What I want to do: have a shortcode that adds itself to a list/map - and have that map displayed on a page (with a list of all the “things” on one page).
Is there a good way to do it?
I almost know one way: I could use .Site.Store.SetInMap to put things into a map. And I can read stuff out of the map.
But… the thing I don’t know how to do:
Force the page that I want the “result” to be on to be the last one built (so all the other pages have put their stuff into the map first).
As far as I know, there is no way to control the order in which pages are built, so some lucky page is last, and would have the complete list. Other pages will only have the information from the pages that were rendered before it.
Is there either (1) some way to designate a page to be built last (so my scheme of using .Site.Store would work) or (2) some better way to do this?
That’s magic! I was able to get a (miniature) test with defer to work.
Just curious… why did you suggest step 1?
In my test, I did it by hacking the existing “single page” template - replacing the existing {{ .Content }} with:
git clone --single-branch -b hugo-forum-topic-54816 https://github.com/jmooring/hugo-testing hugo-forum-topic-54816
cd hugo-forum-topic-54816
hugo server
Files of interest:
layouts/_shortcodes/capture.html
layouts/snippets/section.html
The “capture” shortcode uses the page store to hold an array of all snippets captured on the page. Each element in the array is a map which holds an anchor id and the shortcode context.
The “snippets” template ranges over all pages on this site, forces rendering of each one so that the page stores are updated, then lists the snippets with a link to where they were captured. I like this approach because you can easily filter and sort the snippet listing.
Thank you!
I really appreciate your generousity with your time and knowledge.
There is a lot for me to learn from this! I get the basic idea (loop over the pages and look into their individual stores). I will need to read it carefully and think about how this applies. (and a ton of other lessons I am taking from it).
Some quick questions… (good answers lead to better questions)
why the underscore shortcodes “layout/_shortcodes” and not “layout/shortcodes”?
how does the section snippet know not to force itself to render (and cause a loop)?
do the pages have their front matter before they are rendered (so, for example, could I wrap the inside of the loop with something like `{{ if .Params.collectstuff }}`` (so I could manually enable which pages this happens on).
And a bunch of why this and not that…
If I understand correctly, the “layout/snippets/section.html” page is a specialized layout that applies to the “_index.md” page of the snippets section. could this have been a shortcode that was put on some random page in the middle?
Is there some best practice to make sure that this specialized section.html page tracks changes to the “generic” used by default? (or in this case just in layouts)
How would this compare with the defer strategy? It seems less magical (although, the force rendering trick is pretty magical too).
Again, thanks for your generosity. This is really amazingly helpful.
New template system in v0.146.0 and later. See notes.
Yes, but perform a debug.Timer A/B test. I think you’ll find that looking for the existence of a page store is plenty fast, and might be faster than another conditional. I tested HasShortcode and was surprised at the performance difference.
always a hundred ways to die I like @jmooring approach
to answer the question:
I consider that page something special , I would also have expected it’s rather a list page than a normal page in the middle of all the others.
using a dedicated template
shows clearly that this is special
avoids deadlocks if you don’t set the params correctly when adding a new page
makes each template clearer
no condition to evaluate 100 times for one hit
One more note about performance. For the test site I added 2000 pages that don’t call the “capture” shortcode. There’s wasn’t a material difference in build time, meaning that looping over every page in the site didn’t make a lot of difference when compared to just looping over those with page stores.
For a site with a few hundred pages I wouldn’t bother with the A/B testing, unless you have some really nasty shortcodes that do stupid questionable things.
Disregard my performance notes above. My test was unrealistic. It is significantly faster to test for HasShortcodebefore you force content rendering, for example:
{{ range site.Pages }} {{/* filter and sort as needed */}}
{{ if .HasShortcode "capture" }}
{{ $noop := .WordCount }} {{/* force content rendering */}}
{{ with .Store.Get "things" }}
<ul>
{{ range . }}
<li>
{{ .ctx.Inner | .ctx.Page.RenderString }}
<a href="{{ .ctx.Page.RelPermalink }}#{{ .anchor }}">(go to source)</a>
</li>
{{ end }}
</ul>
{{ end }}
{{ end }}
{{ end }}
I like