aifc module is deprecated (and removed in Python 3.13)

[mgorny]

This package still uses the aifc module that is deprecated since Python 3.11, and was removed in Python 3.13.

As a result, the test suite fails with 3.13.0b2:

$ tox -e py313
py313: install_deps> python -I -m pip install pytest
.pkg: _optional_hooks> python /usr/lib/python3.12/site-packages/pyproject_api/_backend.py True flit_core.buildapi
.pkg: get_requires_for_build_sdist> python /usr/lib/python3.12/site-packages/pyproject_api/_backend.py True flit_core.buildapi
.pkg: build_sdist> python /usr/lib/python3.12/site-packages/pyproject_api/_backend.py True flit_core.buildapi
py313: install_package> python -I -m pip install --force-reinstall --no-deps /tmp/audioread/.tox/.tmp/package/2/audioread-3.0.2.tar.gz
py313: commands[0]> pytest
========================================================= test session starts =========================================================
platform linux -- Python 3.13.0b2, pytest-8.2.2, pluggy-1.5.0
cachedir: .tox/py313/.pytest_cache
rootdir: /tmp/audioread
configfile: pyproject.toml
collected 4 items                                                                                                                     

test/test_audioread.py FFFF                                                                                                     [100%]

============================================================== FAILURES ===============================================================
__________________________________________________ test_audioread_early_exit[test-1] __________________________________________________

audiofile = <conftest.AudiofileSpec object at 0x7efcb006cad0>

    def test_audioread_early_exit(audiofile):
        """Abort the read before it is completed.
    
        This test guards against regressions such as
        https://github.com/beetbox/audioread/pull/78
    
        """
>       with audioread.audio_open(audiofile.path) as a:

test/test_audioread.py:29: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
.tox/py313/lib/python3.13/site-packages/audioread/__init__.py:123: in audio_open
    backends = available_backends()
.tox/py313/lib/python3.13/site-packages/audioread/__init__.py:80: in available_backends
    from . import rawread
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

    # This file is part of audioread.
    # Copyright 2011, Adrian Sampson.
    #
    # Permission is hereby granted, free of charge, to any person obtaining
    # a copy of this software and associated documentation files (the
    # "Software"), to deal in the Software without restriction, including
    # without limitation the rights to use, copy, modify, merge, publish,
    # distribute, sublicense, and/or sell copies of the Software, and to
    # permit persons to whom the Software is furnished to do so, subject to
    # the following conditions:
    #
    # The above copyright notice and this permission notice shall be
    # included in all copies or substantial portions of the Software.
    
    """Uses standard-library modules to read AIFF, AIFF-C, and WAV files."""
>   import aifc
E   ModuleNotFoundError: No module named 'aifc'

.tox/py313/lib/python3.13/site-packages/audioread/rawread.py:16: ModuleNotFoundError
__________________________________________________ test_audioread_early_exit[test-2] __________________________________________________

audiofile = <conftest.AudiofileSpec object at 0x7efcb001bc50>

    def test_audioread_early_exit(audiofile):
        """Abort the read before it is completed.
    
        This test guards against regressions such as
        https://github.com/beetbox/audioread/pull/78
    
        """
>       with audioread.audio_open(audiofile.path) as a:

test/test_audioread.py:29: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
.tox/py313/lib/python3.13/site-packages/audioread/__init__.py:123: in audio_open
    backends = available_backends()
.tox/py313/lib/python3.13/site-packages/audioread/__init__.py:80: in available_backends
    from . import rawread
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

    # This file is part of audioread.
    # Copyright 2011, Adrian Sampson.
    #
    # Permission is hereby granted, free of charge, to any person obtaining
    # a copy of this software and associated documentation files (the
    # "Software"), to deal in the Software without restriction, including
    # without limitation the rights to use, copy, modify, merge, publish,
    # distribute, sublicense, and/or sell copies of the Software, and to
    # permit persons to whom the Software is furnished to do so, subject to
    # the following conditions:
    #
    # The above copyright notice and this permission notice shall be
    # included in all copies or substantial portions of the Software.
    
