128 lines
3.4 KiB
Python
128 lines
3.4 KiB
Python
"""Recursive-descent parser for calc expressions.
|
|
|
|
Grammar:
|
|
expr := term (('+' | '-') term)*
|
|
term := factor (('*' | '/') factor)*
|
|
factor := '-' factor | primary
|
|
primary := NUMBER | '(' expr ')'
|
|
|
|
Node shapes:
|
|
Num(value) — a numeric literal; value is int or float
|
|
BinOp(op, left, right) — op is one of '+' '-' '*' '/'
|
|
Unary(op, operand) — op is '-'
|
|
"""
|
|
|
|
|
|
class ParseError(Exception):
|
|
pass
|
|
|
|
|
|
class Num:
|
|
__slots__ = ('value',)
|
|
|
|
def __init__(self, value):
|
|
self.value = value
|
|
|
|
def __repr__(self):
|
|
return f'Num({self.value!r})'
|
|
|
|
def __eq__(self, other):
|
|
return isinstance(other, Num) and self.value == other.value
|
|
|
|
|
|
class BinOp:
|
|
__slots__ = ('op', 'left', 'right')
|
|
|
|
def __init__(self, op, left, right):
|
|
self.op = op
|
|
self.left = left
|
|
self.right = right
|
|
|
|
def __repr__(self):
|
|
return f'BinOp({self.op!r}, {self.left!r}, {self.right!r})'
|
|
|
|
def __eq__(self, other):
|
|
return (isinstance(other, BinOp) and self.op == other.op
|
|
and self.left == other.left and self.right == other.right)
|
|
|
|
|
|
class Unary:
|
|
__slots__ = ('op', 'operand')
|
|
|
|
def __init__(self, op, operand):
|
|
self.op = op
|
|
self.operand = operand
|
|
|
|
def __repr__(self):
|
|
return f'Unary({self.op!r}, {self.operand!r})'
|
|
|
|
def __eq__(self, other):
|
|
return (isinstance(other, Unary) and self.op == other.op
|
|
and self.operand == other.operand)
|
|
|
|
|
|
class _Parser:
|
|
def __init__(self, tokens):
|
|
self._tokens = tokens
|
|
self._pos = 0
|
|
|
|
def _peek(self):
|
|
return self._tokens[self._pos]
|
|
|
|
def _consume(self, kind=None):
|
|
tok = self._tokens[self._pos]
|
|
if kind and tok.kind != kind:
|
|
raise ParseError(f'expected {kind}, got {tok.kind!r}')
|
|
self._pos += 1
|
|
return tok
|
|
|
|
def parse(self):
|
|
if self._peek().kind == 'EOF':
|
|
raise ParseError('empty input')
|
|
node = self._expr()
|
|
if self._peek().kind != 'EOF':
|
|
raise ParseError(f'unexpected token {self._peek().kind!r}')
|
|
return node
|
|
|
|
def _expr(self):
|
|
node = self._term()
|
|
while self._peek().kind in ('PLUS', 'MINUS'):
|
|
op = self._consume().value
|
|
right = self._term()
|
|
node = BinOp(op, node, right)
|
|
return node
|
|
|
|
def _term(self):
|
|
node = self._factor()
|
|
while self._peek().kind in ('STAR', 'SLASH'):
|
|
op = self._consume().value
|
|
right = self._factor()
|
|
node = BinOp(op, node, right)
|
|
return node
|
|
|
|
def _factor(self):
|
|
if self._peek().kind == 'MINUS':
|
|
self._consume()
|
|
operand = self._factor()
|
|
return Unary('-', operand)
|
|
return self._primary()
|
|
|
|
def _primary(self):
|
|
tok = self._peek()
|
|
if tok.kind == 'NUMBER':
|
|
self._consume()
|
|
return Num(tok.value)
|
|
if tok.kind == 'LPAREN':
|
|
self._consume()
|
|
node = self._expr()
|
|
if self._peek().kind != 'RPAREN':
|
|
raise ParseError(f'unclosed parenthesis, got {self._peek().kind!r}')
|
|
self._consume()
|
|
return node
|
|
raise ParseError(f'unexpected token {tok.kind!r}')
|
|
|
|
|
|
def parse(tokens):
|
|
"""Parse a token list from tokenize() and return an AST node."""
|
|
return _Parser(tokens).parse()
|