From 35db5ff31e17d3c1b85e1bb6658d3fa074a2bfb3 Mon Sep 17 00:00:00 2001
From: Daniel Bachler <daniel@danielbachler.de>
Date: Fri, 13 Oct 2023 13:37:17 +0200
Subject: [PATCH] :hammer: tweak related research query

This deduplicates by url, sorts authors and only uses full chart embeds and ignores plain links to charts
---
 db/wpdb.ts | 20 ++++++++++++++++----
 1 file changed, 16 insertions(+), 4 deletions(-)

diff --git a/db/wpdb.ts b/db/wpdb.ts
index 5df0030fc84..0be61f0b173 100644
--- a/db/wpdb.ts
+++ b/db/wpdb.ts
@@ -653,6 +653,7 @@ export const getRelatedChartsForVariable = async (
 
 interface RelatedResearchQueryResult {
     linkTargetSlug: string
+    componentType: string
     chartSlug: string
     title: string
     postSlug: string
@@ -670,6 +671,7 @@ export const getRelatedResearchAndWritingForVariable = async (
             select
                 distinct
                 pl.target as linkTargetSlug,
+                pl.componentType as componentType,
                 c.slug as chartSlug,
                 p.title as title,
                 p.slug as postSlug,
@@ -694,6 +696,7 @@ export const getRelatedResearchAndWritingForVariable = async (
             	pg.id = p.gdocSuccessorId
             where
                 pl.linkType = 'grapher'
+                and componentType = 'src' -- this filters out links in tags and keeps only embedded charts
                 and cd.variableId = ?
                 and cd.property in ('x', 'y') -- ignore cases where the indicator is size, color etc
                 and p.status = 'publish' -- only use published wp charts
@@ -707,6 +710,7 @@ export const getRelatedResearchAndWritingForVariable = async (
             select
                 distinct
                 pl.target as linkTargetSlug,
+                pl.componentType as componentType,
                 c.slug as chartSlug,
                 p.content ->> '$.title' as title,
                 p.slug as postSlug,
@@ -729,6 +733,7 @@ export const getRelatedResearchAndWritingForVariable = async (
                 pv.url = concat('https://ourworldindata.org/', p.slug )
             where
                 pl.linkType = 'grapher'
+                and componentType = 'chart' -- this filters out links in tags and keeps only embedded charts
                 and cd.variableId = ?
                 and cd.property in ('x', 'y') -- ignore cases where the indicator is size, color etc
                 and p.published = 1`,
@@ -741,14 +746,17 @@ export const getRelatedResearchAndWritingForVariable = async (
     // but it seemed easier to understand if we do the sort here
     const sorted = sortBy(combined, (post) => -post.pageviews)
 
-    return sorted.map((post) => {
+    const allSortedRelatedResearch = sorted.map((post) => {
         const parsedAuthors = JSON.parse(post.authors)
         // The authors in the gdocs table are just a list of strings, but in the wp_posts table
         // they are a list of objects with an "author" key and an "order" key. We want to normalize this so that
         // we can just use the same code to display the authors in both cases.
-        const authors = parsedAuthors.map((author: any) =>
-            !isString(author) ? author.author : author
-        )
+        let authors
+        if (parsedAuthors.length > 0 && !isString(parsedAuthors[0])) {
+            authors = sortBy(parsedAuthors, (author) => author.order).map(
+                (author: any) => author.author
+            )
+        } else authors = parsedAuthors
         return {
             title: post.title,
             url: `/${post.postSlug}`,
@@ -757,6 +765,10 @@ export const getRelatedResearchAndWritingForVariable = async (
             imageUrl: post.thumbnail,
         }
     })
+    // the queries above use distinct but because of the information we pull in if the same piece of research
+    // uses different charts that all use a single indicator we would get duplicates for the post to link to so
+    // here we deduplicate by url. The first item is retained by uniqBy, latter ones are discarded.
+    return uniqBy(allSortedRelatedResearch, "url")
 }
 
 export const getRelatedArticles = async (