Topic with Japanese in URL doesn't redirect if URL doesn't perfectly match

seanblue · 10월 16, 2019, 11:10오후

Does that mean that when people switch from encoded to ascii it doesn’t regenerate the slugs? If so it seems like the one time process you wrote to clear the slugs wouldn’t work for those cases.

Arta_S · 10월 16, 2019, 11:28오후

Hello,

The pull did not fix the issue completely. If the slug contains a + sign, it will show the 400 error.
To confirm this, I have switched back to “None” for slug generation, edited the title and removed the + sign from title and switched back to “encoded” it fixed the 400 error.

The scenario is when a title includes a plus sign, it returns 400 error. (There could be other characters that could lead to this. I only tested the + sign)

Thank you

Falco · 10월 17, 2019, 12:23오전

Is that on a topic slug or in a category slug?

Arta_S · 10월 17, 2019, 12:25오전

A topic slug. Didn’t try on the categories.

Falco · 10월 17, 2019, 12:33오전

The encoded slug generator strips all + signs so I have no idea how you get that:

github.com/discourse/discourse

lib/slug.rb

main


      
            end
          
            def self.ascii_generator(string)
              I18n.with_locale(SiteSetting.default_locale) { string.tr("'", "").parameterize }
            end
          
            def self.encoded_generator(string, downcase: true)
              # This generator will sanitize almost all special characters,
              # including reserved characters from RFC3986.
              # See also URI::REGEXP::PATTERN.
              string = string.strip.gsub(/\s+/, "-").gsub(CHAR_FILTER_REGEXP, "")
          
              string = string.downcase if downcase
          
              CGI.escape(string)
            end
          
            def self.none_generator(string)
              ""
            end
          end

Just tested locally and they are removed just fine for me.

Arta_S · 10월 17, 2019, 12:36오전

Perhaps the scenario is different on your local.
My website language is Persian. Can you switch to Persian and then test?

To duplicate this, create a topic with the following subject:
Iranian cats + belong to a good breed

My slug turns into an unreadable format and returns a 400 error:

https://domain.ltd/t/%D8%A7%DA%A9%D8%A7%D9%86%D8%AA-%D9%BE%D8%B1%D9%85%DB%8C%D9%88%D9%85-%D9%88%D8%A8-%D8%B3%D8%A7%DB%8C%D8%AA-%D9%85%D9%88%D8%B2-%D8%B1%D8%A7%DB%8C%DA%AF%D8%A7%D9%86-%D8%A7%D8%B4%D8%AA%D8%B1%D8%A7%DA%A9-%D8%B4%D8%AE%D8%B5%DB%8C-%D8%A8%D8%B1%D8%A7%DB%8C-%DA%A/2067/40

Falco · 10월 17, 2019, 12:39오전

Only you are aren’t using the “og” + but a different persian symbol that looks like it? Can you share a string that repros the problem?

Arta_S · 10월 17, 2019, 12:41오전

I edited my previous reply. Just to confirm it is a + .

Falco · 10월 17, 2019, 12:47오전

That is weird, for me the slug turns into

%DA%AF%D8%B1%D8%A8%D9%87-%D9%87%D8%A7%DB%8C-%D8%A7%DB%8C%D8%B1%D8%A7%D9%86%DB%8C-%D8%A7%D8%B2-%D9%86%DA%98%D8%A7%D8%AF-%D8%AE%D9%88%D8%A8%DB%8C-%D8%A8%D8%B1%D8%AE%D9%88%D8%B1%D8%AF%D8%A7%D8%B1-%D9%87%D8%B3%D8%AA%D9%86%D8%AF

Are you sure you are running latest and refreshed the page?

Arta_S · 10월 17, 2019, 12:51오전

I am running the latest version.

I am confused now. I created a new topic with the same condition as the topic with corrupt slug. It works just fine. Just like you stated, probably it isn’t the + that is causing the problem.

Please allow me sometime to install a dummy and make some tests. I will update this topic with more information.

Thank you

Falco · 10월 17, 2019, 12:54오전

Did you get that URL from the share button? It was double encoding and I just fixed that on the latest merge.

Arta_S · 10월 17, 2019, 1:06오전

Check this screencast.

Falco · 10월 17, 2019, 1:20오전

Oh I see the problem. It only happens with slugs who are above 255 chars and only in 66% of the cases. I will work in a fix tomorrow.

chrispanag · 10월 17, 2019, 3:28오후

Happens with us too:

eg: Γνώμη για αγορά μπουζουκιού μέσω ίντερνετ και εκτίμηση - #2 από nikos_politis - Μουσικά όργανα - Ρεμπέτικο Φόρουμ

Strangely, it only happens when you hit the topic from the “latest” list. When you link to it from another topic, it works just fine.

The full link:
https://rembetiko.gr/t/%CE%B3%CE%BD%CF%89%CE%BC%CE%B7-%CE%B3%CE%B9%CE%B1-%CE%B1%CE%B3%CE%BF%CF%81%CE%B1-%CE%BC%CF%80%CE%BF%CF%85%CE%B6%CE%BF%CF%85%CE%BA%CE%B9%CE%BF%CF%85-%CE%BC%CE%B5%CF%83%CF%89-%CE%B9%CE%BD%CF%84%CE%B5%CF%81%CE%BD%CE%B5%CF%84-%CE%BA%CE%B1%CE%B9-%CE%B5%CE%BA%C/38511/2

Falco · 10월 17, 2019, 3:31오후

Same thing, that slug is over 255 chars.

chrispanag · 10월 17, 2019, 3:35오후

While investigating, I found some categories that can’t load at all.

