| 70 | |
| 71 | |
| 72 | class Scanner: |
| 73 | __slots__ = ("tokens", "current") |
| 74 | |
| 75 | def __init__(self, input: str) -> None: |
| 76 | self.tokens = self.lex(input) |
| 77 | self.current = next(self.tokens) |
| 78 | |
| 79 | def lex(self, input: str) -> Iterator[Token]: |
| 80 | pos = 0 |
| 81 | while pos < len(input): |
| 82 | if input[pos] in (" ", "\t"): |
| 83 | pos += 1 |
| 84 | elif input[pos] == "(": |
| 85 | yield Token(TokenType.LPAREN, "(", pos) |
| 86 | pos += 1 |
| 87 | elif input[pos] == ")": |
| 88 | yield Token(TokenType.RPAREN, ")", pos) |
| 89 | pos += 1 |
| 90 | else: |
| 91 | match = re.match(r"(:?\w|:|\+|-|\.|\[|\]|\\|/)+", input[pos:]) |
| 92 | if match: |
| 93 | value = match.group(0) |
| 94 | if value == "or": |
| 95 | yield Token(TokenType.OR, value, pos) |
| 96 | elif value == "and": |
| 97 | yield Token(TokenType.AND, value, pos) |
| 98 | elif value == "not": |
| 99 | yield Token(TokenType.NOT, value, pos) |
| 100 | else: |
| 101 | yield Token(TokenType.IDENT, value, pos) |
| 102 | pos += len(value) |
| 103 | else: |
| 104 | raise ParseError( |
| 105 | pos + 1, |
| 106 | f'unexpected character "{input[pos]}"', |
| 107 | ) |
| 108 | yield Token(TokenType.EOF, "", pos) |
| 109 | |
| 110 | def accept(self, type: TokenType, *, reject: bool = False) -> Optional[Token]: |
| 111 | if self.current.type is type: |
| 112 | token = self.current |
| 113 | if token.type is not TokenType.EOF: |
| 114 | self.current = next(self.tokens) |
| 115 | return token |
| 116 | if reject: |
| 117 | self.reject((type,)) |
| 118 | return None |
| 119 | |
| 120 | def reject(self, expected: Sequence[TokenType]) -> "NoReturn": |
| 121 | raise ParseError( |
| 122 | self.current.pos + 1, |
| 123 | "expected {}; got {}".format( |
| 124 | " OR ".join(type.value for type in expected), |
| 125 | self.current.type.value, |
| 126 | ), |
| 127 | ) |
| 128 | |
| 129 | |