| 1 |
<?php |
| 2 |
|
| 3 |
namespace Linguise\Script; |
| 4 |
|
| 5 |
use Linguise\Vendor\Linguise\Script\Core\Response; |
| 6 |
use Linguise\Vendor\Linguise\Script\Core\Request; |
| 7 |
use Linguise\WordPress\AttributeHandler; |
| 8 |
use Linguise\WordPress\FragmentHandler; |
| 9 |
|
| 10 |
defined('LINGUISE_SCRIPT_TRANSLATION') || die(''); |
| 11 |
|
| 12 |
/** |
| 13 |
* Configuration |
| 14 |
*/ |
| 15 |
class Configuration |
| 16 |
{ |
| 17 |
/** |
| 18 |
* WordPress CMS |
| 19 |
* |
| 20 |
* @var string |
| 21 |
*/ |
| 22 |
public static $cms = 'wordpress'; |
| 23 |
|
| 24 |
/** |
| 25 |
* List of WordPress strings we found to translate |
| 26 |
* |
| 27 |
* @var array |
| 28 |
*/ |
| 29 |
protected static $WPOriginalStrings = []; |
| 30 |
|
| 31 |
/** |
| 32 |
* Extracted json data from WordPress vars |
| 33 |
* |
| 34 |
* @var array |
| 35 |
*/ |
| 36 |
protected static $WPJsonData = []; |
| 37 |
|
| 38 |
/** |
| 39 |
* List of nested variables we have json_decoded |
| 40 |
* We'll need to json_encode it back before rendering |
| 41 |
* |
| 42 |
* @var array |
| 43 |
*/ |
| 44 |
protected static $WPJsonDecoded = []; |
| 45 |
|
| 46 |
/** |
| 47 |
* List of script and variables to look into |
| 48 |
* We'll add the script to look for depending on the enabled plugins |
| 49 |
* |
| 50 |
* @var array |
| 51 |
* dev can add their own WP vars by hooking into ConfigurationLocal.php and fill this array as their convenience |
| 52 |
*/ |
| 53 |
protected static $WPKnownVars = []; |
| 54 |
|
| 55 |
/** |
| 56 |
* Hook onBeforeTranslation |
| 57 |
* |
| 58 |
* @return void |
| 59 |
*/ |
| 60 |
public static function onBeforeTranslation() |
| 61 |
{ |
| 62 |
$request = Request::getInstance(); |
| 63 |
|
| 64 |
if (strpos($request->getPathname(), 'wp-json/oembed/1.0/embed') !== false) { |
| 65 |
return; |
| 66 |
} |
| 67 |
|
| 68 |
self::excludeAdminBar(); |
| 69 |
|
| 70 |
// Extract WordPress translation from Scripts |
| 71 |
self::wpFragmentsScriptExtraction(); |
| 72 |
} |
| 73 |
|
| 74 |
/** |
| 75 |
* Hook onAfterTranslation |
| 76 |
* |
| 77 |
* @return void |
| 78 |
*/ |
| 79 |
public static function onAfterTranslation() |
| 80 |
{ |
| 81 |
$request = Request::getInstance(); |
| 82 |
|
| 83 |
if (strpos($request->getPathname(), 'wp-json/oembed/1.0/embed') !== false) { |
| 84 |
return; |
| 85 |
} |
| 86 |
|
| 87 |
// Insert back WordPress translation into Scripts |
| 88 |
self::wpFragmentsScriptInsertion(); |
| 89 |
|
| 90 |
/** |
| 91 |
* Hotfix for some of urls with non latin char become invalid (HTML entity–encoded Unicode characters) |
| 92 |
* And causing some of SEO tool complaining, Semrush mark hreflang as broken. |
| 93 |
* TODO: can be removed if core translate the URL properly |
| 94 |
*/ |
| 95 |
// Normalize canonical link href value |
| 96 |
self::normalizeCanonical(); |
| 97 |
} |
| 98 |
|
| 99 |
/** |
| 100 |
* Do not translate the adminbar |
| 101 |
* |
| 102 |
* @return void |
| 103 |
*/ |
| 104 |
protected static function excludeAdminBar() |
| 105 |
{ |
| 106 |
$response = Response::getInstance(); |
| 107 |
$content = $response->getContent(); |
| 108 |
// Use regex to handle cases where other attributes (e.g., data-rocket-location-hash) appear before id="wpadminbar" |
