[X] My Assistant

So did you conciously change the behaviour of the flag? E.g. using it with the /tag/ endpoint it seems to work as expected:
http://ehentaihip.com/tag/cnaruto
http://ehentaihip.com/tag/cnaruto&skip_mastertags=1

But it doesn't work with the f_search option in the URL anymore:
http://ehentaihip.com/?f_search=cnaruto
http://ehentaihip.com/?f_search=cnaruto&skip_mastertags=1 <- still gives the results of the master.

I'm very sure that before, the latter option would also work. Without that, I don't see how the flag has any merit whatsoever (to a normal user).

It doesn't have any merit to a normal user and is not supposed to, it's an internal flag for use with internal tools and has never been advertised as anything else.

I did recently correct an issue where it would still attempt to apply it to normal searches if you provided it manually. Not just for the hell of it, but because it would corrupt the search caches so that if you then did the search normally, it would use the previously cached result and still not include the master tags.

It doesn't have any merit to a normal user and is not supposed to, it's an internal flag for use with internal tools and has never been advertised as anything else.

Maybe you didn't advertise it as such but the wiki certainly recommended the option to normal users (and still does in other places, though I trust Shank will take care of them soon).

To present an actual use case I had today, I wanted to find all unlsaved temp tags starting with "viper", to find un-namespaced tags in the obscure viper series of video games. I know the search tool under My Tags does something similar but it has limited yield and includes slaved temp tags, so it's not as efficient. But if there is a technical reason to throw it out, that's okay. Still sad to see the functionality go, it was a good friend of many years.

certainly recommended the option

Already removed

(and still does in other places, though I trust Shank will take care of them soon).

Now taken care of

Skimming through some old posts has reminded me that there was some demand for being able to do something like "find all galleries with no tags in the artist namespace". I still think that could be useful. Before this big search engine update, it was said to be not viable due to technical limitations. Is it any more viable now?

Skimming through some old posts has reminded me that there was some demand for being able to do something like "find all galleries with no tags in the artist namespace". I still think that could be useful. Before this big search engine update, it was said to be not viable due to technical limitations. Is it any more viable now?

It's probably possible to support it, but there are no plans to add it in the near future.

can support add uploader to watched?

Hey, I'm trying to use the OR operator like this:
parody:naruto$ ~parody:original$
What I'm aiming for is to filter doujishis with a parody of Naruto, otherwise just show me the original ones. I think I might be using it incorrectly . In the end, I'm trying to find a way to search for both manga and doujishis, but if it's a doujishi, only show me the original ones. tnks for any kind of help

parody:naruto$ ~parody:original$
What I'm aiming for is to filter doujishis with a parody of Naruto, otherwise just show me the original ones.

Try: ~parody:naruto$ ~parody:original$
If you are doing an OR search, you need at least two search items with "~" for it to make sense.

In the end, I'm trying to find a way to search for both manga and doujishis, but if it's a doujishi, only show me the original ones. tnks for any kind of help

I don't think this is possible, you will have to do two separate searches (one for all manga, one for only original doujinshi).

Try: ~parody:naruto$ ~parody:original$
If you are doing an OR search, you need at least two search items with "~" for it to make sense.
I don't think this is possible, you will have to do two separate searches (one for all manga, one for only original doujinshi).

Tnks! that one works on the naruto example, a shame about the second one but it is no a big problem!

An issue with the search engine came up in the Renaming/Reclassing thread. Someone uploaded a set of galleries named like,
𝙈𝙀𝘿𝙐𝙎𝘼|𝙍𝙄𝘿𝙀𝙍 (𝙁𝘼𝙏𝙀 𝙎𝙀𝙍𝙄𝙀𝙎) [𝙋𝙄𝙓𝙄𝙑] 7 {𝘼𝙞 𝙂𝙚𝙣𝙚𝙧𝙖𝙩𝙚𝙙}

using the [codepoints.net] Mathematical Alphanumeric Symbols Unicode block, Mathematical Bold Italic characters, for extra style.

The search engine does not normalize those to regular english text, so they aren't searchable.

http://ehentaihip.com/?f_search=𝙈𝙀𝘿𝙐𝙎𝘼
"No hits found." (They all got renamed; it would normally be bottom priority because it shouldn't affect the search engine.)

http://ehentaihip.com/?f_search=MEDUSA
"Found 369 results."

Startpage, DuckDuckGo and Wikipedia will all find Medusa if searched with 𝙈𝙀𝘿𝙐𝙎𝘼.

There are a lot of characters in Unicode that are similar to ASCII alphanumericals, and even if we were to try to find all of them, people are always going to find new ones. We do remap the ones that are ｃｏｍｍｏｎｌｙ　ｕｓｅｄ　ｉｎ　Ｊａｐａｎｅｓｅ for searching/indexing purposes, but I don't really feel like spending the time trying to create a map of all the other variants.

Incidentally, you say it works on Wikipedia, but Mediawiki doesn't seem to be doing it, so I can't just grab a map from there. Could be Wikipedia runs on a newer version though.

I think there's probably an open source library that will handle normalizing unicode text like that.

[www.unicode.org] Unicode Standard Annex #15 - Unicode Normalization Forms

Maybe this.
[github.com] Unicode Normalization forms according to UAX#15 rules

[unicode.org] Unicode FAQ - Normalization
NFKD normalization form looks appropriate for search.

Example of NFKD normalization using an online Unicode transformer.
[util.unicode.org] https://util.unicode.org/UnicodeJsps/transf...%F0%9D%99%99%7D

EHwiki search box doesn't do it,
https://ehwiki.org/index.php?search=%F0%9D%...&fulltext=1

But Wikipedia does.
[en.wikipedia.org] https://en.wikipedia.org/w/index.php?fullte...earch&ns0=1

MediaWiki's documentation page:
[www.mediawiki.org] Unicode normalization considerations

Since 1.4, MediaWiki applies normalization form C (NFC) to Unicode text input.

[www.mediawiki.org] utfnormal

utfnormal is a library that contains Unicode normalization routines. It includes pure PHP implementations, and automatically uses the php-intl extension if installed.

This post has been edited by Necromusume: Oct 27 2023, 23:23

Small oddity:

Searching o:comic$ in western gives "50,000+ results".
Seacrhing other:comic$ in western gives "about 70,000 results".

Both searches seem to yield exactly the same (as expected) but for some reason the result count estimation is more specific for the second one.

Both searches seem to yield exactly the same (as expected) but for some reason the result count estimation is more specific for the second one.

Count estimates are refined as the result set is explored (which is basically free), so most likely the second one just has had more pages sampled.

I think the new engine is really subtle...