added ast definition and parser

README.md

@@ -4,3 +4,5 @@ IMProved aims to build an interpreter for IMP (based on
 [this](http://jayconrod.com/posts/37/a-simple-interpreter-from-scratch-in-python-part-1))
 that is fully documented, occassionally improved and
 embeddable into Python.
+
+As you may notice, I am bad at documentation.

improved/ast.py

@@ -0,0 +1,422 @@
+"""The AST definition"""
+import operator
+
+class Equality(object):
+    """
+        A baseclass for all AST nodes.
+        It implements basic checks for equality.
+    """
+    def __eq__(self, o):
+        """
+            Checks for equality. Returns True when the
+            class dictionaries are the same and the
+            classes are based on the same baseclass.
+            Otherwise returns False.
+
+            o -- the other object
+
+            returns -- a Boolean
+        """
+        return isinstance(o, self.__class__) and self.__dict__ == o.__dict__
+
+    def __ne__(self, o):
+        """
+            Checks for inequality.
+            Simply negates the equality check.
+
+            o -- the other object
+
+            returns -- a Boolean
+        """
+        return not self.__eq__(o)
+
+class Statement(Equality):
+    """The baseclass of all statements."""
+    pass
+
+class ArithmeticExp(Equality):
+    """The baseclass of all arithmetic expressions."""
+    pass
+
+class BooleanExp(Equality):
+    """The baseclass of all boolean expressions."""
+    pass
+
+class IntArithmeticExp(ArithmeticExp):
+    """The AST node for integer arithmetic expressions."""
+    def __init__(self, i):
+        """
+            The initialization method.
+
+            i -- the integer value
+        """
+        self.i = i
+
+    def __repr__(self):
+        """
+            A representation of the node for debug purposes.
+
+            returns -- a String of the form "IntArithmeticStatement(i)"
+        """
+        return 'IntArithmeticExp(%d)' % self.i
+
+    def eval(self, env):
+        """
+            Evaluates the node. Returns the value.
+
+            env -- the environment in which to evaluate the node
+
+            returns -- an Integer
+        """
+        return self.i
+
+class VarArithmeticExp(ArithmeticExp):
+    """The AST node for all variable arithmetic expressions."""
+    def __init__(self, name):
+        """
+            The initialization method.
+
+            name -- the variable name
+        """
+        self.name = name
+
+    def __repr__(self):
+        """
+            A representation of the node for debug purposes.
+
+            returns -- a String of the form "VarArithmeticStatement(name)"
+        """
+        return 'VarArithmeticExp(%s)' % self.name
+
+    def eval(self, env):
+        """
+            Evaluates the node. Returns the variable's value.
+
+            env -- the environment in which to evaluate the node
+
+            returns -- the variable's value
+        """
+        return env.get(self.name, 0)
+
+class BinArithmeticExp(ArithmeticExp):
+    """The AST node for all binary arithmetic expressions."""
+
+    operators = {
+        '+': operator.add,
+        '-': operator.sub,
+        '*': operator.mul,
+        '/': operator.truediv,
+        '**': operator.pow,
+    }
+
+    def __init__(self, op, left, right):
+        """
+            The initialization method.
+
+            op    -- the operation to perform
+            left  -- the left operand
+            right -- the right operand
+        """
+        self.op = op
+        self.left = left
+        self.right = right
+
+    def __repr__(self):
+        """
+            A representation of the node for debug purposes.
+
+            returns -- a String of the form
+                       "BinArithmeticStatement(operation, left, right)"
+        """
+        return 'BinArithmeticExp(%s, %s, %s)' % (self.op, self.left, self.right)
+
+    def eval(self, env):
+        """
+            Evaluates the node. Evaluates the left and right operands
+            and performs the arithmetic epxression on them.
+
+            env -- the environment in which to evaluate the node
+
+            returns -- the evaluated value
+        """
+        left_value = self.left.eval(env)
+        right_value = self.right.eval(env)
