Add some logic so the upgrader can know how to handle the new locale …

…for UTF-8 migration.

Sources/Lang.php

@@ -622,6 +622,90 @@ public static function numberFormat(int|float|string $number, ?int $decimals = n
 			self::$thousands_separator,
 		);
 	}
+
+	/**
+	 * Mapper for old Language names to locales.
+	 * This is the full array.  Languages can map to:
+	 * - null: No translation, langauge is removed and no UTF-8 upgrade possible.
+	 * - Same locales: Merging language into the same locale.
+	 * 
+	 * This MAY NOT map the same name to multiple locales.
+	 * This is used to support Upgrading from SMF 2.1 and below.
+	 * 
+	 * @return array Key/Value array with Key being the old name and value the locale.
+	 */
+	public static function oldLanguageMap(): array
+	{
+		/**
+		 *  Some notes:
+		 * - english_pirate isn't a real language, so we use the _x_ to mark it as a 'private language'.
+		 * - Various languages had 'informal', these are not a locale, so we typically just mapped these to the 'root'
+		 * */ 
+		return [
+			'albanian' => 'sq_AL',
+			'arabic' => 'ar_001', // 001 is the region code the whole world, so this means modern standard Arabic.
+			'bulgarian' => 'bg_BG',
+			'cambodian' => 'km_KH',
+			'catalan' => 'ca_ES',
+			'chinese-simplified' => 'zh_Hans', // Hans indicates the script rather than a country.
+			'chinese-traditional' => 'zh_Hant', // Hant indicates the script rather than a country.
+			'croatian' => 'hr_HR',
+			'czech' => 'cs_CZ',
+			'czech_informal' => 'cs',
+			'danish' => 'da_DK',
+			'dutch' => 'nl_NL',
+			'english' => 'en_US',
+			'english_british' => 'en_GB',
+			'english_pirate' => 'en_x_pirate',
+			'esperanto' => 'eo',
+			'finnish' => 'fi_FI',
+			'french' => 'fr_FR',
+			'galician' => 'gl_ES',
+			'german' => 'de_DE',
+			'german_informal' => 'de',
+			'greek' => 'el_GR',
+			'hebrew' => 'he_IL',
+			'hungarian' => 'hu_HU',
+			'indonesian' => 'id_ID',
+			'italian' => 'it_IT',
+			'japanese' => 'ja_JP',
+			'lithuanian' => 'lt_LT',
+			'macedonian' => 'mk_MK',
+			'malay' => 'ms_MY',
+			'norwegian' => 'nb_NO',
+			'persian' => 'fa_IR',
+			'polish' => 'pl_PL',
+			'portuguese_brazilian' => 'pt_BR',
+			'portuguese_pt' => 'pt_PT',
+			'romanian' => 'ro_RO',
+			'russian' => 'ru_RU',
+			'serbian_cyrillic' => 'sr_Cyrl', // Cyrl indicates Cyrillic script.
+			'serbian_latin' => 'sr_Latn', // Latn indicates Latin script.
+			'slovak' => 'sk_SK',
+			'slovenian' => 'sl_SI',
+			'spanish_es' => 'es_ES',
+			'spanish_latin' => 'es_419', // 419 is the region code for Latin America.
+			'swedish' => 'sv_SE',
+			'thai' => 'th_TH',
+			'turkish' => 'tr_TR',
+			'ukrainian' => 'uk_UA',
+			'urdu' => 'ur_PK',
+			'vietnamese' => 'vi_VN',
+		];
+	}
+
+	/**
+	 * Given a old language naming, this will return the locale name.
+	 * This is used to support upgrading from SMF 2.1 and below.
+	 * This is also used to support compatibility for customizations. 
+	 * 
+	 * @param string $lang Language name
+	 * @return ?string Locale is returned if found, null otherwise.
+	 */
+	public static function getLocaleFromLanguageName(string $lang): ?string
+	{
+		return self::oldLanguageMap()[$lang] ?? null;
+	}
 }
 
 // Export properties to global namespace for backward compatibility.

other/upgrade.php

@@ -3278,6 +3278,17 @@ function ConvertUtf8()
 			'yoruba' => 'UTF-8',
 		];
 
+		// Map in the new locales. We do it like this, because we want to try our best to capture
+		// the correct charset no mater what the status of the language upgrade is.
+		foreach ($lang_charsets as $key => $value) {
+			// This could be more efficient, but its upgrade logic.
+			$locale = Lang::getLocaleFromLanguageName($key);
+
+			if ($locale !== null) {
+				$lang_charsets[$locale] = $value;
+			}
+		}
+
 		// Default to ISO-8859-1 unless we detected another supported charset
 		$upcontext['charset_detected'] = (isset($lang_charsets[Config::$language], $charsets[strtr(strtolower($upcontext['charset_detected']), ['utf' => 'UTF', 'iso' => 'ISO'])])) ? $lang_charsets[Config::$language] : 'ISO-8859-1';