Dvc.org: broken external links at community tutorials page

[sudodoki]

At https://dvc.org/doc/tutorials/community you see all the links are pointing to dvc site and 404 although content/docs/tutorials/community.md is featuring proper external links.

Tried investigating root cause, might be something in redirects-list.json & src/utils/shared/redirects.js

Also, not 100% relevant (and sure :)), but in gatsby-config both external-link-plugin & gatsby-remark-external-links are used which might be doing similar things

[jorgeorpinel]

Great find! It applies to the first 3 links and to the last one. Investigating... ⏳

[jorgeorpinel]

@iAdramelk @pavelgrinchenko this doesn't seem to be a problem with the markdown or with the redirect list. The links in code are correct so the link checker doesn't catch these 404s. Maybe it's a Gatsby problem indeed?

[jorgeorpinel]

p.s. seems to be replacing host names that have subdomains (e.g. blog.codecentric.de or www.linkedin.com) for the server host name.

[jorgeorpinel]

p.p.s. it's affecting all external links everywhere in the website e.g. https://dvc.org/doc/understanding-dvc/resources#articles so definitely a docs engine problem.

[sudodoki]

@jorgeorpinel not sure it's broken the same way in develop & build mode, but

This is happening for
1) external links
2) that end with trailing /
Because of redirect rule being applied and returning only relative part of URL

Details:
<Link> component is calling getRedirect here and
this "^/(.+)/$" pattern is being checked against path part of external URLs omitting the host, returning relative urls that are being rendered instead of original external url.

So as a quickfix, can just remove trailing / from external links to avoid this and might need to modify redirect matching/using results logic

[iAdramelk]

@pavelgrinchenko Can you please check it?

I think @sudodoki is right, we should probably reapply host for the links if it's not equal to the current site or return original link without removing / for such links, but maybe I'm missing some edge case there.

[pavelgrinchenko]

@jorgeorpinel I made PR with fix, could you review it

[shcheklein]

@pavelgrinchenko thanks! 🙏 Merged.

@sudodoki

external-link-plugin & gatsby-remark-external-links

names are confusing indeed, buy yes, we need both - one to support custom HTML in blog (it is a a custom plugin we wrote), second one - is regular gatsby that deals with ext links.

Good catch, btw. Thanks for the report.

[shcheklein]

@pavelgrinchenko it looks like it is still broken :(

[sudodoki]

@shcheklein at which page are you seeing same behavior? https://dvc.org/doc/tutorials/community links from original issue seem fixed.

And the one mentioned in thread https://dvc.org/doc/understanding-dvc/resources#articles as well

[shcheklein]

@sudodoki links are fixed, but when I click them they still redirect me to dvc.org Not Found. Is it only me? 🤔

[pavelgrinchenko]

@shcheklein I forgot to add domain check to click handler... Here is another fix

[pavelgrinchenko]

@shcheklein Hm, seems this logic outdated there at all as we are already replacing original href with redirected when it's necessary. So I removed it from the click handler

[jorgeorpinel]

Awesome, great fix. Thanks guys! And thanks again @sudodoki

[jorgeorpinel]

Oops only the URLs are fixed. It works OK when you Cmd/Ctrl+click but regular click changes the page to dvc.org 404 still... Cc @pavelgrinchenko

p.s. I tried with no browser cache

[shcheklein]

It works for me now. Not sure what was the issue after the deployment, but it works. @jorgeorpinel could you please confirm and reopen if it is still broken?

[jorgeorpinel]

It works OK for me now as well.