Rename downloaded files based on metadata

[tanoem]

Is your feature request related to a problem? Please describe.
On June 11th Nexus made a change where downloaded files no longer have the extra information in their filename to uniquely identify them. For example, before it would look like this:

MCM Unlocked - Latest Version-180186-2-1-1-1780439910.zip

But now files download like this:

MCM Unlocked - Latest Version.zip

The lack of mod ID, version, and timestamp mean there is a high potential for filename collisions between different versions and even different mods that name their files something simple like "Main" or "Bug Fix". The end result is that managing downloads without unique filenames became very difficult. After some feedback they have announced they will be adding back only the version number but not the mod ID or timestamp:

https://forums.nexusmods.com/topic/13539100-changes-to-downloaded-file-names/

This can still lead to collisions for mods using simple names or mod authors that reuse the same version number in a new upload.

Describe the solution you'd like
Rather than directly using the filename given when downloading the file from Nexus, it may be worth using the metadata which is stored in the associated ".meta" files to rename the files that are downloaded in order to avoid the potential for filename collisions. Even if Nexus decides to bring back the original naming scheme, this feature still may be good to have as future proofing in case this ever gets changed again in the future.

Describe alternatives you've considered
Someone has created an MO2 plugin to rename downloads: https://www.nexusmods.com/site/mods/1998

Unfortunately many people may not be aware this is available and it would be good to integrate this feature into Mod Organizer itself.

Additional context
Without the unique filenames, Mod Organizer will regularly prompt the user saying the file already exists when downloading new versions.

[JonathanFeenstra]

Looks like we'll need to change our getDownloadFileName method to ensure uniqueness. I'd like to know how Vortex handles it, since it would probably make sense to implement a similar approach in MO2.

@Pickysaurus do you have any suggestions on how to approach this? We could add mod ids and timestamps to every download filename from now on, but then duplicate mod files that were downloaded before the change would no longer be recognised as such. Then we'd also need to implement a different way of checking for duplicates that doesn't rely on the filename.

[Pickysaurus]

The best way to identify a file is to send the hash to the API. It's a feature of the V1 API. Eventually you'll also be able to use other hashes too.

MO2 can name the file however works best for the file system. You could reconstruct the old style of file name from the MD5 lookup data if you wanted to. Would the older files already have their identity information in the .meta file? Or is that strictly limited to things that come in as NXM links?

[JonathanFeenstra]

The best way to identify a file is to send the hash to the API.

Wouldn't that require consuming an extra API request with every download?

Would the older files already have their identity information in the .meta file? Or is that strictly limited to things that come in as NXM links?

The information in the meta files can either come from downloading through NXM links or from querying metadata for a (manual) download (by requesting it from the Nexus API through the MD5 hash).

It sounds like we'll need to adapt the logic to distinguish between the old and new filename format. When it's the new format, we'll need to rename it either by trying to reconstruct the old style filename (which may cost an extra API request if I understand correctly), or using some other unique format if there's another more straightforward way to identify duplicate downloads.

How is this handled in Vortex?

[Pickysaurus]

Wouldn't that require consuming an extra API request with every download?

Yes, however the API limits are set up with this in mind. Vortex MD5 checks any file it gets iirc regardless of it coming in via NXM or manual import. So it wouldn't use any more than Vortex.

If you find it's a problem or slows MO2 down a lot we can talk about batch endpoints or adding the additional hashes that require less computation.

If you want to reconstruct the file name that might be something you can do as part of exchanging the NXM link for a download link and starting the download? NXM links do give you the game domain, mod ID and file ID if you want to use them.

[JonathanFeenstra]

I see, downloading through NXM links already provides enough data to uniquely identify each download, so I think we can simply adjust the check for duplicates to use that info without needing an extra API request for the MD5 lookup. Then the way we format the filenames doesn't matter much as long as it's unique so it doesn't overwrite existing downloads. Downloads without meta files would still require manual querying to get the required data, but that should be an edge case.