Danbooru

Danbooru 2 Issues Topic

Posted under General

This topic has been locked.

I find it annoying now that all of my saved searches are now on my profile page. Can we have an option to toggle which ones get shown and which ones don't?

Also, however you displayed each of those saved searches, it's screwed up. It's apparently supposed to show the combined searches of each category but I'm seeing multiple instances of the same category with the same results.

Updated

Yeah... 'bout that,

Inspired by tapnek, I went and looked at my profile page. Subscriptions, all present and correct; just dandy, nothing could be finer.

The saved searches repeats the same block of 10 posts around 210 times.

tapnek said:

I find it annoying now that all of my saved searches are now on my profile page. Can we have an option to toggle which ones get shown and which ones don't?

Also, however you displayed each of those saved searches, it's screwed up. It's apparently supposed to show the combined searches of each category but I'm seeing multiple instances of the same category with the same results.

Yeah, I concur. I even brought this up in the Saved Searches topic (forum #118149). I have over 30 saved searches for my tag gardening efforts that I don't want to see on my profile page. Please make this toggle-able, like Tapnek said...

It seems to be a problem with parsing categories of saved searches: rather than a single search per category, you get a search of the whole category for each individual search in it (so, say, if you have a category with 5 artists separately, you will see your whole artists category 5 times).

EDIT: Oh, and also I'm not too fond of it cluttering my profile. Please let me hide it.

Provence said:

Would it be possible to implement adding the source via the tag script?
I just tried it with that: source:https://exhentai.org/g/969119/722513d6e3/ on post #2464634
This could become neccessary if there are a lot of posts uploaded but without source.
So if I find the source, I've to manually fill in the source field. Since this works with parents, ratings, pools and every tag category, I thought it might also work with sources, but apparantly no?

Just tested this out with straight up editing, and it also holds true.

JSON Query after source: added to Tags
{"post_id":1283978,"tags":"1girl bangs black_legwear blouse brown_eyes brown_hair brown_shoes character_sheet concept_art cross full_body girls_und_panzer green_skirt grey_background kneehighs loafers long_sleeves looking_at_viewer miniskirt neckerchief nishizumi_miho official_art pleated_skirt ribbon school_uniform serafuku shoes short_hair simple_background skirt smile socks source:http://girls-und-panzer.jp/chara_miho.html standing sugimoto_isao text translated white_blouse","updater_id":23799,"updated_at":"2016-08-25T16:19:19.346Z","rating":"s","parent_id":null,"source":"","id":16221665,"added_tags":"source:http://girls-und-panzer.jp/chara_miho.html","removed_tags":"","obsolete_added_tags":"","obsolete_removed_tags":"","unchanged_tags":"1girl bangs black_legwear blouse brown_eyes brown_hair brown_shoes character_sheet concept_art cross full_body girls_und_panzer green_skirt grey_background kneehighs loafers long_sleeves looking_at_viewer miniskirt neckerchief nishizumi_miho official_art pleated_skirt ribbon school_uniform serafuku shoes short_hair simple_background skirt smile socks standing sugimoto_isao text translated white_blouse rating:s","updater_name":"BrokenEagle98"}
JSON Query after source: added to Source and Tags
{"post_id":1283978,"tags":"1girl bangs black_legwear blouse brown_eyes brown_hair brown_shoes character_sheet concept_art cross full_body girls_und_panzer green_skirt grey_background kneehighs loafers long_sleeves looking_at_viewer miniskirt neckerchief nishizumi_miho official_art pleated_skirt ribbon school_uniform serafuku shoes short_hair simple_background skirt smile socks source:http://girls-und-panzer.jp/chara_miho.html standing sugimoto_isao text translated white_blouse","updater_id":23799,"updated_at":"2016-08-25T16:21:26.569Z","rating":"s","parent_id":null,"source":"http://girls-und-panzer.jp/chara_miho.html","id":16221671,"added_tags":"source:http://girls-und-panzer.jp/chara_miho.html source:http://girls-und-panzer.jp/chara_miho.html","removed_tags":"","obsolete_added_tags":"","obsolete_removed_tags":"","unchanged_tags":"1girl bangs black_legwear blouse brown_eyes brown_hair brown_shoes character_sheet concept_art cross full_body girls_und_panzer green_skirt grey_background kneehighs loafers long_sleeves looking_at_viewer miniskirt neckerchief nishizumi_miho official_art pleated_skirt ribbon school_uniform serafuku shoes short_hair simple_background skirt smile socks standing sugimoto_isao text translated white_blouse rating:s","updater_name":"BrokenEagle98"}
JSON Query after source: added to Source
{"post_id":1283978,"tags":"1girl bangs black_legwear blouse brown_eyes brown_hair brown_shoes character_sheet concept_art cross full_body girls_und_panzer green_skirt grey_background kneehighs loafers long_sleeves looking_at_viewer miniskirt neckerchief nishizumi_miho official_art pleated_skirt ribbon school_uniform serafuku shoes short_hair simple_background skirt smile socks standing sugimoto_isao text translated white_blouse","updater_id":23799,"updated_at":"2016-08-25T16:21:57.553Z","rating":"s","parent_id":null,"source":"http://girls-und-panzer.jp/chara_miho.html","id":16221672,"added_tags":"source:http://girls-und-panzer.jp/chara_miho.html","removed_tags":"","obsolete_added_tags":"","obsolete_removed_tags":"","unchanged_tags":"1girl bangs black_legwear blouse brown_eyes brown_hair brown_shoes character_sheet concept_art cross full_body girls_und_panzer green_skirt grey_background kneehighs loafers long_sleeves looking_at_viewer miniskirt neckerchief nishizumi_miho official_art pleated_skirt ribbon school_uniform serafuku shoes short_hair simple_background skirt smile socks standing sugimoto_isao text translated white_blouse rating:s","updater_name":"BrokenEagle98"}

As can be seen, only by adding the source to the Source entry does the 'source' field get updated. When it is added to both Source and Tags entries, it creates two identical strings in the 'added_tags' field which should be impossible, so there's also some input checking that is being missed since the existence of two identical tags should not be allowed.

