Guess reasonably sized settlements

Refs #31

package.json

@@ -8,6 +8,7 @@
     "@js-temporal/polyfill": "^0.4.4",
     "@observablehq/framework": "^1.9.0",
     "@vvo/tzdb": "^6.141.0",
+    "all-the-cities": "^3.1.0",
     "d3-dsv": "^3.0.1",
     "d3-time-format": "^4.1.0",
     "diacritics": "^1.3.0",
@@ -20,6 +21,7 @@
     "@dotenvx/dotenvx": "^1.5.0",
     "@trivago/prettier-plugin-sort-imports": "^4.3.0",
     "@tsconfig/node20": "^20.1.4",
+    "@types/all-the-cities": "^3.1.3",
     "@types/diacritics": "^1.3.3",
     "@types/node": "^20.14.9",
     "@typescript-eslint/eslint-plugin": "^7.14.1",

src/lib/Iso3166.ts

@@ -1,5 +1,6 @@
 import { Schema } from '@effect/schema'
 import { rawTimeZones } from '@vvo/tzdb'
+import cities from 'all-the-cities'
 import diacritics from 'diacritics'
 import { Array, Option, type Predicate, String, flow } from 'effect'
 import iso3166 from 'i18n-iso-countries'
@@ -12,6 +13,8 @@ export const isAlpha2Code: Predicate.Refinement<unknown, Alpha2Code> = (u): u is
 
 export const Alpha2CodeSchema: Schema.Schema<Alpha2Code, string> = Schema.String.pipe(Schema.filter(isAlpha2Code))
 
+const cities10000 = Array.filter(cities, city => city.population >= 10_000)
+
 export const guessCountry: (location: string) => Option.Option<Alpha2Code> = flow(
   String.replaceAll('.', ''),
   String.replaceAll(/\((.+?)\)/g, ', $1'),
@@ -40,6 +43,15 @@ export const guessCountry: (location: string) => Option.Option<Alpha2Code> = flo
           timeZone => timeZone.countryCode,
         ),
       ),
+      Option.orElse(() =>
+        Option.map(
+          Array.findFirst(
+            cities10000,
+            city => diacritics.remove(location).toLowerCase() === diacritics.remove(city.name).toLowerCase(),
+          ),
+          city => city.country,
+        ),
+      ),
     ),
   ),
   Option.filter(isAlpha2Code),

test/Iso3166.test.ts

@@ -11,12 +11,14 @@ describe('guessCountry', () => {
     ['Colorado - United States', 'US'],
     ['Czech Republic', 'CZ'],
     ['Czechia', 'CZ'],
+    ['Düsseldorf', 'DE'],
     ['GHANA', 'GH'],
     ['London', 'GB'],
     ['London, Ontario', 'CA'],
     ['London, UK', 'GB'],
     ['London, United Kingdom', 'GB'],
     ['New York City and Los Angeles', 'US'],
+    ['Norwich', 'GB'],
     ['Québec', 'CA'],
     ['Rio de Janeiro, Brazil.', 'BR'],
     ['Sunnyvale, California', 'US'],
@@ -31,18 +33,12 @@ describe('guessCountry', () => {
     expect(actual).toStrictEqual(Option.some(expected))
   })
 
-  test.for([
-    'Babol iran',
-    'Düsseldorf',
-    'Fayetteville GA (near Atlanta)',
-    'Mars',
-    'Norwich',
-    'Southeast Asia',
-    'The UK',
-    'Yale university',
-  ])("doesn't guess %s", input => {
-    const actual = _.guessCountry(input)
+  test.for(['Babol iran', 'Fayetteville GA (near Atlanta)', 'Mars', 'Southeast Asia', 'The UK', 'Yale university'])(
+    "doesn't guess %s",
+    input => {
+      const actual = _.guessCountry(input)
 
-    expect(actual).toStrictEqual(Option.none())
-  })
+      expect(actual).toStrictEqual(Option.none())
+    },
+  )
 })