As mentioned previously, I recently converted my website over to Eleventy. I'm an IndieWeb participant and, moreover, my website is fairly complex so I ended up writing (or, in some cases, pilfering) a fair amount of code to make it all work together in an acceptable manner. I thought I would document the process via a series of blog posts in case anyone else finds it of value.

None of this is intended to be an Eleventy tutorial and you'll get the most from these posts if you already know something about how Eleventy works.

With that out of the way, we'll start with my basic setup.

Post Format

My posts, like a lot of people's, are all written in markdown with yaml frontmatter. A typical blog entry looks like this:

---
date: 2022-03-01T13:57:07-05:00
title: 'Much Ado About (Indieweb) Bookmarks'
blurb: 'The ins and outs of owning your bookmarks on the #indieweb'
tags:
  - indieweb
mp-syndicate-to:
  - twitter
---

Like many Indieweb participants, I post...

There's nothing really surprising here and I suspect the format would be familiar to Jekyll users. I have a standard title, date and tags attribute. The mp-syndicate-to and blurb attributes will be covered later when I talk about syndication.

I also post notes. They look like this:

---
date: 2022-02-24T01:27:35Z
---

I am grateful that Wordle exists if for no other reason...

There's no title here, which marks it as a note instead of an article. More on that later.

I post bookmarks, too:

---
date: 2022-01-16T18:10:55.755Z
bookmark-of: 'https://blog.mollywhite.net/its-not-still-the-early-days/'
---

As you can see, that one contains no title and no content. Most of my bookmarks (and likes, and reposts) are formatted like this. The bookmark-of attribute identifies this as a bookmark post. More on that later.

Permalinks

By default, Eleventy will create an output folder for every non-layout input template, with the main content tucked away in a corresponding index.html file, so you can access the page with URL that looks like <content-path>/.

I understand the rationale, but I still find it jarring to have a whole folder for every output URL, so one of the first things I did was change the permalink scheme.

My posts are all located in src/posts/feed/ which has the following feed.json directory data file tucked away inside:

src/posts/feed/feed.json:

{
  "permalink": " {{ page | postPermalink }}",
  "layout": "post.njk"
}

You can see that the permalink is calculated via a postPermalink filter, defined in my .eleventy.js file as so:

.eleventy.js:

const dayjs = require('dayjs');
const utc = require('dayjs/plugin/utc')
const timezone = require('dayjs/plugin/timezone') // dependent on utc plugin
dayjs.extend(utc)
dayjs.extend(timezone)


function getSlug(fslug) {
  // test for my note files from micropub, which look like this:
  // 20200810123845.md
  if (/^\d\d\d\d\d\d\d\d\d\d\d\d\d\d(.*)/.test(fslug)) {
    // strip off the date portion, only pick up the time portion of the timestamp
    return fslug.substring(8);
  } else {
    // otherwise just return the filename
    return fslug;
  }
}

module.exports = (eleventyConfig) => {
  ...
  eleventyConfig.addFilter("postPermalink", page => {
    const date_part = dayjs(page.date).tz('America/Montreal').format('YYYY/MM/DD');
    const slug = getSlug(page.fileSlug);
    return `${date_part}/${slug}.html`;
  });
  ...
};

From reading this, you can see that my preferred permalink structure is /year/month/day/slug.html (inspired by Cool URIs don't change), and you can probably tell that I live in Montreal. Because I host my website on Netlify, you can drop the html part of the permalink when accessing the URL in a browser, and it becomes /year/month/day/slug.

One thing to notice, which I will likely delve into when I start talking about my Micropub client, is that when I publish notes via Micropub, the file name is the current timestamp, to the second. I want the slug to be just the time portion of that timestamp, so my code here will extract that part. This is very specific to my use case and may not be of interest to many others.

With this configuration in place, every post will end up at a URL directly off my domain name, with no extra decorations that may end up changing over time. My "introduction to the IndieWeb" article, for example, is located at /2019/12/08/intro-to-indie-web and I feel like there's a good chance this URL would survive any reorganization of my website.

Post Organization - Recap and Implementation

I implied above that my post feed consists of several different kinds of posts. I've written about the different post types recognized by the IndieWeb community, so I won't rehash that here. I will, however, rehash the post organization scheme that I eventually settled on for my personal site:

• Posts. My blog entries, notes, and standalone photos. If you're interested in what's going on in my life this would be the feed to follow.
• Blog. Just my blog entries, because I'm old-fashioned that way.
• Bookmarks. My "read later" link pile. Whether or not I actually read these links later is a separate conversation. Leave me alone :-)
• Everything. All my posts, including everything I've already mentioned but also my likes, reposts, and replies. IndieWeb enthusiants might be interested in this feed.

I wanted all these groups to show up as distinct, subscribable feeds on my website. To do this, I made use of a feature of Eleventy called collections, which allows you to group content in arbitrary ways.

