Wikipedia oneboxing with url encoded section hash

Benjamin_D · August 4, 2021, 2:44pm

I haven’t categorized the topic as bug because it might be by design, not sure…
so this works
https://fr.wikipedia.org/wiki/Th%C3%A9ologie#Platon

this doesn’t
https://fr.wikipedia.org/wiki/Th%C3%A9ologie#La_th%C3%A9ologie_selon_Aristote

Weirdly the span ids are La_th.C3.A9ologie_selon_Aristote and La_théologie_selon_Aristote so I think there is something here: discourse/lib/onebox/engine/wikipedia_onebox.rb at 2f28ba318c3f650b1a0a76530021bc21bc8ab6f6 · discourse/discourse · GitHub

I sure hope so!

Falco · August 4, 2021, 2:49pm

Both of your oneboxes work just fine for me in Android

Benjamin_D · August 4, 2021, 2:56pm

The links work yes, but for the second one the excerpt should be:

La poésie et les origines de la pensée… etc

ps. you had me for a second here!

Falco · August 4, 2021, 3:52pm

Oh you want the onebox summary to reflect the sub-section? That is controlled by

github.com

discourse/discourse/blob/main/lib/onebox/engine/wikipedia_onebox.rb#L22


      
          
          private
          
          def data
            paras = []
            text = ""
          
            # Detect section Hash in the url and retrive the related paragraphs. if no hash provided the first few paragraphs will be used
            # Author Lidlanca
            # Date 9/8/2014
            if (m_url_hash = @url.match(%r{#([^/?]+)})) # extract url hash
              m_url_hash_name = m_url_hash[1]
            end
          
            if m_url_hash.nil? # no hash found in url
              paras = raw.search("p") # default get all the paras
            else
              section_header_title = raw.xpath("//span[@id='#{CGI.unescape(m_url_hash_name)}']")
          
              if section_header_title.empty?
                paras = raw.search("p") # default get all the paras

That is not matching

<h3>
  <span id="La_th.C3.A9ologie_selon_Aristote"></span>
  <span id="La_théologie_selon_Aristote">La théologie selon Aristote</span>
</h3>

The code is over 7 years old, so it’s very possible that the Wikipedia markup changed in the meanwhile. It’s a pr-welcome if someone wants to work on this.

jbalsas · August 11, 2021, 9:20pm

I’ve sent a possible fix at Fix/wikipedia oneboxing with url encoded hash by jbalsas · Pull Request #14015 · discourse/discourse · GitHub</titl.

I think unescaping the m_url_hash_name should be enough to cover all cases.

Hope this is the right process to contribute!

Topic		Replies	Views
Support Wikipedia anchors in onebox Feature	6	1901	September 10, 2014
Why are some WordPress links Oneboxed, and others are not? Feature	18	4018	March 29, 2017
Oneboxing of sites with hash (#) in URL not working Bug	9	1838	July 29, 2017
[quote] & <blockquote> aren't parsing a URL in a citation as per BBCode & HTML Support	13	3186	July 4, 2020
Mini (Inline) Onebox Support RFC Feature rfc	21	5502	April 21, 2022

Wikipedia oneboxing with url encoded section hash

Related topics