This PR is motivated from https://github.com/just-the-docs/just-the-docs/pull/1259#issuecomment-1655899503. It adds a new workflow (`CI / Validate HTML (3.1)`) that validates the output of `bundle exec jekyll build`. It does this with two separate tools:
1. The [`html5validator-action`](https://github.com/Cyb3r-Jak3/html5validator-action), which is a wrapper (Docker image + argument forwarding) around the [Nu HTML checker](https://github.com/validator/validator), which is what is used by the [W3C markup validation service](https://validator.w3.org/)
2. [`html-proofer`](https://github.com/gjtorikian/html-proofer), which performs auxiliary checks on the validity of script, image, and link *values*, but not the markup itself
- note: prior versions of `html-proofer` did use nokogiri to also validate HTML, but the author has elected to remove that feature in versions 4+
I then fix a few issues that are flagged by these tools. I'll split this into,
**changes affecting users**:
- strictly incorrect: in `_layouts/minimal.html`, a `<div>` had duplicate `id`s. I've removed the incorrect one, which is related to...
- semantically wrong (but not technically incorrect): in both `minimal` and `default` layouts, we had two `<div>` tags with `id="main-content-wrap"`. These don't do anything; the associated styling is with the *class* `main-content-wrap`. I've elected to remove these `id`s to avoid confusion and keep the layouts in sync; however, **this is technically a breaking change**
- observe that `#main-content` is used for the "skip to main content" feature, which I missed in an earlier iteration of this PR
**changes affecting only our documentation**
- a broken link to mermaid docs (I've changed it to a valid one)
- an incorrectly-specified `aux_link` to our own repository
- various links that point to the bare URL `another-page`, which is clearly invalid; I've changed these to point to our homepage
- an incorrect header link
- various links to `http://example.com`, which I've changed to point to our homepage
- an incorrect link to `@flyx`'s profile for the AsciiDoctor gist
- a handful of (otherwise-valid) `http` links that should be `https`: the lunr docs, and patrick's personal website
The commit history shows the Nu validator flagging issues in CI properly in commits [4128b23](4128b23ef2) and [3527220](35272203ba).
## relevant configuration
- I exclude `github.com` URLs from external link checks with `html-proofer`. This is because GitHub does not like it when we ping too frequently, and rate limits us, which in turn provides many false positives. This is aligned with their documentation, which uses this ignore.
- I've pinned the hash for the 3rd-party action that wraps the W3C markup validation service. This aligns with #1148, but means that we'll have to keep an eye on it for updates.
In #1058, I noted:
> Tangentially related work:
> ...
> - better annotate new features (motivated by writing these docs)
> - we should add "New" to new features :)
> - we should note when a feature was introduced (I think this is a core part of most software documentation)
> - we should annotate things that are "Advanced" in so far as the average Just the Docs user will not use them / they require significant Jekyll knowledge
>
This came up again in https://github.com/just-the-docs/just-the-docs/discussions/1136#discussioncomment-4716253, so I think it's best for us to resolve this sooner rather than later.
This PR is me doing that. I:
- have added a headings-level "New" label to every new heading introduced since `v0.3`
- added, when possible, inline YAML comments when new configuration options have been introduced
I did this by scanning through the CHANGELOG and selecting each feature that is either tagged with `Add` and has documentation.
I may have also missed any new features, so some double-checking would be helpful!
This is a prototype for review and discussion. My use and testing of this PR is on top of 6d9d41359c46882d9b64a446d5a83fac5b3e20a7. The changes are trival to rebase to `main` and I'm happy to do so if this prototype moves forward.
* Feature request details in linked issue, fixesjust-the-docs/just-the-docs#1067
* I welcome feedback and all discussion
* A draft doc site of mine using this PR is at https://docs.hidale.com/
To use the prototype, the two include files need to be customized. Here are mine from the draft website 9c0d836408
Co-authored-by: Matt Wang <matt@matthewwang.me>
The default is for hyphens to separate tokens in search terms: `gem-based` is equivalent to `gem based`.
This adds `search_tokenizer_separator` as a site configuation option, to support search for hyphenated words.
Because `page.title` (and `page.url`) is already escaped and page titles served by lunrjs do not need escaping, ampersands need to be unescaped to display properly in search results.
The code snippet on the search documentation is missing the `relUrl` parameter in each search result, and the `url` parameter is not an absolute URL.
This updates to match the formatted output of `search.rake`
