Better map tooltip

evals/countries.py

@@ -2,6 +2,7 @@ import re
 import xml.etree.ElementTree as ET
 from collections import defaultdict
 
+import pycountry
 from language_data.population_data import LANGUAGE_SPEAKING_POPULATION
 from language_data.util import data_filename
 
@@ -31,18 +32,21 @@ def population(bcp_47):
 def make_country_table(language_table):
     countries = defaultdict(list)
     for lang in language_table.itertuples():
-        for country, pop in population(lang.bcp_47).items():
+        for country, speaker_pop in population(lang.bcp_47).items():
             countries[country].append(
                 {
                     "name": lang.language_name,
                     "bcp_47": lang.bcp_47,
-                    "population": pop,
+                    "population": speaker_pop,
                     "score": lang.average,
                 }
             )
     for country, languages in countries.items():
-        pop = sum(entry["population"] for entry in languages)
-        score = sum(entry["score"] * entry["population"] for entry in languages) / pop
+        speaker_pop = sum(entry["population"] for entry in languages)
+        score = (
+            sum(entry["score"] * entry["population"] for entry in languages)
+            / speaker_pop
+        )
         countries[country] = {
             "score": score,
             "languages": languages,

frontend/src/components/WorldMap.js

@@ -2,6 +2,16 @@ import { useRef, useEffect, useState } from 'react'
 import * as topojson from 'topojson-client'
 import * as Plot from '@observablehq/plot'
 
+function smoothProgressBar(fraction, { color = false } = {}) {
+  const blocks = ['▏','▎','▍','▌','▋','▊','▉','█'];
+  const width = 10;
+  const totalUnits = width * 8;
+  const filledUnits = Math.round(fraction * totalUnits);
+  const fullBlocks = Math.floor(filledUnits / 8);
+  const remainder = filledUnits % 8;
+  return '█'.repeat(fullBlocks) + (remainder > 0 ? blocks[remainder - 1] : '')
+}
+
 const WorldMap = ({ data }) => {
   const containerRef = useRef()
   const [mapData, setMapData] = useState()
@@ -29,17 +39,20 @@ const WorldMap = ({ data }) => {
             const score = data.countries[d.properties?.ISO_A2_EH]?.score
             return score
           },
-          title: d => `<b>${d.properties?.ISO_A2_EH}</b> (${d.properties?.NAME_EN})`,
+          title: d => {
+            const languages = data.countries[d.properties?.ISO_A2_EH]?.languages.toSorted((a, b) => b.population - a.population)
+            const pop = languages?.map(a => a.population).reduce((prev, a) => prev + a, 0)
+            const langstring = languages?.slice(0, 10).map(a=> `${smoothProgressBar(a.population / pop)} ${a.name}`).join('\n\n') + (languages?.length > 10 ? `\n\n...` : '')
+            return `${d.properties.ADMIN}\n\n${langstring}`
+          },
           tip: true
         })
       ],
       color: {
-        range: ["red", "blue"],
+        scheme: 'Blues',
         unknown: 'gray',
-        // type: 'linear',
         label: 'Score',
         legend: true,
-        // percent: true,
         domain: [0, 0.5]
       }
     })

results.json

@@ -3048,7 +3048,7 @@
       "family": "Eskimo-Aleut",
       "flores_path": null,
       "fleurs_tag": null,
-      "commonvoice_hours": 0.0,
+      "commonvoice_hours": 7.6,
       "commonvoice_locale": "esu",
       "in_benchmark": false,
       "task": null,
@@ -4584,7 +4584,7 @@
       "family": "Eskimo-Aleut",
       "flores_path": null,
       "fleurs_tag": null,
-      "commonvoice_hours": 3.2,
+      "commonvoice_hours": 7.2,
       "commonvoice_locale": "ipk",
       "in_benchmark": false,
       "task": null,
@@ -5752,7 +5752,7 @@
       "family": "Indo-European",
       "flores_path": null,
       "fleurs_tag": null,
-      "commonvoice_hours": 3.4,
+      "commonvoice_hours": 5.3,
       "commonvoice_locale": "kw",
       "in_benchmark": false,
       "task": null,
@@ -9224,7 +9224,7 @@
       "family": null,
       "flores_path": null,
       "fleurs_tag": null,
-      "commonvoice_hours": 1.4,
+      "commonvoice_hours": 1.8,
       "commonvoice_locale": "sei",
       "in_benchmark": false,
       "task": null,
@@ -10952,7 +10952,7 @@
       "family": "Austroasiatic",
       "flores_path": "vie_Latn",
       "fleurs_tag": "vi_vn",
-      "commonvoice_hours": 5.9,
+      "commonvoice_hours": 6.0,
       "commonvoice_locale": "vi",
       "in_benchmark": true,
       "task": null,