| 1 |
<?php |
| 2 |
|
| 3 |
/* |
| 4 |
* This file is part of the Symfony package. |
| 5 |
* |
| 6 |
* (c) Fabien Potencier <fabien@symfony.com> |
| 7 |
* |
| 8 |
* For the full copyright and license information, please view the LICENSE |
| 9 |
* file that was distributed with this source code. |
| 10 |
*/ |
| 11 |
|
| 12 |
namespace Symfony\Component\String\Slugger; |
| 13 |
|
| 14 |
use Symfony\Component\String\AbstractUnicodeString; |
| 15 |
use Symfony\Component\String\UnicodeString; |
| 16 |
use Symfony\Contracts\Translation\LocaleAwareInterface; |
| 17 |
|
| 18 |
if (!interface_exists(LocaleAwareInterface::class)) { |
| 19 |
throw new \LogicException('You cannot use the "Symfony\Component\String\Slugger\AsciiSlugger" as the "symfony/translation-contracts" package is not installed. Try running "composer require symfony/translation-contracts".'); |
| 20 |
} |
| 21 |
|
| 22 |
/** |
| 23 |
* @author Titouan Galopin <galopintitouan@gmail.com> |
| 24 |
*/ |
| 25 |
class AsciiSlugger implements SluggerInterface, LocaleAwareInterface |
| 26 |
{ |
| 27 |
private const LOCALE_TO_TRANSLITERATOR_ID = [ |
| 28 |
'am' => 'Amharic-Latin', |
| 29 |
'ar' => 'Arabic-Latin', |
| 30 |
'az' => 'Azerbaijani-Latin', |
| 31 |
'be' => 'Belarusian-Latin', |
| 32 |
'bg' => 'Bulgarian-Latin', |
| 33 |
'bn' => 'Bengali-Latin', |
| 34 |
'de' => 'de-ASCII', |
| 35 |
'el' => 'Greek-Latin', |
| 36 |
'fa' => 'Persian-Latin', |
| 37 |
'he' => 'Hebrew-Latin', |
| 38 |
'hy' => 'Armenian-Latin', |
| 39 |
'ka' => 'Georgian-Latin', |
| 40 |
'kk' => 'Kazakh-Latin', |
| 41 |
'ky' => 'Kirghiz-Latin', |
| 42 |
'ko' => 'Korean-Latin', |
| 43 |
'mk' => 'Macedonian-Latin', |
| 44 |
'mn' => 'Mongolian-Latin', |
| 45 |
'or' => 'Oriya-Latin', |
| 46 |
'ps' => 'Pashto-Latin', |
| 47 |
'ru' => 'Russian-Latin', |
| 48 |
'sr' => 'Serbian-Latin', |
| 49 |
'sr_Cyrl' => 'Serbian-Latin', |
| 50 |
'th' => 'Thai-Latin', |
| 51 |
'tk' => 'Turkmen-Latin', |
| 52 |
'uk' => 'Ukrainian-Latin', |
| 53 |
'uz' => 'Uzbek-Latin', |
| 54 |
'zh' => 'Han-Latin', |
| 55 |
]; |
| 56 |
|
| 57 |
private $defaultLocale; |
| 58 |
private $symbolsMap = [ |
| 59 |
'en' => ['@' => 'at', '&' => 'and'], |
| 60 |
]; |
| 61 |
|
| 62 |
/** |
| 63 |
* Cache of transliterators per locale. |
| 64 |
* |
| 65 |
* @var \Transliterator[] |
| 66 |
*/ |
| 67 |
private $transliterators = []; |
| 68 |
|
| 69 |
/** |
| 70 |
* @param array|\Closure|null $symbolsMap |
| 71 |
*/ |
| 72 |
public function __construct(string $defaultLocale = null, $symbolsMap = null) |
| 73 |
{ |
| 74 |
if (null !== $symbolsMap && !\is_array($symbolsMap) && !$symbolsMap instanceof \Closure) { |
| 75 |
throw new \TypeError(sprintf('Argument 2 passed to "%s()" must be array, Closure or null, "%s" given.', __METHOD__, \gettype($symbolsMap))); |
| 76 |
} |
| 77 |
|
| 78 |
$this->defaultLocale = $defaultLocale; |
| 79 |
$this->symbolsMap = $symbolsMap ?? $this->symbolsMap; |
| 80 |
} |
| 81 |
|
| 82 |
/** |
| 83 |
* {@inheritdoc} |
| 84 |
*/ |
| 85 |
public function setLocale($locale) |
| 86 |
{ |
| 87 |
$this->defaultLocale = $locale; |
| 88 |
} |
| 89 |
|
| 90 |
/** |
| 91 |
* {@inheritdoc} |
| 92 |
*/ |
| 93 |
public function getLocale() |
| 94 |
{ |
| 95 |
return $this->defaultLocale; |
| 96 |
} |
| 97 |
|
| 98 |
/** |
| 99 |
* {@inheritdoc} |
| 100 |
*/ |
| 101 |
