Merge pull request #3480 from owid/search-utils

enhance(search): Search utils

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

@@ -35,6 +35,7 @@ export {
     isTouchDevice,
     type Json,
     csvEscape,
+    escapeRegExp,
     urlToSlug,
     trimObject,
     fetchText,
@@ -240,6 +241,7 @@ export {
     countries,
     type Country,
     getCountryBySlug,
+    getRegionByNameOrVariantName,
     isCountryName,
     continents,
     type Continent,

packages/@ourworldindata/utils/src/regions.test.ts

@@ -1,6 +1,10 @@
 #! /usr/bin/env jest
 
-import { isCountryName, getCountryBySlug } from "./regions.js"
+import {
+    isCountryName,
+    getCountryBySlug,
+    getRegionByNameOrVariantName,
+} from "./regions.js"
 
 it("isCountryName", () => {
     expect(isCountryName("United States")).toEqual(true)
@@ -15,3 +19,30 @@ it("getCountryBySlug", () => {
     })
     expect(getCountryBySlug("not-a-country")).toEqual(undefined)
 })
+
+it("getRegionByNameOrVariantName", () => {
+    expect(getRegionByNameOrVariantName("United States")).toMatchObject({
+        name: "United States",
+        slug: "united-states",
+        code: "USA",
+    })
+
+    expect(getRegionByNameOrVariantName("USA")).toMatchObject({
+        name: "United States",
+        slug: "united-states",
+        code: "USA",
+    })
+
+    // Test case-insensitivity
+    expect(getRegionByNameOrVariantName("UNITED KINGDOM")).toMatchObject({
+        name: "United Kingdom",
+        slug: "united-kingdom",
+        code: "GBR",
+    })
+
+    expect(getRegionByNameOrVariantName("uae")).toMatchObject({
+        name: "United Arab Emirates",
+        slug: "united-arab-emirates",
+        code: "ARE",
+    })
+})

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

@@ -67,6 +67,18 @@ const countriesBySlug: Record<string, Country> = Object.fromEntries(
     countries.map((country) => [country.slug, country])
 )
 
+const regionsByNameOrVariantNameLowercase: Map<string, Region> = new Map(
+    regions.flatMap((region) => {
+        const names = [region.name.toLowerCase()]
+        if ("variantNames" in region && region.variantNames) {
+            names.push(
+                ...region.variantNames.map((variant) => variant.toLowerCase())
+            )
+        }
+        return names.map((name) => [name, region])
+    })
+)
+
 const currentAndHistoricalCountryNames = regions
     .filter(({ regionType }) => regionType === "country")
     .map(({ name }) => name.toLowerCase())
@@ -76,3 +88,8 @@ export const isCountryName = (name: string): boolean =>
 
 export const getCountryBySlug = (slug: string): Country | undefined =>
     countriesBySlug[slug]
+
+export const getRegionByNameOrVariantName = (
+    nameOrVariantName: string
+): Region | undefined =>
+    regionsByNameOrVariantNameLowercase.get(nameOrVariantName.toLowerCase())

site/search/SearchUtils.tsx

@@ -0,0 +1,24 @@
+import {
+    Region,
+    getRegionByNameOrVariantName,
+    regions,
+    escapeRegExp,
+} from "@ourworldindata/utils"
+
+const allCountryNamesAndVariants = regions.flatMap((c) => [
+    c.name,
+    ...(("variantNames" in c && c.variantNames) || []),
+])
+
+// A RegExp that matches any country, region and variant name. Case-independent.
+const regionNameRegex = new RegExp(
+    `\\b(${allCountryNamesAndVariants.map(escapeRegExp).join("|")})\\b`,
+    "gi"
+)
+
+export const extractRegionNamesFromSearchQuery = (query: string) => {
+    const matches = query.matchAll(regionNameRegex)
+    const regionNames = Array.from(matches, (match) => match[0])
+    if (regionNames.length === 0) return null
+    return regionNames.map(getRegionByNameOrVariantName) as Region[]
+}