feat(author): bake author gdocs

baker/SiteBaker.tsx

@@ -104,6 +104,7 @@ import {
 } from "../db/model/Variable.js"
 import { Knex } from "knex"
 import { getBakePath } from "@ourworldindata/components"
+import { GdocAuthor } from "../db/model/Gdoc/GdocAuthor.js"
 
 type PrefetchedAttachments = {
     linkedDocuments: Record<string, OwidGdocMinimalPostInterface>
@@ -137,6 +138,7 @@ const nonWordpressSteps = [
     "gdriveImages",
     "dods",
     "dataInsights",
+    "authors",
 ] as const
 
 const otherSteps = ["removeDeletedPosts"] as const
@@ -302,7 +304,7 @@ export class SiteBaker {
         picks?: [string[], string[], string[], string[]]
     ): Promise<PrefetchedAttachments> {
         if (!this._prefetchedAttachmentsCache) {
-            const publishedGdocs = await GdocPost.getPublishedGdocs().then(
+            const publishedGdocs = await GdocPost.getPublishedGdocPosts().then(
                 (fullGdocs) => fullGdocs.map(fullGdocToMinimalGdoc)
             )
             const publishedGdocsDictionary = keyBy(publishedGdocs, "id")
@@ -441,7 +443,7 @@ export class SiteBaker {
             postSlugs.push(post.slug)
         }
 
-        const gdocPosts = await GdocPost.getPublishedGdocs()
+        const gdocPosts = await GdocPost.getPublishedGdocPosts()
 
         for (const post of gdocPosts) {
             postSlugs.push(post.slug)
@@ -483,7 +485,7 @@ export class SiteBaker {
     async bakeGDocPosts(slugs?: string[]) {
         await db.getConnection()
         if (!this.bakeSteps.has("gdocPosts")) return
-        const publishedGdocs = await GdocPost.getPublishedGdocs()
+        const publishedGdocs = await GdocPost.getPublishedGdocPosts()
 
         const gdocsToBake =
             slugs !== undefined
@@ -765,6 +767,63 @@ export class SiteBaker {
             await this.stageWrite(outPath, html)
         }
     }
+
+    private async bakeAuthors() {
+        if (!this.bakeSteps.has("authors")) return
+
+        const publishedAuthors = await GdocAuthor.getPublishedAuthors()
+
+        for (const publishedAuthor of publishedAuthors) {
+            const attachments = await this.getPrefetchedGdocAttachments([
+                publishedAuthor.linkedDocumentIds,
+                publishedAuthor.linkedImageFilenames,
+                publishedAuthor.linkedChartSlugs.grapher,
+                publishedAuthor.linkedChartSlugs.explorer,
+            ])
+
+            // We don't need these to be attached to the gdoc in the current
+            // state of author pages. We'll keep them here as documentation
+            // of intent, until we need them.
+            // publishedAuthor.linkedCharts = {
+            //     ...attachments.linkedCharts.graphers,
+            //     ...attachments.linkedCharts.explorers,
+            // }
+
+            // Attach documents metadata linked to in the "featured work" section
+            publishedAuthor.linkedDocuments = attachments.linkedDocuments
+
+            // Attach image metadata for the profile picture and the "featured work" images
+            publishedAuthor.imageMetadata = attachments.imageMetadata
+
+            // Attach image metadata for the “latest work" images
+            await publishedAuthor.loadLatestWorkImages()
+
+            await publishedAuthor.validate()
+            if (
+                publishedAuthor.errors.filter(
+                    (e) => e.type === OwidGdocErrorMessageType.Error
+                ).length
+            ) {
+                await logErrorAndMaybeSendToBugsnag(
+                    `Error(s) baking "${
+                        publishedAuthor.slug
+                    }" :\n  ${publishedAuthor.errors
+                        .map((error) => error.message)
+                        .join("\n  ")}`
+                )
+            }
+            try {
+                await this.bakeOwidGdoc(publishedAuthor)
+            } catch (e) {
+                logErrorAndMaybeSendToBugsnag(
+                    `Error baking author with id "${publishedAuthor.id}" and slug "${publishedAuthor.slug}": ${e}`
+                )
+            }
+        }
+
+        this.progressBar.tick({ name: "✅ baked author pages" })
+    }
+
     // Pages that are expected by google scholar for indexing
     private async bakeGoogleScholar() {
         if (!this.bakeSteps.has("googleScholar")) return
@@ -927,6 +986,7 @@ export class SiteBaker {
         await this.validateGrapherDodReferences()
         await this.bakeGDocPosts()
         await this.bakeDataInsights()
+        await this.bakeAuthors()
         await this.bakeDriveImages()
     }
 

