UnicodeDecodeError with binary tags in mov files

[Dobatymo]

Overview

A UnicodeDecodeError is raised when opening some files.

Expected behavior

It should not fail.

Actual behavior

Traceback:

Traceback (most recent call last):
  File "asd.py", line 153, in __init__
    container = av.open(path, "r")
  File "av\container\core.pyx", line 365, in av.container.core.open
  File "av\container\input.pyx", line 70, in av.container.input.InputContainer.__cinit__
  File "av\utils.pyx", line 31, in av.utils.avdict_to_dict
  File "av\utils.pyx", line 14, in av.utils._decode
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xff in position 0: invalid start byte

Investigation

I noticed the problem when trying to open various .mov files. I think the problem might be the binary data in com.apple.quicktime.artwork tags.

Reproduction

Just call av.open("asd.mov", "r"). I have some example files, but cannot post them publicly.

Versions

• OS: Windows 7 x64
• PyAV runtime:

PyAV v7.0.2.dev0
git origin: [email protected]:mikeboers/PyAV
git commit: v6.2.0-132-gd9bebbd
library configuration: --disable-static --enable-shared --enable-gpl --enable-version3 --enable-sdl2 --enable-fontconfig --enable-gnutls --e
nable-iconv --enable-libass --enable-libdav1d --enable-libbluray --enable-libfreetype --enable-libmp3lame --enable-libopencore-amrnb --enabl
e-libopencore-amrwb --enable-libopenjpeg --enable-libopus --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libtheora --enable-
libtwolame --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libzimg --enable
-lzma --enable-zlib --enable-gmp --enable-libvidstab --enable-libvorbis --enable-libvo-amrwbenc --enable-libmysofa --enable-libspeex --enabl
e-libxvid --enable-libaom --enable-libmfx --enable-amf --enable-ffnvcodec --enable-cuvid --enable-d3d11va --enable-nvenc --enable-nvdec --en
able-dxva2 --enable-avisynth --enable-libopenmpt
library license: GPL version 3 or later
libavcodec     58. 54.100
libavdevice    58.  8.100
libavfilter     7. 57.100
libavformat    58. 29.100
libavutil      56. 31.100
libswresample   3.  5.100
libswscale      5.  5.100

Research

I have done the following:

• [x] Checked the PyAV documentation
• [x] Searched on Google
• [x] Searched on Stack Overflow
• [x] Looked through old GitHub issues
• [ ] Asked on PyAV Gitter
• [ ] ... and waited 72 hours for a response.

[jlaine]

You're going to have to provide an example file, for which you hold the copyright, otherwise there is nothing we can do here.

[jlaine]

Is there any documentation about binary tags I can read somewhere? I was convinced metadata only held strings, but if that's not the case we will probably need a different approach. One idea might be to allow passing a MetadataEncoder class which would need to implement these methods:

class MetadataEncoder:
    def decode(self, key: str, value: bytes) -> Any:
        ...

    def encode(self, key: str, value: Any) -> bytes:
        ...

The default implementation would be the one we currently have, with encode turning str into bytes and decode doing the opposite.

[Dobatymo]

There is a table here for apple metadata https://developer.apple.com/library/archive/documentation/QuickTime/QTFF/Metadata/Metadata.html#//apple_ref/doc/uid/TP40000939-CH1-SW43

com.apple.quicktime.artwork seems to be a binary blob and com.apple.quicktime.rating.user is even a BE Float32. So probably the only easy solution would be to return bytes and have the user do all the decoding manually or provide a MetadataEncoder which seems like a good idea.

Otherwise a mapping for tag names to data types would be needed.

The above docs mention value types, which would be great for mapping. But I don't know where they are stored.

EDIT: I managed to create a simple mp4 file with the com.apple.quicktime.artwork tag which shows this error: com.apple.quicktime.artwork.zip

[Dobatymo]

Just a reminder that I attached an example file a while back in case this was missed.

[mikeboers]

Thanks @Dobatymo. Unfortunately, it seems both of us primary maintainers are really distracted with paid work.

[github-actions[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[Dobatymo]

Just commenting since I don't want this issue to become stale...

[github-actions[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[Dobatymo]

Just commenting since I don't want this issue to become stale...

[nrhodes]

From perusing the code, looks as though metadata_errors may control this. Even though it is documented as only being used for encoding, it's also used for decoding as well. Try: av.open(path, "r", metadata_errors="ignore")

[github-actions[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[Dobatymo]

With av==10.0.0 the file can be opened, but the metadata seems to be simply ignored, even with metadata_errors="strict"

[CheshireCC]

From perusing the code, looks as though metadata_errors may control this. Even though it is documented as only being used for encoding, it's also used for decoding as well. Try: av.open(path, "r", metadata_errors="ignore")

thx，that's really helpful