✨ data profiles in search PR review comments

Author: ikesau

adminSiteServer/apiRoutes/gdocs.ts

@@ -563,6 +563,9 @@ export async function deleteGdoc(
     if (gdoc.published && checkIsGdocPostExcludingFragments(gdoc)) {
         await removeIndividualGdocPostFromIndex(gdoc)
     }
+    if (gdoc.published && checkIsProfile(gdoc)) {
+        await removeIndividualProfileFromIndex(gdoc as unknown as GdocProfile)
+    }
     if (gdoc.published) {
         if (!tombstone && gdocSlug && gdocSlug !== "/") {
             // Assets have TTL of one week in Cloudflare. Add a redirect to make sure
@@ -640,6 +643,16 @@ export async function getPreviewGdocIndexRecords(
         gdocJson.publishedAt = fallbackDate
         gdocJson.updatedAt ??= fallbackDate
 
+        if (checkIsProfile(gdocJson)) {
+            const payload: PagesIndexRecordsResponse = {
+                records: [],
+                count: 0,
+                message:
+                    "Profile preview is not supported — profiles are indexed at publish time for all entities",
+            }
+            return payload
+        }
+
         // Only generate records for posts (excluding fragments)
         if (
             !checkIsGdocPostExcludingFragments(gdocJson) &&

baker/algolia/configureAlgolia.ts

@@ -42,7 +42,15 @@ export const configureAlgolia = async () => {
         indexLanguages: ["en"],
 
         // see https://www.algolia.com/doc/guides/managing-results/relevance-overview/in-depth/ranking-criteria/
-        ranking: ["typo", "words", "exact", "proximity", "attribute", "custom"],
+        ranking: [
+            "typo",
+            "words",
+            "filters",
+            "exact",
+            "proximity",
+            "attribute",
+            "custom",
+        ],
         alternativesAsExact: [
             "ignorePlurals",
             "singleWordSynonym",

baker/algolia/utils/pages.ts

@@ -19,11 +19,7 @@ import {
     articulateEntity,
 } from "@ourworldindata/utils"
 import { getAlgoliaClient } from "../configureAlgolia.js"
-import {
-    PageRecord,
-    OwidGdocProfileInterface,
-    OwidGdocBaseInterface,
-} from "@ourworldindata/types"
+import { PageRecord, OwidGdocProfileInterface } from "@ourworldindata/types"
 import { getAnalyticsPageviewsByUrlObj } from "../../../db/model/Pageview.js"
 import { PAGES_INDEX } from "../../../site/search/searchUtils.js"
 import type { Hit, SearchClient } from "@algolia/client-search"
@@ -57,7 +53,10 @@ const computePageScore = (record: Omit<PageRecord, "score">): number => {
 }
 
 const getThumbnailUrl = (
-    gdoc: OwidGdocBaseInterface,
+    gdoc:
+        | OwidGdocPostInterface
+        | OwidGdocDataInsightInterface
+        | OwidGdocProfileInterface,
     cloudflareImages: Record<string, DbEnrichedImage>
 ): string => {
     if (gdoc.content.type === OwidGdocType.DataInsight) {

packages/@ourworldindata/components/src/GdocsUtils.ts

@@ -90,7 +90,10 @@ export function getPrefixedGdocPath(
                         OwidGdocType.TopicPage,
                         OwidGdocType.LinearTopicPage,
                         OwidGdocType.AboutPage,
-                        OwidGdocType.Announcement
+                        OwidGdocType.Announcement,
+                        // Profile slugs already contain the full path
+                        // (e.g. profile/energy/usa)
+                        OwidGdocType.Profile
                     ),
                 },
             },
@@ -108,12 +111,6 @@ export function getPrefixedGdocPath(
             },
             () => `${prefix}/team/${gdoc.slug}`
         )
-        .with(
-            {
-                content: { type: OwidGdocType.Profile },
-            },
-            () => `${prefix}/profile/${gdoc.slug}`
-        )
         .with(
             {
                 content: {

site/search/Autocomplete.tsx

@@ -34,7 +34,6 @@ import {
     getItemUrlForFilter,
     getPageTypeNameAndIcon,
     SEARCH_BASE_PATH,
-    extractFiltersFromQuery,
 } from "./searchUtils.js"
 import {
     getUserCountryInformation,
@@ -383,13 +382,18 @@ const createFiltersSource = (
 })
 
