Include featured image in wpPost -> ArchieML migration

baker/postUpdatedHook.ts

@@ -53,18 +53,36 @@ const syncPostToGrapher = async (
             left join wp_term_taxonomy t on t.term_taxonomy_id = r.term_taxonomy_id
             WHERE t.taxonomy = 'author' and p.ID = ?
             group by p.ID
-        )
+        ),
+        post_featured_image AS (
+        SELECT
+            p.ID,
+            (
+            SELECT
+                meta_value
+            FROM
+                wp_postmeta
+            WHERE
+                post_id = p.ID
+                AND meta_key = '_thumbnail_id') AS featured_image_id
+        FROM
+            wp_posts p
+        WHERE
+            p.ID = ?
+            )
         -- finally here we select all the fields from posts_with_authors and
         -- then we join in the first_revision to get the created_at field
         select
             p.*,
             pwa.authors,
-            fr.created_at as created_at
+            fr.created_at as created_at,
+            regexp_replace((SELECT guid FROM wp_posts WHERE ID = fi.featured_image_id), '^https://owid.cloud/(app|wp-content)/', 'https://ourworldindata.org/wp-content/') AS featured_image
         from wp_posts p
         left join post_ids_with_authors pwa on pwa.id = p.id
-        left join first_revision fr on fr.post_id = pwa.id
+        left join first_revision fr on fr.post_id = p.id
+        left join post_featured_image fi on fi.ID = p.id
         where p.id = ?`,
-        [postId, postId, postId]
+        [postId, postId, postId, postId]
     )
     const dereferenceReusableBlocksFn = await buildReusableBlocksResolver()
 
@@ -80,6 +98,7 @@ const syncPostToGrapher = async (
               type: wpPost.post_type,
               status: wpPost.post_status,
               content: wpPost.post_content,
+              featured_image: wpPost.featured_image || "",
               published_at:
                   wpPost.post_date_gmt === zeroDateString
                       ? null

db/migrateWpPostsToArchieMl.ts

@@ -90,7 +90,8 @@ const migrate = async (): Promise<void> => {
         "authors",
         "excerpt",
         "created_at_in_wordpress",
-        "updated_at"
+        "updated_at",
+        "featured_image"
     ).from(db.knexTable(Post.postsTable)) //.where("id", "=", "29766"))
 
     for (const post of posts) {
@@ -159,6 +160,7 @@ const migrate = async (): Promise<void> => {
                     subtitle: post.excerpt,
                     excerpt: post.excerpt,
                     authors: parsePostAuthors(post.authors),
+                    "featured-image": post.featured_image,
                     dateline: dateline,
                     // TODO: this discards block level elements - those might be needed?
                     refs: undefined,

db/migration/1685653967284-PostsFeaturedImage.ts

@@ -0,0 +1,17 @@
+import { MigrationInterface, QueryRunner } from "typeorm"
+
+export class PostsFeaturedImage1685653967284 implements MigrationInterface {
+    public async up(queryRunner: QueryRunner): Promise<void> {
+        queryRunner.query(`
+        ALTER TABLE posts
+        ADD COLUMN featured_image VARCHAR(1024) NOT NULL DEFAULT "";    
+        `)
+    }
+
+    public async down(queryRunner: QueryRunner): Promise<void> {
+        queryRunner.query(`
+        ALTER TABLE posts
+        DROP COLUMN featured_image;    
+        `)
+    }
+}

db/syncPostsToGrapher.ts

@@ -156,16 +156,33 @@ const syncPostsToGrapher = async (): Promise<void> => {
             from wp_posts p
 			left join posts_authors pa on p.ID = pa.id
             group by p.ID
-        )
+        ),
+        post_featured_image AS (
+        SELECT
+            p.ID,
+            (
+            SELECT
+                meta_value
+            FROM
+                wp_postmeta
+            WHERE
+                post_id = p.ID
+                AND meta_key = '_thumbnail_id') AS featured_image_id
+        FROM
+            wp_posts p
+            )
         -- Finally collect all the fields we want to keep - this is everything from wp_posts, the authors from the
         -- posts_with_authors CTE and the created_at from the first_revision CTE
         select
             p.*,
             pwa.authors as authors,
-            fr.created_at as created_at
+            fr.created_at as created_at,
+            -- select the featured image url and normalize the to point to our full domain at the wp-content folder
+            regexp_replace((SELECT guid FROM wp_posts WHERE ID = fi.featured_image_id), '^https://owid.cloud/(app|wp-content)/', 'https://ourworldindata.org/wp-content/') AS featured_image
         from wp_posts p
 		left join post_ids_with_authors pwa   on p.ID = pwa.ID
         left join first_revision fr on fr.post_id = pwa.ID
+        left join post_featured_image fi on fi.ID = p.id
         where p.post_type in ('post', 'page') AND post_status != 'trash'
         `
     )
@@ -202,6 +219,7 @@ const syncPostsToGrapher = async (): Promise<void> => {
             excerpt: post.post_excerpt,
             created_at_in_wordpress:
                 post.created_at === zeroDateString ? null : post.created_at,
+            featured_image: post.featured_image || "",
         }
     }) as PostRow[]
 

packages/@ourworldindata/utils/src/owidTypes.ts

@@ -181,6 +181,7 @@ export interface PostRow {
     authors: string
     excerpt: string
     created_at_in_wordpress: Date | null
+    featured_image: string
 }
 
 export interface PostRowWithGdocPublishStatus extends PostRow {