Philomena open beta / breaking API changes

when you click on "view all" next to "recent forum posts" on profile page and then click next page it shows posts from random people but it also changed the search from user_id:334626 to created_at.gte:1 week ago

More demons, more birds!

@RicardW
@byte[]

Here you go:

[example] GET /api/v1/json/tags/:slug

Navigation is done exclusively with the slug. The linked tags (in aliased_tags, etc.) are slugs.

1. Tags with apostrophes in them give 404 errors. Example

2. Text parser for description fields breaks links that end with .php extension. Example: Example

The code for the above link was "Example":https://example.com/index.php

3. Profile badges are drawn out of chronological order.

4. No way to access list of associated user links nor the creation link from user's page. User links , User links/New

Not sure I understand the API changes…

Once the migration to Philomena is done, the old API stays under that /v1/ route. But do the new API routes stay the same as they are now? Because I can see that this has the exact same structure ans is on the exact same route as the old API was. The current API docs are also the same as the new API docs.

While we're talking about the API, will the new backend (if not now then maybe in the future) have OAuth or something that'd let users using 3rd party apps vote on and fave pictures, comment, stuff like that?

More Philomenas!

@DarthBarf
Yeah I'm trying to figure it out. It works in local development environments but not in production (the best kind of bug).


@Angius
That's the backwards-compatible fill for old API clients, as mentioned in the OP. The new API is exclusively under /api/v1/json (and /api/v1/rss for the watched feed). If you're confused, please have a look at the router:

https://github.com/derpibooru/philomena/blob/5f9fe0cb98cee76c3b58b932f7103dea4975a859/lib/philomena_web/router.ex#L77-85

@Angius
I'll also throw in that the API docs are just a blob of HTML in the database and so they're rendering the same. I haven't updated the page yet.

@byte[]
First, thanks for doing that so quickly! Also, thanks for resolving confusion about slugs vs tag ids that was in the previous API.

There might be an escaping problem with tag slugs in general in the new codebase. I looked in my local tag json directory and started testing anything which looked like it had a slug with a lot of escaping. Sure enough, there's at least a problem with the url escaping of '(' and ')'. It looks like this is a more general problem with tags though. If you go to https://philomena.derpibooru.org/1790981 and click on the bass guitar tag, you get a valid page, but if you click on you spin me round (like a record) (dead or alive) , you get a 404.

@byte[]

Gotcha, so old API is under the current routes, and the new API is under /v1/.

@RicardW
It's because of the parentheses

I already reported it in here and is being/has been looked into
Also happens with ! and according to Darthbarf it also happens with apostrophes

@DarthBarf
@RicardW
@Rainboom Dash

Tag slug problem is now fixed, I hope.

@byte[]
When I first saw this I immediately became scared of a massive makeover and/or functionality changes. (not saying it's always a bad thing, but I've seen many sites confuse changes and improvements are not synonymous). Glad that fear isn't being realized and best of luck with it.

I forgot that Elixir doesn't sort dates according to their natural order, it sorts them by the order of the fields in the struct. Badges now appear in the correct order.

- In settings Philomena caps the comments per page at 50, whereas it's 100 on the main site. Is this an intended change?

- On images with locked comment section, the comment reply links are not formatted.
e.g. >>409163 NSFW

- Using the next/previous link on a image without any query supplied (/images/2175695/navigate?rel=next) will just lead back to itself.

- Is there/will there be a new API equivalent for accessing an image's next/previous page?

[Current API] /next/2175695.json?q=featured+image

I'm compiling changes to the API in a gist. At least the changes that affect my project. It's public, so feel free to submit any revisions and use it in general.

That out of the way, random image seems to be broken. Old API returned just an ID of a random image, the new API returns basically the entire search results.

[Old API] /search.json?q=pinkie+pie&random_image=y
[New API] /api/v1/json/search?q=pinkie+pie&random_image=y

@Marker
- No, I'll change it back tomorrow.
- Parser bug. Will fix.
- Oops.
- Sure

@Angius
There is no random_image parameter in the new API. Set sf to random and pick the first image out of the response.

There is also some other stuff that was not intended to be removed (like totals). I'll add them back.

- When searching for comments, the "Search" button changes to say "false" during the search. It is correctly grayed out, just the text is incorrect. The same also happens during forum post search.

- "Most commented-on images" does not work (just returns the same results as the front page).

- Site statistics page is missing Galleries, Commissions, Moderation and Uploads by day sections

@JP
"false": That would be because my finger slipped or something

When i click on "rankings" i arrive on a blank page with just "bad request" writed

Tag implication lists have no commas (on the Rails side there are commas) separating tags; combined with the decreased line height, it makes tag implication lists really messy and almost impossible to read.