Prevent EmbedParser to follow infinite redirect for Twitter links

fix according to oscarotero/Embed#520 (comment) remp/helpdesk#2594
remp2020 · Mar 28, 2024 · 6a5d169 · 6a5d169
1 parent ede45c6
commit 6a5d169
Show file tree

Hide file tree

Showing 4 changed files with 33 additions and 12 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -28,6 +28,7 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/) and this p
 - Fixed issue with oversize images in MS Outlook. remp/remp#1330
 - Fixed issue with persistent embed cookies stored in system `tmp` folder, which were shared across releases. remp/helpdesk#2587
   - Each release now stores embed cookies in its own temp folder.
+- Added ability to set custom CURL settings for `EmbedParser`. remp/helpdesk#2594 
 
 ## Archive
 

diff --git a/Mailer/app/Models/Generators/EmbedParser.php b/Mailer/app/Models/Generators/EmbedParser.php
@@ -12,6 +12,11 @@ public function setTwitterLinkText(string $twitterLinkText = null): void
         $this->twitterLinkText = $twitterLinkText;
     }
 
+    private function isTwitterLink($link): bool
+    {
+        return str_contains($link, 'twitt');
+    }
+
     public function createEmbedMarkup(string $link, ?string $title = null, ?string $image = null, bool $isVideo = false): string
     {
         $html = "<br>";
@@ -20,7 +25,7 @@ public function createEmbedMarkup(string $link, ?string $title = null, ?string $
 
         if (!is_null($image) && !is_null($title)) {
             $html .= "<img src='{$image}' alt='{$title}' style='outline:none;text-decoration:none;-ms-interpolation-mode:bicubic;max-width:100%;clear:both;display:inline;width:100%;height:auto;'>";
-        } elseif (preg_match('/twitt/', $link)) {
+        } elseif ($this->isTwitterLink($link)) {
             return "<br><a style=\"display: block;margin: 0 0 20px;padding: 7px 10px;text-decoration: none;text-align: center;font-weight: bold;font-family:'Helvetica Neue', Helvetica, Arial;color: #249fdc; background: #ffffff; border: 3px solid #249fdc;margin: 16px 0 16px 0\" href=\"{$link}\" target=\"_blank\">{$this->twitterLinkText}</a>";
         } else {
             $html .= "<span style='text-decoration: underline; color: #1F3F83;'>" . $link . "</span>";

diff --git a/Mailer/extensions/mailer-module/src/Models/Generators/EmbedParser.php b/Mailer/extensions/mailer-module/src/Models/Generators/EmbedParser.php
@@ -12,18 +12,31 @@ class EmbedParser
 {
     protected ?string $videoLinkText;
 
-    public function __construct(private CurlClient $curlClient)
-    {
-    }
+    protected array $curlSettings = [];
 
     public function setVideoLinkText(?string $videoLinkText = null): void
     {
         $this->videoLinkText = $videoLinkText;
     }
 
+    public function setCurlSettings(array $settings)
+    {
+        $this->curlSettings = $settings;
+    }
+
     private function fetch(string $url): ?array
     {
-        $embed = new Embed(new Crawler($this->curlClient));
+        $curlSettings = [
+            // Twitter may generate infinite redirect (2024/03),
+            // fix according to https://github.com/oscarotero/Embed/issues/520#issue-1782756560
+            'follow_location' => !$this->isTwitterLink($url),
+            ...$this->curlSettings,
+        ];
+
+        $curlClient = new CurlClient();
+        $curlClient->setSettings($curlSettings);
+
+        $embed = new Embed(new Crawler($curlClient));
         $embed = $embed->get($url);
 
         $oEmbed = $embed->getOEmbed();
@@ -38,17 +51,22 @@ private function fetch(string $url): ?array
             'link' => $embed->url->__toString(),
             'title' => $embed->title ?? '',
             'image' => $image,
-            'isVideo' => $type === 'video'
+            'isVideo' => $type === 'video',
         ];
     }
 
+    private function isTwitterLink($link)
+    {
+        return str_contains($link, 'twitt');
+    }
+
     public function parse(string $link): ?string
     {
         $link = trim($link);
 
         if (preg_match('/^(?:(?:https?:)?\/\/)?(?:www\.)?facebook\.com\/[a-zA-Z0-9.]+\/videos\/(?:[a-zA-Z0-9.]+\/)?([0-9]+)/', $link)
             || str_contains($link, 'youtu')
-            || str_contains($link, 'twitt')
+            || $this->isTwitterLink($link)
         ) {
             try {
                 if ($data = $this->fetch($link)) {

diff --git a/Mailer/extensions/mailer-module/src/config/config.neon b/Mailer/extensions/mailer-module/src/config/config.neon
@@ -333,15 +333,12 @@ services:
 	- Remp\MailerModule\Models\Generators\WordpressHelpers
 	embedParser:
 		factory: Remp\MailerModule\Models\Generators\EmbedParser
+		setup:
+			- setCurlSettings([cookies_path: %tempDir%/embed-cookies.txt])
 	articleLocker:
 		factory: Remp\MailerModule\Models\Generators\ArticleLocker
 
 	# healtCheck - inject temp for storage check
 	- Remp\MailerModule\Presenters\HealthPresenter(%tempDir%)
 
 	- Remp\MailerModule\Components\MailLinkStats\MailLinkStats
-
-	embedCurlClient:
-		factory: Embed\Http\CurlClient
-		setup:
-			- setSettings([cookies_path: %tempDir%/embed-cookies.txt])