Language names via intl library. Fixes #773

4 files changed, 551 insertions, 0 deletions

diff --git a/library/intl/scripts/country/generate.php b/library/intl/scripts/country/generate.php
new file mode 100644
index 000000000..879dc0a25
--- /dev/null
+++ b/library/intl/scripts/country/generate.php
@@ -0,0 +1,162 @@
+<?php
+
+/**
+ * Generates the json files stored in resources/country.
+ */
+
+set_time_limit(0);
+
+// Downloaded from http://unicode.org/Public/cldr/26/json-full.zip
+$enCountries = '../json-full/main/en/territories.json';
+$codeMappings = '../json-full/supplemental/codeMappings.json';
+$telephoneCodeData = '../json-full/supplemental/telephoneCodeData.json';
+if (!file_exists($enCountries)) {
+    die("The $enCountries file was not found");
+}
+if (!file_exists($codeMappings)) {
+    die("The $codeMappings file was not found");
+}
+if (!file_exists($telephoneCodeData)) {
+    die("The $telephoneCodeData file was not found");
+}
+if (!function_exists('collator_create')) {
+    // Reimplementing intl's collator would be a huge undertaking, so we
+    // use it instead to presort the generated locale specific data.
+    die('The intl extension was not found.');
+}
+
+$ignoredCountries = array(
+    'AN', // Netherlands Antilles, no longer exists.
+    'BV', 'HM', 'CP', // Uninhabited islands.
+    'EU', 'QO', // European Union, Outlying Oceania. Not countries.
+    'ZZ', // Unknown region
+);
+
+// Locales listed without a "-" match all variants.
+// Locales listed with a "-" match only those exact ones.
+$ignoredLocales = array(
+    // Interlingua is a made up language.
+    'ia',
+    // Valencian differs from its parent only by a single character (è/é).
+    'ca-ES-VALENCIA',
+    // Those locales are 90% untranslated.
+    'aa', 'as', 'az-Cyrl', 'az-Cyrl-AZ', 'bem', 'dua', 'gv', 'haw', 'ig', 'ii',
+    'kkj', 'kok', 'kw', 'lkt', 'mgo', 'nnh', 'nr', 'nso', 'om', 'os', 'pa-Arab',
+    'pa-Arab-PK', 'qu', 'rw', 'sah', 'smn', 'ss', 'ssy', 'st', 'tg', 'tn', 'ts',
+    'uz-Arab', 'uz-Arab-AF', 've', 'vo', 'xh', 'yi',
+    // Special "grouping" locales.
+    'root', 'en-US-POSIX', 'en-001', 'en-150', 'es-419',
+);
+
+// Assemble the base data. Use the "en" data to get a list of countries.
+$telephoneCodeData = json_decode(file_get_contents($telephoneCodeData), true);
+$telephoneCodeData = $telephoneCodeData['supplemental']['telephoneCodeData'];
+$codeMappings = json_decode(file_get_contents($codeMappings), true);
+$codeMappings = $codeMappings['supplemental']['codeMappings'];
+$countryData = json_decode(file_get_contents($enCountries), true);
+$countryData = $countryData['main']['en']['localeDisplayNames']['territories'];
+$baseData = array();
+foreach ($countryData as $countryCode => $countryName) {
+    if (is_numeric($countryCode) || in_array($countryCode, $ignoredCountries)) {
+        // Ignore continents, regions, uninhabited islands.
+        continue;
+    }
+    if (strpos($countryCode, '-alt-') !== FALSE) {
+        // Ignore alternative names.
+        continue;
+    }
+
+    $baseData[$countryCode]['code'] = $countryCode;
+    // Countries are not guaranteed to have an alpha3 and/or numeric code.
+    if (isset($codeMappings[$countryCode]['_alpha3'])) {
+        $baseData[$countryCode]['three_letter_code'] = $codeMappings[$countryCode]['_alpha3'];
+    }
+    if (isset($codeMappings[$countryCode]['_numeric'])) {
+        $baseData[$countryCode]['numeric_code'] = $codeMappings[$countryCode]['_numeric'];
+    }
+
+    // Determine the telephone code for this country.
