Support currency locale overrides from CSV (#82)

* Add currency locale overrides where data is missing

* Regenerate currencies to update locales

Author: ericluria

cleanse/cleanse.go

@@ -787,6 +787,28 @@ func LoadCurrencyLocales() map[string]string {
 	return table
 }
 
+
+// For some reason, some countries are missing from the underlying CLDR "Numbers" data that we ingest.
+// It's easier to add this manual override than figure out how to fix the source data.
+func LoadLocaleOverrides() []common.Locale {
+	results := []common.Locale{}
+
+	for _, record := range readCsv("data/overrides/locales.csv") {
+		results = append(results, common.Locale{
+			Id:       record["id"],
+			Name:     record["name"],
+			Country:  record["country"],
+			Language: record["language"],
+			Numbers: common.LocaleNumbers{
+				Decimal: record["numbers_decimal"],
+				Group:   record["numbers_group"],
+			},
+		})
+	}
+
+	return results
+}
+
 func LoadContinents() []Continent {
 	return []Continent{
 		Continent{

data/final/currencies.json

@@ -301,7 +301,8 @@
     "symbols": {
       "primary": "CZK",
       "narrow": "Kč"
-    }
+    },
+    "default_locale": "cs-CZ"
   },
   {
     "name": "Danish Krone",
@@ -374,7 +375,8 @@
     "number_decimals": 2,
     "symbols": {
       "primary": "ETB"
-    }
+    },
+    "default_locale": "am-ET"
   },
   {
     "name": "Euro",
@@ -706,7 +708,8 @@
     "number_decimals": 2,
     "symbols": {
       "primary": "MVR"
-    }
+    },
+    "default_locale": "dv-MV"
   },
   {
     "name": "Mauritania Ouguiya",
@@ -789,7 +792,8 @@
     "symbols": {
       "primary": "NPR",
       "narrow": "Rs"
-    }
+    },
+    "default_locale": "ne-NP"
   },
   {
     "name": "New Belarusian Ruble",
@@ -1046,7 +1050,8 @@
     "number_decimals": 2,
     "symbols": {
       "primary": "RSD"
-    }
+    },
+    "default_locale": "sr-RS"
   },
   {
     "name": "Seychelles Rupee",

data/final/locales.json

@@ -49,6 +49,16 @@
       "group": ","
     }
   },
+  {
+    "id": "am-ET",
+    "name": "Amharic - Ethiopia",
+    "country": "ETH",
+    "language": "am",
+    "numbers": {
+      "decimal": ".",
+      "group": ","
+    }
+  },
   {
     "id": "ar-DZ",
     "name": "Arabic - Algeria",
@@ -489,6 +499,16 @@
       "group": "."
     }
   },
+  {
+    "id": "cs-CZ",
+    "name": "Czech - Czechia",
+    "country": "CZE",
+    "language": "cs",
+    "numbers": {
+      "decimal": ",",
+      "group": "."
+    }
+  },
   {
     "id": "da-DK",
     "name": "Danish - Denmark",
@@ -519,6 +539,16 @@
       "group": "."
     }
   },
+  {
+    "id": "dv-MV",
+    "name": "Divehi - Maldives",
+    "country": "MDV",
+    "language": "dv",
+    "numbers": {
+      "decimal": ".",
+      "group": ","
+    }
+  },
   {
     "id": "nl-AW",
     "name": "Dutch - Aruba",
@@ -569,6 +599,16 @@
       "group": "."
     }
   },
+  {
+    "id": "nl-SR",
+    "name": "Dutch - Suriname",
+    "country": "SUR",
+    "language": "nl",
+    "numbers": {
+      "decimal": ",",
+      "group": "."
+    }
+  },
   {
     "id": "nl",
     "name": "Dutch - The Netherlands",
@@ -1589,6 +1629,16 @@
       "group": ","
     }
   },
+  {
+    "id": "et-EE",
+    "name": "Estonian - Estonia",
+    "country": "EST",
+    "language": "et",
+    "numbers": {
+      "decimal": ".",
+      "group": ","
+    }
+  },
   {
     "id": "et",
     "name": "Estonian - Ethiopia",
@@ -2419,6 +2469,16 @@
       "group": "."
     }
   },
+  {
+    "id": "lv-LV",
+    "name": "Latvian - Latvia",
+    "country": "LVA",
+    "language": "lv",
+    "numbers": {
+      "decimal": ".",
+      "group": ","
+    }
+  },
   {
     "id": "lv",
     "name": "Latvian - Latvia",
@@ -2459,6 +2519,16 @@
       "group": "."
     }
   },
+  {
+    "id": "lt-LT",
+    "name": "Lithuanian - Lithuania",
+    "country": "LTU",
+    "language": "lt",
+    "numbers": {
+      "decimal": ",",
+      "group": "."
+    }
+  },
   {
     "id": "lt",
     "name": "Lithuanian - Lithuania",
@@ -2569,6 +2639,16 @@
       "group": ","
     }
   },
+  {
+    "id": "ne-NP",
+    "name": "Nepali - Nepal",
+    "country": "NPL",
+    "language": "ne",
+    "numbers": {
+      "decimal": ".",
+      "group": ","
+    }
+  },
   {
     "id": "ne",
     "name": "Nepali - Niger",
@@ -2619,6 +2699,16 @@
       "group": ","
     }
   },
+  {
+    "id": "fa-TJ",
+    "name": "Persian - Tajikistan",
+    "country": "TJK",
+    "language": "fa",
+    "numbers": {
+      "decimal": ".",
+      "group": ","
+    }
+  },
   {
     "id": "pl",
     "name": "Polish - Poland",
@@ -2889,6 +2979,16 @@
       "group": "."
     }
   },
+  {
+    "id": "sr-RS",
+    "name": "Serbian - Serbia",
+    "country": "SRB",
+    "language": "sr",
+    "numbers": {
+      "decimal": ",",
+      "group": "."
+    }
+  },
   {
     "id": "sn",
     "name": "Shona - Senegal",
@@ -3379,6 +3479,16 @@
       "group": " "
     }
   },
+  {
+    "id": "tk-TM",
+    "name": "Turkmen - Turkmenistan",
+    "country": "TKM",
+    "language": "tk",
+    "numbers": {
+      "decimal": ",",
+      "group": "."
+    }
+  },
   {
     "id": "uk-UA",
     "name": "Ukrainian - Ukraine",