public function slug(string $string, string $separator = '-', string $locale = null): AbstractUnicodeString |
| 102 |
{ |
| 103 |
$locale = $locale ?? $this->defaultLocale; |
| 104 |
|
| 105 |
$transliterator = []; |
| 106 |
if ($locale && ('de' === $locale || 0 === strpos($locale, 'de_'))) { |
| 107 |
// Use the shortcut for German in UnicodeString::ascii() if possible (faster and no requirement on intl) |
| 108 |
$transliterator = ['de-ASCII']; |
| 109 |
} elseif (\function_exists('transliterator_transliterate') && $locale) { |
| 110 |
$transliterator = (array) $this->createTransliterator($locale); |
| 111 |
} |
| 112 |
|
| 113 |
if ($this->symbolsMap instanceof \Closure) { |
| 114 |
// If the symbols map is passed as a closure, there is no need to fallback to the parent locale |
| 115 |
// as the closure can just provide substitutions for all locales of interest. |
| 116 |
$symbolsMap = $this->symbolsMap; |
| 117 |
array_unshift($transliterator, static function ($s) use ($symbolsMap, $locale) { |
| 118 |
return $symbolsMap($s, $locale); |
| 119 |
}); |
| 120 |
} |
| 121 |
|
| 122 |
$unicodeString = (new UnicodeString($string))->ascii($transliterator); |
| 123 |
|
| 124 |
if (\is_array($this->symbolsMap)) { |
| 125 |
$map = null; |
| 126 |
if (isset($this->symbolsMap[$locale])) { |
| 127 |
$map = $this->symbolsMap[$locale]; |
| 128 |
} else { |
| 129 |
$parent = self::getParentLocale($locale); |
| 130 |
if ($parent && isset($this->symbolsMap[$parent])) { |
| 131 |
$map = $this->symbolsMap[$parent]; |
| 132 |
} |
| 133 |
} |
| 134 |
if ($map) { |
| 135 |
foreach ($map as $char => $replace) { |
| 136 |
$unicodeString = $unicodeString->replace($char, ' '.$replace.' '); |
| 137 |
} |
| 138 |
} |
| 139 |
} |
| 140 |
|
| 141 |
return $unicodeString |
| 142 |
->replaceMatches('/[^A-Za-z0-9]++/', $separator) |
| 143 |
->trim($separator) |
| 144 |
; |
| 145 |
} |
| 146 |
|
| 147 |
private function createTransliterator(string $locale): ?\Transliterator |
| 148 |
{ |
| 149 |
if (\array_key_exists($locale, $this->transliterators)) { |
| 150 |
return $this->transliterators[$locale]; |
| 151 |
} |
| 152 |
|
| 153 |
// Exact locale supported, cache and return |
| 154 |
if ($id = self::LOCALE_TO_TRANSLITERATOR_ID[$locale] ?? null) { |
| 155 |
return $this->transliterators[$locale] = \Transliterator::create($id.'/BGN') ?? \Transliterator::create($id); |
| 156 |
} |
| 157 |
|
| 158 |
// Locale not supported and no parent, fallback to any-latin |
| 159 |
if (!$parent = self::getParentLocale($locale)) { |
| 160 |
return $this->transliterators[$locale] = null; |
| 161 |
} |
| 162 |
|
| 163 |
// Try to use the parent locale (ie. try "de" for "de_AT") and cache both locales |
| 164 |
if ($id = self::LOCALE_TO_TRANSLITERATOR_ID[$parent] ?? null) { |
| 165 |
$transliterator = \Transliterator::create($id.'/BGN') ?? \Transliterator::create($id); |
| 166 |
} |
| 167 |
|
| 168 |
return $this->transliterators[$locale] = $this->transliterators[$parent] = $transliterator ?? null; |
| 169 |
} |
| 170 |
|
| 171 |
private static function getParentLocale(?string $locale): ?string |
| 172 |
{ |
| 173 |
if (!$locale) { |
| 174 |
return null; |
| 175 |
} |
| 176 |
if (false === $str = strrchr($locale, '_')) { |
| 177 |
// no parent locale |
| 178 |
return null; |
| 179 |
} |
| 180 |
|
| 181 |
return substr($locale, 0, -\strlen($str)); |
| 182 |
} |
| 183 |
} |
| 184 |
|