Django/Python: How to read a file and validate that it is an audio file? [duplicate]

Answer 1

An alternative to the other posted answer which does header parsing. This means someone could still include other data behind a valid header.

Is to verify the whole file which costs more CPU but also has a stricter policy. A library that can do this is python audiotools and the relevant API method is AudioFile.verify.

Used like this:

import audiotools

f = audiotools.open(filename)
try:
    result = f.verify()
except audiotools.InvalidFile:
    # Invalid file.
    print("Invalid File")
else:
    # Valid file.
    print("Valid File")

A warning is that this verify method is quite strict and can actually flag badly encoded files as invalid. You have to decide yourself if this is a proper method or not for your use case.

Answer 2

The idea is to use an utility such as sndhdr to try to read a header of the file.

from sndhdr import what
import os
from settings import MEDIA_ROOT
...
if what( os.path.join( MEDIA_ROOT, file.name ) ) == None:
# header parse failed, so, it could be a non-audio file.

In this case you must be sure that the utility could not recognize a non-audio file as an audio file.

I never use the sndhdr utility, so, it may be better to use another one. For example there is also mutagen project.

Update.

The more professional approach.

1. Let a user upload his files without the such validation.
2. Create a special model field associated with one file: is_checked. Set the field False by default.
3. Don't let other users to downoad the files which have is_checked == False
4. Organize an asynchronous task that runs an anti-malware soft that checked unchecked files. The software must be called as a function from your pythonic task. The function must answer the question: is a particular file has a malware part.
5. If the current file has the malware part delete it and delete the record about the file. Otherwise set up is_checked = True from your task.

Why a separated task? Because the such malware checking can be a long operation.