+    if (in_array($countryCode, array('IC', 'EA'))) {
+        // "Canary Islands" and "Ceuta and Melilla" use Spain's.
+        $baseData[$countryCode]['telephone_code'] = $telephoneCodeData['ES'][0]['telephoneCountryCode'];
+    } elseif ($countryCode == 'XK') {
+        // Kosovo uses three telephone codes. Use Serbia's until that gets resolved.
+        $baseData[$countryCode]['telephone_code'] = $telephoneCodeData['RS'][0]['telephoneCountryCode'];
+    } elseif (isset($telephoneCodeData[$countryCode])) {
+        $baseData[$countryCode]['telephone_code'] = $telephoneCodeData[$countryCode][0]['telephoneCountryCode'];
+    }
+}
+
+// Write out base.json.
+ksort($baseData);
+$json = json_encode($baseData, JSON_PRETTY_PRINT | JSON_UNESCAPED_UNICODE);
+file_put_contents('base.json', $json);
+
+// Gather available locales.
+$locales = array();
+if ($handle = opendir('../json-full/main')) {
+    while (false !== ($entry = readdir($handle))) {
+        if (substr($entry, 0, 1) != '.') {
+            $entryParts = explode('-', $entry);
+            if (!in_array($entry, $ignoredLocales) && !in_array($entryParts[0], $ignoredLocales)) {
+                $locales[] = $entry;
+            }
+        }
+    }
+    closedir($handle);
+}
+
+// Create the localizations.
+$countries = array();
+foreach ($locales as $locale) {
+    $data = json_decode(file_get_contents('../json-full/main/' . $locale . '/territories.json'), true);
+    $data = $data['main'][$locale]['localeDisplayNames']['territories'];
+    foreach ($data as $countryCode => $countryName) {
+        if (isset($baseData[$countryCode])) {
+            // This country name is untranslated, use the english version.
+            if ($countryCode == $countryName) {
+                $countryName = $countryData[$countryCode];
+            }
+
+            $countries[$locale][$countryCode] = array(
+                'name' => $countryName,
+            );
+        }
+    }
+}
+
+// Identify localizations that are the same as the ones for the parent locale.
+// For example, "fr-FR" if "fr" has the same data.
+$duplicates = array();
+foreach ($countries as $locale => $localizedCountries) {
+    if (strpos($locale, '-') !== FALSE) {
+        $localeParts = explode('-', $locale);
+        array_pop($localeParts);
+        $parentLocale = implode('-', $localeParts);
+        $diff = array_udiff($localizedCountries, $countries[$parentLocale], function ($first, $second) {
+            return ($first['name'] == $second['name']) ? 0 : 1;
+        });
+
+        if (empty($diff)) {
+            // The duplicates are not removed right away because they might
+            // still be needed for other duplicate checks (for example,
+            // when there are locales like bs-Latn-BA, bs-Latn, bs).
+            $duplicates[] = $locale;
+        }
+    }
+}
+// Remove the duplicates.
+foreach ($duplicates as $locale) {
+    unset($countries[$locale]);
+}
+
+// Write out the localizations.
+foreach ($countries as $locale => $localizedCountries) {
+    $collator = collator_create($locale);
+    uasort($localizedCountries, function($a, $b) use ($collator) {
+        return collator_compare($collator, $a['name'], $b['name']);
+    });
+
+    $json = json_encode($localizedCountries, JSON_PRETTY_PRINT | JSON_UNESCAPED_UNICODE);
+    file_put_contents($locale . '.json', $json);
+}
diff --git a/library/intl/scripts/currency/generate.php b/library/intl/scripts/currency/generate.php
new file mode 100644
index 000000000..723906f7a
--- /dev/null
+++ b/library/intl/scripts/currency/generate.php
@@ -0,0 +1,153 @@
+<?php
+
+/**
+ * Generates the json files stored in resources/currency.
+ *
+ * The ISO currency list is used as a base, since it doesn't contain
+ * deprecated currencies, unlike CLDR (v25 has 139 deprecated entries).
