[Feature Request]: Exact Searches

[AdamGaffney96]

I've read the documentation

• I have read the how to open an issue guide, particularly the feature request section.

Your Feature Request

Is your feature request related to a problem? Please describe.

The issue I'm having is related to the search functionality. As you can see from the below screenshot, I have a bunch of seasons of a YouTube show downloaded. Unfortunately due to errors I was getting with playlist downloads specifically (I think related to the removal of the About section I've seen mentioned here in other places), I had to download all these episodes individually. What this means is that they're not easily grouped for me to just move through them one-by-one.
What I wanted to do was just search for "Game changer season 2" to get Season 2 of the show, as they're very consistent with their video titling, however seemingly due to a lack of exact searching, I'm getting a slightly annoying issue where other seasons are showing up. The issue is clearly that Episode 2 of all other seasons clearly have every keyword I'm looking for in them, so these pull up alongside actual Season 2, and even some other episodes of the show that just happen to have "2" in the title somewhere. I can't find anything in the docs that refer to exact searching, the way you would use quotation marks in Google to look for exact word order and phrasing.

Describe the solution you'd like

I'm listing this as a feature request as I'm just not sure if this is actually already built-in or not and I just can't find it. So ideally it's already in there and I'd like someone to just point me in the direction of it as I can't find it in the docs. Otherwise, I would like a way to search for an exact ordered list, perhaps using quotes "season 2" or just using another keyword like exact: season 2 for example.

Additional context

I am currently on v0.4.2 on TubeArchivist. I updated my docker-compose to try pull the latest, but I was getting errors and TubeArchivist was no longer loading so I've reverted back to my last working version until I get a chance to test that again. I'm hoping I'm just missing where the feature is listed in the docs, but if not it'd be great if it was implemented (and if it is implemented already in newer versions please do let me know).

Any help would be appreciated, thank you!

Your help is needed!

• Yes I will work on this in the next few days or weeks.

[bbilly1]

Current searching uses bool_prefix, to that doesn't matter in whitch order the terms are, that's why you see the results you see.

We could tweak the score and rank nearer matches higher, or implement as you suggest some exact matching to use phrase search instead.

Search will need pagination as a first priority, so additional improvements are quite far out.

[kureta]

I have a related but different problem. I can open a new issue if necessary. The problem is, when I search for the word "Laplace", first 8 results are "replace", "replacement", "replacing", the word "Laplace" comes as the 9th result, then another "replace", then the second match containing the word "Laplace". This doesn't make sense to me but I don't know anything about elastic search. I guess it is looking up using a word dictionary and fuzzy matches with the words in that dictionary are prioritized before exact matches with a non-existing word.

[bbilly1]

I guess it is looking up using a word dictionary and fuzzy matches with the words in that dictionary are prioritized before exact matches with a non-existing word.

You can already configure fuzzy matching. Take a look at the docs.

[kureta]

I see. Just checked the docs but this doesn't exactly solve the issue. I want fuzzy search. "Laplace" should also match "Laplacian" but as I increase fuzziness "replace" overtakes "laplace". I can understand "replace" overtaking "laplacian" but "laplace" is a perfect match, why does it fall below replace?

[bbilly1]

This is a good starting point of you want to learn more how ES handles search relevance: https://www.elastic.co/what-is/search-relevance