Some links are misinterpreted

[CamilleRoux]

Some links generated by Amazon are not correctly interpreted. Here is an exemple:
https://www.amazon.fr/Partition-intérieure-jazz-musiques-improvisées/dp/2907891030?_mk_fr_FR=%C3%85M%C3%85%C5%BD%C3%95%C3%91&dchild=1&keywords=partition+int%C3%A9rieure&qid=1625013895&sr=8-1&linkCode=ll1&tag=theoriemusicale-21&linkId=87d44c3dedd4b919e02195911a7a2b0d&language=fr_FR&ref=as_li_ss_tl

Here is a screen capture:

Here is the link:
https://www.amazon.fr/Partition-int%C3%A9rieure-jazz-musiques-improvis%C3%A9es/dp/2907891030?__mk_fr_FR=%C3%85M%C3%85%C5%BD%C3%95%C3%91&dchild=1&keywords=partition+int%C3%A9rieure&qid=1625013895&sr=8-1&linkCode=ll1&tag=theoriemusicale-21&linkId=87d44c3dedd4b919e02195911a7a2b0d&language=fr_FR&ref_=as_li_ss_tl

The problem seems to come from the double underscore

PS: Great book by the way

[sam (Sam Saffron)]

@Vitaly is this something we should report on the markdown.it repo?

[Vitaly]

That’s still known issue, linkify doesn't accept urls with underscores · Issue #38 · markdown-it/markdown-it · GitHub, workarounded with <link> (angle brackets around link). Nothing new to report.

Still no deadlines. Good news is: i finally rolled out to production my new forum software https://rcopen.com/ (why a lot of resources been spent to create markdown-it). So, chances to fix this bug changed from “far infinity” to “some future”

[Vitaly]

@sam, help required. I need info, can links from real world end with _, ~, - or +.

http://example.com/?sdf,wer- and so on.

That’s important to know before markdown-it's linkifier rewrite. Could you grep some huge postings databases for such cases and let me know if anything found or not found? Probably, those links were posted as <...> for workaround. So, regexp pattern for db scan will be something like /<http.+[+~_-]>/g (not tested)

I have no access to such big volume of markdown texts to reach acceptable confidence. May be you could help, or know anyone who could help?

[sam (Sam Saffron)]

@Vitaly

I need info, can links from real world end with _ , ~ , - or +

We don’t have access to customer data, but on meta

- : medium rare (181)
_ : rare (58)
+ : very rare (8)
~ : almost never (2)

There is also this corpus you can query: https://data.stackexchange.com/ which may help you get more data (you can query Stack sites)

[Vitaly]

Interesting. Could you send me to PM all those links from meta db? Need to see visually

OR, if possible - link to postings, where those links were found

@sam

There is also this corpus you can query: https://data.stackexchange.com/ which may help you get more data (you can query Stack sites)

Thank you for info. Will take a look.

[JammyDodger]

I believe this is most recently being tracked in:

Links broken with (at least) two underscores in URL

Could we fix this by doing the work in our paste handler, and if we’re pasting a URL, percent-encode underscores in the query string?

I’ll close this off in favour of that one.