Cloud Support with Api?

Are Finstad 4 years ago updated by Tom 4 years ago 3

i think as the cloud gets more and more used nowadays people would want to host their files on there for long time storage.

thought is that Ubooquity could use the api of the cloud service to request file downloads, metadata and thumbnails.
could technically also upload and replace the files in storage if metadata changes are made by Ubooquity.

i find that i use my local storage less and less but there is a limit in how well NetDrive works for systems as a program like that is quite limited in how much it can load at a time.

a program using the api to just load what it needs in the other hand can be more useful with time.

Top Cloud Picks that should be supported with something like this:
Amazon Cloud (best value for the money still)
Google Drive
OneDrive (Microsoft)

im sure there is many others. but those are the ones that is worth it in the long run.

with support for this we could host or comics, manga or books on the cloud without issues.

ofc the only issue would be larger files and so on. might have to have a way to issue a download of the file to the local cache. unless you can make it open the archives without the need for a download.


When Ubooquity indexes files, it opens them to extract the cover and the metadata.

To do that, it needs the complete file (even more so when it serves pages for the online reader), doing so on a file hosted in a cloud is unofortunately not possible.


For cbz (zip files) I made a working prototype in python the basically reads a zip over any remote resource supporting Range access (tested with Google Drive and Hubic). Here there's a gist, based on the original zipfile python module:


Also Pdf files, when "linearized", should also be accessible in "ranges", at least theoretically.

I found that in practice it works generally good (I have actually indexed a library of ~4000 files, more than 200Gb of cbz) but suffers a lot from cloud services performance, that could be unpredictable. Even Google Drive sometimes requires several seconds to access a single zipped image, probably because of authentication infrastructure overhead. When thing go smooth, page download requires ~1-2 second (for pages of ~6-700k) that's enough for most common reading needs.

Interesting, thanks !

Given the limitations (zip only and performances), I don't know if trying to build something efficient using remote resources would be really doable, but at least now I know there is a possibility.