102 lines
2.6 KiB
Python
102 lines
2.6 KiB
Python
from dataclasses import dataclass
|
|
from typing import Any, List
|
|
|
|
|
|
class ParseError(Exception):
|
|
pass
|
|
|
|
|
|
@dataclass
|
|
class Num:
|
|
value: Any
|
|
|
|
def __repr__(self):
|
|
return f"Num({self.value!r})"
|
|
|
|
|
|
@dataclass
|
|
class BinOp:
|
|
op: str
|
|
left: Any
|
|
right: Any
|
|
|
|
def __repr__(self):
|
|
return f"BinOp({self.op!r}, {self.left!r}, {self.right!r})"
|
|
|
|
|
|
@dataclass
|
|
class Unary:
|
|
op: str
|
|
operand: Any
|
|
|
|
def __repr__(self):
|
|
return f"Unary({self.op!r}, {self.operand!r})"
|
|
|
|
|
|
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 is not None and tok.kind != kind:
|
|
raise ParseError(f"expected {kind}, got {tok.kind!r} ({tok.value!r})")
|
|
self._pos += 1
|
|
return tok
|
|
|
|
def _parse_expr(self):
|
|
node = self._parse_term()
|
|
while self._peek().kind in ('PLUS', 'MINUS'):
|
|
op = self._consume().value
|
|
right = self._parse_term()
|
|
node = BinOp(op, node, right)
|
|
return node
|
|
|
|
def _parse_term(self):
|
|
node = self._parse_unary()
|
|
while self._peek().kind in ('STAR', 'SLASH'):
|
|
op = self._consume().value
|
|
right = self._parse_unary()
|
|
node = BinOp(op, node, right)
|
|
return node
|
|
|
|
def _parse_unary(self):
|
|
if self._peek().kind == 'MINUS':
|
|
op = self._consume().value
|
|
operand = self._parse_unary()
|
|
return Unary(op, operand)
|
|
return self._parse_primary()
|
|
|
|
def _parse_primary(self):
|
|
tok = self._peek()
|
|
if tok.kind == 'NUMBER':
|
|
self._consume()
|
|
return Num(tok.value)
|
|
if tok.kind == 'LPAREN':
|
|
self._consume()
|
|
node = self._parse_expr()
|
|
if self._peek().kind != 'RPAREN':
|
|
raise ParseError(f"expected ')', got {self._peek().kind!r}")
|
|
self._consume()
|
|
return node
|
|
raise ParseError(f"unexpected token {tok.kind!r} ({tok.value!r})")
|
|
|
|
|
|
def parse(tokens: list):
|
|
"""Parse a token list produced by lexer.tokenize() into an AST.
|
|
|
|
Returns one of: Num(value), BinOp(op, left, right), Unary(op, operand).
|
|
Raises ParseError on malformed input.
|
|
"""
|
|
p = _Parser(tokens)
|
|
if p._peek().kind == 'EOF':
|
|
raise ParseError("empty input")
|
|
node = p._parse_expr()
|
|
if p._peek().kind != 'EOF':
|
|
raise ParseError(f"unexpected token {p._peek().kind!r} ({p._peek().value!r})")
|
|
return node
|