deal with missing storageFilename

[jstanley0]

Apparently my Google Photos archive has media objects that are missing storageFilename, which blows up the "Process duplicates" task. This looks like a reasonable way to skip them.

[mtalcott]

Thanks for the PR @jstanley0! 🙏

This would indeed skip over the problematic mediaItems in the processing task. However, it's not the only place in the application that expects the storageFilename to be present - the task results page that shows a list of duplicates also expects it to be present, so it can display a thumbnail image.

Rather than filtering these out everywhere, instead I'll opt to skip over them (delete their records) when we're unable to download the thumbnail image in the first place. PR to come!

[jstanley0]

yeah, it took an hour or so to actually get to that point, and I discovered that problem after I sent the PR.

I had guessed that files skipped here wouldn't be considered as possible duplicates (and am still a bit confused how they manage to break the results page)

unrelated, but I didn't want to file an issue when I believe I already know the answer - but in almost all of my dupes, I have an old copy uploaded by Backup and Sync that is exempt from the storage quota, and a new copy uploaded by Drive that isn't (and these are byte-for-byte identical, btw, not differently-compressed). it would be nice to preferentially keep the one that's exempt from the storage quota, but looking at the media item API result I don't think I can distinguish them

[mtalcott]

unrelated, but I didn't want to file an issue when I believe I already know the answer - but in almost all of my dupes, I have an old copy uploaded by Backup and Sync that is exempt from the storage quota, and a new copy uploaded by Drive that isn't (and these are byte-for-byte identical, btw, not differently-compressed). it would be nice to preferentially keep the one that's exempt from the storage quota, but looking at the media item API result I don't think I can distinguish them

I hadn't considered that. It would certainly be better to keep the copy that doesn't count toward storage quota. Bummer there isn't an obvious way to distinguish in the media item API's info, I'm not surprised given the other limitations I've seen but that'd be a great addition.