+ */
+
+set_time_limit(0);
+
+// Downloaded from http://www.currency-iso.org/en/home/tables/table-a1.html
+$isoCurrencies = '../c2.xml';
+// Downloaded from http://unicode.org/Public/cldr/26/json-full.zip
+$cldrCurrencies = '../json-full/main/en-US/currencies.json';
+$currencyData = '../json-full/supplemental/currencyData.json';
+if (!file_exists($isoCurrencies)) {
+    die("The $isoCurrencies file was not found");
+}
+if (!file_exists($cldrCurrencies)) {
+    die("The $cldrCurrencies file was not found");
+}
+if (!file_exists($currencyData)) {
+    die("The $currencyData file was not found");
+}
+if (!function_exists('collator_create')) {
+    // Reimplementing intl's collator would be a huge undertaking, so we
+    // use it instead to presort the generated locale specific data.
+    die('The intl extension was not found.');
+}
+
+// Locales listed without a "-" match all variants.
+// Locales listed with a "-" match only those exact ones.
+$ignoredLocales = array(
+    // Interlingua is a made up language.
+    'ia',
+    // Valencian differs from its parent only by a single character (è/é).
+    'ca-ES-VALENCIA',
+    // Those locales are 90% untranslated.
+    'aa', 'as', 'az-Cyrl', 'az-Cyrl-AZ', 'bem', 'dua', 'gv', 'haw', 'ig', 'ii',
+    'kkj', 'kok', 'kw', 'lkt', 'mgo', 'nnh', 'nr', 'nso', 'om', 'os', 'pa-Arab',
+    'pa-Arab-PK', 'qu', 'rw', 'sah', 'smn', 'ss', 'ssy', 'st', 'tg', 'tn', 'ts',
+    'uz-Arab', 'uz-Arab-AF', 've', 'vo', 'xh', 'yi',
+    // Special "grouping" locales.
+    'root', 'en-US-POSIX', 'en-001', 'en-150', 'es-419',
+);
+
+// Assemble the base data.
+$baseData = array();
+$currencyData = json_decode(file_get_contents($currencyData), true);
+$currencyData = $currencyData['supplemental']['currencyData']['fractions'];
+$isoData = simplexml_load_file($isoCurrencies);
+foreach ($isoData->CcyTbl->CcyNtry as $currency) {
+    $attributes = (array) $currency->CcyNm->attributes();
+    if (!empty($attributes) && !empty($attributes['@attributes']['IsFund'])) {
+        // Ignore funds.
+        continue;
+    }
+    $currency = (array) $currency;
+    if (empty($currency['Ccy'])) {
+        // Ignore placeholders like "Antarctica".
+        continue;
+    }
+    if (substr($currency['CtryNm'], 0, 2) == 'ZZ' || in_array($currency['Ccy'], array('XUA', 'XSU', 'XDR'))) {
+        // Ignore special currencies.
+        continue;
+    }
+
+    $currencyCode = $currency['Ccy'];
+    $baseData[$currencyCode] = array(
+        'code' => $currencyCode,
+        'numeric_code' => $currency['CcyNbr'],
+    );
+    // Take the fraction digits from CLDR, not ISO, because it reflects real
+    // life usage more closely. If the digits aren't set, that means that the
+    // default value (2) should be used.
+    if (isset($currencyData[$currencyCode]['_digits'])) {
+        $fractionDigits = $currencyData[$currencyCode]['_digits'];
+        if ($fractionDigits != 2) {
+            $baseData[$currencyCode]['fraction_digits'] = $fractionDigits;
+        }
+    }
+}
+
+// Write out base.json.
+ksort($baseData);
+$json = json_encode($baseData, JSON_PRETTY_PRINT | JSON_UNESCAPED_UNICODE);
+file_put_contents('base.json', $json);
+
+// Gather available locales.
+$locales = array();
+if ($handle = opendir('../json-full/main')) {
+    while (false !== ($entry = readdir($handle))) {
+        if (substr($entry, 0, 1) != '.') {
+            $entryParts = explode('-', $entry);
+            if (!in_array($entry, $ignoredLocales) && !in_array($entryParts[0], $ignoredLocales)) {
+                $locales[] = $entry;
+            }
+        }
+    }
+    closedir($handle);
+}
+
+// Create the localizations.