data/overrides/locales.csv

@@ -0,0 +1,24 @@
+"id","name","country","language","numbers_decimal","numbers_group"
+"en-KY","English - Cayman Islands","CYM","en",".",","
+"cs-CZ","Czech - Czechia","CZE","cs",",","."
+"es-SV","Spanish - El Salvador","SLV","es",".",","
+"et-EE","Estonian - Estonia","EST","et",".",","
+"am-ET","Amharic - Ethiopia","ETH","am",".",","
+"en-FK","English - Falkland Islands","FLK","en",".",","
+"en-GI","English - Gibraltar","GIB","en",".",","
+"fr-HT","French - Haiti","HTI","fr",",","."
+"lv-LV","Latvian - Latvia","LVA","lv",".",","
+"lt-LT","Lithuanian - Lithuania","LTU","lt",",","."
+"fr-MG","French - Madagascar","MDG","fr",",","."
+"dv-MV","Divehi - Maldives","MDV","dv",".",","
+"ar-MR","Arabic - Mauritania","MRT","ar",".",","
+"pt-MZ","Portuguese - Mozambique","MOZ","pt",",","."
+"ne-NP","Nepali - Nepal","NPL","ne",".",","
+"es-PA","Spanish - Panama","PAN","es",",","."
+"pt-ST","Portuguese - São Tomé and Príncipe","STP","pt",",","."
+"sr-RS","Serbian - Serbia","SRB","sr",",","."
+"fa-TJ","Persian - Tajikistan","TJK","fa",".",","
+"en-SH","English - Saint Helena","SHN","en",".",","
+"nl-SR","Dutch - Suriname","SUR","nl",",","."
+"tk-TM","Turkmen - Turkmenistan","TKM","tk",",","."
+"zh-TW","Chinese - Taiwan","TWN","zh",".",","

final/final.go

@@ -184,21 +184,6 @@ func commonLocales(data CleansedDataSet) []common.Locale {
 		"se":  "sw",
 	}
 
-	// For some reason, Taiwan is missing from the underlying CLDR "Numbers" data that we ingest.
-	// It's easier to add this manual override than figuring out how to fix the source data.
-	overrides := []common.Locale{
-		common.Locale{
-			Id:       "zh-TW",
-			Name:     "Chinese - Taiwan",
-			Country:  "TWN",
-			Language: "zh",
-			Numbers: common.LocaleNumbers{
-				Decimal: ".",
-				Group:   ",",
-			},
-		},
-	}
-
 	unsupportedCountryCodes := common.UnsupportedCountryCodes()
 
 	for _, n := range data.Numbers {
@@ -294,7 +279,7 @@ func commonLocales(data CleansedDataSet) []common.Locale {
 		}
 	}
 
-	all = append(all, overrides...)
+	all = append(all, cleanse.LoadLocaleOverrides()...)
 
 	uniqueLocales := uniqueLocaleIds(all)
 	sortLocales(uniqueLocales)