| 109 |
$content = preg_replace('/<div\b[^>]*\bid="wpadminbar"/i', '$0 translate="no"', $content); |
| 110 |
$response->setContent($content); |
| 111 |
} |
| 112 |
|
| 113 |
/** |
| 114 |
* Decoed href value of hreflang |
| 115 |
* |
| 116 |
* @return void |
| 117 |
*/ |
| 118 |
protected static function normalizeCanonical() |
| 119 |
{ |
| 120 |
$response = Response::getInstance(); |
| 121 |
$content = $response->getContent(); |
| 122 |
|
| 123 |
$content = self::normalizeCanonicalLink($content); |
| 124 |
|
| 125 |
$response->setContent($content); |
| 126 |
} |
| 127 |
|
| 128 |
/** |
| 129 |
* Normalize canonical link href values by decoding HTML entities. |
| 130 |
* Converts things like: |
| 131 |
* https://example.com/ar/صفحة/ |
| 132 |
* into: |
| 133 |
* https://example.com/ar/صفحة/ |
| 134 |
* |
| 135 |
* @param string $html The HTML content |
| 136 |
* |
| 137 |
* @return string The modified HTML content |
| 138 |
*/ |
| 139 |
protected static function normalizeCanonicalLink(string $html): string |
| 140 |
{ |
| 141 |
return preg_replace_callback( |
| 142 |
'/(<link\s+rel="canonical"[^>]*?href=")([^"]+)(")/i', |
| 143 |
function ($matches) { |
| 144 |
// Decode HTML entities (&#xNNNN; or &) |
| 145 |
$decoded = html_entity_decode($matches[2], ENT_QUOTES | ENT_HTML5, 'UTF-8'); |
| 146 |
|
| 147 |
// Optional: percent-encode path segments for safety |
| 148 |
$parts = parse_url($decoded); |
| 149 |
if (isset($parts['path'])) { |
| 150 |
$has_trailing_slash = (substr($parts['path'], -1) === '/'); |
| 151 |
|
| 152 |
$segments = array_map('rawurlencode', explode('/', trim($parts['path'], '/'))); |
| 153 |
$joined_path = implode('/', $segments); |
| 154 |
|
| 155 |
if (empty($joined_path)) { |
| 156 |
$parts['path'] = $has_trailing_slash ? '/' : ''; |
| 157 |
} else { |
| 158 |
$parts['path'] = '/' . $joined_path . ($has_trailing_slash ? '/' : ''); |
| 159 |
} |
| 160 |
} |
| 161 |
|
| 162 |
// Rebuild URL |
| 163 |
$rebuilt = |
| 164 |
(isset($parts['scheme']) ? $parts['scheme'] . '://' : '') . |
| 165 |
($parts['host'] ?? '') . |
| 166 |
($parts['path'] ?? '') . |
| 167 |
(isset($parts['query']) ? '?' . $parts['query'] : '') . |
| 168 |
(isset($parts['fragment']) ? '#' . $parts['fragment'] : ''); |
| 169 |
|
| 170 |
return $matches[1] . $rebuilt . $matches[3]; |
| 171 |
}, |
| 172 |
$html |
| 173 |
); |
| 174 |
} |
| 175 |
|
| 176 |
/** |
| 177 |
* Parse the content to extract JavaScript JSON encoded translatable variables |
| 178 |
* |
| 179 |
* @return void |
| 180 |
*/ |
| 181 |
protected static function wpFragmentsScriptExtraction() |
| 182 |
{ |
| 183 |
$response = Response::getInstance(); |
| 184 |
$content = $response->getContent(); |
| 185 |
|
| 186 |
if (self::isXml($content)) { |
| 187 |
return; |
| 188 |
} |
| 189 |
|
| 190 |
$all_fragments = FragmentHandler::findWPFragments($content); |
| 191 |
$attrs_fragments = AttributeHandler::findWPFragments($content); |
| 192 |
$merged_fragments = array_merge($all_fragments, $attrs_fragments); |
| 193 |
|
| 194 |
$newContent = $content; |
| 195 |
foreach ($merged_fragments as $frag_key => $frag_json) { |
| 196 |
foreach ($frag_json as $frag_param => $frag_values) { |
| 197 |
