use Illuminate\Database\Eloquent\Model;
use Illuminate\Support\Arr;
use Illuminate\Support\Str;
-use LanguageDetector\LanguageDetector;
+use LanguageDetection\Language;
class ChatLog extends Model {
return $this->belongsTo(User::class);
}
+ public function getTextWithoutEmotes() {
+ $text = $this->text_content;
+ if (isset($this->tags['emotes']) && !empty($this->tags['emotes'])) {
+ $emotes = explode('/', $this->tags['emotes']);
+ foreach ($emotes as $emote) {
+ $set = explode(':', $emote);
+ $positions = explode(',', $set[1]);
+ foreach ($positions as $position) {
+ $coords = explode('-', $position);
+ for ($i = intval($coords[0]); $i <= intval($coords[1]); ++$i) {
+ $text[$i] = ' ';
+ }
+ }
+ }
+ }
+ return trim(preg_replace('/\s+/', ' ', $text));
+ }
+
public function evaluate() {
$this->evaluateUser();
$this->evaluateChannel();
}
if ($this->command == 'PRIVMSG') {
- if ($this->isKnownBot()) {
+ if (static::isKnownBot($this->nick)) {
$this->type = 'bot';
} else if (substr($this->params[0], 0, 1) == '#') {
$this->type = 'chat';
throw new \Exception('unidentified message');
}
- public function isKnownBot() {
- return in_array(strtolower($this->nick), [
+ public static function isKnownBot($nick) {
+ return in_array(strtolower($nick), [
'birrellthesquirrel',
'funtoon',
'nidbot2000',
'nightbot',
'pokemoncommunitygame',
'speedgaming',
+ 'starbase47',
'streamelements',
'wizebot',
'zockerstuebchen',
if (Str::contains($rawText, ['haha', 'hehe', 'hihi', 'kekw', 'lol', 'lul', 'xd'])) {
return 'lol';
}
- if (Str::startsWith($rawText, ['ahoi', 'hallo', 'hello', 'hi', 'huhu']) || Str::endsWith($rawText, ['hi', 'wave'])) {
+ if (Str::startsWith($rawText, ['ahoi', 'hallo', 'hello', 'hi ', 'huhu']) || Str::endsWith($rawText, ['hi', 'wave'])) {
return 'hi';
}
if (Str::contains($rawText, ['pog', 'wow'])) {
if (Str::contains($rawText, ['hype'])) {
return 'hype';
}
+ if (Str::startsWith($rawText, 'o7') || Str::endsWith($rawText, 'o7') || Str::contains($rawText, 'salut')) {
+ return 'o7';
+ }
return 'unclassified';
}
$languages[] = 'en';
}
}
- $detector = LanguageDetector::detect($this->text_content, $languages);
- $scores = $detector->getScores();
- $lang = strval($detector->getLanguage());
+ $detector = (new Language($languages))->detect($this->getTextWithoutEmotes());
+ $scores = $detector->close();
+ $lang = strval($detector);
//var_dump($scores, $lang, $this->text_content);
- if (is_array($scores) && isset($scores[$lang]) && $scores[$lang] > 0.35) {
+ if (!empty($lang) && $scores[$lang] > 0.4) {
$this->detected_language = $lang;
}
}
- protected function scanForSpam() {
- if (substr($this->text_content, 0, 1) == '!') {
+ public static function spammyText($raw_text) {
+ $text = strtolower($raw_text);
+ if (substr($text, 0, 1) == '!') {
return true;
}
- if (strpos($this->text_content, '$') !== false) {
+ if (strpos($text, '$') !== false) {
return true;
}
- if (strpos($this->text_content, '€') !== false) {
+ if (strpos($text, '€') !== false) {
return true;
}
- if (strpos($this->text_content, '@') !== false) {
+ if (strpos($text, '@') !== false) {
return true;
}
- if (strpos($this->text_content, '://') !== false) {
+ if (strpos($text, '://') !== false) {
return true;
}
- if (is_numeric($this->text_content)) {
+ if (strpos($text, 'followers') !== false) {
+ return true;
+ }
+ if (strpos($text, 'horstie') !== false) {
return true;
}
- if (strpos($this->text_content, 'followers') !== false) {
+ if (strpos($text, 'promotion') !== false) {
return true;
}
- if (strpos($this->text_content, 'promotion') !== false) {
+ if (strpos($text, 'viewers') !== false) {
return true;
}
- if (strpos($this->text_content, 'viewers') !== false) {
+ if (strpos($text, 'view ers') !== false) {
return true;
}
- if (strpos($this->text_content, 'view ers') !== false) {
+ if (strpos($text, 'vielen dank für den raid') !== false) {
+ return true;
+ }
+ if (strpos($text, 'willkommen auf starbase 47') !== false) {
return true;
}
return false;
}
+ protected function scanForSpam() {
+ if (is_numeric($this->text_content)) {
+ return true;
+ }
+ return static::spammyText($this->text_content);
+ }
+
protected $casts = [
'banned' => 'boolean',
'params' => 'array',