Add post-processing/metadata middlewares/plugins on new content collections

[spaceemotion]

Body

Summary

For markdown files, we have the global remarkPlugins array for any additional steps when reading markdown files.
However, any loaders that load other file types (like .json or .mdoc) don't have a way to add plugins during the loading when using the glob or file loaders.

Background & Motivation

(Our project already uses the new loaders from the Astro 5.0 beta)

We are working with Markdoc files for our website, as they are more flexible than regular Markdown/MDX. However, certain recipes like the Add last modified time only work with direct file access and as a rehype plugin.

Since we load our Markdoc files via the glob loader, they don't go through the Markdown processing pipeline. Figuring out the last modification time via git requires knowing the actual file, which could be done as a callback in the loading step.

Technically, we could build our own loader, but that involves a lot of extra steps that seem quite convoluted (like making sure the file watchers work with vite). In case the generateId option is used, there's also no way to figure out the original file from a collection entry's ID. Having a way to inject custom meta data in the native file loaders would help.

Goals

• A plugin/middleware step in the native file loaders with access to the file path, ID and data of the entry
• The plugins would be independent of the entry IDs, but would be called after the ID has been generated, and the schema validated

Example

const docs = defineCollection({
  loader: glob({
    base: './src/content/docs',
    pattern: [
      '**/*.md',
      '!_*.md',
    ],
    
    metadata: [
      ({ filepath }) => {
        // figure out last date via git
        const lastModification = execSync(`git log -1 --pretty="format:%cI" "${filepath}"`);

        return {
          // Callbacks can return multiple results
          modifiedAt: lastModification,
        };
      },
    ],
  }),
});