Recognise regions of countries

Refs #31

.eslintrc.json

@@ -24,7 +24,7 @@
     "import/no-cycle": "error",
     "import/no-duplicates": ["error", { "prefer-inline": true }],
     "import/no-extraneous-dependencies": ["error"],
-    "import/no-internal-modules": ["error", { "allow": ["*/lib/*"] }],
+    "import/no-internal-modules": ["error", { "allow": ["*/lib/*", "world_countries_lists/data/**/*.json"] }],
     "import/no-named-as-default": "off",
     "import/no-named-as-default-member": "off",
     "quotes": ["error", "single", { "avoidEscape": true }]

package.json

@@ -9,14 +9,17 @@
     "@observablehq/framework": "^1.9.0",
     "d3-dsv": "^3.0.1",
     "d3-time-format": "^4.1.0",
+    "diacritics": "^1.3.0",
     "effect": "^3.3.5",
     "i18n-iso-countries": "^7.11.2",
-    "iso-639-1": "^3.1.2"
+    "iso-639-1": "^3.1.2",
+    "world_countries_lists": "^2.9.0"
   },
   "devDependencies": {
     "@dotenvx/dotenvx": "^1.5.0",
     "@trivago/prettier-plugin-sort-imports": "^4.3.0",
     "@tsconfig/node20": "^20.1.4",
+    "@types/diacritics": "^1.3.3",
     "@types/node": "^20.14.9",
     "@typescript-eslint/eslint-plugin": "^7.14.1",
     "@typescript-eslint/parser": "^7.14.1",

src/lib/Iso3166.ts

@@ -1,6 +1,8 @@
 import { Schema } from '@effect/schema'
+import diacritics from 'diacritics'
 import { Array, Option, type Predicate, String, flow } from 'effect'
 import iso3166 from 'i18n-iso-countries'
+import subDivisions from 'world_countries_lists/data/subdivisions/subdivisions.json'
 
 export type Alpha2Code = iso3166.Alpha2Code
 
@@ -14,6 +16,18 @@ export const guessCountry: (location: string) => Option.Option<Alpha2Code> = flo
   String.replaceAll(/\((.+?)\)/g, ', $1'),
   String.replaceAll(/( and | - )/gi, ', '),
   location => Array.prepend(Array.map(location.split(',').reverse(), String.trim), location),
-  Array.findFirst(Option.liftNullable(location => iso3166.getAlpha2Code(location, 'en'))),
+  Array.findFirst(location =>
+    Option.fromNullable(iso3166.getAlpha2Code(location, 'en')).pipe(
+      Option.orElse(() =>
+        Option.map(
+          Array.findFirst(
+            subDivisions,
+            subDivision => diacritics.remove(location).toLowerCase() === subDivision.name.toLowerCase(),
+          ),
+          subDivision => subDivision.country,
+        ),
+      ),
+    ),
+  ),
   Option.filter(isAlpha2Code),
 )

test/Iso3166.test.ts

@@ -5,37 +5,37 @@ import * as _ from '../src/lib/Iso3166.js'
 describe('guessCountry', () => {
   test.for([
     ['Algiers ( Algeria)', 'DZ'],
+    ['Baylor College of Medicine, Houston, Texas', 'US'],
     ['Chicago, IL, USA', 'US'],
     ['Colorado - United States', 'US'],
     ['Czech Republic', 'CZ'],
     ['Czechia', 'CZ'],
     ['GHANA', 'GH'],
+    ['London, Ontario', 'CA'],
     ['London, UK', 'GB'],
     ['London, United Kingdom', 'GB'],
+    ['Québec', 'CA'],
     ['Rio de Janeiro, Brazil.', 'BR'],
+    ['Sunnyvale, California', 'US'],
     ['UK', 'GB'],
     ['U.K.', 'GB'],
     ['United Kingdom', 'GB'],
     ['united states', 'US'],
+    ['Washington, DC', 'US'],
   ])('guesses %s', ([input, expected]) => {
     const actual = _.guessCountry(input)
 
     expect(actual).toStrictEqual(Option.some(expected))
   })
 
   test.for([
-    'Baylor College of Medicine, Houston, Texas',
     'Beijing University of Technology, Beijing 100124, PR China',
     'Fayetteville GA (near Atlanta)',
     'London',
-    'London, Ontario',
     'Mars',
     'New York City and Los Angeles',
-    'Québec',
     'Southeast Asia',
-    'Sunnyvale, California',
     'The UK',
-    'Washington, DC',
   ])("doesn't guess %s", input => {
     const actual = _.guessCountry(input)
 

tsconfig.json

@@ -1,7 +1,8 @@
 {
   "extends": "@tsconfig/node20/tsconfig.json",
   "compilerOptions": {
-    "exactOptionalPropertyTypes": true
+    "exactOptionalPropertyTypes": true,
+    "resolveJsonModule": true
   },
   "include": ["src", "test", "*.ts"]
 }