baker/algolia/indexToAlgolia.tsx

@@ -197,7 +197,7 @@ function generateGdocRecords(
 // Generate records for countries, WP posts (not including posts that have been succeeded by Gdocs equivalents), and Gdocs
 const getPagesRecords = async (knex: Knex<any, any[]>) => {
     const pageviews = await getAnalyticsPageviewsByUrlObj(knex)
-    const gdocs = await GdocPost.getPublishedGdocs()
+    const gdocs = await GdocPost.getPublishedGdocPosts()
     const publishedGdocsBySlug = keyBy(gdocs, "slug")
     // TODO: the knex instance should be handed down as a parameter
     const slugsWithPublishedGdocsSuccessors =

baker/sitemap.ts

@@ -70,7 +70,7 @@ export const makeSitemap = async (
         undefined,
         (postrow) => !alreadyPublishedViaGdocsSlugsSet.has(postrow.slug)
     )
-    const gdocPosts = await GdocPost.getPublishedGdocs()
+    const gdocPosts = await GdocPost.getPublishedGdocPosts()
 
     const publishedDataInsights = await db.getPublishedDataInsights(knex)
     const dataInsightFeedPageCount = calculateDataInsightIndexPageCount(

db/model/Gdoc/GdocAuthor.ts

@@ -1,4 +1,4 @@
-import { Entity, Column } from "typeorm"
+import { Entity, Column, Raw } from "typeorm"
 import {
     OwidGdocErrorMessage,
     OwidGdocAuthorInterface,
@@ -8,6 +8,7 @@ import {
     OwidGdocErrorMessageType,
     DbEnrichedLatestWork,
     DEFAULT_GDOC_FEATURED_IMAGE,
+    OwidGdocType,
 } from "@ourworldindata/utils"
 import { GdocBase } from "./GdocBase.js"
 import { htmlToEnrichedTextBlock } from "./htmlToEnriched.js"
@@ -35,7 +36,11 @@ export class GdocAuthor extends GdocBase implements OwidGdocAuthorInterface {
         return blocks
     }
 
-    _loadSubclassAttachments = async (): Promise<void> => {
+    _loadSubclassAttachments = (): Promise<void> => {
+        return this.loadLatestWorkImages()
+    }
+
+    loadLatestWorkImages = async (): Promise<void> => {
         if (!this.content.title) return
         const knex = db.knexInstance()
 
@@ -109,4 +114,16 @@ export class GdocAuthor extends GdocBase implements OwidGdocAuthorInterface {
         }
         return errors
     }
+
+    static async getPublishedAuthors(): Promise<GdocAuthor[]> {
+        return GdocAuthor.find({
+            where: {
+                published: true,
+                content: Raw(
+                    (content) =>
+                        `${content}->"$.type" = '${OwidGdocType.Author}'`
+                ),
+            },
+        })
+    }
 }

db/model/Gdoc/GdocPost.ts

@@ -210,7 +210,7 @@ export class GdocPost extends GdocBase implements OwidGdocPostInterface {
         return parseDetails(gdoc.content.details)
     }
 
-    static async getPublishedGdocs(): Promise<GdocPost[]> {
+    static async getPublishedGdocPosts(): Promise<GdocPost[]> {
         // #gdocsvalidation this cast means that we trust the admin code and
         // workflow to provide published articles that have all the required content
         // fields (see #gdocsvalidationclient and pending #gdocsvalidationserver).
@@ -246,7 +246,7 @@ export class GdocPost extends GdocBase implements OwidGdocPostInterface {
     /**
      * Excludes published listed Gdocs with a publication date in the future
      */
-    static async getListedGdocs(): Promise<GdocPost[]> {
+    static async getListedGdocPosts(): Promise<GdocPost[]> {
         return GdocPost.findBy({
             published: true,
             publicationContext: OwidGdocPublicationContext.listed,

db/model/Post.ts

@@ -243,7 +243,7 @@ const selectHomepagePosts: FilterFnPostRestApi = (post) =>
 export const getBlogIndex = memoize(
     async (knex: Knex<any, any[]>): Promise<IndexPost[]> => {
         await db.getConnection() // side effect: ensure connection is established
-        const gdocPosts = await GdocPost.getListedGdocs()
+        const gdocPosts = await GdocPost.getListedGdocPosts()
         const wpPosts = await Promise.all(
             await getPostsFromSnapshots(
                 knex,