initial support motion photos

[sickelap]

This adds support for parsing and extracting video from Google Pixel and Samsung motion photos.
Sample photo attached to #797 did not have XMP metadata and exiftool was unable to parse it. Parsing manually does pretty good job. Extracted videos are added as embedded_media property from Photo.main_file.

Adding support for HEIF/HEIC (a.k.a. Apple Live Photos) is more involved, but hopefully we will have it done as well in the future.

PR for frontend will follow

[derneuere]

We now have a couple of other migrations, which means, that this line needs to be updated

[derneuere]

I would prefer it if we could maintain consistency in how we save technical files for LibrePhotos. "embedded_media" files are extracted in the folder where the original file lives. This breaks the regular pattern where we save needed files in the protected_media folder. I think adding a folder named embedded_media and saving embedded_media there would make the most sense.

I think there are also a couple of changes needed for calculating the thumbnails. The thumbnail in the list is probably a video and the big thumbnail an image. We also would need a new type of live / motion photos, which you could then add to PigSerializer.

I am looking forward to your frontend pull request!

[sonarcloud]

SonarCloud Quality Gate failed.   

0 Bugs
0 Vulnerabilities
3 Security Hotspots
9 Code Smells

No Coverage information
0.0% Duplication

[sickelap]

I totally agree with embedded media location. As those are not primary files but rather extracted from the original, it makes sense to keep them in a private location. In fact, this was bugging me too. Now, extracted media files will be placed into the /protected_media/embedded_media/.

I don't think that messing with thumbnails is necessary because of what I have in mind for the frontend.
The idea I am working towards is not to display embedded media as standalone content. Instead, when opening the motion photo in light box, we will have a “Play” button (similar to how it's done in google photos). I also don't think we need to have a dedicated type for embedded media and the extracted content is a M2M relationship to the Photo.main_file. This makes implementation simpler (I think) and not as intrusive as it would have been otherwise.

[derneuere]

Alright, looking forward to the frontend pr :)