Script for regenerating Commerce 1.x currencies based on CLDR

generate_currencies.php

<?php

/**
 * Regenerates the Commerce 1.x currency list from CLDR.
 *
 * Assumes the existence of commerce/ and intl/ folders.
 * The intl/ folder should contain the commerceguys/intl
 * library with the CLDR dataset downloaded via
 * scripts/fetch_data.sh
 */

include __DIR__ . '/commerce/includes/commerce.currency.inc';
include __DIR__ .  '/intl/vendor/autoload.php';

$dataDir = 'intl/scripts/assets/';
$commerceCurrencies = commerce_commerce_currency_info();
$isoCurrencies = getIsoCurrencies();
$addedCurrencies = array_keys(array_diff_key($isoCurrencies, $commerceCurrencies));
$removedCurrencies = array_keys(array_diff_key($commerceCurrencies, $isoCurrencies));

$currencyData = json_decode(file_get_contents($dataDir . 'cldr-core/supplemental/currencyData.json'), true);
$currencyData = $currencyData['supplemental']['currencyData'];
$territoryInfo = json_decode(file_get_contents($dataDir . 'cldr-core/supplemental/territoryInfo.json'), true);
$territoryInfo = $territoryInfo['supplemental']['territoryInfo'];

$newCommerceCurrencies = [];
foreach ($isoCurrencies as $currencyCode => $numericCode) {
  // Determine the currency's default locale (currency -> country -> locale).
  // This can't work for EUR because it's used by multiple countries.
  $locale = 'en';
  $fallbackLocale = 'en';
  $languageCode = NULL;
  if ($currencyCode == 'EUR') {
    $locale = 'fr';
  }
  else {
    $countryCode = getCurrencyCountry($currencyCode);
    if ($countryCode) {
      $languageCode = getCountryLanguage($countryCode);
      $locale = $languageCode . '-' . $countryCode;
      $fallbackLocale = $languageCode;
    }
  }

  $englishData = json_decode(file_get_contents($dataDir . 'cldr-numbers-full/main/en/currencies.json'), true);
  $name = $englishData['main']['en']['numbers']['currencies'][$currencyCode]['displayName'];

  $currency = getCurrency($currencyCode, $locale, $fallbackLocale);
  $symbol = $currency['symbol'];

  $numberFormat = getNumberFormat($locale, $fallbackLocale);
  $currencyPattern = $numberFormat['currencyFormats-numberSystem-latn']['standard'];
  // Ignore the negative portion of the currency pattern.
  $currencyPattern = strtok($currencyPattern, ';');
  $numberPosition = strpos($currencyPattern, '0.00');
  $symbolPosition = strpos($currencyPattern, '¤');
  $symbolSpacer = ' ';
  if ($symbolPosition < $numberPosition) {
    $symbolPlacement = 'before';
    $symbolSpacer = mb_substr($currencyPattern, $symbolPosition + 1, 1, 'UTF-8');
    $symbolEdge =  mb_substr($symbol, -1, 1, 'UTF-8');
  }
  else {
    $symbolPlacement = 'after';
    $symbolSpacer = mb_substr($currencyPattern, $symbolPosition - 2, 1, 'UTF-8');
    $symbolEdge =  mb_substr($symbol, 0, 1, 'UTF-8');
  }
  if (in_array($symbolSpacer, ['#', '0', '¤'])) {
    $symbolSpacer = '';
  }
  // CLDR quote: 
  // "By default a space is automatically added between letters in a currency symbol and adjacent numbers."
  if ($symbolSpacer == '' && ctype_alpha($symbolEdge)) {
     $symbolSpacer = ' ';
  }

  $decimals = 2;
  if (isset($currencyData['fractions'][$currencyCode]['_digits'])) {
    $decimals = $currencyData['fractions'][$currencyCode]['_digits'];
  }
  // Override for Hungary, which has digits only in theory.
  if ($currencyCode == 'HUF') {
    $decimals = 0;
  }

  $newCommerceCurrencies[$currencyCode] = [
    'code' => $currencyCode,
    'numeric_code' => $numericCode,
    'symbol' => $symbol,
    'name' => 't(' . $name . ')t',
    'symbol_placement' => $symbolPlacement,
    'symbol_spacer' => $symbolSpacer,
    // The symbol should always be used, not the code.
    'code_placement' => 'hidden',
    // CLDR has no information on minor_unit and major_unit. Use the old data when available.
    'minor_unit' => isset($commerceCurrencies[$currencyCode]['minor_unit']) ? 't(' . $commerceCurrencies[$currencyCode]['minor_unit'] . ')t' : '',
    'major_unit' => isset($commerceCurrencies[$currencyCode]['major_unit']) ? 't(' . $commerceCurrencies[$currencyCode]['major_unit'] . ')t' : '',
    'decimals' =>  (int) $decimals,
    'thousands_separator' => $numberFormat['symbols-numberSystem-latn']['group'],
    'decimal_separator' => $numberFormat['symbols-numberSystem-latn']['decimal'],
  ];
  // Strip default values.
  $defaults = [
    'symbol' => '',
    'minor_unit' => '',
    'decimals' => 2,
    'rounding_step' => 0,
    'thousands_separator' => ',',
    'decimal_separator' => '.',
    'symbol_placement' => 'hidden',
    'symbol_spacer' => ' ',
    'code_placement' => 'after',
    'code_spacer' => ' ',
  ];
  foreach ($newCommerceCurrencies[$currencyCode] as $key => $value) {
    if (isset($defaults[$key]) && $defaults[$key] == $value) {
      unset($newCommerceCurrencies[$currencyCode][$key]);
    }
  }
  // 1.x doesn't support arabic number formatting, so it's best to
  // fallback to the default behavior, just show the currency code
  // after the number.
  if ($languageCode == 'ar') {
    unset($newCommerceCurrencies[$currencyCode]['symbol_placement']);
    unset($newCommerceCurrencies[$currencyCode]['symbol_spacer']);
    unset($newCommerceCurrencies[$currencyCode]['code_placement']);
  }
}
// CLDR doesn't seem to have a pt-BR number format, and the default pt one
// uses no symbol spacer. Confirmed on http://www.telhanorte.com.br/
// and http://www.cec.com.br/ that a spacer is needed, so unset the empty
// string to use the default.
unset($newCommerceCurrencies['BRL']['symbol_spacer']);
// Probably not even used, but we keep it for BC.
$newCommerceCurrencies['CHF']['rounding_step'] = '0.05';
// Fix the symbol for HRK ("kn" instead of "HRK").
$newCommerceCurrencies['HRK']['symbol'] = 'kn';
// @todo Get commerce to stop re-sorting the currencies by code, then sort
// by name here.
ksort($newCommerceCurrencies);
file_put_php('currencies.php', $newCommerceCurrencies);

