| 1 |
<?php |
| 2 |
|
| 3 |
namespace Gettext\Languages; |
| 4 |
|
| 5 |
use Exception; |
| 6 |
|
| 7 |
/** |
| 8 |
* Holds the CLDR data. |
| 9 |
*/ |
| 10 |
class CldrData |
| 11 |
{ |
| 12 |
/** |
| 13 |
* Super-special plural category: this should always be present for any language. |
| 14 |
* |
| 15 |
* @var string |
| 16 |
*/ |
| 17 |
const OTHER_CATEGORY = 'other'; |
| 18 |
|
| 19 |
/** |
| 20 |
* The list of the plural categories, sorted from 'zero' to 'other'. |
| 21 |
* |
| 22 |
* @var string[] |
| 23 |
*/ |
| 24 |
public static $categories = array('zero', 'one', 'two', 'few', 'many', self::OTHER_CATEGORY); |
| 25 |
|
| 26 |
/** |
| 27 |
* The loaded CLDR data. |
| 28 |
* |
| 29 |
* @var array |
| 30 |
*/ |
| 31 |
private static $data; |
| 32 |
|
| 33 |
/** |
| 34 |
* @var array |
| 35 |
*/ |
| 36 |
private static $plurals; |
| 37 |
|
| 38 |
/** |
| 39 |
* Returns a dictionary containing the language names. |
| 40 |
* The keys are the language identifiers. |
| 41 |
* The values are the language names in US English. |
| 42 |
* |
| 43 |
* @return string[] |
| 44 |
*/ |
| 45 |
public static function getLanguageNames() |
| 46 |
{ |
| 47 |
return self::getData('languages'); |
| 48 |
} |
| 49 |
|
| 50 |
/** |
| 51 |
* Return a dictionary containing the territory names (in US English). |
| 52 |
* The keys are the territory identifiers. |
| 53 |
* The values are the territory names in US English. |
| 54 |
* |
| 55 |
* @return string[] |
| 56 |
*/ |
| 57 |
public static function getTerritoryNames() |
| 58 |
{ |
| 59 |
return self::getData('territories'); |
| 60 |
} |
| 61 |
|
| 62 |
/** |
| 63 |
* Return a dictionary containing the script names (in US English). |
| 64 |
* The keys are the script identifiers. |
| 65 |
* The values are the script names in US English. |
| 66 |
* |
| 67 |
* @param bool $standAlone set to true to retrieve the stand-alone script names, false otherwise |
| 68 |
* |
| 69 |
* @return string[] |
| 70 |
*/ |
| 71 |
public static function getScriptNames($standAlone) |
| 72 |
{ |
| 73 |
return self::getData($standAlone ? 'standAloneScripts' : 'scripts'); |
| 74 |
} |
| 75 |
|
| 76 |
/** |
| 77 |
* A dictionary containing the plural rules. |
| 78 |
* The keys are the language identifiers. |
| 79 |
* The values are arrays whose keys are the CLDR category names and the values are the CLDR category definition. |
| 80 |
* |
| 81 |
* @example The English key-value pair is somethink like this: |
| 82 |
* <code><pre> |
| 83 |
* "en": { |
| 84 |
* "pluralRule-count-one": "i = 1 and v = 0 @integer 1", |
| 85 |
* "pluralRule-count-other": " @integer 0, 2~16, 100, 1000, 10000, 100000, 1000000, … @decimal 0.0~1.5, 10.0, 100.0, 1000.0, 10000.0, 100000.0, 1000000.0, …" |
| 86 |
* } |
| 87 |
* </pre></code> |
| 88 |
* |
| 89 |
* @return array |
| 90 |
*/ |
| 91 |
public static function getPlurals() |
| 92 |
{ |
| 93 |
return self::getData('plurals'); |
| 94 |
} |
| 95 |
|
| 96 |
/** |
| 97 |
* Return a list of superseded language codes. |
| 98 |
* |
| 99 |
* @return array keys are the former language codes, values are the new language/locale codes |
| 100 |
*/ |
| 101 |
public static function getSupersededLanguages() |
| 102 |
{ |
| 103 |
return self::getData('supersededLanguages'); |
| 104 |
} |
| 105 |
|
| 106 |
/** |
