parse_number.py

from enum import Enum, auto

"""
START:
    SIGN -> SIGN
    DIGIT -> INT
    "0" -> ZERO
    "." -> P_DOT
    ERROR -> ERROR

SIGN:
    DIGIT -> INT
    "0" -> ZERO
    "." -> P_DOT
    ERROR -> ERROR

ZERO:
    "x" | "X" -> HEX
    "o" | "O" -> OCT
    "b" | "B" -> BIN
    "." -> P_DOT
    DIGIT -> INT  # (Warning: Invalid octal sequence, e.g., 09)
    "e" | "E" -> EXP
    END -> END

INT:
    DIGIT -> INT
    "." -> DOT
    "e" | "E" -> EXP
    END -> END

DOT:
    DIGIT -> FRAC
    ERROR -> ERROR

FRAC:
    DIGIT -> FRAC
    "e" | "E" -> EXP
    END -> END

EXP:
    "+" | "-" -> EXP_SIGN
    DIGIT -> EXP_INT
    ERROR -> ERROR

EXP_SIGN:
    DIGIT -> EXP_INT
    ERROR -> ERROR

EXP_INT:
    DIGIT -> EXP_INT
    END -> END

HEX:
    HEX_DIGIT -> HEX
    END -> END

OCT:
    OCT_DIGIT -> OCT
    END -> END

BIN:
    "0" | "1" -> BIN
    END -> END

END:
    ERROR -> ERROR

ERROR:
    ERROR -> ERROR
"""


class State(Enum):
    START = auto()
    SIGN = auto()
    ZERO = auto()
    INT = auto()
    P_DOT = auto()
    DOT = auto()
    FRAC = auto()
    EXP = auto()
    EXP_SIGN = auto()
    EXP_INT = auto()
    HEX = auto()
    OCT = auto()
    BIN = auto()
    END = auto()
    ERROR = auto()


class TokenType(Enum):
    NUMBER = auto()
    ERROR = auto()


class NumberParser:
    def __init__(self):
        self.state = State.START

    def transition(self, char: str):
        if self.state == State.START:
            if char in "+-":
                self.state = State.SIGN
            elif char == "0":
                self.state = State.ZERO
            elif char.isdigit():
                self.state = State.INT
            elif char == ".":
                self.state = State.P_DOT
            else:
                self.state = State.ERROR

        elif self.state == State.SIGN:
            if char == "0":
                self.state = State.ZERO
            elif char.isdigit():
                self.state = State.INT
            elif char == ".":
                self.state = State.P_DOT
            else:
                self.state = State.ERROR

        elif self.state == State.ZERO:
            if char in "xX":
                self.state = State.HEX
            elif char in "oO":
                self.state = State.OCT
            elif char in "bB":
                self.state = State.BIN
            elif char == ".":
                self.state = State.P_DOT
            elif char.isdigit():
                self.state = State.INT  # Warning: Invalid octal sequence
            elif char in "eE":
                self.state = State.EXP
            else:
                self.state = State.END

        elif self.state == State.INT:
            if char.isdigit():
                self.state = State.INT
            elif char == ".":
                self.state = State.DOT
            elif char in "eE":
                self.state = State.EXP
            else:
                self.state = State.END

        elif self.state == State.P_DOT:
            if char.isdigit():
                self.state = State.FRAC
            else:
                self.state = State.ERROR

        elif self.state == State.DOT:
            if char.isdigit():
                self.state = State.FRAC
            else:
                self.state = State.ERROR

        elif self.state == State.FRAC:
            if char.isdigit():
                self.state = State.FRAC
            elif char in "eE":
                self.state = State.EXP
            else:
                self.state = State.END

        elif self.state == State.EXP:
            if char in "+-":
                self.state = State.EXP_SIGN
            elif char.isdigit():
                self.state = State.EXP_INT
            else:
                self.state = State.ERROR

        elif self.state == State.EXP_SIGN:
            if char.isdigit():
                self.state = State.EXP_INT
            else:
                self.state = State.ERROR

        elif self.state == State.EXP_INT:
            if char.isdigit():
                self.state = State.EXP_INT
            else:
                self.state = State.END

        elif self.state == State.HEX:
            if char.isdigit() or char.lower() in "abcdef":
                self.state = State.HEX
            else:
                self.state = State.END

        elif self.state == State.OCT:
            if char in "01234567":
                self.state = State.OCT
            else:
                self.state = State.END

        elif self.state == State.BIN:
            if char in "01":
                self.state = State.BIN
            else:
                self.state = State.END

        else:
            self.state = State.ERROR

    def parse(self, token: str):
        self.state = State.START

        prev_char = ""
        for char in token:
            if char == '_' and prev_char != "" and (prev_char.isdigit() or prev_char.lower() in "abcdef_"):
                prev_char = char
                continue
            prev_char = char

            self.transition(char)
            if self.state == State.ERROR:
                return TokenType.ERROR

        if self.state in {State.INT, State.FRAC, State.EXP_INT, State.HEX, State.OCT, State.BIN, State.ZERO, State.DOT}:
            return TokenType.NUMBER
        return TokenType.ERROR


def test_parser():
    parser = NumberParser()
    good_cases = [
        "0",
        "0.0",
        "-0",
        "+0",
        "123",
        "-123",
        "+123",
        "0.2",
        "2.123456",
        ".2",
        "2.",
        "-2.5",
        "+2.5",
        "1e3",
        "1e+3",
        "1e-3",
        "-1e-3",
        "+1e3",
        "0x3F",
        "0o77",
        "0b1010",
        "-0x3A",
        "+0b101",
        "0e0",
        "-0e0",
        "+0e0",
        "123.456",
        "1e1000",
        "1e-1000",
        "1e+1000",
        "1_000",
        "1_000_000",
        "1_000.000_1",
        "1_000_000e10",
        "0b0",
        "0x0",
        "0o0",
        "-1.",
        "+1.",
        "01",
    ]

    bad_cases = [
        "invalid",
        "1e",
        "0b102",
        "0xG",
        "0o08",
        "0o9",
        "0bb1",
        "0.2.",
        "0.2.2",
        "-",
        "+",
        ".",
        "1e+",
        "1e-",
        "e3",
        "1.2.3",
        "1e+1.5",
        "+-123",
        "-+123",
        "0b123",
        "0o08",
        "0b102",
        "0xG",
        "0._2",
        "_3",
        "FF",
        "1e3e2"
    ]

    print("-" * 20, "GOOD", "-" * 20)
    for case in good_cases:
        result = parser.parse(case)
        print(f"Token: {case:16} -> Type: {result.name}")

    print("-" * 20, "BAD", "-" * 20)
    for case in bad_cases:
        result = parser.parse(case)
        print(f"Token: {case:16} -> Type: {result.name}")


test_parser()