if ($count = count($addedCurrencies)) {
  echo "Added $count currencies: " . implode(', ', $addedCurrencies) . "\n";
} else {
    echo "No currencies were added. \n";
}

if ($count = count($removedCurrencies)) {
  echo "Removed $count currencies: " . implode(', ', $removedCurrencies) . "\n";
} else {
  echo "No currencies were removed. \n";
}

$updatedCurrencies = [];
foreach ($commerceCurrencies as $currencyCode => $currency) {
  if (isset($newCommerceCurrencies[$currencyCode]) && array_diff($newCommerceCurrencies[$currencyCode], $currency)) {
    $updatedCurrencies[] = $currencyCode;
  }
}
if ($count = count($updatedCurrencies)) {
  echo "Updated $count currencies: " . implode(', ', $updatedCurrencies) . "\n";
} else {
  echo "No currencies were updated. \n";
}

/**
 * t() shim to allow commerce_commerce_currency_info() to work.
 */
function t($string) 
{
  return $string;
}

/**
 * Gets the ISO currencies.
 *
 * ISO only lists active currencies, making it practical to use as a base.
 *
 * @return array
 *   Format: $currencyCode => $numericCode.
 */
function getIsoCurrencies() 
{
  global $dataDir;
  
  $isoCurrencies = [];
  $isoData = simplexml_load_file($dataDir . 'c2.xml');
  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'], ['XUA', 'XSU', 'XDR'])) {
      // Ignore special currencies.
      continue;
    }
    if (in_array($currency['Ccy'], ['SVC', 'ZWL'])) {
      // No longer used, but still included in the ISO list.
      continue;
    }

    $isoCurrencies[$currency['Ccy']] = $currency['CcyNbr'];
  }

  return $isoCurrencies;
}

/**
 * Gets the country for a specific currency.
 */
function getCurrencyCountry($currencyCode) 
{
  global $currencyData;

  foreach ($currencyData['region'] as $countryCode => $data) {
    foreach ($data as $index => $currencies) {
      if (isset($currencies[$currencyCode]) && empty($currencies[$currencyCode]['_to'])) {
        return $countryCode;
      }
    }
  }
}

/**
 * Gets the language for a specific country.
 */
function getCountryLanguage($countryCode) 
{
  global $territoryInfo;

  $languages = $territoryInfo[$countryCode]['languagePopulation'];
  // Sort by population percent date.
  uasort($languages, function ($a, $b) {
    $a = $a['_populationPercent'];
    $b = $b['_populationPercent'];
    if ($a == $b) {
        return 0;
    }
    return ($a > $b) ? -1 : 1;
  });
  $languageCode = key($languages);
  $languageCode = str_replace('_', '-', $languageCode);

  return $languageCode;
}

function getCurrency($currencyCode, $locale, $fallbackLocale) 
{
  global $dataDir;

  $usedLocale = NULL;
  $localeCandidates = [$locale, $fallbackLocale, 'en'];
  foreach ($localeCandidates as $localeCandidate) {
    if (file_exists($dataDir . 'cldr-numbers-full/main/' . $localeCandidate . '/currencies.json')) {
      $usedLocale = $localeCandidate;
      break;
    }
  }
  $data = json_decode(file_get_contents($dataDir . 'cldr-numbers-full/main/' . $usedLocale . '/currencies.json'), true);
  $currency = $data['main'][$usedLocale]['numbers']['currencies'][$currencyCode];

  return $currency;
}

function getNumberFormat($locale, $fallbackLocale) 
{
  global $dataDir;

  $usedLocale = NULL;
  $localeCandidates = [$locale, $fallbackLocale, 'en'];
  foreach ($localeCandidates as $localeCandidate) {
    if (file_exists($dataDir . 'cldr-numbers-full/main/' . $localeCandidate . '/numbers.json')) {
      $usedLocale = $localeCandidate;
      break;
    }
  }
  $data = json_decode(file_get_contents($dataDir . 'cldr-numbers-full/main/' . $usedLocale . '/numbers.json'), true);
  $numberFormat = $data['main'][$usedLocale]['numbers'];

  return $numberFormat;
}

/**
 * Converts the provided data into php and writes it to the disk.
 */
function file_put_php($filename, $data) 
{
    $data = var_export($data, true);
    $data = str_replace(['array (', "=> \n  "], ['array(', '=> '], $data);
    $data = str_replace('=>   array(', '=> array(', $data);
    $data = '<?php' . "\n\n" . '$data = ' . $data . ';';
    // Put back t().
    $data = str_replace(["'t(", ")t',"], ["t('", "'),"], $data);
    file_put_contents($filename, $data);
}