fix IRC message parser

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

diff --git a/app/Models/ChatLog.php b/app/Models/ChatLog.php
index 792e9e7a9b59c3aa212edf5b59786c7dfec6b6ae..3b4df15f2ac9596846729b91fc564c5780ca68b8 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 {
 
@@ -20,6 +20,24 @@ 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();
@@ -125,11 +143,11 @@ 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;
                }
        }