switch to (slightly) better language detection model

[alttp.git] / app / Models / ChatLog.php

diff --git a/app/Models/ChatLog.php b/app/Models/ChatLog.php
index 8564ff097b55abbe7eb12e4072568222bb60c944..e17076aee164871121663d9976cdbc796699238f 100644 (file)
--- a/app/Models/ChatLog.php
+++ b/app/Models/ChatLog.php
@@ -6,7 +6,7 @@ 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 {
 
@@ -86,7 +86,7 @@ class ChatLog extends Model {
                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'])) {
@@ -125,16 +125,17 @@ 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->text_content);
+               $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($text) {
+       public static function spammyText($raw_text) {
+               $text = strtolower($raw_text);
                if (substr($text, 0, 1) == '!') {
                        return true;
                }
@@ -153,6 +154,9 @@ class ChatLog extends Model {
                if (strpos($text, 'followers') !== false) {
                        return true;
                }
+               if (strpos($text, 'horstie') !== false) {
+                       return true;
+               }
                if (strpos($text, 'promotion') !== false) {
                        return true;
                }
@@ -162,11 +166,17 @@ class ChatLog extends Model {
                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($text)) {
+               if (is_numeric($this->text_content)) {
                        return true;
                }
                return static::spammyText($this->text_content);