Work in progress: Externalized page metadata

[bep]

Also see #5510 -- a concurrent batch variant (whatever that would look like) could also be considered.

This issue describes a way to add or modify Page metadata in batches from external file(s). This is the information you currently only set in Page front matter.

Some use cases:

• The default page sort is ByWeight (with fall back to date and then title). Setting weight in front matter to get a particular order can be cumbersome if you have thousands of articles. With this implemented, you can re-arrange the pages by assigning weight in batches by Glob pattern matching.
• Tag all pages with both funny and cat in the path with the tag Funny Cats
• Add category Other to pages with no category set.

Some examples in TOML:

# Re-arrange the blog posts so dog articles come first.
# Note that the `:counter` will restart at each `meta` element.
[[meta]]
match = "/*/blog/**"
matchFirst = """
**dog*
**cat*
**
"""
counterStart = 10 # default is 1
counterIncrement =  10 # default is 1
replace = true # default is false
[[meta.data]]
weight = ":counter"

# Add a category and a title to blog posts with none set.
[[meta]]
match = "/*/blog/**"
[[meta.data]]
title = "Blog article #:counter"
categories = ["Other"]

• There can be many meta definitions, and all pages will be tested against all of them.
• A page that matches both the Glob pattern in match and matchFirst will be assigned the given metadata. if replace = true any existing value will be replaced.
• The match pattern is primarily "motivated by speed", but it also allows for simple and coarse-grained patterns in matchFirst.
• The values in meta.data map directly to what you would normally set in page front matter.
• The patterns in matchFirst sets the order of the 1-based :counter placeholder. In the example above, the articles about dog will get weight 10, 20, 30, and then come the cats and finally all the other.
• In the example, matchFirst is a multiline string, but we will accept string arrays, too.
  We will match against the canonical path* (always prefixed by the current language, so one can do /{en,jp}/blog/** etc).

*: It would be tempting to match against .RelPermalink, but that will typically change across languages. We need to define the term canonical path as part of this issue.

Some references:

• This proposal is inspired by how we do "page resources" metadata, see https://gohugo.io/content-management/page-resources/#the-counter-placeholder-in-name-and-title
• Having this configuration in the site configuration will make more sense once Add a /config dir #5422 is ready (should be next very soon)
• This proposal will also make more sense/add more value if you also consider Add remote support in resources.Get #5255 and Build pages from data source #5074

[regisphilibert]

Great vision for the future here! :)

[moorereason]

This is cool.

My brain wants to do

weight = “{{ mul .Counter 10 }}”

Or is that a bridge too far?

[regisphilibert]

I think we'll eventually need template functions and context like @moorereason suggests here when we need to assign value to pages built from external data source:

title = "{{ .Data.name }}"
subtitle = “{{ .Data.fields.sub_title }}”

[bep]

My brain wants to do

My brain wants to do that, too. And we will eventually get there, but I'm not sure "Go templates" is the correct medicine -- as they are basically text templates (e.g. presentational), and also not very fast.

See https://discourse.gohugo.io/t/tools-libraries-that-can-drive-new-hugo-features/7891/48?u=bep

[bep]

title = "{{ .Data.name }}"
subtitle = “{{ .Data.fields.sub_title }}”

OK, so I suspect the above would be highly ineffective at scale. But, it got me thinking a little.

[[meta]]
match = "/*/blog/**"
script = """
.Scratch.Add "counter" 1
$data := index .Site.Data .Page.Section
.Set "name" $data.name
.Set "weight" (.Scratch.Get "counter")
"""

The .Scratch would be scoped per matchFirst (or if not set: match).

The above would probably only work for TOML? I.e. the multiline stuff. But it would be pretty flexible, a.

Note that the field names above are very temporary (meta? maybe frontmatters).

[regisphilibert]

The script value could ideally be stored in a separate file to improve coder's experience.

Or better yet one of those new custom function being discussed.(#5222). The func could even return an object built from the "page/data/source" context to allow setting multiple key/value pair to be consumed by meta.data.

# Add a category and a title to blog posts with none set.
[[meta]]
match = "/*/blog/**"
[[meta.data]]
script = "func getBlogMeta ."

I'm really bad at writing toml.

[moorereason]

Dood. I missed to golua library. Hugo+Lua would be very interesting. I’ll have to ponder the implications.

[bep]

But looking at the above attempts of "scripting in config" looks ... ugly. We risk ending up ...

https://gitlab.com/gitlab-org/gitlab-ce/blob/master/.gitlab-ci.yml

[stale]

This issue has been automatically marked as stale because it has not had recent activity. The resources of the Hugo team are limited, and so we are asking for your help.
If this is a bug and you can still reproduce this error on the master branch, please reply with all of the information you have about it in order to keep the issue open.
If this is a feature request, and you feel that it is still relevant and valuable, please tell us why.
This issue will automatically be closed in the near future if no further activity occurs. Thank you for all your contributions.

[github-actions]

This issue has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.