base_n.py

from functools import cached_property
from timeit import timeit


class BaseN:
    def __init__(self, charset: str) -> None:
        self.charset = charset
        self.reverse_charset = {c: i for i, c in enumerate(charset)}
        assert len(self.charset) == len(self.reverse_charset), "charset must be unique"

    @cached_property
    def base(self):
        return len(self.charset)

    def encode(self, s: bytes):
        int_char = int.from_bytes(s, "big")
        result = []  # we can use deque to improve performance
        while int_char > 0:
            result.append(self.charset[int_char % self.base])
            int_char //= self.base
        return "".join(result)

    def decode(self, s: str):
        result = 0
        for i in reversed(s):
            result *= self.base
            result += self.reverse_charset[i]
        return result.to_bytes((result.bit_length() + 7) // 8, "big")


base3050 = BaseN(
    charset="".join(chr(i) for i in range(0x4E00, 0x4E00 + 3050)),
)

target_str = b"Hello World!" * 10


def test():
    encoded = base3050.encode(target_str)
    decoded = base3050.decode(encoded)
    assert decoded == target_str


print(f"{timeit(test, number=10000)*1000:.3f}ms")


encoded = base3050.encode(b"Hello World!")
print(encoded)  # 厯嘩劺刊僾呡句姑丂

decoded = base3050.decode(encoded)
print(decoded)  # b'Hello World!'