Where exactly does Search search?

[WoodyBABL]

Using 10.11.2.

Was trying to put together a collection of Christmas episodes, so I searched for Christmas... naturally. 

I noticed that an episode of Newhart was missing. The episode description contains "24 guests are snowed in on Christmas Eve." 

I thought maybe it needed to have a tag..So I added the tag Christmas to the episode metadata.

Did a search for Christmas and the episode still didn't show up. Tried the search through the Android TV client and the Android phone client. Any idea why the search is ignoring it?

[crobibero]

The title. That is it

[WoodyBABL]

Wow. That seems limiting.

Just now I went to the Shows library and filtered by the tag Christmas. I see that only picks up if the tag exists for the series not individual episodes.

[kandykarter]

Jellyfin's built-in search is getting better, but it's pretty limited, still. There's a plugin where you can connect it to an instance of Meillisearch, which works well (at least for me, on docker, linux, etc, your mileage may vary) and tends to improve results quite a bit.

My favourite example is that if I search "opera" in the standard Jellyfin search, the movie Opera (1987) is the third result. With the Meillisearch plugin, it's the first.

https://github.com/arnesacnussem/jellyfi...eilisearch

[WoodyBABL]

Thanks. I'll give that a look. It's interesting that the stock search will hit Titles and People, but not Tags.

[kandykarter]

They did implement tag search a couple versions ago but for some reason it made searches REALLY slow, so they shelved it. I suspect it's probably going to be back in the mix soon now that the efcore refactor is done.

[giannisg2004]

Thanks. I'll give that a look. It's interesting that the stock search will hit Titles and People, but not Tags.

Another vote for Meilisearch from me. It is great. Aside from title, it searches by overview (i.e. the keyword appearing on the short plot summary you see when you click on the movie), by production year, by studio, by genre, by tags. And it doesn't need exact matches, it implements a kind of fuzzy search, so typos and spelling mistakes are still providing results.

And the best part? It's actually quite a bit faster than the stock search.

The only downside? It needs some resources for running its database and index. My Meilisearch container, with the (optional) WebUI installed and running, sits at 1.5GB RAM servicing a 360MB Jellyfin index.

But if you have the RAM to spare, it's a no-brainer.