$frag_as_html = FragmentHandler::intoHTMLFragments($frag_key, $frag_param, $frag_values); |
| 198 |
$newContent = str_replace('</body>', $frag_as_html . '</body>', $newContent); |
| 199 |
} |
| 200 |
} |
| 201 |
|
| 202 |
$response->setContent($newContent); |
| 203 |
} |
| 204 |
|
| 205 |
/** |
| 206 |
* Insert back all translated strings |
| 207 |
* |
| 208 |
* @return void |
| 209 |
*/ |
| 210 |
protected static function wpFragmentsScriptInsertion() |
| 211 |
{ |
| 212 |
$response = Response::getInstance(); |
| 213 |
$content = $response->getContent(); |
| 214 |
|
| 215 |
if (self::isXml($content)) { |
| 216 |
return; |
| 217 |
} |
| 218 |
|
| 219 |
$all_fragments = FragmentHandler::intoJSONFragments($content); |
| 220 |
try { |
| 221 |
$translatedHtml = FragmentHandler::injectTranslatedWPFragments($content, $all_fragments); |
| 222 |
} catch (\LogicException $e) { |
| 223 |
$translatedHtml = $content; |
| 224 |
} |
| 225 |
$attrs_fragments = AttributeHandler::intoJSONFragments($translatedHtml); |
| 226 |
try { |
| 227 |
$translatedHtml = AttributeHandler::injectTranslatedWPFragments($translatedHtml, $attrs_fragments); |
| 228 |
} catch (\LogicException $e) { |
| 229 |
$translatedHtml = $content; |
| 230 |
} |
| 231 |
|
| 232 |
$response->setContent($translatedHtml); |
| 233 |
} |
| 234 |
|
| 235 |
/** |
| 236 |
* Remove token after user set new password |
| 237 |
* |
| 238 |
* @return void |
| 239 |
*/ |
| 240 |
public static function onBeforeRedirect() |
| 241 |
{ |
| 242 |
$response = Response::getInstance(); |
| 243 |
$cookies = $response->getCookies(); |
| 244 |
foreach ($cookies as $index => $cookie) { |
| 245 |
if (strpos($cookie, 'wp-resetpass') !== false) { |
| 246 |
/** |
| 247 |
* If set cookie into translated page sometimes not working for Cyrillic alphabet, i.e: Russian |
| 248 |
* Set path into / to work in translated and non translated URL. |
| 249 |
*/ |
| 250 |
$cookies[$index]->setPath('/'); |
| 251 |
} |
| 252 |
} |
| 253 |
} |
| 254 |
|
| 255 |
/** |
| 256 |
* We should translate JSON request to oembed |
| 257 |
* |
| 258 |
* @return boolean |
| 259 |
*/ |
| 260 |
public static function onJsonShouldTranslate() |
| 261 |
{ |
| 262 |
$request = Request::getInstance(); |
| 263 |
|
| 264 |
if (strpos($request->getPathname(), 'wp-json/oembed/1.0/embed') !== false) { |
| 265 |
return true; |
| 266 |
} |
| 267 |
|
| 268 |
return false; |
| 269 |
} |
| 270 |
|
| 271 |
/** |
| 272 |
* Check if content is XML |
| 273 |
* |
| 274 |
* @param string $content The request content |
| 275 |
* |
| 276 |
* @return boolean |
| 277 |
*/ |
| 278 |
public static function isXml($content) |
| 279 |
{ |
| 280 |
$patterns = [ |
| 281 |
'/^[\s]*<\?xml[\s\S]*?>[\s]*<(urlset|sitemapindex)[\s\S]*?>[\s\S]*?<\/(urlset|sitemapindex)>/mi', |
| 282 |
'/^<(urlset|sitemapindex)/', |
| 283 |
'/^[\s]*(<\?xml[^>]*>)?<rss[^>]*xmlns:g="http:\/\/base\.google\.com\/ns(?:\/[^"]*)?"[^>]*>/mi', |
| 284 |
'/<\?xml[^>]*>\s*<rss[^>]*>([\s\S]*?<\/item>\s*<\/channel>\s*<\/rss>)/mi', |
| 285 |
]; |
| 286 |
|
| 287 |
foreach ($patterns as $pattern) { |
| 288 |
if (preg_match($pattern, $content)) { |
| 289 |
return true; |
| 290 |
} |
| 291 |
} |
| 292 |
|
| 293 |
return false; |
| 294 |
} |
| 295 |
} |
| 296 |
|
