from dataclasses import dataclass from typing import Union class LexError(Exception): pass @dataclass class Token: kind: str value: Union[int, float, str, None] _SINGLE = { '+': 'PLUS', '-': 'MINUS', '*': 'STAR', '/': 'SLASH', '(': 'LPAREN', ')': 'RPAREN', } def tokenize(src: str) -> list: tokens = [] i = 0 while i < len(src): ch = src[i] if ch in ' \t': i += 1 elif ch in _SINGLE: tokens.append(Token(_SINGLE[ch], ch)) i += 1 elif ch.isdigit() or ch == '.': start = i has_dot = False while i < len(src) and (src[i].isdigit() or (src[i] == '.' and not has_dot)): if src[i] == '.': has_dot = True i += 1 raw = src[start:i] value = float(raw) if has_dot else int(raw) tokens.append(Token('NUMBER', value)) else: raise LexError(f"unexpected character {ch!r} at position {i}") tokens.append(Token('EOF', None)) return tokens