Properly validate files before upload

[ivanistheone]

• ricecooker version: 0.6.39
• Python version: 3.7
• Operating System: macOS

Problem

Currently Ricecooker will allow the upload of empty or otherwise invalid files to Studio. The validation performed is only "shallow validation" to make sure file extension is OK.

Proposed solution

We could follow the example of ThumbnailFile to implement "deep validation" checks as part of the process_file method of each file type:

• PDF try to read using PyPDF2
• ePub try to open .epub file with ebooklib.epub.read_epub
• audio and video could check codecs with ffmpeg[1,2,3]
• Check that files are not too large and give a warning.

[ivanistheone]

Deep validation checks can also handle this: #139

[ivanistheone]

This will partially get caught if --thumbnail generation paths are used, but would be good to make these checks even if no thumbnail generation requested.

Related issue: large file size deep validation check #139

[ivanistheone]

Another problem from today: an exercise that contains .webp data was processed without raising errors, and only errored out during upload to Studio (see).

Some "exercise files validation checks" seem in order so we can fail fast.

[rtibbles]

Note that webp images are properly converted before upload now.