+        try:
+            return operators[self.op](left_value, right_value)
+        except KeyError:
+            raise RuntimeError('unknown operator: ' + self.op)
+
+class RelationExp(BooleanExp):
+    """The AST node for relational boolean expressions."""
+
+    operators = {
+        '<': operator.gt,
+        '<=': operator.ge,
+        '>': operator.lt,
+        '>=': operator.le,
+        '=': operator.eq,
+        '!=': operator.ne,
+    }
+
+    def __init__(self, op, left, right):
+        """
+            The initialization method.
+
+            op    -- the operation to perform
+            left  -- the left operand
+            right -- the right operand
+        """
+        self.op = op
+        self.left = left
+        self.right = right
+
+    def __repr__(self):
+        """
+            A representation of the node for debug purposes.
+
+            returns -- a String of the form
+                       "RelationExp(operation, left, right)"
+        """
+        return 'RelationExp(%s, %s, %s)' % (self.op, self.left, self.right)
+
+    def eval(self, env):
+        """
+            Evaluates the node. Evaluates the left and right operands
+            and performs the relational epxression on them.
+
+            env -- the environment in which to evaluate the node
+
+            returns -- the evaluated value
+        """
+        left_value = self.left.eval(env)
+        right_value = self.right.eval(env)
+        try:
+            return operators[self.op](left_value, right_value)
+        except KeyError:
+            raise RuntimeError('unknown operator: ' + self.op)
+
+class AndExp(BooleanExp):
+    """The AST node for and expressions"""
+    def __init__(self, left, right):
+        """
+            The initializer method.
+
+            left  -- the left operand
+            right -- the right operand
+        """
+        self.left = left
+        self.right = right
+
+    def __repr__(self):
+        """
+            A representation of the node for debug purposes.
+
+            returns -- a String of the form "AndExp(left, right)"
+        """
+        return 'AndExp(%s, %s)' % (self.left, self.right)
+
+    def eval(self, env):
+        """
+            Evaluates the node. Evaluates the left and right operands
+            and performs a logical and on them.
+            Supports short circuiting of logic.
+
+            env -- the environment in which to evaluate the node
+
+            returns -- the evaluated value
+        """
+        return self.left.eval(env) and self.right.eval(env)
+
+class OrExp(BooleanExp):
+    """The AST node for or expressions"""
+    def __init__(self, left, right):
+        """
+            The initializer method.
+
+            left  -- the left operand
+            right -- the right operand
+        """
+        self.left = left
+        self.right = right
+
+    def __repr__(self):
+        """
+            A representation of the node for debug purposes.
+
+            returns -- a String of the form "OrExp(left, right)"
+        """
+        return 'OrExp(%s, %s)' % (self.left, self.right)
+
+    def eval(self, env):
+        """
+            Evaluates the node. Evaluates the left and right operands
+            and performs a logical or on them.
+            Supports short circuiting of logic.
+
+            env -- the environment in which to evaluate the node
+
+            returns -- the evaluated value
+        """
+        return self.left.eval(env) or  self.right.eval(env)
+
+class NotExp(BooleanExp):
+    """The AST node for not expressions"""
+    def __init__(self, exp):
+        """
+            The initializer method.
+
+            exp -- the operand to negate
+        """
+        self.exp = exp
+
+    def __repr__(self):
+        """
+            A representation of the node for debug purposes.
+
+            returns -- a String of the form "NotExp(operand)"
+        """
+        return 'NotExp(%s)' % self.exp
+
+    def eval(self, env):
+        """
+            Evaluates the node. Evaluates the operand and  performs a
+            logical not on it.
+
+            env -- the environment in which to evaluate the node
+
+            returns -- the evaluated value
+        """