 /**
- * Creates a profile source for a specific country. The country name is
- * determined in `getSources` (which already runs `extractFiltersFromQuery`)
- * and passed in here so the work isn't duplicated. The full autocomplete
- * query is forwarded to Algolia so that matching words are highlighted.
+ * Creates a profile source that boosts the user's geolocated country
+ * using Algolia's `optionalFilters`. This avoids running expensive
+ * client-side country detection on every keystroke while still ensuring:
+ * - "energy" → "Energy in Canada" (boosted by geolocation)
+ * - "canada" → Canada profiles (matched by Algolia on title)
+ * - "energy france" → "Energy in France" (matched naturally)
+ *
+ * Requires the `filters` ranking criterion in the index settings
+ * (see configureAlgolia.ts).
  */
 const createProfileSource = (
-    countryName: string
+    countryName: string | undefined
 ): AutocompleteSource<BaseItem> => ({
     sourceId: "profiles",
     onSelect: algoliaOnSelect,
@@ -405,7 +409,11 @@ const createProfileSource = (
                     params: {
                         query,
                         filters: `type:${OwidGdocType.Profile}`,
-                        facetFilters: [[`availableEntities:${countryName}`]],
+                        ...(countryName && {
+                            optionalFilters: [
+                                `availableEntities:${countryName}`,
+                            ],
+                        }),
                         hitsPerPage: 1,
                     },
                 },
@@ -506,39 +514,12 @@ export function Autocomplete({
             getSources({ query }) {
                 const sources: AutocompleteSource<BaseItem>[] = []
                 if (query) {
-                    const filtersSource = createFiltersSource(
-                        allTopics,
-                        synonymMap
-                    )
-
-                    // Detect country in query using extractFiltersFromQuery
-                    // which retains exact matches (unlike suggestFiltersFromQuerySuffix)
-                    const detectedFilters = extractFiltersFromQuery(
-                        query,
-                        listedRegionsNames(),
-                        allTopics,
-                        [],
-                        { threshold: 0.75, limit: 3 },
-                        synonymMap
-                    )
-                    const detectedCountry = detectedFilters.find(
-                        (f) => f.type === FilterType.COUNTRY
-                    )
-
                     sources.push(
-                        filtersSource,
+                        createFiltersSource(allTopics, synonymMap),
+                        createProfileSource(userCountryNameRef.current),
                         AlgoliaPagesSource,
                         AlgoliaChartsSource
                     )
-
-                    const countryName =
-                        detectedCountry?.name ?? userCountryNameRef.current
-                    if (countryName) {
-                        const profileSource = createProfileSource(countryName)
-                        // Country in query: profile before pages; geolocation fallback: after pages
-                        const insertIndex = detectedCountry ? 1 : 2
-                        sources.splice(insertIndex, 0, profileSource)
-                    }
                 } else {
                     sources.push(FeaturedSearchesSource)
                 }

site/search/SearchTemplatesWriting.tsx

@@ -1,12 +1,12 @@
-import { match } from "ts-pattern"
-import { useSearchContext } from "./SearchContext.js"
-import { SearchTopicType } from "@ourworldindata/types"
-import { SearchWritingResults } from "./SearchWritingResults.js"
-import { SearchDataInsightsResults } from "./SearchDataInsightsResults.js"
-import { SearchWritingTopicsResults } from "./SearchWritingTopicsResults.js"
+import { match } from "ts-pattern";
+import { useSearchContext } from "./SearchContext.js";
+import { SearchTopicType } from "@ourworldindata/types";
+import { SearchWritingResults } from "./SearchWritingResults.js";
+import { SearchDataInsightsResults } from "./SearchDataInsightsResults.js";
+import { SearchWritingTopicsResults } from "./SearchWritingTopicsResults.js";
 
 export const SearchTemplatesWriting = () => {
-    const { templateConfig } = useSearchContext()
+    const { templateConfig } = useSearchContext();
 
     return (
         match([
@@ -97,5 +97,5 @@ export const SearchTemplatesWriting = () => {
             // Writing + No Topic + No Country + No Query
             .with([null, false, false], () => <SearchWritingTopicsResults />)
             .exhaustive()
-    )
-}
+    );
+};

site/search/SearchWritingResults.tsx

@@ -26,18 +26,20 @@ import { SearchWritingResultsSkeleton } from "./SearchWritingResultsSkeleton.js"
 import { SearchHorizontalDivider } from "./SearchHorizontalDivider.js"
 import { useSearchContext } from "./SearchContext.js"
 
-type PageHit = TopicPageHit | ProfileHit
+type TopicOrProfileHit = TopicPageHit | ProfileHit
 
-function isPageHit(hit: FlatArticleHit | PageHit): hit is PageHit {
+function isTopicOrProfileHit(
+    hit: FlatArticleHit | TopicOrProfileHit
+): hit is TopicOrProfileHit {
     return (
         hit.type === OwidGdocType.TopicPage ||
         hit.type === OwidGdocType.LinearTopicPage ||
         hit.type === OwidGdocType.Profile
     )
 }
 
-function renderPageHit(
-    hit: PageHit,
+function renderTopicOrProfileHit(
+    hit: TopicOrProfileHit,
     index: number,
     hasLargeTopic: boolean,
     analytics: ReturnType<typeof useSearchContext>["analytics"]
@@ -84,7 +86,7 @@ function SingleColumnResults({
 }) {
     const { analytics } = useSearchContext()
 