The first step in this process involves identifying the type of each post, so I can put it in the right bucket. Is it a blog entry? A note? A bookmark?

One approach I've seen is to just put each type of post into its own folder - blog entries go into an article folder, notes go into a notes folder, etc. This is simple and it works, but it means that you need to remember to do this every time you create a new post. I felt that this was a burden that would discourage frequent posting and, in any case, I suspected the process could be automated so I opted instead to write some custom code to handle it.

I ended up writing a simplified (some might say "half-assed") version of the post discovery algorithm to attach a special "postType" data attribute to each post. The complete post discovery algorithm is meant to work with the entire gamut of microformatted posts seen "in the wild" and was unnecessary in the context of my eleventy setup because we're dealing my content, on my website, over which I have full control.

So I wrote a postType utility. As I said, it's geared towards my own content but my posts are formatted in a vaguely standard manner so I suspect this code would be at least semi-useful to anyone using Eleventy:

utils/post-type.js:

const validUrl = require("valid-url");

/**
 * Returns the "value" of the data passed in.
 * 
 * - if it's an array, use the first item, otherwise just use the item passed in
 * - If there's a "value" field in the data, use that, otherwise just use the value itelf
 *
 */
function getPropValue(value) {
  return value ? getValue(Array.isArray(value) ? value[0] : value) : undefined;
}

function getValue(value) {
  return value.value || value;
}

function postType(item) {
  if (!item || !item.data) {
    return undefined;
  }

  if (item.data.type === "h-event") {
    return "event";
  }

  const props = item.data;
  const propNames = Object.keys(props);

  if (propNames.includes( 'rsvp' ) &&
     (props.rsvp.includes( 'yes' ) ||
      props.rsvp.includes( 'no' ) ||
      props.rsvp.includes( 'maybe' ) ||
      props.rsvp.includes( 'interested' ))) {
    return 'rsvp';
  }

  const impliedTypes = [
    ["repost-of", "repost"],
    ["like-of", "like"],
    ["bookmark-of", "bookmark"],
    ["in-reply-to", "reply"],
    ["video", "video"],
    ["photo", "photo"],
  ];
  
  for (let i = 0; i < impliedTypes.length; i++) {
    const [propertyName, impliedType] = impliedTypes[i];
    if (propNames.includes(propertyName) && validUrl.isUri(getPropValue(props[propertyName]))) {
      return impliedType;
    }
  }

  return item.data.title ? 'article' : 'note';
}

module.exports = {
  postType
}

This is a Javascript adaptation of the post discovery function from the mf2util python package, which I used in the previous version of my site. I simplified it for my use case, but opted to leave in the event and rsvp support in case I ever implement that for my site.

The postType attribute will be one of event, rsvp, repost, like, bookmark, reply, video, photo, article or note. It's attached to each post via Eleventy's computed data feature - I have a feed.11tydata.js file in my feed folder, alongside the feed.json file mentioned above:

src/posts/feed/feed.11tydata.js:

const pt = require('../../../utils/post-type.js');

module.exports = {
  eleventyComputed: {
    postType: (data) => pt.postType({data}),
  }
};

The next step is to group the posts into my desired feeds. To this end, I added this to my .eleventy.js file:

.eleventy.js:

function postTypes(collection, postTypes) {
  return collection.filter((item) => postTypes.includes(item.data.postType));
}

function getPosts(collection) {
  return collection.getFilteredByGlob("./src/posts/feed/**/*.md").reverse();
}

module.exports = (eleventyConfig) => {
  eleventyConfig.addCollection("all", (collection) =>
    getPosts(collection)
  );

  eleventyConfig.addCollection("posts", (collection) =>
    postTypes(getPosts(collection), ["article", "note", "photo"])
  );
  
  eleventyConfig.addCollection("blog", (collection) =>
    postTypes(getPosts(collection), ["article"])
  );

  eleventyConfig.addCollection("bookmarks", (collection) =>
    postTypes(getPosts(collection), ["bookmark"])
  );
  ...
}

Things to note:

• The getPosts function returns a reverse-chronologically ordered list of all my posts.
• The postTypes function will select out all posts with a type that matches one of the types in the supplied list.
• Eleventy's collection API, plus the two helper functions mentioned above, are used to create the actual feeds that I want.

After that, you just have to hook these collections into your templates. I won't get into that here.

More To Come

This website, as I mentioned before, isn't simple. In the end, I ended up writing code to:

• provide tag and archive pages for all my posts
• unfurl external URLs, so we can see context for likes and bookmarks
• receive and display webmentions
• send webmentions and syndicate posts

It's a lot, and I'm not even getting into the actual web design of the site, which might deserve its own set of posts.

In any case, I hope someone gets something out of this. Stay tuned to go further down the rabbit hole.