+        return not self.exp.eval(env)
+
+class AssignStatement(Statement):
+    """The AST node for assignments"""
+    def __init__(self, name, rexp):
+        """
+            The initialization method.
+
+            name -- the variable name
+            rexp -- the expression that evaluates to the name
+        """
+        self.name = name
+        self.rexp = rexp
+
+    def __repr__(self):
+        """
+            A representation of the node for debug purposes.
+
+            returns -- a String of the form "AssignStatement(name, expression)"
+        """
+        return 'AssignStatement(%s, %s)' % (self.name, self.rexp)
+
+    def eval(self, env):
+        """
+            Evaluates the node. Evaluates the statement
+            for the expression and saves it within the
+            environment (under the given name).
+
+            env -- the environment in which to evaluate the node
+        """
+        env[self.name] = self.rexp.eval(env)
+
+
+class CompoundStatement(Statement):
+    """The AST node for compound statements"""
+    def __init__(self, first, second):
+        """
+            The initialization method.
+
+            first  -- the first statement
+            second -- the second statement
+        """
+        self.first = first
+        self.second = second
+
+    def __repr__(self):
+        """
+            A representation of the node for debug purposes.
+
+            returns -- a String of the form "CompoundStatement(first, second)"
+        """
+        return 'CompoundStatement(%s, %s)' % (self.first, self.second)
+
+    def eval(self, env):
+        """
+            Evaluates the node. Evaluates the statements
+            one after the other.
+
+            env -- the environment in which to evaluate the node
+        """
+        self.first.eval(env)
+        self.second.eval(env)
+
+
+class IfStatement(Statement):
+    """The AST node for if statements"""
+    def __init__(self, condition, on_true, on_false):
+        """
+            The initialization method.
+
+            condition -- the condition that needs to be evaluated
+            on_true   -- the statement that is to be executed when
+                         the condition is met
+            on_false  -- the statement that is to be executed when
+                         the condition is not met
+        """
+        self.condition = condition
+        self.on_true = on_true
+        self.on_false = on_false
+
+    def __repr__(self):
+        """
+            A representation of the node for debug purposes.
+
+            returns -- a String of the form "IfStatement(condition, on_true, on_false)"
+        """
+        return 'IfStatement(%s, %s, %s)' % (self.condition, self.on_true, self.on_false)
+
+    def eval(self, env):
+        """
+            Evaluates the node. Evaluates the condition.
+            If it is met, the on_true block is executed.
+            If not and the on_false block exists, it is executed.
+
+            env -- the environment in which to evaluate the node
+        """
+        if self.condition.eval(env):
+            self.on_true.eval(env)
+        else:
+            if self.on_false:
+                self.on_false.eval(env)
+
+
+class WhileStatement(Statement):
+    """The AST node for while statements"""
+    def __init__(self, condition, body):
+        """
+            The initialization method.
+
+            condition -- the condition that needs to be evaluated
+            body      -- the statement that is to be executed while
+                         the condition is met
+        """
+        self.condition = condition
+        self.body = body
+
+    def __repr__(self):
+        """
+            A representation of the node for debug purposes.
+
+            returns -- a String of the form
+                       "WhileStatement(condition, body)"
+        """
+        return 'WhileStatement(%s, %s)' % (self.condition, self.body)
+
+    def eval(self, env):
+        """
+            Evaluates the node. Evaluates the condition.
+            While it is met, the body block is executed.
+
+            env -- the environment in which to evaluate the node
+        """
+        while self.condition.eval(env):
+            self.body.eval(env)