-    const allHits: (FlatArticleHit | PageHit)[] = [
+    const allHits: (FlatArticleHit | TopicOrProfileHit)[] = [
         ...profiles,
         ..._.zip(articlePages, topicPages).flatMap(
             ([articlePage, topicPage]) => [
@@ -96,8 +98,13 @@ function SingleColumnResults({
     return (
         <div className="search-writing-results__single-column">
             {allHits.map((hit, index) => {
-                if (isPageHit(hit)) {
-                    return renderPageHit(hit, index, hasLargeTopic, analytics)
+                if (isTopicOrProfileHit(hit)) {
+                    return renderTopicOrProfileHit(
+                        hit,
+                        index,
+                        hasLargeTopic,
+                        analytics
+                    )
                 } else {
                     return (
                         <SearchFlatArticleHit
@@ -131,13 +138,17 @@ function MultiColumnResults({
     const { analytics } = useSearchContext()
 
     // Profiles appear before topic pages in the tiles
-    const allPageHits: PageHit[] = [...profiles, ...topics]
+    const allTopicOrProfileHits: TopicOrProfileHit[] = [...profiles, ...topics]
 
     // Calculate interleaved layout: 4 topics for every 5 articles (ratio
     // maintained proportionally).
     const interleavedCount = Math.round((articles.length * 4) / 5)
-    const interleavedPageHits = allPageHits.slice(0, interleavedCount)
-    const remainingPageHits = allPageHits.slice(interleavedCount)
+    const interleavedTopicOrProfileHits = allTopicOrProfileHits.slice(
+        0,
+        interleavedCount
+    )
+    const remainingTopicOrProfileHits =
+        allTopicOrProfileHits.slice(interleavedCount)
     return (
         <div className="search-writing-results__grid">
             {articles.length > 0 && (
@@ -156,19 +167,24 @@ function MultiColumnResults({
                     ))}
                 </div>
             )}
-            {interleavedPageHits.length > 0 && (
+            {interleavedTopicOrProfileHits.length > 0 && (
                 <div className="search-writing-results__topics">
-                    {interleavedPageHits.map((hit, index) =>
-                        renderPageHit(hit, index, hasLargeTopic, analytics)
+                    {interleavedTopicOrProfileHits.map((hit, index) =>
+                        renderTopicOrProfileHit(
+                            hit,
+                            index,
+                            hasLargeTopic,
+                            analytics
+                        )
                     )}
                 </div>
             )}
-            {remainingPageHits.length > 0 && (
+            {remainingTopicOrProfileHits.length > 0 && (
                 <div className="search-writing-results__overflow">
-                    {remainingPageHits.map((hit, index) =>
-                        renderPageHit(
+                    {remainingTopicOrProfileHits.map((hit, index) =>
+                        renderTopicOrProfileHit(
                             hit,
-                            interleavedPageHits.length + index,
+                            interleavedTopicOrProfileHits.length + index,
                             hasLargeTopic,
                             analytics
                         )
@@ -196,11 +212,13 @@ export const SearchWritingResults = ({
         queryFn: (liteSearchClient, state, offset, length) => {
             return queryProfiles(liteSearchClient, state, offset, length)
         },
-        firstPageSize: 4,
+        firstPageSize: 2,
         laterPageSize: 4,
         enabled: showProfiles,
     })
 
+    const profileSlots = Math.min(profilesQuery.totalResults, 2)
+
     const articlesQuery = useInfiniteSearchOffset<
         SearchFlatArticleResponse,
         FlatArticleHit
@@ -209,11 +227,16 @@ export const SearchWritingResults = ({
         queryFn: (liteSearchClient, state, offset, length) => {
             return queryArticles(liteSearchClient, state, offset, length)
         },
-        firstPageSize: 2,
+        firstPageSize: 4 - profileSlots,
         laterPageSize: 6,
     })
 
     const noArticles = articlesQuery.totalResults === 0
+    const articleSlots = Math.min(articlesQuery.totalResults, 4 - profileSlots)
+    const topicFirstPageSize = 6 - profileSlots - articleSlots
+
+    const dependenciesLoaded =
+        !articlesQuery.isLoading && (!showProfiles || !profilesQuery.isLoading)
 
     const topicsQuery = useInfiniteSearchOffset<
         SearchTopicPageResponse,
@@ -223,9 +246,9 @@ export const SearchWritingResults = ({
         queryFn: (liteSearchClient, state, offset, length) => {
             return queryTopicPages(liteSearchClient, state, offset, length)
         },
-        firstPageSize: noArticles ? 6 : 2,
+        firstPageSize: topicFirstPageSize,
         laterPageSize: noArticles ? 6 : 4,
-        enabled: hasTopicPages && !articlesQuery.isLoading,
+        enabled: hasTopicPages && dependenciesLoaded,
     })
 
     const hasLargeTopic = topicsQuery.totalResults === 1