X-Git-Url: https://git.localhorst.tv/?a=blobdiff_plain;f=app%2FModels%2FChatLog.php;h=063e8036485cdf5cc31ef793215db38f63a90b4c;hb=7d1ef8619d7513184340f73e99efa20481b86d75;hp=79eab97debff1a56a6ee081efbdf59951f22a5d9;hpb=abdc2ea1ade1fb12ceacc28660890750e69ae36f;p=alttp.git

diff --git a/app/Models/ChatLog.php b/app/Models/ChatLog.php
index 79eab97..063e803 100644
--- a/app/Models/ChatLog.php
+++ b/app/Models/ChatLog.php
@@ -2,11 +2,12 @@
 
 namespace App\Models;
 
+use App\TwitchBot\TokenizedMessage;
 use Illuminate\Database\Eloquent\Factories\HasFactory;
 use Illuminate\Database\Eloquent\Model;
 use Illuminate\Support\Arr;
 use Illuminate\Support\Str;
-use LanguageDetector\LanguageDetector;
+use LanguageDetection\Language;
 
 class ChatLog extends Model {
 
@@ -20,6 +21,28 @@ class ChatLog extends Model {
 		return $this->belongsTo(User::class);
 	}
 
+	public function tokenize() {
+		return TokenizedMessage::fromLog($this);
+	}
+
+	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();
@@ -43,10 +66,11 @@ class ChatLog extends Model {
 			}
 			$this->text_content = $this->params[1];
 			$this->detectLanguage();
-			if ($this->scanForSpam()) {
+			$tokenized = $this->tokenize();
+			if ($tokenized->isSpammy()) {
 				$this->banned = true;
 			}
-			$this->classification = static::classify($this->text_content);
+			$this->classification = $tokenized->classify();
 			return;
 		}
 
@@ -68,45 +92,11 @@ class ChatLog extends Model {
 		]);
 	}
 
-	public static function classify($text) {
-		if (empty($text)) {
-			return 'unclassified';
-		}
-		if (is_numeric(trim($text))) {
-			return 'number';
-		}
-		$rawText = strtolower(preg_replace('/[^\w]/', '', $text));
-		$tokenizedText = preg_split('/\s+/', strtolower(trim($text)));
-		if (Str::startsWith($rawText, 'gg') || Str::endsWith($rawText, 'gg')) {
-			return 'gg';
-		}
-		if (Str::contains($rawText, ['glgl', 'glhf', 'hfgl'])) {
-			return 'gl';
-		}
-		if (Str::contains($rawText, ['haha', 'hehe', 'hihi', 'kekw', 'lol', 'lul', 'xd'])) {
-			return 'lol';
-		}
-		if (Str::startsWith($rawText, ['ahoi', 'hallo', 'hello', 'hi', 'huhu']) || Str::endsWith($rawText, ['hi', 'wave'])) {
-			return 'hi';
-		}
-		if (Str::contains($rawText, ['pog', 'wow'])) {
-			return 'pog';
-		}
-		if (Str::contains($rawText, ['hype'])) {
-			return 'hype';
-		}
-		if (Str::startsWith($rawText, 'o7') || Str::endsWith($rawText, 'o7') || Str::contains($rawText, 'salut')) {
-			return 'o7';
-		}
-		return 'unclassified';
-	}
-
 	protected function evaluateUser() {
 	}
 
 	protected function evaluateChannel() {
 		if (empty($this->params)) {
-			$this->channel()->associate(null);
 			return;
 		}
 		$cname = $this->params[0];
@@ -114,7 +104,12 @@ class ChatLog extends Model {
 			$cname = '#'.$cname;
 		}
 		$channel = Channel::firstWhere('twitch_chat', '=', $cname);
-		$this->channel()->associate($channel);
+		if (!is_null($channel)) {
+			$this->channel()->associate($channel);
+			if (empty($this->twitch_category) && now()->sub(15, 'minute')->isBefore($this->created_at)) {
+				$this->twitch_category = $channel->twitch_category;
+			}
+		}
 	}
 
 	protected function detectLanguage() {
@@ -125,63 +120,15 @@ class ChatLog extends Model {
 				$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;
 		}
 	}
 
-	public static function spammyText($raw_text) {
-		$text = strtolower($raw_text);
-		if (substr($text, 0, 1) == '!') {
-			return true;
-		}
-		if (strpos($text, '$') !== false) {
-			return true;
-		}
-		if (strpos($text, '€') !== false) {
-			return true;
-		}
-		if (strpos($text, '@') !== false) {
-			return true;
-		}
-		if (strpos($text, '://') !== false) {
-			return true;
-		}
-		if (strpos($text, 'followers') !== false) {
-			return true;
-		}
-		if (strpos($text, 'horstie') !== false) {
-			return true;
-		}
-		if (strpos($text, 'promotion') !== false) {
-			return true;
-		}
-		if (strpos($text, 'viewers') !== false) {
-			return true;
-		}
-		if (strpos($text, 'view ers') !== false) {
-			return true;
-		}
-		if (strpos($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',