Improved language detection

This commit is contained in:
Michael 2022-09-04 07:32:46 +00:00
parent dd59ad9032
commit 4304aa9eec

View file

@ -919,8 +919,6 @@ class Item
$item['gravity'] = self::getGravity($item);
$item['language'] = self::getLanguage($item);
$default = ['url' => $item['author-link'], 'name' => $item['author-name'],
'photo' => $item['author-avatar'], 'network' => $item['network']];
$item['author-id'] = ($item['author-id'] ?? 0) ?: Contact::getIdForURL($item['author-link'], 0, null, $default);
@ -1108,6 +1106,8 @@ class Item
// Check for hashtags in the body and repair or add hashtag links
$item['body'] = self::setHashtags($item['body']);
$item['language'] = self::getLanguage($item);
$notify_type = Delivery::POST;
// Filling item related side tables
@ -1869,6 +1869,8 @@ class Item
return '';
}
$naked_body = self::getDominantLanguage($naked_body);
$availableLanguages = DI::l10n()->getAvailableLanguages();
// See https://github.com/friendica/friendica/issues/10511
// Persian is manually added to language detection until a persian translation is provided for the interface, at
@ -1884,6 +1886,33 @@ class Item
return '';
}
/**
* Check if latin or non latin are dominant in the body and only return the dominant one
*
* @param string $body
* @return string
*/
private static function getDominantLanguage(string $body): string
{
$latin = '';
$non_latin = '';
for ($i = 0; $i < mb_strlen($body); $i++) {
$character = mb_substr($body, $i, 1);
$ord = mb_ord($character);
// We add the most common characters to both strings.
if (($ord <= 64) || ($ord >= 91 && $ord <= 96) || ($ord >= 123 && $ord <= 191) || in_array($ord, [215, 247]) || ($ord >= 697 && $ord <= 735) || ($ord > 65535)) {
$latin .= $character;
$non_latin .= $character;
} elseif ($ord < 768) {
$latin .= $character;
} else {
$non_latin .= $character;
}
}
return (mb_strlen($latin) > mb_strlen($non_latin)) ? $latin : $non_latin;
}
public static function getLanguageMessage(array $item): string
{
$iso639 = new \Matriphe\ISO639\ISO639;