokens = []     
     for lineNumber, line in enumerate(lines):    
         current = 0
         while current < len(line):
@@ -209,4 +254,4 @@ def doTokenize(lines):
 
 def tokenize(lines):
     tokens = doTokenize(lines)
-    return list(filter(lambda x: x.type != TokenType.COMMENT, tokens))
+    return Tokens([ token for token in tokens if token.type != TokenType.COMMENT])