+$currencies = array();
+foreach ($locales as $locale) {
+    $data = json_decode(file_get_contents('../json-full/main/' . $locale . '/currencies.json'), true);
+    $data = $data['main'][$locale]['numbers']['currencies'];
+    foreach ($data as $currencyCode => $currency) {
+        if (isset($baseData[$currencyCode])) {
+            $currencies[$locale][$currencyCode] = array(
+                'name' => $currency['displayName'],
+                'symbol' => $currency['symbol'],
+            );
+        }
+    }
+}
+
+// Identify localizations that are the same as the ones for the parent locale.
+// For example, "fr-FR" if "fr" has the same data.
+$duplicates = array();
+foreach ($currencies as $locale => $localizedCurrencies) {
+    if (strpos($locale, '-') !== FALSE) {
+        $localeParts = explode('-', $locale);
+        array_pop($localeParts);
+        $parentLocale = implode('-', $localeParts);
+        $diff = array_udiff($localizedCurrencies, $currencies[$parentLocale], function ($first, $second) {
+            return ($first['name'] == $second['name']) ? 0 : 1;
+        });
+
+        if (empty($diff)) {
+            // The duplicates are not removed right away because they might
+            // still be needed for other duplicate checks (for example,
+            // when there are locales like bs-Latn-BA, bs-Latn, bs).
+            $duplicates[] = $locale;
+        }
+    }
+}
+// Remove the duplicates.
+foreach ($duplicates as $locale) {
+    unset($currencies[$locale]);
+}
+
+// Write out the localizations.
+foreach ($currencies as $locale => $localizedCurrencies) {
+    $collator = collator_create($locale);
+    uasort($localizedCurrencies, function($a, $b) use ($collator) {
+        return collator_compare($collator, $a['name'], $b['name']);
+    });
+
+    $json = json_encode($localizedCurrencies, JSON_PRETTY_PRINT | JSON_UNESCAPED_UNICODE);
+    file_put_contents($locale . '.json', $json);
+}
diff --git a/library/intl/scripts/language/generate.php b/library/intl/scripts/language/generate.php
new file mode 100644
index 000000000..e11c8df49
--- /dev/null
+++ b/library/intl/scripts/language/generate.php
@@ -0,0 +1,129 @@
+<?php
+
+/**
+ * Generates the json files stored in resources/language.
+ *
+ * CLDR lists about 515 languages, many of them dead (like Latin or Old English).
+ * In order to decrease the list to a reasonable size, only the languages
+ * for which CLDR itself has translations are listed.
+ */
+
+set_time_limit(0);
+
+// Downloaded from http://unicode.org/Public/cldr/26/json-full.zip
+$enLanguages = '../json-full/main/en/languages.json';
+if (!file_exists($enLanguages)) {
+    die("The $enLanguages file was not found");
+}
+if (!function_exists('collator_create')) {
+    // Reimplementing intl's collator would be a huge undertaking, so we
+    // use it instead to presort the generated locale specific data.
+    die('The intl extension was not found.');
+}
+
+// Locales listed without a "-" match all variants.
+// Locales listed with a "-" match only those exact ones.
+$ignoredLocales = array(
+    // Interlingua is a made up language.
+    'ia',
+    // Valencian differs from its parent only by a single character (è/é).
+    'ca-ES-VALENCIA',
+    // Those locales are 90% untranslated.
+    'aa', 'as', 'az-Cyrl', 'az-Cyrl-AZ', 'bem', 'dua', 'gv', 'haw', 'ig', 'ii',
+    'kkj', 'kok', 'kw', 'lkt', 'mgo', 'nnh', 'nr', 'nso', 'om', 'os', 'pa-Arab',
+    'pa-Arab-PK', 'qu', 'rw', 'sah', 'smn', 'ss', 'ssy', 'st', 'tg', 'tn', 'ts',
+    'uz-Arab', 'uz-Arab-AF', 've', 'vo', 'xh', 'yi',
+    // Special "grouping" locales.
+    'root', 'en-US-POSIX', 'en-001', 'en-150', 'es-419',
+);
+
+$languages = array();
+// Load the "en" data first so that it can be used as a fallback for
+// untranslated language names in other locales.