| 107 |
* Retrieve the name of a language, as well as if a language code is deprecated in favor of another language code. |
| 108 |
* |
| 109 |
* @param string $id the language identifier |
| 110 |
* |
| 111 |
* @return array|null Returns an array with the keys 'id' (normalized), 'name', 'supersededBy' (optional), 'territory' (optional), 'script' (optional), 'baseLanguage' (optional), 'categories'. If $id is not valid returns null. |
| 112 |
*/ |
| 113 |
public static function getLanguageInfo($id) |
| 114 |
{ |
| 115 |
$result = null; |
| 116 |
$matches = array(); |
| 117 |
if (preg_match('/^([a-z]{2,3})(?:[_\-]([a-z]{4}))?(?:[_\-]([a-z]{2}|[0-9]{3}))?(?:$|-)/i', $id, $matches)) { |
| 118 |
$languageId = strtolower($matches[1]); |
| 119 |
$scriptId = (isset($matches[2]) && ($matches[2] !== '')) ? ucfirst(strtolower($matches[2])) : null; |
| 120 |
$territoryId = (isset($matches[3]) && ($matches[3] !== '')) ? strtoupper($matches[3]) : null; |
| 121 |
$normalizedId = $languageId; |
| 122 |
if (isset($scriptId)) { |
| 123 |
$normalizedId .= '_' . $scriptId; |
| 124 |
} |
| 125 |
if (isset($territoryId)) { |
| 126 |
$normalizedId .= '_' . $territoryId; |
| 127 |
} |
| 128 |
// Structure precedence: see Likely Subtags - http://www.unicode.org/reports/tr35/tr35-31/tr35.html#Likely_Subtags |
| 129 |
$variants = array(); |
| 130 |
$variantsWithScript = array(); |
| 131 |
$variantsWithTerritory = array(); |
| 132 |
if (isset($scriptId) && isset($territoryId)) { |
| 133 |
$variantsWithTerritory[] = $variantsWithScript[] = $variants[] = "{$languageId}_{$scriptId}_{$territoryId}"; |
| 134 |
} |
| 135 |
if (isset($scriptId)) { |
| 136 |
$variantsWithScript[] = $variants[] = "{$languageId}_{$scriptId}"; |
| 137 |
} |
| 138 |
if (isset($territoryId)) { |
| 139 |
$variantsWithTerritory[] = $variants[] = "{$languageId}_{$territoryId}"; |
| 140 |
} |
| 141 |
$variants[] = $languageId; |
| 142 |
$allGood = true; |
| 143 |
$scriptName = null; |
| 144 |
$scriptStandAloneName = null; |
| 145 |
if (isset($scriptId)) { |
| 146 |
$scriptNames = self::getScriptNames(false); |
| 147 |
if (isset($scriptNames[$scriptId])) { |
| 148 |
$scriptName = $scriptNames[$scriptId]; |
| 149 |
$scriptStandAloneNames = self::getScriptNames(true); |
| 150 |
$scriptStandAloneName = $scriptStandAloneNames[$scriptId]; |
| 151 |
} else { |
| 152 |
$allGood = false; |
| 153 |
} |
| 154 |
} |
| 155 |
$territoryName = null; |
| 156 |
if (isset($territoryId)) { |
| 157 |
$territoryNames = self::getTerritoryNames(); |
| 158 |
if (isset($territoryNames[$territoryId])) { |
| 159 |
if ($territoryId !== '001') { |
| 160 |
$territoryName = $territoryNames[$territoryId]; |
| 161 |
} |
| 162 |
} else { |
| 163 |
$allGood = false; |
| 164 |
} |
| 165 |
} |
| 166 |
$languageName = null; |
| 167 |
$languageNames = self::getLanguageNames(); |
| 168 |
foreach ($variants as $variant) { |
| 169 |
if (isset($languageNames[$variant])) { |
| 170 |
$languageName = $languageNames[$variant]; |
| 171 |
if (isset($scriptName) && (!in_array($variant, $variantsWithScript))) { |
| 172 |
$languageName = $scriptName . ' ' . $languageName; |
| 173 |
} |
| 174 |
if (isset($territoryName) && (!in_array($variant, $variantsWithTerritory))) { |
| 175 |
$languageName .= ' (' . $territoryNames[$territoryId] . ')'; |
