Re-introduce check for comment similarity to annoy spammers. Do not a…

…llow comments with email addresses.
BeWelcome · May 12, 2024 · 51edc14 · 51edc14
1 parent 0d7ac8c
commit 51edc14
Show file tree

Hide file tree

Showing 4 changed files with 181 additions and 25 deletions.
diff --git a/src/Controller/CommentController.php b/src/Controller/CommentController.php
@@ -152,7 +152,6 @@ public function addComment(
         $memberPreference = $loggedInMember->getMemberPreference($preference);
         $showCommentGuideline = ('0' === $memberPreference->getValue());
 
-        /** @var Member $loggedInMember */
         $form = $this->createForm(CommentType::class, null, [
             'to_member' => $member,
             'show_comment_guideline' => $showCommentGuideline,
@@ -162,30 +161,38 @@ public function addComment(
         if ($form->isSubmitted() && $form->isValid()) {
             /** @var Comment $comment */
             $comment = $form->getData();
-            $comment->setToMember($member);
-            $comment->setFromMember($loggedInMember);
-            if (CommentQualityType::NEGATIVE === $comment->getQuality()) {
-                $comment->setAdminAction(CommentAdminActionType::ADMIN_CHECK);
-                $comment->setEditingAllowed(false);
-            }
-            $entityManager->persist($comment);
+            if (
+                $commentModel->checkCommentSpam($loggedInMember, $comment)
+                || $commentModel->checkForEmailAddress($comment)
+            ) {
+                $form->addError(new FormError($this->translator->trans('commentsomethingwentwrong')));
+            } else {
+                $comment->setToMember($member);
+                $comment->setFromMember($loggedInMember);
 
-            // Mark comment guidelines as read and hide the checkbox for the future
-            $memberPreference->setValue('1');
-            $entityManager->persist($memberPreference);
-            $entityManager->flush();
+                if (CommentQualityType::NEGATIVE === $comment->getQuality()) {
+                    $comment->setAdminAction(CommentAdminActionType::ADMIN_CHECK);
+                    $comment->setEditingAllowed(false);
+                }
+                $entityManager->persist($comment);
 
-            $mailer->sendNewCommentNotification($comment);
+                // Mark comment guidelines as read and hide the checkbox for the future
+                $memberPreference->setValue('1');
+                $entityManager->persist($memberPreference);
+                $entityManager->flush();
 
-            $this->addTranslatedFlash(
-                'notice',
-                'flash.comment.added',
-                [
-                    'username' => $member->getUsername(),
-                ]
-            );
+                $mailer->sendNewCommentNotification($comment);
 
-            return $this->redirectToRoute('profile_comments', ['username' => $member->getUsername()]);
+                $this->addTranslatedFlash(
+                    'notice',
+                    'flash.comment.added',
+                    [
+                        'username' => $member->getUsername(),
+                    ]
+                );
+
+                return $this->redirectToRoute('profile_comments', ['username' => $member->getUsername()]);
+            }
         }
 
         return $this->render('/profile/comment.add.html.twig', [

diff --git a/src/Model/CommentModel.php b/src/Model/CommentModel.php
@@ -60,28 +60,127 @@ public function checkIfNewExperience(Comment $original, Comment $updated): bool
         $newExperience = false;
         try {
             $maxlen = max(strlen($updatedText), strlen($originalText));
-            $calculator = new LevenshteinDistance(false, 0, 1000**2);
+            $calculator = new LevenshteinDistance(false, 0, 1000 ** 2);
             $iteration = 0;
             $maxIteration = $maxlen / 1000;
             while ($iteration < $maxIteration && !$newExperience) {
-
                 $currentUpdatedText = substr($updatedText, $iteration * 1000, 1000);
                 $currentOriginalText = substr($originalText, $iteration * 1000, 1000);
                 $levenshteinDistance = ($calculator->calculate(
                     $currentUpdatedText,
-                    $currentOriginalText)
+                    $currentOriginalText
+                )
                 )['distance'];
 
                 if ($levenshteinDistance >= max(strlen($currentUpdatedText), strlen($currentOriginalText)) / 7) {
                     $newExperience = true;
                 }
                 $iteration++;
             }
-        } catch(Throwable $e) {
+        } catch (Throwable $e) {
             // ignore exception and just return false (likely consumed too much memory)
             return $newExperience;
         }
 
