I will be attempting to make a federated anime tracker this summer, but I am not quite sure what features people would want and how I would get the details for animes, mangas, etc.

For the latter: What I thought was to either scrape other anime websites continuosly in the background, but this most likely is against the ToS of every anime tracking website, such as AniList or MAL. (I actually asked anidb.net for special access to their DB because apparently you can request access to it, but I've been left on read by the two staff members) My second idea was to make it an anime tracker website where animes are only user-submitted. And the user submissions would be approved by assigned moderators. However, I think this would be quite inconvenient. I'd like to get your opinions and/or ideas for this.

For the former: So if you have any requests or suggestions, please drop it down in the comments section.

Thanks in advance.

you are viewing a single comment's thread
view the rest of the comments

[–] muhyb@programming.dev 8 points 6 months ago (1 children)

It's a slow system. MAL takes user submissions because they already have a big database and those submissions help filling the cracks. If you don't have a database to begin with it slows things down, especially if these submissions will go by trust-based. Doing this by randoms is also very risky.

However, I'm not completely oppose to that idea because it can be helpful for some areas. My suggestion is start with a basic dog tag system, where the anime name, alternative names, status (airing, completed), season and start date, studios (also licensors and producers), age rating etc. These information needs to be scraped for the fastest way to form a quick database, they are publicly available (even on Wikipedia) so it should be fine to scrape. You can even go full Wikipedia after got only the names. User submissions could be useful for the introduction / summary parts of the titles at this stage. For only names (and basic tags), you can scrape AniDB from this list. It's just a search query so shouldn't be against their ToS.

You can also check Kitsu for ideas, I like their DB request system. Pretty basic but can be done differently with the power of ActivityPub.

[–] Asudox@lemmy.world 2 points 6 months ago* (last edited 6 months ago) (2 children)

I see. I can use the Jikan API to scrape animes and mangas which will take approximately 1-2 days after I get approval then. Oh and I forgot to mention, the federation part isn't really how people think it will be I guess. The only federation that will be done will be the reviews, threads and the comments in them. With every anime/manga/vn, etc. being a new community which will contain those threads and reviews. And because of that, I don't really know if this project is something people would want to self host. I guess I could provide full dumps of the database every month or something but I suppose that would be expensive. Then there are images as well, which will take hundreds of GBs easily even in compressed form.

[–] muhyb@programming.dev 3 points 6 months ago (1 children)

Noticed the edit:

For hosting images, you can go alternatives like some Lemmy sites do: Mirror everything automatically to Internet Archive.

I think people would want to self host, because they will get all the anime/manga titles with their communities out of the box and can moderate their own sites while their users can react to any other community via federation.

[–] Asudox@lemmy.world 4 points 6 months ago (1 children)

Wouldn't Internet Archive be a bit slow? And also, I don't want to stress their servers.

[–] muhyb@programming.dev 2 points 6 months ago (1 children)

That's a noble concern. I just gave an example there since some communities do that but yeah it would be better if not done I guess.

Also mentioned on other reply but you can ask this to !selfhosted@lemmy.world and probably will get an optimized answer regarding that issue.

[–] Asudox@lemmy.world 2 points 6 months ago (1 children)

Thanks.

[–] muhyb@programming.dev 2 points 6 months ago

You're welcome, and good luck!

[–] muhyb@programming.dev 3 points 6 months ago (1 children)

Didn't know about Jikan API. After a quick look at their docs, I think it should be a steady source for scraping.

For features, can be done a lot with ActivityPub. Of course the most wanted features would be a watchlist / episode tracker (and possibly an importing from the lists people already have, I switched to Kitsu from MAL that way) but just thinking about federating the all anime/manga titles with basically their own communities out of the box sounds great. Good luck with the project!

[–] Asudox@lemmy.world 3 points 6 months ago* (last edited 6 months ago) (1 children)

Yeah, though I am not sure how the federated instance admins would react. I am planning for every anime, manga, vn, etc. to have their own communities. This means about over 100k communities being made in an instant. Maybe instead of creating the communities in an instant, creating them when user activity first happens would be more fit. But this would also restrict other platforms' users being able to comment on never heard or new anime entries until someone from the anime tracking platform comments or reviews them.

Thanks btw.

[–] muhyb@programming.dev 1 points 6 months ago (1 children)

Both have ups and downs. Assuming these lists will be in the code, do you have an estimation how big would that be? If you think they won't strangle the code, just go with it. Something like storing them in JSON and loading them when needed could be better for optimization though.

You can also do some best of both worlds, like not creating the communities beforehand but make the titles searchable from the database open for all users. That might require a bigger traffic from hosting side though, but it should be OK since these will be spread to all self hosted communities.

I think you can also ask some of your questions to selfhosted communities.

[–] Asudox@lemmy.world 1 points 6 months ago (1 children)

What do you mean by "lists in the code"? Which lists and do you mean by "in the code" hardcoded?

[–] muhyb@programming.dev 2 points 6 months ago (1 children)

Well, since 100k titles would make a pretty big database, at least storing the metadata like years, seasons, genres etc as hardcoded could make it run faster than going full-fledged JSON, at least I meant that. However this will be an open source project and there will be localizations, so now it doesn't look like a good idea to me somehow.

[–] Asudox@lemmy.world 2 points 6 months ago (1 children)

What? What do you mean full-fledged JSON. I won't be storing the animes in JSON, but in a PostgreSQL DB. I don't understand what you mean...

[–] muhyb@programming.dev 2 points 6 months ago (1 children)

Sorry for the confusion. Don't take my words literally, I'm not an expert on databases. I use JSON for my job and it's enough for me, so use what database your project needs. But by full-fledged I meant completely leaning on databases which may slow some things down for the server side. However I now checked PostgreSQL and it seems much better for large datasets, so I don't think that would be a concern for you.

[–] Asudox@lemmy.world 2 points 6 months ago (1 children)

Oh, I see. Ok, I get it now. JSON indeed would be very slow for this project. I also do think of maybe caching some stuff for this reason.

[–] muhyb@programming.dev 2 points 6 months ago (1 children)

Yeah, that would be nice for the running seasons I guess. Maybe even a couple years worth seasons since most of the traffic will come from those.

[–] Asudox@lemmy.world 2 points 6 months ago

Yep, that's what I was thinking.