Enrich auto-generated BlogPosting JSON-LD with missing recommended fields

[Vallhalen]

Feature Request

EmDash auto-generates BlogPosting JSON-LD via buildBlogPostingJsonLd() in src/page/jsonld.ts. This is great — it prevents XSS bugs from hand-rolled serialization and ensures every article page has structured data.

However, the generated schema is missing several fields that Google recommends for Article/BlogPosting rich results. This forces developers to add a second, manual BlogPosting block with the missing fields — creating duplicate JSON-LD blocks on every article page.

Currently generated

{
  "@type": "BlogPosting",
  "headline": "...",
  "description": "...",
  "image": "...",
  "url": "...",
  "datePublished": "...",
  "dateModified": "...",
  "author": { "@type": "Person", "name": "..." },
  "publisher": { "@type": "Organization", "name": "..." },
  "mainEntityOfPage": { "@type": "WebPage", "@id": "..." }
}

Missing fields (recommended by Google)

Field	Why it matters
@id	Links this entity to other schema blocks on the page (e.g., https://example.com/blog/post-slug)
inLanguage	Required for multilingual sites, helpful for AI citation systems
articleSection	Maps to post categories — improves topical signals
author.@id	Links to a Person entity defined elsewhere on the site
author.url	Author profile page URL
publisher.@id	Links to Organization entity
publisher.logo	Required by Google for publisher rich results (ImageObject with url)

Proposal

Extend buildBlogPostingJsonLd() and PublicPageContext to support these fields. Possible approach:

1. Add @id — derive from page.canonical (already available)
2. Add inLanguage — derive from page.locale (already available)
3. Add articleSection — new optional field in articleMeta or PublicPageContext
4. Extend author — allow articleMeta.author to be an object { name, url, id } instead of just a string
5. Extend publisher — allow siteName to include logo URL, or add a publisherLogo field

This way the auto-generated schema would be rich enough that developers don't need to add their own BlogPosting block, eliminating duplicates.

Current workaround

We add a manual <script type="application/ld+json"> with a complete BlogPosting in our [slug].astro template. This works but creates two BlogPosting blocks per page — the EmDash one (id: "primary") and ours (raw script tag). Google handles it but it's noisy.

Related

• The same approach could enrich buildWebSiteJsonLd() with @id, description, inLanguage, publisher, and potentialAction (SearchAction) — currently it only has name and url.

[Vallhalen]

Status check on 0.13.0 — partially addressed. buildBlogPostingJsonLd() now emits author, publisher, and mainEntityOfPage (great — those were the biggest gaps). Current 0.13 output:

{ "@context":"https://schema.org", "@type":"BlogPosting", "headline":"…", "description":"…",
  "image":"…", "url":"…", "datePublished":"…", "dateModified":"…",
  "author":{"@type":"Person","name":"…"},
  "publisher":{"@type":"Organization","name":"…"},
  "mainEntityOfPage":{"@type":"WebPage","@id":"…"} }

Still missing from the original request (so keeping this open):

• top-level @id (entity linking)
• inLanguage (derivable from page.locale — matters for multilingual + AI citation)
• articleSection (post categories → topical signal)
• author.@id / author.url (link to Person entity / profile)
• publisher.@id / publisher.logo (ImageObject — required by Google for publisher rich results)

We still add a manual BlogPosting block to fill these. The inLanguage + publisher.logo ones would give the most SEO value for the least effort, since locale and site logo are already in context.

[vhscom]

Tip: you can avoid the duplicate-block problem without waiting on this feature. Instead of adding your own <script> tag, return your extra fields from a page:metadata hook as {kind: "jsonld", id: "primary", ...}. EmDash builds the page's metadata by combining plugin output, site output, and its own defaults in that order, and it only keeps the first block it sees for a given id. Since core's own BlogPosting block also uses id: "primary", a hook returning the same id replaces it instead of adding a second one, so you end up with one full block instead of two.

For example,

// src/plugins/enrich-blogposting/index.ts
//
// Descriptor: tells EmDash this plugin exists and how to load it.
// This file is what you import into astro.config.mjs.

import type { PluginDescriptor } from "emdash";

export function enrichBlogPostingPlugin(): PluginDescriptor {
  return {
    id: "enrich-blogposting",
    version: "1.0.0",
    format: "standard",
    entrypoint: new URL("./runtime.ts", import.meta.url).pathname,
    options: {},
  };
}

// src/plugins/enrich-blogposting/runtime.ts
//
// Hook logic: loaded via the descriptor's `entrypoint` above.
//
// The trick: core's own BlogPosting block is contributed with id: "primary".
// EmDash keeps only the FIRST metadata block it sees for a given id, and
// plugin contributions are checked before core's own defaults. Returning
// our block with the same id: "primary" replaces core's block instead of
// adding a second <script> tag next to it.

import type { PageMetadataContribution, PageMetadataEvent } from "emdash";

function cleanJsonLd(obj: Record<string, unknown>): Record<string, unknown> {
  const cleaned: Record<string, unknown> = {};
  for (const [key, value] of Object.entries(obj)) {
    if (value === undefined || value === null) continue;
    cleaned[key] =
      typeof value === "object" && !Array.isArray(value)
        ? cleanJsonLd(value as Record<string, unknown>)
        : value;
  }
  return cleaned;
}

export default {
  hooks: {
    "page:metadata": {
      handler: (event: PageMetadataEvent): PageMetadataContribution | null => {
        const { page } = event;
        if (page.pageType !== "article" || !page.canonical) return null;

        const ogTitle = page.seo?.ogTitle ?? page.pageTitle ?? page.title;
        const description = page.seo?.ogDescription || page.description;
        const ogImage = page.seo?.ogImage || page.image || null;
        const { publishedTime, modifiedTime, author } = page.articleMeta ?? {};
        const siteName = page.siteName;

        // Swap these for your own site's real values.
        const authorUrl = "https://example.com/authors/jane-doe";
        const publisherLogoUrl = "https://example.com/logo.png";
        const category = "Tutorials";

        return {
          kind: "jsonld",
          id: "primary", // must match core's id to replace, not duplicate
          graph: cleanJsonLd({
            "@context": "https://schema.org",
            "@type": "BlogPosting",
            "@id": page.canonical,
            headline: ogTitle,
            description,
            image: ogImage || undefined,
            url: page.canonical,
            inLanguage: page.locale || "en",
            articleSection: category,
            datePublished: publishedTime || undefined,
            dateModified: modifiedTime || publishedTime || undefined,
            author: author
              ? { "@type": "Person", name: author, url: authorUrl }
              : undefined,
            publisher: siteName
              ? {
                  "@type": "Organization",
                  name: siteName,
                  logo: { "@type": "ImageObject", url: publisherLogoUrl },
                }
              : undefined,
            mainEntityOfPage: { "@type": "WebPage", "@id": page.canonical },
          }),
        };
      },
    },
  },
};

// astro.config.mjs — register the plugin
import { enrichBlogPostingPlugin } from "./src/plugins/enrich-blogposting/index.ts";

export default defineConfig({
  integrations: [
    emdash({
      plugins: [
        // ...your other plugins
        enrichBlogPostingPlugin(),
      ],
    }),
  ],
});

[MA2153]

Converted this to a discussion as it is a feature request.