| 176 |
} |
| 177 |
break; |
| 178 |
} |
| 179 |
} |
| 180 |
if (!isset($languageName)) { |
| 181 |
$allGood = false; |
| 182 |
} |
| 183 |
$baseLanguage = null; |
| 184 |
if (isset($scriptId) || isset($territoryId)) { |
| 185 |
if (isset($languageNames[$languageId]) && ($languageNames[$languageId] !== $languageName)) { |
| 186 |
$baseLanguage = $languageNames[$languageId]; |
| 187 |
} |
| 188 |
} |
| 189 |
$plural = null; |
| 190 |
$plurals = self::getPlurals(); |
| 191 |
foreach ($variants as $variant) { |
| 192 |
if (isset($plurals[$variant])) { |
| 193 |
$plural = $plurals[$variant]; |
| 194 |
break; |
| 195 |
} |
| 196 |
} |
| 197 |
if (!isset($plural)) { |
| 198 |
$allGood = false; |
| 199 |
} |
| 200 |
$supersededBy = null; |
| 201 |
$supersededBys = self::getSupersededLanguages(); |
| 202 |
foreach ($variants as $variant) { |
| 203 |
if (isset($supersededBys[$variant])) { |
| 204 |
$supersededBy = $supersededBys[$variant]; |
| 205 |
break; |
| 206 |
} |
| 207 |
} |
| 208 |
if ($allGood) { |
| 209 |
$result = array(); |
| 210 |
$result['id'] = $normalizedId; |
| 211 |
$result['name'] = $languageName; |
| 212 |
if (isset($supersededBy)) { |
| 213 |
$result['supersededBy'] = $supersededBy; |
| 214 |
} |
| 215 |
if (isset($scriptStandAloneName)) { |
| 216 |
$result['script'] = $scriptStandAloneName; |
| 217 |
} |
| 218 |
if (isset($territoryName)) { |
| 219 |
$result['territory'] = $territoryName; |
| 220 |
} |
| 221 |
if (isset($baseLanguage)) { |
| 222 |
$result['baseLanguage'] = $baseLanguage; |
| 223 |
} |
| 224 |
$result['categories'] = $plural; |
| 225 |
} |
| 226 |
} |
| 227 |
|
| 228 |
return $result; |
| 229 |
} |
| 230 |
|
| 231 |
/** |
| 232 |
* Returns the loaded CLDR data. |
| 233 |
* |
| 234 |
* @param string $key Can be 'languages', 'territories', 'plurals', 'supersededLanguages', 'scripts', 'standAloneScripts' |
| 235 |
* |
| 236 |
* @return array |
| 237 |
*/ |
| 238 |
private static function getData($key) |
| 239 |
{ |
| 240 |
if (!isset(self::$data)) { |
| 241 |
$fixKeys = function ($list, &$standAlone = null) { |
| 242 |
$result = array(); |
| 243 |
$standAlone = array(); |
| 244 |
$match = null; |
| 245 |
foreach ($list as $key => $value) { |
| 246 |
$variant = ''; |
| 247 |
if (preg_match('/^(.+)-alt-(short|variant|stand-alone|long|menu)$/', $key, $match)) { |
| 248 |
$key = $match[1]; |
| 249 |
$variant = $match[2]; |
| 250 |
} |
| 251 |
$key = str_replace('-', '_', $key); |
| 252 |
switch ($key) { |
| 253 |
case 'root': // Language: Root |
| 254 |
case 'und': // Language: Unknown Language |
| 255 |
case 'zxx': // Language: No linguistic content |
| 256 |
case 'ZZ': // Territory: Unknown Region |
| 257 |
case 'Zinh': // Script: Inherited |
| 258 |
case 'Zmth': // Script: Mathematical Notation |
| 259 |
case 'Zsym': // Script: Symbols |
| 260 |
case 'Zxxx': // Script: Unwritten |
| 261 |
case 'Zyyy': // Script: Common |
| 262 |
case 'Zzzz': // Script: Unknown Script |
| 263 |
break; |
| 264 |
default: |
| 265 |
switch ($variant) { |
| 266 |
case 'stand-alone': |
| 267 |
$standAlone[$key] = $value; |
| 268 |
break; |
| 269 |
case '': |
| 270 |
$result[$key] = $value; |
| 271 |
break; |
| 272 |
} |
| 273 |
break; |
| 274 |
} |
| 275 |
} |
| 276 |
|
| 277 |
return $result; |
| 278 |
}; |
| 279 |
$data = array(); |