    """Uses standard-library modules to read AIFF, AIFF-C, and WAV files."""
>   import aifc
E   ModuleNotFoundError: No module named 'aifc'

.tox/py313/lib/python3.13/site-packages/audioread/rawread.py:16: ModuleNotFoundError
_____________________________________________________ test_audioread_full[test-1] _____________________________________________________

audiofile = <conftest.AudiofileSpec object at 0x7efcafd8c2d0>

    def test_audioread_full(audiofile):
        """Read the audio data from the file."""
>       with audioread.audio_open(audiofile.path) as a:

test/test_audioread.py:38: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
.tox/py313/lib/python3.13/site-packages/audioread/__init__.py:123: in audio_open
    backends = available_backends()
.tox/py313/lib/python3.13/site-packages/audioread/__init__.py:80: in available_backends
    from . import rawread
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

    # This file is part of audioread.
    # Copyright 2011, Adrian Sampson.
    #
    # Permission is hereby granted, free of charge, to any person obtaining
    # a copy of this software and associated documentation files (the
    # "Software"), to deal in the Software without restriction, including
    # without limitation the rights to use, copy, modify, merge, publish,
    # distribute, sublicense, and/or sell copies of the Software, and to
    # permit persons to whom the Software is furnished to do so, subject to
    # the following conditions:
    #
    # The above copyright notice and this permission notice shall be
    # included in all copies or substantial portions of the Software.
    
    """Uses standard-library modules to read AIFF, AIFF-C, and WAV files."""
>   import aifc
E   ModuleNotFoundError: No module named 'aifc'

.tox/py313/lib/python3.13/site-packages/audioread/rawread.py:16: ModuleNotFoundError
_____________________________________________________ test_audioread_full[test-2] _____________________________________________________

audiofile = <conftest.AudiofileSpec object at 0x7efcafff7950>

    def test_audioread_full(audiofile):
        """Read the audio data from the file."""
>       with audioread.audio_open(audiofile.path) as a:

test/test_audioread.py:38: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
.tox/py313/lib/python3.13/site-packages/audioread/__init__.py:123: in audio_open
    backends = available_backends()
.tox/py313/lib/python3.13/site-packages/audioread/__init__.py:80: in available_backends
    from . import rawread
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

    # This file is part of audioread.
    # Copyright 2011, Adrian Sampson.
    #
    # Permission is hereby granted, free of charge, to any person obtaining
    # a copy of this software and associated documentation files (the
    # "Software"), to deal in the Software without restriction, including
    # without limitation the rights to use, copy, modify, merge, publish,
    # distribute, sublicense, and/or sell copies of the Software, and to
    # permit persons to whom the Software is furnished to do so, subject to
    # the following conditions:
    #
    # The above copyright notice and this permission notice shall be
    # included in all copies or substantial portions of the Software.
    
    """Uses standard-library modules to read AIFF, AIFF-C, and WAV files."""
>   import aifc
E   ModuleNotFoundError: No module named 'aifc'

.tox/py313/lib/python3.13/site-packages/audioread/rawread.py:16: ModuleNotFoundError
======================================================= short test summary info =======================================================
FAILED test/test_audioread.py::test_audioread_early_exit[test-1] - ModuleNotFoundError: No module named 'aifc'
FAILED test/test_audioread.py::test_audioread_early_exit[test-2] - ModuleNotFoundError: No module named 'aifc'
FAILED test/test_audioread.py::test_audioread_full[test-1] - ModuleNotFoundError: No module named 'aifc'
FAILED test/test_audioread.py::test_audioread_full[test-2] - ModuleNotFoundError: No module named 'aifc'
========================================================== 4 failed in 0.08s ==========================================================
py313: exit 1 (0.38 seconds) /tmp/audioread> pytest pid=97746
.pkg: _exit> python /usr/lib/python3.12/site-packages/pyproject_api/_backend.py True flit_core.buildapi
  py313: FAIL code 1 (5.46=setup[5.08]+cmd[0.38] seconds)
  evaluation failed :( (5.57 seconds)

[lucas42]

It's worth mentioning that the sunau module is also deprecated and being removed at the same time.
(There was discussion about wave, but looks like they'd decided to keep it in python core)

There's an attempt to make the deprecated libraries available as pip dependencies: https://github.com/youknowone/python-deadlib However, it seems there's currently a couple of issues with that repository which means it won't work with 3.13 yet.

