Release same-origin blog images

scripts/sync-blog-posts.mjs

@@ -2,6 +2,7 @@ import { mkdir, writeFile } from 'node:fs/promises';
 
 const BLOG_REPO = 'IndopenSource/Blog-IndopenSource';
 const OUT_FILE = new URL('../src/data/blog-posts.json', import.meta.url);
+const BLOG_ASSET_DIR = new URL('../public/blog-assets/', import.meta.url);
 const token = process.env.GITHUB_TOKEN || process.env.GH_TOKEN;
 
 async function requestJson(url) {
@@ -156,6 +157,38 @@ function resolveThumbnail(value, articlePath, branch) {
   return new URL(value, `https://raw.githubusercontent.com/${BLOG_REPO}/${branch}/${articleDirectory}/`).toString();
 }
 
+async function mirrorBlogAsset(value, articlePath, branch) {
+  const resolved = resolveThumbnail(value, articlePath, branch);
+  if (!resolved.startsWith(`https://raw.githubusercontent.com/${BLOG_REPO}/`)) return resolved;
+
+  const assetName = new URL(resolved).pathname.split('/').pop()?.replace(/[^A-Za-z0-9._-]/g, '');
+  if (!assetName) return resolved;
+
+  const [year, month] = articlePath.split('/').slice(1, 3);
+  const assetDirectory = new URL(`${year}/${month}/`, BLOG_ASSET_DIR);
+  const response = await fetch(resolved);
+  if (!response.ok) {
+    console.warn(`Could not mirror blog asset ${resolved}: HTTP ${response.status}`);
+    return resolved;
+  }
+
+  await mkdir(assetDirectory, { recursive: true });
+  await writeFile(new URL(assetName, assetDirectory), Buffer.from(await response.arrayBuffer()));
+  return `/blog-assets/${year}/${month}/${assetName}`;
+}
+
+async function mirrorContentAssets(content, articlePath, branch) {
+  const rawAssetUrls = [...new Set(
+    content.match(/https:\/\/raw\.githubusercontent\.com\/IndopenSource\/Blog-IndopenSource\/[^\s"'<>)]*/g) || []
+  )];
+
+  for (const url of rawAssetUrls) {
+    content = content.replaceAll(url, await mirrorBlogAsset(url, articlePath, branch));
+  }
+
+  return content;
+}
+
 async function getCommitMeta(path) {
   const commits = await requestJson(`https://api.github.com/repos/${BLOG_REPO}/commits?path=${encodeURIComponent(path)}&per_page=100`);
   const firstCommit = commits.at(-1);
@@ -269,6 +302,8 @@ for (const path of articleFiles) {
         )
       ]
     : commitMeta.authors.length ? commitMeta.authors : [resolvedAuthor.author];
+  const thumbnail = await mirrorBlogAsset(data.thumbnail || data.image || data.cover || '', path, defaultBranch);
+  const mirroredContent = await mirrorContentAssets(content, path, defaultBranch);
 
   posts.push({
     slug: slugFromPath(path),
@@ -282,8 +317,11 @@ for (const path of articleFiles) {
     status: data.status || 'draft',
     lang: data.lang === 'en' ? 'en' : 'id',
     translationKey: data.translationKey || '',
-    thumbnail: resolveThumbnail(data.thumbnail || data.image || data.cover || '', path, defaultBranch),
-    content,
+    thumbnail,
+    thumbnailWidth: Number(data.thumbnailWidth) || 0,
+    thumbnailHeight: Number(data.thumbnailHeight) || 0,
+    thumbnailType: data.thumbnailType || '',
+    content: mirroredContent,
     sourceUrl: file.html_url,
     // The visible "release" date is the editorial frontmatter `date` (the date
     // the author intended), falling back to the first commit only when absent.

