rewrite some names into common names, to match data-countries

martinlindhe · Oct 18, 2015 · 0bd5199 · 0bd5199
1 parent 1d4c8d5
commit 0bd5199
Show file tree

Hide file tree

Showing 3 changed files with 33 additions and 15 deletions.
diff --git a/data/calling_codes.csv b/data/calling_codes.csv
@@ -139,8 +139,8 @@ code,country
 77,Kazakhstan
 254,Kenya
 686,Kiribati
-850,"Korea, North"
-82,"Korea, South"
+850,"North Korea"
+82,"South Korea"
 965,Kuwait
 996,Kyrgyzstan
 856,Laos
@@ -270,8 +270,8 @@ code,country
 998,Uzbekistan
 678,Vanuatu
 58,Venezuela
-3906698,"Vatican City State (Holy See)"
-379,"Vatican City State (Holy See)"
+3906698,"Vatican City"
+379,"Vatican City"
 84,Vietnam
 1808,"Wake Island, USA"
 681,"Wallis and Futuna"

diff --git a/data/calling_codes.json b/data/calling_codes.json
@@ -560,11 +560,11 @@
         "code": "686"
     },
     {
-        "country": "Korea, North",
+        "country": "North Korea",
         "code": "850"
     },
     {
-        "country": "Korea, South",
+        "country": "South Korea",
         "code": "82"
     },
     {
@@ -1084,11 +1084,11 @@
         "code": "58"
     },
     {
-        "country": "Vatican City State (Holy See)",
+        "country": "Vatican City",
         "code": "3906698"
     },
     {
-        "country": "Vatican City State (Holy See)",
+        "country": "Vatican City",
         "code": "379"
     },
     {

diff --git a/tools/update_calling_codes.php b/tools/update_calling_codes.php
@@ -55,6 +55,30 @@ function write_json($fileName, $list)
 }
 
 
+function extractLinks($s)
+{
+    preg_match_all('/\[\[[\w\d\s,+|]+\]\]/', $s, $matches);
+
+    return $matches[0];
+}
+
+function mapCountry($c)
+{
+    if ($c == 'Korea, South') {
+        return 'South Korea';
+    }
+
+    if ($c == 'Korea, North') {
+        return 'North Korea';
+    }
+
+    if ($c == 'Vatican City State (Holy See)') {
+        return 'Vatican City';
+    }
+
+    return $c;
+}
+
 $res = (new MartinLindhe\MediawikiClient\Client)
     ->server('en.wikipedia.org')
     ->cacheTtlSeconds(3600) // 1 hour
@@ -83,13 +107,6 @@ function write_json($fileName, $list)
 
 $data = substr($x, $pos, $pos2 - $pos);
 
-function extractLinks($s)
-{
-    preg_match_all('/\[\[[\w\d\s,+|]+\]\]/', $s, $matches);
-
-    return $matches[0];
-}
-
 $list = [];
 
 $rows = explode("\n", $data);
@@ -100,6 +117,7 @@ function extractLinks($s)
     }
 
     $country = cleanText($rows[$i++]);
+    $country = mapCountry($country);
 
 
     $codes = extractLinks($rows[$i++]);