[lucas42]

I've submitted a pull request to the python-deadlib library to try to get its aifc fork from the python core working on python 3.13: youknowone/python-deadlib#9

Another potential approach would be to handle the ModuleNotFoundError more gracefully here. That would reduce the range of file types supported on python3.13, but is still better than a hard error, even when aifc support isn't needed!

[lucas42]

The python-deadlib modules have now been fixed up and can be used to replace the missing modules from python core.

So a workaround to this issue is pip install standard-aifc standard-sunau

[montvid]

The library does not build on python 3.12 too Anjok07/ultimatevocalremovergui#1578 (comment)

pipx >(analyze_pip_output:311): pip seemed to fail to build package:
    audioread==3.0.0

Some possibly relevant errors from pip install:
    error: subprocess-exited-with-error
    ModuleNotFoundError: No module named 'imp'

[lucas42]

@montvid I don't think your error is related to this issue.

Looks like that the imp issue might have been fixed in 3.0.1 (see #136 )
I'd guess ultimatevocalremovergui just needs to update its dependencies to the latest release.

[ISSOtm]

I wonder if using https://github.com/scientific-python/lazy-loader could help? Then the modules would only be loaded (and thus fail to be found) on demand.

[lucas42]

@ISSOtm That approach would mean some uses of audioread would work in python 3.13, but others would error. Whilst it's tempting (and would certainly solve my personal use of it), I think it'd cause confusing results to people unfamiliar with the internal implementation. The question of "is this library supported in python 3.13?" would need an answer of "it depends", followed by details of the nuance.
Given the effort needed to introduce lazy-loader, it seems just as simple to fully support python 3.13, and in doing so be much clearer to users of the library.

[bmcfee]

@ISSOtm That approach would mean some uses of audioread would work in python 3.13, but others would error. Whilst it's tempting (and would certainly solve my personal use of it), I think it'd cause confusing results to people unfamiliar with the internal implementation. The question of "is this library supported in python 3.13?" would need an answer of "it depends", followed by details of the nuance. Given the effort needed to introduce lazy-loader, it seems just as simple to fully support python 3.13, and in doing so be much clearer to users of the library.

Isn't this always the case though? The formats supported by any particular installation of audioread generally depend on several factors that can't be strictly determined by python dependencies. For example, if it calls out to the ffmpeg backend, the supported formats will depend on whatever that version of ffmpeg happened to be compiled with.

[lucas42]

Isn't this always the case though? The formats supported by any particular installation of audioread generally depend on several factors that can't be strictly determined by python dependencies. For example, if it calls out to the ffmpeg backend, the supported formats will depend on whatever that version of ffmpeg happened to be compiled with.

In my mind, which runtime versions a library supports feels more fundamental than other programmes it interacts with. But taking a step back, I concede that distinction is kinda arbitrary.

So I guess the real question is whether people prefer to face more errors at install time, or make installs easier with the potential for more runtime errors. As a user of the library, I'm in favour of install time errors I find them much easier to identify and debug than runtime errors. But I understand others may disagree.

[bmcfee]

As a user of the library, I'm in favour of install time errors I find them much easier to identify and debug than runtime errors.

I'm with you on that. 😅 There's a limit to what python package specs can realistically accomplish here though.

For whatever it's worth, I like the solution of having a 3.13-specific dependency for aifc and sunau (as done in #145 ). One variation on this could be putting it in an extras_require section, so you could do eg pip install audioread[legacy] to be extra-sure that these formats will be supported at install-time, but otherwise let them drop or be supported by other backends.