Grimmory not respecting sidecar settings

[secondsabre]

This might be related to one of the other sidecar-related issues, but they didn't seem to match exactly and I wanted to draw attention to this issue in particular; apologies if it's a duplicate. Ran all these tests on a new deployment of Grimmory v3.3.1 on my NAS: books and app both reside on the same system and DISK_TYPE=LOCAL.

Situation: Working on rebuilding/reimporting my library from Calibre. I built a python script to generate Grimmory-compliant sidecar files for each of my ebooks by parsing the metadata.opf, and it seems to work just fine. Syntax is correct, structure is good, everything matches against a sidecar that's generated by Grimmory itself.

When I got to actually create my library (import the files), no matter if I tell Grimmory to use "sidecar only" or even "none", it always seems to grab the metadata from the ePub files themselves. Sidecar is enabled in the settings, and I've also disabled all metadata import from other online sources, so it must be pulling from the ePubs themselves, right?

If I go into the library options and use the bulk "Import Sidecar" button, we get a success message, but the information doesn't actually show in the metadata until I close the Grimmory tab and reload it from scratch. Cache is off, tried force-refresh, all that kinda stuff, but closing and re-opening seems to be the only way to trigger the changes.

And even then, it accepts the info from the sidecare, but still retains any info that it grabbed from the ebook itself. For example, I map my calibre "tags" (which are actually genres) to grimmory's tags, with the intent to use Grimmory's scraper to fill in the genres on that side. After reloading, the tags are correct, but the genres that came in with the book itself remains, so a Fetch Metadata > Replace Missing Only won't load up new genres because the field is already populated.

Basically, when you create a library and tell Grimmory to only load meta from sidecars, or tell it to not grab meta at all, it should respect that to allow users better control over how their library is popualted.