From 5faca281b1c591caa1402a399c64dc67ffc44cb8 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Phil=20H=C3=B6fer?= <phil@suma-ev.de>
Date: Fri, 14 Jan 2022 12:11:23 +0000
Subject: [PATCH 1/2] Implement porn filter for Scopia

---
 app/Models/parserSkripte/Scopia.php | 80 +++++++++++++++++++++++++----
 1 file changed, 70 insertions(+), 10 deletions(-)

diff --git a/app/Models/parserSkripte/Scopia.php b/app/Models/parserSkripte/Scopia.php
index 8403d07a..6c9960ab 100644
--- a/app/Models/parserSkripte/Scopia.php
+++ b/app/Models/parserSkripte/Scopia.php
@@ -36,16 +36,19 @@ class Scopia extends Searchengine
                 $anzeigeLink = $link;
                 $descr = $result->description->__toString();
                 $this->counter++;
-                $this->results[] = new \App\Models\Result(
-                    $this->engine,
-                    $title,
-                    $link,
-                    $anzeigeLink,
-                    $descr,
-                    $this->engine->{"display-name"},
-                    $this->engine->homepage,
-                    $this->counter
-                );
+                if(! $this->containsPornContent($title.$descr)) { //see note at filtering method
+                    $this->results[] = new \App\Models\Result(
+                        $this->engine,
+                        $title,
+                        $link,
+                        $anzeigeLink,
+                        $descr,
+                        $this->engine->{"display-name"},
+                        $this->engine->homepage,
+                        $this->counter
+                    );
+                }
+                
             }
         } catch (\Exception $e) {
             Log::error("A problem occurred parsing results from $this->name:");
@@ -54,6 +57,63 @@ class Scopia extends Searchengine
         }
     }
 
+    private function containsPornContent($text) {
+        // Returns true if pornographic content is detected
+        // We noticed scopia often serving pornographic results for non-pornographic queries. After much deliberation we decided to filter pornographic from scopia. Those will have to be supplied by other search engines.
+
+        $words = [
+            "fisting" => 60,
+            "live cam" => 60,
+            "fick" => 60,
+            "anal" => 60,
+            "dildo" => 60,
+            "masturbat" => 60,
+            "gangbang" => 60,
+            "fotze" => 60,
+            "porn" => 50,
+            "anus" => 50,
+            "penetration" => 50,
+            "cuckold" => 50,
+            "orgasmus" => 50,
+            "milf" => 50,
+            "dilf" => 50,
+            "voyeur" => 40,
+            "fuck" => 40,
+            "nude" => 40,
+            "muschi" => 40,
+            "sex" => 40,
+            "nackt" => 40,
+            "amateur" => 30,
+            "schlampe" => 30,
+            "eroti" => 30,
+            "dick" => 30,
+            "teen" => 30,
+            "hardcore" => 30,
+            "fetisch" => 30,
+            "pussy" => 30,
+            "pussies" => 30,
+            "cheat" => 20,
+            "gratis" => 20,
+            "geil" => 20,
+            "video" => 10,
+            "girl" => 10,
+            "boy" => 10,
+            "weib" => 10,
+            "titt" => 10,
+            "bikini" => 10,
+            "hot " => 10,
+            "pics" => 10,
+            "free" => 10,
+        ];
+        $acc = 0;
+        foreach($words as $word => $score) {
+            if (stristr($text,$word)) {
+                $acc += $score;
+            }
+        }
+        return $acc >= 100;
+    }
+
     public function getNext(\App\MetaGer $metager, $result)
     {
         $result = html_entity_decode($result);
-- 
GitLab


From 3a2b92c5758d4c57ef18f189355740129c064322 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Phil=20H=C3=B6fer?= <phil@suma-ev.de>
Date: Fri, 14 Jan 2022 12:23:59 +0000
Subject: [PATCH 2/2] adjust porn filter and fix typo

---
 app/Models/parserSkripte/Scopia.php | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/app/Models/parserSkripte/Scopia.php b/app/Models/parserSkripte/Scopia.php
index 6c9960ab..937e0347 100644
--- a/app/Models/parserSkripte/Scopia.php
+++ b/app/Models/parserSkripte/Scopia.php
@@ -59,11 +59,12 @@ class Scopia extends Searchengine
 
     private function containsPornContent($text) {
         // Returns true if pornographic content is detected
-        // We noticed scopia often serving pornographic results for non-pornographic queries. After much deliberation we decided to filter pornographic from scopia. Those will have to be supplied by other search engines.
+        // We noticed scopia often serving pornographic results for non-pornographic queries. After much deliberation we decided to filter pornographic results from scopia. Those will have to be supplied by other search engines.
 
         $words = [
             "fisting" => 60,
             "live cam" => 60,
+            "telefonsex" => 60,
             "fick" => 60,
             "anal" => 60,
             "dildo" => 60,
@@ -84,6 +85,7 @@ class Scopia extends Searchengine
             "sex" => 40,
             "nackt" => 40,
             "amateur" => 30,
+            "webcam" => 30,
             "schlampe" => 30,
             "eroti" => 30,
             "dick" => 30,
-- 
GitLab