There's a difference between how DText is previewed from how it is rendered.

Observe:

Header5

Text

Header6

Text

h5. Header5
Text
h6. Header6
Text

On the DText preview, the above appear to be the same. When rendered, they are different.

[[kantai collection]]

[[kantai collection]]

On the DText preview, the above appear to be the same. When rendered, there is a mistake in the former item.

To see the above in action with the DText preview, just quote this post, and click the preview (please don't submit it again though... :)

The above behavior was noted when editing the help:dtext wiki entry.

BrokenEagle98 said:

There's a difference between how DText is previewed from how it is rendered.

Observe:

Header5

Text

Header6

Text

h5. Header5
Text
h6. Header6
Text

On the DText preview, the above appear to be the same. When rendered, they are different.

[[kantai collection]]

[[kantai collection]]

On the DText preview, the above appear to be the same. When rendered, there is a mistake in the former item.

To see the above in action with the DText preview, just quote this post, and click the preview (please don't submit it again though... :)

The above behavior was noted when editing the help:dtext wiki entry.

This has to do with new DText parser implementation, based on state machine rather than regexp. @albert

On an unrelated note, Artcomm listing fails with "something broke" and no debug info. Can't investigate, empty listing on debug machine works fine, listing on testbooru does work as well.

Type-kun said:

On an unrelated note, Artcomm listing fails with "something broke" and no debug info. Can't investigate, empty listing on debug machine works fine, listing on testbooru does work as well.

These posts in particular are failing:

http://danbooru.donmai.us/artist_commentaries?search[post_id]=2469261
http://danbooru.donmai.us/artist_commentaries?search[post_id]=2469237
http://danbooru.donmai.us/artist_commentaries?search[post_id]=2469172
http://danbooru.donmai.us/artist_commentaries?search[post_id]=2469154

Other posts work, and JSON responses work too. The common feature here is that translated_title/translated_description are null instead of the empty string (as it's supposed to be). So it's probably a null deref in the view. Also the view was recently updated to use the new DText implementation, so that's probably the root cause, it returning nulls instead of the empty string. I don't have my dev environment set up any more though so I can't test further.

HaxtonFale said:

I've only now noticed that if you pop out the tagging box, the font for tags is different - no longer monospaced.

So that's what it is. I always thought something slightly off about the popout edit box. Anyway, put this into your custom CSS in your user settings:

.ui-widget #form textarea {
  font-family: monospace;
  font-size: 1.2em;
}

Jarlath said:

I'm seeing each saved search plus the main "saved searches" repeating able 30 times here. It looks like it gave up after the second category and is repeating it forever.

Looks like the bug is just that the categories are not being uniquified here.

RE: hiding saved searches, I think just giving them CSS classes so they could be hidden with user CSS would be sufficient.

I would consider spaces before a header tag to be invalid syntax and that new parser correctly ignores that case. Is there ever a legitimate case for that?

The nested bold tags and wiki links are, uh, interesting. I think I understand what's going on there and can probably fix it.

evazion said:

Looks like the bug is just that the categories are not being uniquified here.

RE: hiding saved searches, I think just giving them CSS classes so they could be hidden with user CSS would be sufficient.

@evazion
Can you tell me how to do that?

albert said:

I would consider spaces before a header tag to be invalid syntax and that new parser correctly ignores that case. Is there ever a legitimate case for that?

The nested bold tags and wiki links are, uh, interesting. I think I understand what's going on there and can probably fix it.

The spaces before the headers are used primarily in the help:wiki to keep the DText renderer from acting on it, so that the proper DText syntax can be shown.

Unfortunately, the [code] label adds the "<pre>" html label which seems to do some additional text formatting, which is why it isn't used on that wiki page.

For the following, click Quote to see the actual DText being used.

Example with [code]:

[b]Bold Text[/b]

Example with interleaving of [b] labels:

[b]Bold Text[/b]

Would it be possible to have something like a [nodtext] label that does no processing on the internal text, yet also doesn't add additional HTML labels like the [code] label does?

Just for reference, I've had to use the interleaved [b] labels quite a few times, especially when demonstrating how to use DText to a new user.

@Sacriven:

Well right now it's tricky. You can add this to your custom CSS on your settings page:

#c-users > #a-show > div:nth-child(4) ~ .box {
  display: none;
}

But that will hide everything that comes after your favorites. If you have tag subscriptions you'll have bump up the nth-child(4) bit to nth-child(9) to not hide them (assuming you have 5 subscriptions).

Making it not dependent on the number of tag subscriptions would require albert or Type-Kun to change the site to give saved searches their own CSS class.

evazion said:

@Sacriven:

Well right now it's tricky. You can add this to your custom CSS on your settings page:

#c-users > #a-show > div:nth-child(4) ~ .box {
  display: none;
}

But that will hide everything that comes after your favorites. If you have tag subscriptions you'll have bump up the nth-child(4) bit to nth-child(9) to not hide them (assuming you have 5 subscriptions).

Making it not dependent on the number of tag subscriptions would require albert or Type-Kun to change the site to give saved searches their own CSS class.

I've made the necessary changes (and also fixed uniqueness issue). But you'll have to wait until Albert deploys it, as always.