My Blog To-Do List
2023-03-23 - ~10 Minutes
Witness the birth of a new blog! Actually, a new version of a very old blog. There’s still a lot of work to do.
So, here’s my “to-do” list, as it pertains to this blog, along with possible sources of guidance and inspiration.
Update social icons and links in the blog.
Lose Twitter in favor of Mastodon, perhaps? Again, see https://ghost.org/tutorials/add-social-media-icons/ .
Add the OneTab content type.
My old Hugo blog distinguished between a “post” and other types that I called “micropost” and “onetab”. Each type had its own unique metadata fields. For example, each “micropost” had location: and detailed published-date: fields, but no title:. See
https://ghost.org/docs/themes/routing
for possible guidance.
See my previous “Add Micropost and OneTab content types” as well as “Add For Sale and OneTab content types” solutions below for guidance.
Process new .env variables.
Some of this has already been implemented to achieve the site banner that you see on the home page. See https://www.11ty.dev/docs/data-js/#example-exposing-environment-variables for the guidance I used to pull that off.
Add a photo gallery.
Maybe. Another feature that my old Hugo blog supported, although I rarely kept it up-to-date.
Implement my Hugo Timeline module.
Make an 11ty version of https://github.com/SummittDweller/hugo-timeline and implement it here.
Engage Disqus Comments.
Another feature that my old Hugo blog had, and my professional blog still uses. Rinse and repeat.
Everything below this line is DONE! For now.
Pulling Too Much Data
Today (April 20, 2023) I received notice from Mediacom, my ISP, that I’ve used 75% of my monthly data allotment with about a week left in the term. That’s not a huge problem, but I’ve never come close to using even one-half of my data in any previous month. I suspect this is due to all the massive image downloads taking place each time I build a dev copy of this blog.
The yarn start command that I use to build the localhost copy of this blog defines ELEVENTY_ENV=dev. I’m going to see if I can use this environment setting to limit image downloads to the first ten in order to limit the “dev” download requirements. If this works a switch of some kind might need to be defined so that I can override the limit as needed.
☑️ Solved 20-Apr-2023: This played out nicely, but rather than limited the number of images I simply limit the number of posts pulled by .eleventy.js. The code looks like this now:
var environ = process.env.ELEVENTY_ENV;
console.log("ELEVENTY_ENV is: ", environ);
if (environ == "dev") {
post_limit = 6;
} else {
post_limit = 'all';
}
collection = await api.posts
.browse({
filter: "tag:-micropost",
include: "tags,authors",
limit: post_limit
})
...
Implement robust site search.
Just like it says ^^^ there. Using Lunr, Pagefind , something else?
☑️ Solved 20-Apr-2023: This is done and I did indeed use
Pagefind
for the solution. The details can be found elsewhere in this blog at
Searching for a Search Solution
and also in this blog project’s README.md file.
Add For Sale and OneTab content types.
My old Hugo blog distinguished between a “post” and other types that I called “micropost” and “onetab”. Each type had its own unique metadata fields. For example, each “micropost” had location: and detailed published-date: fields, but no title:. See
https://ghost.org/docs/themes/routing
for possible guidance.
See my previous “Add Micropost and OneTab content types” solution below for guidance.
☑️ Solved 6-Apr-2023: Ok, the micropost and for_sale portions of this is done, and documented (somewhat) in
https://github.com/SummittDweller/blog-eleventy-ghost/blob/main/README.md
. Note that in the end I introduced a new for_sale tag (note that it’s all lowercase to set it apart from others) after trying to use other techniques.
I’m pushing the addtion of a onetab content type back up into this list, again.
Improve Front Matter Processing
This effort will make a significant change to the solution shown in “Add and process post front matter” below. Specifically, I’m going to move the “front matter” for all content types out of the excerpt, because it’s difficult/impossible to edit excerpts in my iPhone’s Ghost Publisher software, and put it at the bottom of the content body with a --- seperator marking the start of the data. It should look something like this example:
---
location: Toledo, IA
price: $50.00
note: This is just a note.
Basically, anything found at the bottom of a post, of any TYPE, below a --- delimiter, becomes a key:value pair available in post.data.<key>.
☑️ Solved 5-Apr-2023: This is done but not as originally planned, and it may still need some work. So, in the end I introduced a new for_sale tag (note that like micropost it’s all lowercase to set it apart from others) after trying to use other techniques.
My new “rear matter” solution, detailed in the project’s README.md file, actually uses a dash-slash-dash delimiter at the bottom of the content body, immediately followed by a slash-dash-slash delimiter… all run together in 6 characters, no spaces.
The first delimiter marks the end of the content, while the second one marks the start of the “rear matter”. I can’t show you explicitly how that works here because putting those characters into a post like this will cause errant processing on this post. 🙃
Add Micropost and OneTab content types.
My old Hugo blog distinguished between a “post” and other types that I called “micropost” and “onetab”. Each type had its own unique metadata fields. For example, each “micropost” had location: and detailed published-date: fields, but no title:. See
https://ghost.org/docs/themes/routing
for possible guidance.
☑️ Solved 2-Apr-2023: Ok, the micropost portion of this is done, and it’s documented (somewhat) in
https://github.com/SummittDweller/blog-eleventy-ghost/blob/main/README.md
. Note that in the end I introduced a new micropost tag (note that it’s all lowercase to set it apart from others) after trying to use other techniques like bits of excerpt and internal tags like #micropost.
I’m pushing the addtion of a onetab content type back up into this list, and will add a new for-sale tag and type to the mix.
Support better bulk ingest of content.
