Posted under General
ggraphy said in forum #434305:
Is it supposed to be just detect Japanese text, Chinese text and English text then reorder them in that order for the entire wiki? Sounds doable, though a site-wide implementation could also be more beneficial I think.
I think it's generally best to have "original" names that don't match the tag name prioritized (so Chinese/Korean copyrights get Chinese/Korean names first) so a sitewide forced implementation would interfere with that.
ggraphy said in forum #434305:
Is it supposed to be just detect Japanese text, Chinese text and English text then reorder them in that order for the entire wiki? Sounds doable, though a site-wide implementation could also be more beneficial I think.
For this subset of the wiki edited by AtagoTakaoAzuma, I was planning on having it work based on the fact that each language uses a different set of characters, which is convenient, and I'm sure there's some library out there that will return which glyph set each character is from. (The glyph characters, not fictional characters, to be clear.)
Sitewide: No, I don't see how that could work. You'd need to know the country of origin for every wiki entry. That information isn't available to my knowledge, and there isn't a widespread problem of incorrect OtherName orders outside of this one user's edits anyway (again, as far as I know) so it's not going to be worth trying to get that information just to fix this either.
SilverSpirit said in forum #434331:
Did they offer to make such software? They way I interpreted their comments are that they suggested a custom software would be easier not them offering to make said software. If they are offering to make such software I can tell them what's Japanese and what's Chinese.
I have offered to do this. It'll just be a Python script that talks to the Danbooru API. It won't be a particularly complicated piece of software I don't think.
What I or anyone else who wants to have a go at this needs to know is which wikis are Chinese origin, and so should actually have the Chinese names at the start.
If you are willing to compose a list of Chinese-origin Character and Copyright wiki entries, what would be most useful to me is a list of those tags, in a .txt file, one per line, as they appear on the site.
Depending on how many Chinese wikis there are, this might be quite an undertaking. You can make it a little less work for yourself by listing them with globbing, so for example you could write "*_(wuthering_waves)" so you don't have to list every wiki ending in "_(wuthering_waves)" one by one, as long as the glob only expands to all Chinese wikis.
Probably wait on doing this until there is a plan in place about whether this API approach even makes sense. It would be a real pain for you to work on this and find out we can't even use that work.
zetsubousensei said in forum #434323:
I disagree since people put ship names, birthday tags, or nicknames/abbreviations in other name.
A Chinese character name should have priority over those other ones imo so autosort isn't helpful. I also question how it would handle character and copytags with stylized English titles (or even Chinese copyrights for that matter.)
That does throw a spanner in the works. Are you concerned about that just in the context of this being used whole-site (which I am not advocating for) and think it's an acceptable cost in this specific user's case, or do you think it should stop this user's edits being reordered via the API too?
I don't quite understand what you mean about the stylized English titles or Chinese copyrights.
CrossbowArcanePlus said in forum #434343:
That does throw a spanner in the works. Are you concerned about that just in the context of this being used whole-site (which I am not advocating for) and think it's an acceptable cost in this specific user's case, or do you think it should stop this user's edits being reordered via the API too?
I don't quite understand what you mean about the stylized English titles or Chinese copyrights.
I am specifically addressing ggraphy's suggestion of site-wide implementation. I do not think it'd be helpful to do automatically because of the reasons in that post. It was not in the context of this users edits, which I have not throughly examed
My later point was that if we had a system that automatically prioritized Japanese text in other names then series where the title of the series is in English would be automatically sorted incorrectly, in my opinion. Un-Go's wiki is a good (bad) example of this right now. The series title is written in the Latin Alphabet, and the pixiv results for the proper name are greater than the Katakana, but we still have the katakana name prioritized despite it having fewer results.
Again this is just addressing ggraphy and the idea of site-wide autosort, not specifically the OP topic.
@CrossbowArcanePlus said in forum #434343:
What I or anyone else who wants to have a go at this needs to know is which wikis are Chinese origin, and so should actually have the Chinese names at the start.
If you are willing to compose a list of Chinese-origin Character and Copyright wiki entries, what would be most useful to me is a list of those tags, in a .txt file, one per line, as they appear on the site.
Depending on how many Chinese wikis there are, this might be quite an undertaking. You can make it a little less work for yourself by listing them with globbing, so for example you could write "*_(wuthering_waves)" so you don't have to list every wiki ending in "_(wuthering_waves)" one by one, as long as the glob only expands to all Chinese wikis.
Unfortunately I only have about half an understanding of your suggestion. I have never heard the term "Globbing" before but in terms of the font differences AtagoTakaoAzuma put in the Other Names section I have noticed that they have used a different font for their chinese tags (at least as far as the last two pages of edits are concerned) that make them noticeably different than the default font used for Japanese text here on Danbooru.
SilverSpirit said in forum #434168:
I feel safe in assuming that the majority of those pages are filled with OtherNames wiki edits that contain the Chinese copyright names in front of the Japanese ones for Japanese copyrights.
Oh whoops, I forgot to chime in.
Most of them do yes, some were reverted/fixed but vast majority should still be wrong. Iirc some of the most egregious edits even had the japanese name moved all the way back, which is just great.
Some of the earlier (Jojo) wiki edits had some nonsensical chinese memes in the othernames as well.
I'm 93% certain the new edits (after my neg feedback) are proper edits, but you never know. Most of them appear to be esoteric literally who chinese only gacha slop tags which no names in other languages at all, so at least those are fine.
