Check html before merging

[maelle]

It has some issues again https://validator.w3.org/feed/check.cgi?url=https%3A%2F%2Fropensci.org%2Findex.xml

Cf #114 opened by @jeroen

How could one re-validate it before publishing a blog post? Could it be part of automatic PR checks?

[maelle]

The current issues are "recommendations".

• For the tech note about icons in R, The problem is in https://github.com/ropensci/roweb2/blob/master/content/technotes/2018-05-15-icon.md#icons-in-r- where the img source is right near the header so in the header in html which apparently isn't good for all RSS feed readers? I couldn't figure out how to remove the image from the header without making the image too big/misplaced.

• There's a Spotify iframe in https://ropensci.org/blog/2018/03/13/ode-to-testing/ and "description should not contain iframe tag". Not sure what to do about that. Ignore?

I'll fix the last issue that seems easier.

[maelle]

There will be a second post with an iframe cf #184

[maelle]

Problems that are detected by the RSS feed validator include invalid links, so it'd be good to detect them right away. E.g. that's how I found 401433a

[maelle]

Renaming the issue to reflect the fact that what I want is finding various problems right away not just for the RSS feed.

I saw this similar issue in tidyverse.org, https://github.com/gjtorikian/html-proofer seems to be the tool to have a look at.

[maelle]

• Real-life example https://github.com/gjtorikian/html-proofer#real-life-examples
• Not sure whether this uses Netlify https://github.com/travis-ci/docs-travis-ci-com
• Example of a GitHub repo with deployment from Netlify and html proofing https://github.com/cucumber/docs.cucumber.io

[maelle]

We'd need the html proofing to happen on the new pages introduced by a PR only (e.g. for a blog post, the blog post itself, and the new author pages), because we know there are issues in old pages.

[maelle]

Also interesting https://circleci.com/blog/markdown-proofer/

[maelle]

For blog posts created pages all correspond to a Markdown file so it'd be easier to check those, in particular

• links including internal links (to the website, and ref starting with #)

• do images exist (although that is not especially useful since humans reviewing the PR would see that)

• alt text (not empty, an human would need to check they are informative)

• code linting?

[maelle]

dup of #463