I have a lot of old blog posts to enter here, and research tells me that capability might be lacking. Perhaps I can construct something using https://ghost.org/docs/admin-api/javascript/ or provide an update of https://github.com/dumrauf/post2ghost that works with the newest Ghost API?
☑️ Solved 2-Apr-2023: Well, I got this done after several failures. Ultimately look to https://github.com/SummittDweller/migrate-hugo-to-ghost and more recently, https://github.com/SummittDweller/migrate-html-to-ghost for my solution.
Investigate using “special control” tags.
Control Ghost and Eleventy behavior using tags that with @, #, or some other special character. For example: #Micropost, or $Micropost.
🚫 CANCELED 26-Mar-2023: Now that front matter via the excerpt is working, I don’t think this will be necessary.
Add FontAwesome support for emoji.
I really miss those from my Hugo days, and they would be so handy to have for things like checking items off this list! See https://ghost.org/tutorials/add-social-media-icons/ for guidance.
Looks like this is already avaialble in 11ty, https://www.npmjs.com/package/@vidhill/fortawesome-free-regular-11ty-shortcode , but can I code a shortcode call in Ghost and pass it through? I’ll bet not… NOPE.
☑️ Solved 26-Mar-2023: This turned out to be a bit of a gut 🥊. So, I’ve scrapped the idea in favor of just using CTRL - COMMAND - SPACE on my Mac.
Add and process post front matter.
See https://forum.ghost.org/t/custom-data-fields/8884/4 and https://github.com/nchaulet/ghost-app-frontmatter . Maybe this old project can be resurected and implemented here so that posts, and other types, can have robust “front matter” with custom display.
☑️ Solved 26-Mar-2023: Following the logic of the parseFrontMatter function in
https://github.com/nchaulet/ghost-app-frontmatter/blob/master/index.js
, I added the following code to .eleventy.js.
// Add front matter processing of Ghost posts here,
// pulling data from the `post.excerpt`
// Note that pulling from `post.codeinjection_head` and
// `post.custom_excerpt` did NOT work!
// console.log(post);
// console.log("Before:", post.excerpt);
var parsed = matter(post.excerpt, { delims: ['---', '---'] });
post.data = parsed.data;
var removed = post.excerpt.replace(/---[\S\s]*---/m, '');
post.excerpt = removed;
// console.log("After:", post.excerpt);
// console.log("Data:", post.data);
It looks for typical --- delimiters in the excerpt portion of a post and parses it into data.<element> key:value pairs that are available to my templates.
Display publication and last-mod dates.
Just what the title says. Displaying these things was something my old Hugo blog always did.
☑️ Solved 26-Mar-2023: Another easy addition in src/_includes/partials/card.njk where the Nunjucks now checks for matching published_at and updated_at dates, and reports “Published” if they match, or “Updated” if they don’t. Also, microposts now show differnt data, including location: and the publication date AND time.
Improve the display of tags.
Maybe seperate tags using · symbols · like · this? Perhaps have a look at
https://ghost.org/docs/themes/helpers/tags/
.
☑️ Solved 25-Mar-2023: An easy addition in src/_includes/partials/card.njk where the Nunjucks now looks like this:
<div class="post-card-tags">
{% if post.tags %}
{% for tag in post.tags %}
{{ tag.name }} {% if not loop.last %} · {% endif %}
{% endfor %}
{% endif %}
</div>
Rebuild and deploy the site when content is published.
The site already builds and auto-deploys when changes are pushed to the code/structure, but the forerunner of this site also included a hook that did a build and deploy (per Netlify) when changes to the Ghost content are published. See https://github.com/TryGhost/eleventy-starter-ghost for possible guidance.
Better yet, check out
https://docs.digitalocean.com/reference/api/api-reference/#operation/apps_create_deployment
. It seems the mechanism that’s required here is to make a request, like a curl or POST, to a particular URL (see sample below) perhaps with an auth token included for security.
curl -X POST \
-H "Content-Type: application/json" \
-H "Authorization: Bearer $DIGITALOCEAN_TOKEN" \
"https://api.digitalocean.com/v2/apps/{app_id}/deployments"
Now, the question is, how to make my Ghost install dispatch such a request? Simple, see
https://ghost.org/docs/webhooks/
but note that the request might need to be a POST, not a curl.
Hmmm, this is more difficult than I imagined, so this write-up might soon become a post of its own!
Preliminary Findings
Unfortunately, it looks like Ghost’s webhooks don’t integrate well with DigitalOcean. DO requires a POST request that looks like this in curl form:
curl -X POST \
-H "Content-Type: application/json" \
-H "Authorization: Bearer { $DIGITALOCEAN_TOKEN }" \
-d '{ "force_build":true }' \
"https://api.digitalocean.com/v2/apps/{ app ID }/deployments"
I can’t find a way to build such a request in Ghost since the headers and body of the webhooks it builds don’t look like what you see above. The only thing that can be easily configured in a Ghost webhook is the destination address. So, now I’m going to look into the possibility of using an iOS or MacOS app that triggers a curl like the one you see above. Or, maybe I can add a protected page to the site itself and embed a Javascript POST request within?
☑️ Solved 23-Mar-2023: Please see Built by Webhook + Password Protection for all the gory details.
Force rebuild with “Trigger Site Update” draft.
The idea here is to delay rebuilding the site until triggered by an editor, perhaps using a “special” trigger post. This could help avoid multipe back-to-back rebuilds when lots of content is updated in a short period of time. Of course, if a “special” post were used for this it would have to be implicitly “ignored” and always in a “draft” state.
🚫 CANCELED 23-Mar-2023: No longer needed. Please see Built by Webhook + Password Protection for all the gory details.
Engage spell checking.
Indications from as early as 2015 indicate that this should work straight from the browser, but I’m not so sure?