Python3.4与3.6中TypeError差异问题及兼容方案问询

Let's break down the issue and fix it step by step.

The Problem #

You've got a B64LZMA class that extends str to handle base64-encoded LZMA data. It works fine in Python 3.6+ but throws TypeError: string argument without an encoding in Python 3.4. The problematic line is in the __str__ method:

return bytes(self).decode()

Why the Behavior Difference? #

The root cause is a change in how Python's bytes() function handles subclasses of str:

• Python 3.4: When you call bytes(obj) on a subclass of str, Python treats it as a regular string. It expects you to provide an encoding argument (like bytes(obj, 'utf-8')) and does not invoke the object's __bytes__ method. This is why you get the "string argument without an encoding" error—Python thinks you're trying to convert a string to bytes, not using your custom __bytes__ implementation.
• Python 3.6+: The bytes() function was updated to prioritize calling an object's __bytes__ method first, even if the object is a subclass of str. So bytes(self) correctly triggers your custom decompression logic instead of treating self as a plain string.

How to Fix Compatibility with Python 3.4 #

To make the code work across both versions, avoid relying on bytes(self) to invoke your __bytes__ method. Instead, call __bytes__() directly in the __str__ method:

from base64 import b64decode
from lzma import LZMADecompressor
class B64LZMA(str):
    """A string of base64 encoded, LZMA compressed data."""
    def __bytes__(self):
        """Returns the decompressed data."""
        return LZMADecompressor().decompress(b64decode(self.encode()))
    def __str__(self):
        """Returns the string decoded from __bytes__."""
        # Directly call __bytes__() instead of using bytes(self)
        return self.__bytes__().decode()

TEST_STR = B64LZMA('/Td6WFoAAATm1rRGAgAhARYAAAB0L+WjAQALSGVsbG8gd29ybGQuAGt+oFiSvoAYAAEkDKYY2NgftvN9AQAAAAAEWVo=')
if __name__ == '__main__':
    print(TEST_STR)

Alternatively, you could rewrite the __str__ method to compute the decoded string directly without relying on __bytes__, but calling self.__bytes__() keeps your code DRY (Don't Repeat Yourself) and maintains separation of concerns.

This change ensures that both Python 3.4 and 3.6+ use your custom decompression logic correctly, avoiding the TypeError in older versions.

内容的提问来源于stack exchange，提问作者user3515670