src/components/BlogArticlePage.astro

@@ -32,6 +32,8 @@ const releaseDate = post.releasedAt
   : '';
 const thumbnail = post.thumbnail || '/brand/indopensource-hero.jpg';
 const thumbnailUrl = /^https?:\/\//.test(thumbnail) ? thumbnail : withBase(thumbnail);
+const thumbnailWidth = post.thumbnailWidth || 1280;
+const thumbnailHeight = post.thumbnailHeight || 720;
 const heroAlt = english ? `Article thumbnail: ${post.title}` : `Thumbnail artikel: ${post.title}`;
 const authorByline = post.authorFromFrontmatter
   ? english
@@ -65,6 +67,9 @@ const jsonLd = {
   title={seoTitle}
   description={post.description}
   image={thumbnailUrl}
+  imageWidth={post.thumbnailWidth}
+  imageHeight={post.thumbnailHeight}
+  imageType={post.thumbnailType}
   type="article"
   robots={published ? undefined : 'noindex, nofollow, noarchive'}
   publishedAt={published ? post.releasedAt : undefined}
@@ -99,7 +104,7 @@ const jsonLd = {
       ))}
     </div>
     <figure class="riso-hover riso-shadow mt-8 overflow-hidden border-2 border-ink bg-ink">
-      <img class="aspect-[16/9] w-full object-cover" src={thumbnailUrl} alt={heroAlt} width="1280" height="720" loading="eager" fetchpriority="high" decoding="async" />
+      <img class="aspect-[16/9] w-full object-cover" src={thumbnailUrl} alt={heroAlt} width={thumbnailWidth} height={thumbnailHeight} loading="eager" fetchpriority="high" decoding="async" />
     </figure>
   </PageHeader>
 

src/layouts/BaseLayout.astro

@@ -8,6 +8,9 @@ interface Props {
   description?: string;
   /** Open Graph / Twitter image, absolute or site-relative. */
   image?: string;
+  imageWidth?: number;
+  imageHeight?: number;
+  imageType?: string;
   /** Open Graph `og:type` (e.g. `website`, `article`). */
   type?: string;
   /** Search-engine indexing policy for this page. */
@@ -38,6 +41,9 @@ const {
   title = 'IndopenSource',
   description = 'Roadmap website IndopenSource untuk proyek open source Indonesia.',
   image = '/brand/indopensource-hero.jpg',
+  imageWidth,
+  imageHeight,
+  imageType,
   type = 'website',
   robots = 'index, follow',
   publishedAt,
@@ -210,6 +216,9 @@ const csp = [
     <meta property="og:url" content={canonicalUrl} />
     <meta property="og:image" content={imageUrl} />
     <meta property="og:image:alt" content={title} />
+    {imageWidth && <meta property="og:image:width" content={String(imageWidth)} />}
+    {imageHeight && <meta property="og:image:height" content={String(imageHeight)} />}
+    {imageType && <meta property="og:image:type" content={imageType} />}
     {type === 'article' && publishedAt && <meta property="article:published_time" content={publishedAt} />}
     {type === 'article' && modifiedAt && <meta property="article:modified_time" content={modifiedAt} />}
     <meta name="twitter:card" content="summary_large_image" />

src/lib/content.ts

@@ -43,6 +43,9 @@ export interface BlogPost {
   /** Shared key used to connect translated versions of one article. */
   translationKey?: string;
   thumbnail: string;
+  thumbnailWidth?: number;
+  thumbnailHeight?: number;
+  thumbnailType?: string;
   content: string;
   sourceUrl: string;
   /** Editorial publication date (frontmatter `date`), falling back to commit metadata. */

test/lib.test.mjs

@@ -10,6 +10,7 @@
 // redefining any logic here, so the tests fail if that behaviour regresses.
 
 import { strict as assert } from 'node:assert';
+import { existsSync, readFileSync } from 'node:fs';
 import { describe, it } from 'node:test';
 
 import { projectSlug, rankProjectOwners } from '../src/lib/projects.ts';
@@ -138,6 +139,21 @@ describe('article language routing', () => {
   });
 });
 
+describe('synced article thumbnails', () => {
+  it('uses same-origin files with complete Open Graph metadata', () => {
+    const posts = JSON.parse(readFileSync(new URL('../src/data/blog-posts.json', import.meta.url), 'utf8'));
+
+    for (const post of posts) {
+      assert.match(post.thumbnail, /^\/blog-assets\//);
+      assert.ok(existsSync(new URL(`../public${post.thumbnail}`, import.meta.url)), `missing ${post.thumbnail}`);
+      assert.ok(post.thumbnailWidth > 0);
+      assert.ok(post.thumbnailHeight > 0);
+      assert.match(post.thumbnailType, /^image\//);
+      assert.doesNotMatch(post.content, /raw\.githubusercontent\.com/);
+    }
+  });
+});
+
 describe('renderArticle (frontmatter content)', () => {
   it('renders Markdown to HTML', () => {
     const html = renderArticle('# Title\n\nA **bold** paragraph.');