improved/parser.py

@@ -0,0 +1,130 @@
+"""The parser"""
+
+from functools import reduce
+
+from .ast import *
+from .parser_combinator import *
+from .tokenize import *
+
+ARITHMETIC_PRECEDENCE = [
+    ['*', '/'],
+    ['+', '-'],
+]
+
+BOOLEAN_PRECEDENCE = [
+    ['and'],
+    ['or'],
+]
+
+#Helper
+def precedence(value_parser, precedence_levels, combine):
+    def op_parser(precedence_level):
+        return any_op_in_list(precedence_level) ^ combine
+    parser = value_parser * op_parser(precedence_levels[0])
+    for precedence_level in precedence_levels[1:]:
+        parser = parser * op_parser(precedence_level)
+    return parser
+
+def process_binop(op):
+    return lambda l, r: BinArithmeticExp(op, l, r)
+
+def process_relop(parsed):
+    ((left, op), right) = parsed
+    return RelationExp(op, left, right)
+
+def process_logic(op):
+    if op == 'and':
+        return lambda l, r: AndExp(l, r)
+    elif op == 'or':
+        return lambda l, r: OrExp(l, r)
+    else:
+        raise RuntimeError('unknown logic operator: ' + op)
+
+def process_group(parsed):
+    ((_, p), _) = parsed
+    return p
+
+def any_op_in_list(ops):
+    op_parsers = [keyword(op) for op in ops]
+    parser = reduce(lambda l, r: l | r, op_parsers)
+    return parser
+
+#Parser
+num = Tag(INT) ^ (lambda i: int(i))
+imp_id = Tag(ID)
+
+def keyword(kw):
+    return Reserved(kw, RESERVED)
+
+def arithmetic_group():
+    return keyword('(') + Lazy(arithmetic_exp) + keyword(')') ^ process_group
+
+def arithmetic_value():
+    return ((num ^ (lambda i: IntArithmeticExp(i))) |
+            (id  ^ (lambda v: VarArithmeticExp(v))))
+
+def arithmetic_term():
+    return arithmetic_value() | arithmetic_group()
+
+def arithmetic_exp():
+    return precedence(arithmetic_term(),
+                      ARITHMETIC_PRECEDENCE,
+                      process_binop)
+
+def boolean_not():
+    return keyword('not') + Lazy(boolean_term) ^ (lambda parsed: NotExp(parsed[1]))
+
+def boolean_relop():
+    relops = ['<', '<=', '>', '>=', '=', '!=']
+    return arithmetic_exp() + any_op_in_list(relops) + arithmetic_exp() ^ process_relop
+
+def boolean_group():
+    return keyword('(') + Lazy(boolean_exp) + keyword(')') ^ process_group
+
+def boolean_term():
+    return boolean_not() | boolean_relop() | boolean_group()
+
+def boolean_exp():
+    return precedence(boolean_term(),
+                      BOOLEAN_PRECEDENCE,
+                      process_logic)
+
+def assign_statement():
+    def internal(parsed):
+        ((name, _), exp) = parsed
+        return AssignStatement(name, exp)
+    return imp_id + keyword(':=') + arithmetic_exp() ^ internal
+
+def if_statement():
+    def internal(parsed):
+        (((((_, condition), _), on_true), false_parsed), _) = parsed
+        if false_parsed:
+            (_, on_false) = false_parsed
+        else:
+            on_false = None
+        return IfStatement(condition, on_true, on_false)
+    return (keyword('if') + boolean_exp()
+            + keyword('then') + Lazy(statements)
+            + Opt(keyword('else') + Lazy(statements))
+            + keyword('end') ^ internal)
+
+def while_statement():
+    def internal(parsed):
+        ((((_, condition), _), body), _) = parsed
+        return WhileStatement(condition, body)
+    return (keyword('while') + boolean_exp()
+            + keyword('do') + Lazy(statements)
+            + keyword('end') ^ internal)
+
+def statement():
+    return assign_statement() | if_statement() | while_statement()
+
+def statements():
+    sep = keyword(';') ^ (lambda x: lambda l, r: CompoundStatement(l, r))
+    return Exp(statement(), sep)
+
+def parser():
+    return Phrase(statements())
+
+def imp_parser(tokens):
+    return parser()(tokens, 0)

improved/parser_combinator.py

@@ -155,7 +155,7 @@ class Concat(Parser):
                 return Result(combined_value, right_result.pos)
         return None
 
- class Alternate(Parser):
+class Alternate(Parser):
     """The alternate combinator. Parses using either of two parsers."""
     def __init__(self, left, right):
         """