ajp_anton/jellyfin-multilang
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Release 0.2.5: harden proxy and streamline metadata fetching

Browse Source 
Remove obsolete rule formats, batch SQLite work, preserve artwork source URLs, and coordinate refresh workers with maintenance. Add regression coverage and document backup format changes.
This commit is contained in:
ajp_anton
2026-09-09 01:18:43 +00:00
parent 7eb62b156e
commit e36775f1a3
49 changed files with 2126 additions and 1813 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+17 -3
View File
@@ -6,6 +6,17 @@ languages without changing Jellyfin's stored metadata.
The plugin is built for the Jellyfin 10.11.x stable series on Linux x64 and is
currently early beta software.
## Upgrading to 0.2.5 or 0.3.4
Existing plugin data is upgraded automatically. Make fresh database exports after
upgrading: older translation-database exports cannot be imported by these versions.
User-settings-only exports keep their existing format. Obsolete text-based rules
are no longer supported; current classification rules are unaffected.
The first missing-data scan records the new per-language and per-provider fetch
state, so it may fetch more than subsequent scans. Later scans retry only missing
data when due and repair downloaded artwork without refetching its metadata.
## What It Does
Multilang can fetch and display metadata and artwork in chosen languages for:
@@ -87,9 +98,12 @@ the UI.
Some requests can be slow because Multilang may need to ask Jellyfin for a
larger list than the UI will finally display. This is needed so the plugin can
translate, filter, sort, and slice the list using the user's own rules.
Multilang caches slow transformed responses for a configurable time and also
uses a very short burst cache for repeated requests during navigation. Cache
entries are invalidated when affected items are refreshed.
Multilang caches Jellyfin's responses for a configurable time, sharing the fetched
list across pages and sort orders. Current translations and user rules are applied
each time that list is used. Simultaneous requests share an in-progress fetch, and
a short burst cache also handles fast repeated requests. Library, permission and
watched-state changes invalidate affected caches. Movie and show libraries can
optionally be fetched in advance when a user becomes active.
## Original-Language Support