Merge pull request #11884 from annando/language

Improved language detection
This commit is contained in:
Hypolite Petovan 2022-09-04 15:52:29 -04:00 committed by GitHub
commit 8f283985f0
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
2 changed files with 47 additions and 18 deletions

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;

View file

@ -1411,7 +1411,7 @@ msgstr ""
msgid "Friend Suggestions"
msgstr ""
#: mod/tagger.php:78 src/Content/Item.php:297 src/Model/Item.php:2785
#: mod/tagger.php:78 src/Content/Item.php:297 src/Model/Item.php:2814
msgid "photo"
msgstr ""
@ -2292,7 +2292,7 @@ msgstr ""
msgid "show more"
msgstr ""
#: src/Content/Item.php:288 src/Model/Item.php:2783
#: src/Content/Item.php:288 src/Model/Item.php:2812
msgid "event"
msgstr ""
@ -2639,8 +2639,8 @@ msgid ""
"<a href=\"%1$s\" target=\"_blank\" rel=\"noopener noreferrer\">%2$s</a> %3$s"
msgstr ""
#: src/Content/Text/BBCode.php:1213 src/Model/Item.php:3359
#: src/Model/Item.php:3365 src/Model/Item.php:3366
#: src/Content/Text/BBCode.php:1213 src/Model/Item.php:3388
#: src/Model/Item.php:3394 src/Model/Item.php:3395
msgid "Link to source"
msgstr ""
@ -3782,66 +3782,66 @@ msgstr ""
msgid "Edit groups"
msgstr ""
#: src/Model/Item.php:1895
#: src/Model/Item.php:1924
#, php-format
msgid "Detected languages in this post:\\n%s"
msgstr ""
#: src/Model/Item.php:2787
#: src/Model/Item.php:2816
msgid "activity"
msgstr ""
#: src/Model/Item.php:2789
#: src/Model/Item.php:2818
msgid "comment"
msgstr ""
#: src/Model/Item.php:2792
#: src/Model/Item.php:2821
msgid "post"
msgstr ""
#: src/Model/Item.php:2908
#: src/Model/Item.php:2937
#, php-format
msgid "Content warning: %s"
msgstr ""
#: src/Model/Item.php:3271
#: src/Model/Item.php:3300
msgid "bytes"
msgstr ""
#: src/Model/Item.php:3302
#: src/Model/Item.php:3331
#, php-format
msgid "%2$s (%3$d%%, %1$d vote)"
msgid_plural "%2$s (%3$d%%, %1$d votes)"
msgstr[0] ""
msgstr[1] ""
#: src/Model/Item.php:3304
#: src/Model/Item.php:3333
#, php-format
msgid "%2$s (%1$d vote)"
msgid_plural "%2$s (%1$d votes)"
msgstr[0] ""
msgstr[1] ""
#: src/Model/Item.php:3309
#: src/Model/Item.php:3338
#, php-format
msgid "%d voter. Poll end: %s"
msgid_plural "%d voters. Poll end: %s"
msgstr[0] ""
msgstr[1] ""
#: src/Model/Item.php:3311
#: src/Model/Item.php:3340
#, php-format
msgid "%d voter."
msgid_plural "%d voters."
msgstr[0] ""
msgstr[1] ""
#: src/Model/Item.php:3313
#: src/Model/Item.php:3342
#, php-format
msgid "Poll end: %s"
msgstr ""
#: src/Model/Item.php:3347 src/Model/Item.php:3348
#: src/Model/Item.php:3376 src/Model/Item.php:3377
msgid "View on separate page"
msgstr ""