124 lines
3.0 KiB
Python
124 lines
3.0 KiB
Python
from dataclasses import dataclass
|
|
from typing import Any
|
|
|
|
from calc.lexer import Token
|
|
|
|
|
|
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: list):
|
|
self._tokens = tokens
|
|
self._pos = 0
|
|
|
|
def _peek(self) -> Token:
|
|
return self._tokens[self._pos]
|
|
|
|
def _consume(self, kind: str) -> Token:
|
|
tok = self._peek()
|
|
if tok.kind != kind:
|
|
raise ParseError(
|
|
f"expected {kind!r} but got {tok.kind!r} ({tok.value!r})"
|
|
)
|
|
self._pos += 1
|
|
return tok
|
|
|
|
def _advance(self) -> Token:
|
|
tok = self._tokens[self._pos]
|
|
self._pos += 1
|
|
return tok
|
|
|
|
def parse(self):
|
|
node = self._expr()
|
|
tok = self._peek()
|
|
if tok.kind != 'EOF':
|
|
raise ParseError(
|
|
f"unexpected token {tok.kind!r} ({tok.value!r}) after expression"
|
|
)
|
|
return node
|
|
|
|
# expr := term (('+' | '-') term)*
|
|
def _expr(self):
|
|
node = self._term()
|
|
while self._peek().kind in ('PLUS', 'MINUS'):
|
|
op = self._advance().value
|
|
right = self._term()
|
|
node = BinOp(op, node, right)
|
|
return node
|
|
|
|
# term := unary (('*' | '/') unary)*
|
|
def _term(self):
|
|
node = self._unary()
|
|
while self._peek().kind in ('STAR', 'SLASH'):
|
|
op = self._advance().value
|
|
right = self._unary()
|
|
node = BinOp(op, node, right)
|
|
return node
|
|
|
|
# unary := '-' unary | primary
|
|
def _unary(self):
|
|
if self._peek().kind == 'MINUS':
|
|
op = self._advance().value
|
|
operand = self._unary()
|
|
return Unary(op, operand)
|
|
return self._primary()
|
|
|
|
# primary := NUMBER | '(' expr ')'
|
|
def _primary(self):
|
|
tok = self._peek()
|
|
if tok.kind == 'NUMBER':
|
|
self._advance()
|
|
return Num(tok.value)
|
|
if tok.kind == 'LPAREN':
|
|
self._advance()
|
|
node = self._expr()
|
|
self._consume('RPAREN')
|
|
return node
|
|
raise ParseError(
|
|
f"unexpected token {tok.kind!r} ({tok.value!r}); expected number or '('"
|
|
)
|
|
|
|
|
|
def parse(tokens: list):
|
|
"""Parse a token list produced by `calc.lexer.tokenize` into an AST.
|
|
|
|
Returns one of:
|
|
Num(value)
|
|
BinOp(op, left, right) op in {'+', '-', '*', '/'}
|
|
Unary(op, operand) op == '-'
|
|
|
|
Raises ParseError on malformed input.
|
|
"""
|
|
if not tokens or (len(tokens) == 1 and tokens[0].kind == 'EOF'):
|
|
raise ParseError("empty input")
|
|
return _Parser(tokens).parse()
|