I found that we get these kind of errors:

URIError: URI malformed
    at decodeURIComponent (<anonymous>)
    at a (https://rembetiko.gr/assets/ember_jquery-7dfac344d893a6c10c016353b994db2c6f42af630322ea6da3399c220fcc50ed.js:16:31208)
    at Array.map (<anonymous>)
    at o (https://rembetiko.gr/assets/ember_jquery-7dfac344d893a6c10c016353b994db2c6f42af630322ea6da3399c220fcc50ed.js:16:31136)
    at P.recognize (https://rembetiko.gr/assets/ember_jquery-7dfac344d893a6c10c016353b994db2c6f42af630322ea6da3399c220fcc50ed.js:17:5729)
    at t.applyToState (https://rembetiko.gr/assets/ember_jquery-7dfac344d893a6c10c016353b994db2c6f42af630322ea6da3399c220fcc50ed.js:17:22054)
    at u.n.getTransitionByIntent (https://rembetiko.gr/assets/ember_jquery-7dfac344d893a6c10c016353b994db2c6f42af630322ea6da3399c220fcc50ed.js:17:24304)
    at u.n.transitionByIntent (https://rembetiko.gr/assets/ember_jquery-7dfac344d893a6c10c016353b994db2c6f42af630322ea6da3399c220fcc50ed.js:17:23609)
    at u.n.doTransition (https://rembetiko.gr/assets/ember_jquery-7dfac344d893a6c10c016353b994db2c6f42af630322ea6da3399c220fcc50ed.js:17:25611)
    at u.n.transitionTo (https://rembetiko.gr/assets/ember_jquery-7dfac344d893a6c10c016353b994db2c6f42af630322ea6da3399c220fcc50ed.js:17:30132)

TypeError: Cannot read property 'responseJSON' of undefined
    at n.error (https://rembetiko.gr/assets/application-5d69c42124908eb6b41549efae9db9b1df26741f3a45075c6bfd603bb5267a42.js:71:604)
    at n.C (https://rembetiko.gr/assets/ember_jquery-7dfac344d893a6c10c016353b994db2c6f42af630322ea6da3399c220fcc50ed.js:8:17733)
    at u.c.triggerEvent (https://rembetiko.gr/assets/ember_jquery-7dfac344d893a6c10c016353b994db2c6f42af630322ea6da3399c220fcc50ed.js:8:21275)
    at e.t.trigger (https://rembetiko.gr/assets/ember_jquery-7dfac344d893a6c10c016353b994db2c6f42af630322ea6da3399c220fcc50ed.js:17:13392)
    at u.c.transitionDidError (https://rembetiko.gr/assets/ember_jquery-7dfac344d893a6c10c016353b994db2c6f42af630322ea6da3399c220fcc50ed.js:8:21548)
    at https://rembetiko.gr/assets/ember_jquery-7dfac344d893a6c10c016353b994db2c6f42af630322ea6da3399c220fcc50ed.js:17:11836
    at f (https://rembetiko.gr/assets/ember_jquery-7dfac344d893a6c10c016353b994db2c6f42af630322ea6da3399c220fcc50ed.js:18:1126)
    at T (https://rembetiko.gr/assets/ember_jquery-7dfac344d893a6c10c016353b994db2c6f42af630322ea6da3399c220fcc50ed.js:18:2503)
    at E (https://rembetiko.gr/assets/ember_jquery-7dfac344d893a6c10c016353b994db2c6f42af630322ea6da3399c220fcc50ed.js:18:2402)
    at y (https://rembetiko.gr/assets/ember_jquery-7dfac344d893a6c10c016353b994db2c6f42af630322ea6da3399c220fcc50ed.js:18:1896)

Can those be related?

(you can check for yourself here: https://rembetiko.gr/c/άλλα-θέματα/μικρές-αγγελίες)

Falco · 10월 17, 2019, 8:50오후

@Arta_S and @chrispanag two more fixes incoming, one for long titles and the other for subcategories.

https://github.com/discourse/discourse/commit/2304dcf993f626c86b0cab43d33dc7005e850fb9

and

https://github.com/discourse/discourse/commit/f97b6e85daac2c872a710748dfe004aabb3cff72

chrispanag · 10월 18, 2019, 1:27오전

After upgrading, we still get this error

Falco · 10월 18, 2019, 1:43오전

In this case you will need to rebake the slugs, please try:

cd /var/discourse
./launcher enter app
rails c
Topic.update_all(slug: nil)
exit
exit
exit

hiccup · 10월 18, 2019, 8:55오전

Two days ago from last update I have two problems:

1: 400 Bad Request URL in some topic
2: timeline disappeared in some topic

host Discourse 'standard-plan"

Can you Help Me

sorry i don’t write English well

글		댓글	조회수
Automatic encoding of parsed URL params Development	43	8628	5월 15, 2025
Category doesn’t open to show topic list Support	5	592	11월 5, 2019
404 on various categories Bug	7	1255	10월 16, 2019
Newly added non-ASCII category page does not load Bug	11	1424	7월 29, 2020
When slug generation method is encoded, It redirected many times when we open new tab opens in topic Bug	8	5416	9월 29, 2017

Topic with Japanese in URL doesn't redirect if URL doesn't perfectly match

Related topics