         return $newExperience;
     }
+
+    public function checkCommentSpam(Member $loggedInMember, Comment $comment): bool
+    {
+        $spamCheckParams = [
+            ['duration' => '00:02:00', 'count' => 1],
+            ['duration' => '00:20:00', 'count' => 5],
+            ['duration' => '06:00:00', 'count' => 25],
+        ];
+
+        $check1 = $this->checkCommentsDuration($loggedInMember, $comment, $spamCheckParams[0]);
+        $check2 = $this->checkCommentsDuration($loggedInMember, $comment, $spamCheckParams[1]);
+        $check3 = $this->checkCommentsDuration($loggedInMember, $comment, $spamCheckParams[2]);
+
+        return $check1 || $check2 || $check3;
+    }
+
+    private function checkCommentsDuration(Member $member, Comment $comment, array $params): bool
+    {
+        $duration = $params['duration'];
+        $count = $params['count'];
+
+        $result = false;
+        $commentCount = $this->entityManager
+            ->getConnection()
+            ->executeQuery(
+                "
+                    SELECT
+                        COUNT(*) as cnt
+                    FROM
+                        comments c
+                    WHERE
+                        c.IdFromMember = :memberId
+                        AND TIMEDIFF(NOW(), created) < :duration
+                ",
+                [ ':memberId' => $member->getId(), ':duration' => $duration]
+            )
+            ->fetchOne()
+        ;
+
+        if ($commentCount >= $count) {
+            // Okay limit was hit, check for comment quality
+            // Get all comments written during the given duration
+            $comments = $this->entityManager
+                ->getConnection()
+                ->executeQuery(
+                    "
+                        SELECT
+                            c.TextFree
+                        FROM
+                            comments c
+                        WHERE
+                            c.IdFromMember = :memberId
+                            AND TIMEDIFF(NOW(), created) < :duration
+                    ",
+                    [ ':memberId' => $member->getId(), ':duration' => $duration]
+                )
+                ->fetchAllAssociative()
+            ;
+            $result = $this->checkCommentSimilarity($comments, $comment);
+        }
+
+        return $result;
+    }
+
+    private function checkCommentSimilarity(array $comments, Comment $comment): bool
+    {
+        $similar = 0;
+        $comments[count($comments)] = ['TextFree' => $comment->getTextfree()];
+        $count = count($comments);
+        for ($i = 0; $i < $count - 1; $i++) {
+            for ($j = $i + 1; $j < $count; $j++) {
+                similar_text(
+                    $comments[$i]['TextFree'],
+                    $comments[$j]['TextFree'],
+                    $percent
+                );
+                if ($percent > 95) {
+                    $similar++;
+                }
+            }
+        }
+        return $similar != $count * ($count - 1);
+    }
+
+    public function checkForEmailAddress(Comment $comment): bool
+    {
+        $commentText = $comment->getTextfree();
+        $atPos = strpos($commentText, '@');
+        $whiteSpaceBefore = strrpos(substr($commentText, 0, $atPos), ' ');
+        $whiteSpaceAfter = strpos($commentText, ' ', $atPos);
+        if (false === $whiteSpaceAfter) {
+            $whiteSpaceAfter = strlen($commentText);
+        }
+        $potentialEmailAddress =
+            substr($commentText, $whiteSpaceBefore + 1, $whiteSpaceAfter - $whiteSpaceBefore - 1);
+        $emailAddressFound = filter_var($potentialEmailAddress, FILTER_VALIDATE_EMAIL) !== false;
+
+        return $emailAddressFound;
+    }
 }
diff --git a/templates/profile/comment.form.html.twig b/templates/profile/comment.form.html.twig
@@ -31,6 +31,7 @@
     <h2>{{ 'commentheading'|trans|format(member.username) }}</h2>
     <div class="alert alert-info">{{ 'followcommentguidelines'|trans|raw }}</div>
     {{ form_start(form, { 'attr': { 'novalidate': 'novalidate' } }) }}
+    {{ form_errors(form) }}
     {{ form_row(form.quality) }}
     {{ form_errors(form.quality) }}
     {{ form_row(form.relations) }}

diff --git a/tests/Model/CommentModelTest.php b/tests/Model/CommentModelTest.php
@@ -269,6 +269,55 @@ public function testLongTextWithLowNumberUpdatesIsNotANewExperience()
 
         $this->assertFalse($this->commentModel->checkIfNewExperience($original, $updated));
     }
+
+    public function testCommentWithEmailAddressInTextIsRecognized()
+    {
+        $comment = new Comment();
+        $comment->setTextfree('This is an@email.address in the middle of a text.');
+
+        $emailAddressFound = $this->commentModel->checkForEmailAddress($comment);
+
+        $this->assertTrue($emailAddressFound);
+    }
+    public function testCommentWithEmailAddressAtStartOfTextIsRecognized()
+    {
+        $comment = new Comment();
+        $comment->setTextfree('an@email.address at the start of a text.');
+
+        $emailAddressFound = $this->commentModel->checkForEmailAddress($comment);
+
+        $this->assertTrue($emailAddressFound);
+    }
+    public function testCommentWithEmailAddressAtTheEndOfTextIsRecognized()
+    {
+        $comment = new Comment();
+        $comment->setTextfree('At the end of this text, there is an@email.address');
+
+        $emailAddressFound = $this->commentModel->checkForEmailAddress($comment);
+
+        $this->assertTrue($emailAddressFound);
+    }
+
+    public function testCommentWithTwoEmailAddressesInTextIsRecognized()
+    {
+        $comment = new Comment();
+        $comment->setTextfree('This is an@email.address in the middle of a text. And another one at the end. another@email.net');
+
+        $emailAddressFound = $this->commentModel->checkForEmailAddress($comment);
+
+        $this->assertTrue($emailAddressFound);
+    }
+
+    public function testCommentWithoutEmailAddressInTextIsRecognized()
+    {
+        $comment = new Comment();
+        $comment->setTextfree('This is an @instagram username.');
+
+        $emailAddressFound = $this->commentModel->checkForEmailAddress($comment);
+
+        $this->assertFalse($emailAddressFound);
+    }
+
     private function buildRelations(array $relations): string
     {
         return implode(',', $relations);