Blog Support 1/3: Data fetching

[drwpow]

Changes

Edit: based on feedback, and discussions, we‘ll go with import.meta.collection() for now. So this will be handled in 3 steps:

1. import.meta.collection() support (this PR). Load markdown files, expose to page.
2. Pagination/filtering/sorting/etc. Expanding on Add watch mode #1, basically. This makes pagination possible, but annoying to use.
3. Dynamic routes: needed to make pagination elegant (but not just pagination—any kind of filtering, such as tags, posts by year, searching, etc. etc.)

Testing

• Tests are passing
• Tests updated where necessary

Docs

• Docs / READMEs updated
• Code comments added where helpful

[drwpow]

This adds import glob support, but fortunately in Astro it’s much, much easier. This is basically how it‘s transformed:

- import postData from './posts/*.md';
+ import { __content as postData_0 } from './posts/muppet-babies.md';
+ import { __content as postData_1 } from './posts/muppet-treasure-island.md';
+ // …
+ const postData = [postData_0, postData_1, … postData_8];

Once the import statement is passed off to Astro, Astro does the rest on its own.

[drwpow]

Also this part is a bit hacky, but it works for .md files:

importPath.replace(/\.md$/, '')

Basically, if /pages/post/[file].md was resolved, assign it a URL of the same location but strip the extension.

[drwpow]

The updated syntax is the same:

- const postData = import.meta.collection('./posts/*.md');
+ import { __content as postData_0 } from './posts/muppet-babies.md';
+ import { __content as postData_1 } from './posts/muppet-treasure-island.md';
+ // …
+ const postData = [postData_0, postData_1, … postData_8];

[drwpow]

This is the fix to the issue in Slack: if you have a top-level mustache statement, the compiler would throw before. Now we are simply wrapping it in a Fragment.

[drwpow]

This is the big decision to evaluate in this PR. This is basically saying that we don’t want you to define a top-level schema; we just want you to load things as you need to. That way you can fold in your own data like we do with authors.json here.

[drwpow]

OpenGraph tags, baybeeee

[natemoo-re]

I really love how clean this feels, but it does fall into the overloading native import behavior category. I'd be more comfortable if we aligned with Snowpack's import.meta.glob implementation. With top-level await it should still feel ergonomic, but it's more clear that this is something custom. Thoughts?