| 280 |
$json = json_decode(file_get_contents(__DIR__ . '/cldr-data/main/en-US/languages.json'), true); |
| 281 |
$data['languages'] = $fixKeys($json['main']['en-US']['localeDisplayNames']['languages']); |
| 282 |
$json = json_decode(file_get_contents(__DIR__ . '/cldr-data/main/en-US/territories.json'), true); |
| 283 |
$data['territories'] = $fixKeys($json['main']['en-US']['localeDisplayNames']['territories']); |
| 284 |
$json = json_decode(file_get_contents(__DIR__ . '/cldr-data/supplemental/plurals.json'), true); |
| 285 |
$data['plurals'] = $fixKeys($json['supplemental']['plurals-type-cardinal']); |
| 286 |
$json = json_decode(file_get_contents(__DIR__ . '/cldr-data/main/en-US/scripts.json'), true); |
| 287 |
$data['scripts'] = $fixKeys($json['main']['en-US']['localeDisplayNames']['scripts'], $data['standAloneScripts']); |
| 288 |
$data['standAloneScripts'] = array_merge($data['scripts'], $data['standAloneScripts']); |
| 289 |
$data['scripts'] = array_merge($data['standAloneScripts'], $data['scripts']); |
| 290 |
$data['supersededLanguages'] = array(); |
| 291 |
// Remove the languages for which we don't have plurals |
| 292 |
$m = null; |
| 293 |
foreach (array_keys(array_diff_key($data['languages'], $data['plurals'])) as $missingPlural) { |
| 294 |
if (preg_match('/^([a-z]{2,3})_/', $missingPlural, $m)) { |
| 295 |
if (!isset($data['plurals'][$m[1]])) { |
| 296 |
unset($data['languages'][$missingPlural]); |
| 297 |
} |
| 298 |
} else { |
| 299 |
unset($data['languages'][$missingPlural]); |
| 300 |
} |
| 301 |
} |
| 302 |
// Fix the languages for which we have plurals |
| 303 |
$formerCodes = array( |
| 304 |
'jw' => 'jv', // former Javanese |
| 305 |
'mo' => 'ro_MD', // former Moldavian |
| 306 |
); |
| 307 |
$knownMissingLanguages = array( |
| 308 |
'guw' => 'Gun', |
| 309 |
'hnj' => 'Hmong Njua', |
| 310 |
'lld' => 'Dolomitic Ladin', |
| 311 |
'nah' => 'Nahuatl', |
| 312 |
'smi' => 'Sami', |
| 313 |
); |
| 314 |
foreach (array_keys(array_diff_key($data['plurals'], $data['languages'])) as $missingLanguage) { |
| 315 |
if (isset($formerCodes[$missingLanguage]) && isset($data['languages'][$formerCodes[$missingLanguage]])) { |
| 316 |
$data['languages'][$missingLanguage] = $data['languages'][$formerCodes[$missingLanguage]]; |
| 317 |
$data['supersededLanguages'][$missingLanguage] = $formerCodes[$missingLanguage]; |
| 318 |
} else { |
| 319 |
if (isset($knownMissingLanguages[$missingLanguage])) { |
| 320 |
$data['languages'][$missingLanguage] = $knownMissingLanguages[$missingLanguage]; |
| 321 |
} else { |
| 322 |
throw new Exception("We have the plural rule for the language '{$missingLanguage}' but we don't have its language name"); |
| 323 |
} |
| 324 |
} |
| 325 |
} |
| 326 |
ksort($data['languages'], SORT_STRING); |
| 327 |
ksort($data['territories'], SORT_STRING); |
| 328 |
ksort($data['plurals'], SORT_STRING); |
| 329 |
ksort($data['scripts'], SORT_STRING); |
| 330 |
ksort($data['standAloneScripts'], SORT_STRING); |
| 331 |
ksort($data['supersededLanguages'], SORT_STRING); |
| 332 |
self::$data = $data; |
| 333 |
} |
| 334 |
if (!isset(self::$data[$key])) { |
| 335 |
throw new Exception("Invalid CLDR data key: '{$key}'"); |
| 336 |
} |
| 337 |
|
| 338 |
return self::$data[$key]; |
| 339 |
} |
| 340 |
} |
| 341 |
|
