@Viet 先生、I’m really sorry to bother you with this, but when I try to open the item この前 it says “Quite an explorer, are you?” and does not open…
Other items, including those of higher level, open normally.
I’ve tried logging in and out and pressign Ctrl-F5 - it didn’t help…
8 Likes
Can confirm this happens to me as well.
4 Likes
Same. It’s interesting since other 前 words open fine. I wonder if they tried to delete the vocab item and it only got half deleted.
2 Likes
Can confirm. I just did it in my lessons normally, but can’t access it outside of it
2 Likes
I just tried a few things with the V2 API:
/subjects?ids=3457 → responds with この前 entry
/subjects?slugs=%E3%81%93%E3%81%AE%E5%89%8D → no results
/subjects/3457 → gets the proper entry
It looks like it’s just the slug that’s broken somehow…
3 Likes
Replicated the bug. Not sure what the slug API thing is, but it probably shouldn’t be necessary to see the entry 
1 Like
A slug is a human-readable id. If you were to go to the vocab page for 一人, you’d find that in the URL, so that’s the slug for that vocab.
%E3%81%93%E3%81%AE%E5%89%8D is simply この前 escaped for use in URLs; even if you don’t see it, that’s how that string gets sent to the server.
Now, Wanikani navigates the main app using the slugs (I tried using record IDs, but no good there). So, if the slug is broken, then the link will be too.
2 Likes
I just tried every vocab that uses 前 as a component. In addition to この前, the vocab お前 is also broken.
The Wanikani link, as well as the document_url from the API is broken, and it behaves the same as with my previous API tests.
3 Likes
So, “omae” wa mou shindeiru? 
I would be tempted to say this is a kanji-wide problem, but if this only affects two of many vocab with the kanji, it’s probably not just the 前 slug but something about how those two specifically interact- either with themselves, or with the server. (Of course, “freak inexplicable problem that refuses to fix” is an option too.) Were there any problems before the maintenance work, perhaps?
(disclaimer: I have a little bit of programming experience, but not much, so ignore me if I’m wrong or whatever.)
8 Likes
I only tested the 前-containing vocab, about 16-ish subjects; I don’t have the time to test all 8000-odd subjects. It’s possible that only those two are broken, but I wouldn’t assume that, especially since there’s an easy, automatable method to test for issues (using JSONPath notation for brevity):
- Get a list of all subjects from the v2 API
/subjectsendpoint. - Grab all the wanikani URLs
$.data[*].data.document_url - For each URL, test by requesting. Failing requests redirect, if you can detect that. If you can’t, look for the string
Quite the explorerin the response body
I’ll see if I can work something up to test everything.
3 Likes
Are you sure you have time to do that? Today is bargain day at the supermarket!
5 Likes
Well the rate limiting is slowing me down, but I think I’ve got a script that’ll work. It just needs to be run a bunch of times.
2 Likes
It just ended though...
5 Likes
Spoken like a true C-class hero!
1 Like
It’s a clue!
4 Likes
I think I’ve solved the rate-limiting problem. I’m testing all vocabulary now.
2 Likes
Hoo, boy, everyone. This one was a real head scratcher. After checking a lot of things, it turns out the index for slugs on the database got messed up/out of sync during yesterday’s maintenance window. The database wasn’t finding the subjects, even though the records existed in the database. 
Index rebuilt, subjects once again findable.
11 Likes
We couldn’t let you do maintenance with any bugs.
3 Likes
I mean, we’ve gotta keep it interesting, yeah?
6 Likes