+$languageData = json_decode(file_get_contents($enLanguages), true);
+$languageData = $languageData['main']['en']['localeDisplayNames']['languages'];
+foreach ($languageData as $languageCode => $languageName) {
+    if (strpos($languageCode, '-alt-') === FALSE) {
+        $languages['en'][$languageCode] = array(
+            'code' => $languageCode,
+            'name' => $languageName,
+        );
+    }
+}
+
+// Gather available locales.
+$locales = array();
+if ($handle = opendir('../json-full/main')) {
+    while (false !== ($entry = readdir($handle))) {
+        if (substr($entry, 0, 1) != '.') {
+            $entryParts = explode('-', $entry);
+            if (!in_array($entry, $ignoredLocales) && !in_array($entryParts[0], $ignoredLocales)) {
+                $locales[] = $entry;
+            }
+        }
+    }
+    closedir($handle);
+}
+
+// Remove all languages that aren't an available locale at the same time.
+// This reduces the language list from about 515 to about 185 languages.
+foreach ($languages['en'] as $languageCode => $languageData) {
+    if (!in_array($languageCode, $locales)) {
+        unset($languages['en'][$languageCode]);
+    }
+}
+
+// Load the localizations.
+foreach ($locales as $locale) {
+    $data = json_decode(file_get_contents('../json-full/main/' . $locale . '/languages.json'), true);
+    $data = $data['main'][$locale]['localeDisplayNames']['languages'];
+    foreach ($data as $languageCode => $languageName) {
+        if (isset($languages['en'][$languageCode])) {
+            // This language name is untranslated, use to the english version.
+            if ($languageCode == $languageName) {
+                $languageName = $languages['en'][$languageCode]['name'];
+            }
+
+            $languages[$locale][$languageCode] = array(
+                'code' => $languageCode,
+                'name' => $languageName,
+            );
+        }
+    }
+}
+
+// Identify localizations that are the same as the ones for the parent locale.
+// For example, "fr-FR" if "fr" has the same data.
+$duplicates = array();
+foreach ($languages as $locale => $localizedLanguages) {
+    if (strpos($locale, '-') !== FALSE) {
+        $localeParts = explode('-', $locale);
+        array_pop($localeParts);
+        $parentLocale = implode('-', $localeParts);
+        $diff = array_udiff($localizedLanguages, $languages[$parentLocale], function ($first, $second) {
+            return ($first['name'] == $second['name']) ? 0 : 1;
+        });
+
+        if (empty($diff)) {
+            // The duplicates are not removed right away because they might
+            // still be needed for other duplicate checks (for example,
+            // when there are locales like bs-Latn-BA, bs-Latn, bs).
+            $duplicates[] = $locale;
+        }
+    }
+}
+// Remove the duplicates.
+foreach ($duplicates as $locale) {
+    unset($languages[$locale]);
+}
+
+// Write out the localizations.
+foreach ($languages as $locale => $localizedLanguages) {
+    $collator = collator_create($locale);
+    uasort($localizedLanguages, function($a, $b) use ($collator) {
+        return collator_compare($collator, $a['name'], $b['name']);
+    });
+
+    $json = json_encode($localizedLanguages, JSON_PRETTY_PRINT | JSON_UNESCAPED_UNICODE);
+    file_put_contents($locale . '.json', $json);
+}
diff --git a/library/intl/scripts/number_format/generate.php b/library/intl/scripts/number_format/generate.php
new file mode 100644
index 000000000..4308b4dc7
--- /dev/null
+++ b/library/intl/scripts/number_format/generate.php
@@ -0,0 +1,107 @@
+<?php
+
+/**
+ * Generates the json files stored in resources/number_format.
+ */
+
+set_time_limit(0);
+
+// Downloaded from http://unicode.org/Public/cldr/26/json-full.zip
+if (!is_dir('../json-full/main')) {
+    die("The '../json-full/main' directory was not found");
+}
+
+// Locales listed without a "-" match all variants.
+// Locales listed with a "-" match only those exact ones.
+$ignoredLocales = array(
+    // Interlingua is a made up language.
+    'ia',
+    // Special "grouping" locales.
+    'root', 'en-US-POSIX', 'en-001', 'en-150', 'es-419',
+);
+
+// Gather available locales.
+$locales = array();
+if ($handle = opendir('../json-full/main')) {
+    while (false !== ($entry = readdir($handle))) {
+        if (substr($entry, 0, 1) != '.') {
+            $entryParts = explode('-', $entry);
+            if (!in_array($entry, $ignoredLocales) && !in_array($entryParts[0], $ignoredLocales)) {
+                $locales[] = $entry;
+            }
+        }
+    }
+    closedir($handle);
+}
+
+// Load the data.
+$numberFormats = array();
+foreach ($locales as $locale) {
+    $data = json_decode(file_get_contents('../json-full/main/' . $locale . '/numbers.json'), true);
+    $data = $data['main'][$locale]['numbers'];
+    // Use the default numbering system, if it's supported.
+    if (in_array($data['defaultNumberingSystem'], array('arab', 'arabext', 'beng', 'deva', 'latn'))) {
+        $numberingSystem = $data['defaultNumberingSystem'];
+    } else {
+        $numberingSystem = 'latn';
+    }
+
+    $numberFormats[$locale] = array(
+        'numbering_system' => $numberingSystem,
+        'decimal_pattern' => $data['decimalFormats-numberSystem-' . $numberingSystem]['standard'],
+        'percent_pattern' => $data['percentFormats-numberSystem-' . $numberingSystem]['standard'],
+        'currency_pattern' => $data['currencyFormats-numberSystem-' . $numberingSystem]['standard'],
+        'accounting_currency_pattern' => $data['currencyFormats-numberSystem-' . $numberingSystem]['accounting'],
+    );
+
+    // Add the symbols only if they're different from the default data.
+    $decimalSeparator = $data['symbols-numberSystem-' . $numberingSystem]['decimal'];
+    $groupingSeparator = $data['symbols-numberSystem-' . $numberingSystem]['group'];
+    $plusSign = $data['symbols-numberSystem-' . $numberingSystem]['plusSign'];
+    $minusSign = $data['symbols-numberSystem-' . $numberingSystem]['minusSign'];
+    $percentSign = $data['symbols-numberSystem-' . $numberingSystem]['percentSign'];
+    if ($decimalSeparator != '.') {
+        $numberFormats[$locale]['decimal_separator'] = $decimalSeparator;
+    }
+    if ($groupingSeparator != ',') {
+        $numberFormats[$locale]['grouping_separator'] = $groupingSeparator;
+    }
+    if ($plusSign != '+') {
+        $numberFormats[$locale]['plus_sign'] = $plusSign;
+    }
+    if ($minusSign != '-') {
+        $numberFormats[$locale]['minus_sign'] = $minusSign;
+    }
+    if ($percentSign != '%') {
+        $numberFormats[$locale]['percent_sign'] = $percentSign;
+    }
+}
+
+// Identify localizations that are the same as the ones for the parent locale.
+// For example, "fr-FR" if "fr" has the same data.
+$duplicates = array();
+foreach ($numberFormats as $locale => $formatData) {
+    if (strpos($locale, '-') !== FALSE) {
+        $localeParts = explode('-', $locale);
+        array_pop($localeParts);
+        $parentLocale = implode('-', $localeParts);
+        $diff = array_diff_assoc($formatData, $numberFormats[$parentLocale]);
+
+        if (empty($diff)) {
+            // The duplicates are not removed right away because they might
+            // still be needed for other duplicate checks (for example,
+            // when there are locales like bs-Latn-BA, bs-Latn, bs).
+            $duplicates[] = $locale;
+        }
+    }
+}
+// Remove the duplicates.
+foreach ($duplicates as $locale) {
+    unset($numberFormats[$locale]);
+}
+
+// Write out the data.
+foreach ($numberFormats as $locale => $numberFormat) {
+    $json = json_encode($numberFormat, JSON_PRETTY_PRINT | JSON_UNESCAPED_UNICODE);
+    file_put_contents($locale . '.json', $json);
+}