From c530011b2cd7c37c929beaf9fdaf3890c9650f4a Mon Sep 17 00:00:00 2001
From: Aaron Parecki <aaron@parecki.com>
Date: Fri, 30 Aug 2024 07:29:22 -0700
Subject: [PATCH] add option to ignore rel-alternate as2

---
 controllers/Parse.php                          |  4 ++++
 lib/XRay/Formats/HTML.php                      |  7 ++++++-
 tests/ParseTest.php                            | 18 ++++++++++++++++--
 .../data/source.example.com/rel-alternate-as2  |  2 +-
 .../source.example.com/rel-alternate-priority  |  2 +-
 .../rel-alternate-priority.json                |  2 +-
 6 files changed, 29 insertions(+), 6 deletions(-)

diff --git a/controllers/Parse.php b/controllers/Parse.php
index 867bb5e..237c2df 100644
--- a/controllers/Parse.php
+++ b/controllers/Parse.php
@@ -75,6 +75,10 @@ class Parse {
       $opts['allowIframeVideo'] = $request->get('allow-iframe-video') == 'true';
     }
 
+    if($request->get('ignore-as2')) {
+      $opts['ignore-as2'] = $request->get('ignore-as2') == 'true';
+    }
+
     $url = $request->get('url');
     $html = $request->get('html') ?: $request->get('body');
 
diff --git a/lib/XRay/Formats/HTML.php b/lib/XRay/Formats/HTML.php
index 9219610..ff7e79d 100644
--- a/lib/XRay/Formats/HTML.php
+++ b/lib/XRay/Formats/HTML.php
@@ -114,12 +114,17 @@ class HTML extends Format {
         }
       }
 
-      if(count($alternates['as2'])) {
+      $ignoreAS2 = false;
+      if(isset($opts['ignore-as2']) && $opts['ignore-as2'] == true)
+        $ignoreAS2 = true;
+
+      if(!$ignoreAS2 && count($alternates['as2'])) {
         $relurl = $alternates['as2'][0];
         // Fetch and parse the ActivityStreams JSON link
         $jsonpage = $http->get($relurl, [
           'Accept' => 'application/activity+json,application/json'
         ]);
+
         // Skip and fall back to parsing the HTML if anything about this request fails
         if(!$jsonpage['error'] && $jsonpage['body']) {
           $jsondata = json_decode($jsonpage['body'],true);
diff --git a/tests/ParseTest.php b/tests/ParseTest.php
index bcb274e..cbf6144 100644
--- a/tests/ParseTest.php
+++ b/tests/ParseTest.php
@@ -1167,7 +1167,7 @@ class ParseTest extends PHPUnit\Framework\TestCase
 
         $this->assertEquals('mf2+json', $data['source-format']);
         $this->assertEquals('http://source.example.com/rel-alternate-priority.json', $data['parsed-url']);
-        $this->assertEquals('This should be the content from XRay', $data['data']['content']['text']);
+        $this->assertEquals('This is the content in the MF2 JSON file', $data['data']['content']['text']);
     }
 
     public function testRelAlternatePrioritizesMf2OverAS2()
@@ -1181,7 +1181,21 @@ class ParseTest extends PHPUnit\Framework\TestCase
 
         $this->assertEquals('mf2+json', $data['source-format']);
         $this->assertEquals('http://source.example.com/rel-alternate-priority.json', $data['parsed-url']);
-        $this->assertEquals('This should be the content from XRay', $data['data']['content']['text']);
+        $this->assertEquals('This is the content in the MF2 JSON file', $data['data']['content']['text']);
+    }
+
+    public function testRelAlternateIgnoreAS2AlternateOption()
+    {
+        $url = 'http://source.example.com/rel-alternate-as2';
+        $response = $this->parse(['url' => $url, 'ignore-as2' => true]);
+
+        $body = $response->getContent();
+        $this->assertEquals(200, $response->getStatusCode());
+        $data = json_decode($body, true);
+
+        $this->assertEquals('mf2+html', $data['source-format']);
+        $this->assertArrayNotHasKey('parsed-url', $data);
+        $this->assertEquals('This is the content in the HTML instead of the AS2 JSON', $data['data']['content']['text']);
     }
 
     public function testRelAlternateFallsBackOnInvalidJSON()
diff --git a/tests/data/source.example.com/rel-alternate-as2 b/tests/data/source.example.com/rel-alternate-as2
index a9b5bf1..1f8ebe6 100644
--- a/tests/data/source.example.com/rel-alternate-as2
+++ b/tests/data/source.example.com/rel-alternate-as2
@@ -13,7 +13,7 @@ Connection: keep-alive
   </head>
   <body>
     <div class="h-entry">
-      <p class="p-content">This should not be the content from XRay</p>
+      <p class="p-content">This is the content in the HTML instead of the AS2 JSON</p>
     </div>
   </body>
 </html>
diff --git a/tests/data/source.example.com/rel-alternate-priority b/tests/data/source.example.com/rel-alternate-priority
index bcc882f..6fbbbe8 100644
--- a/tests/data/source.example.com/rel-alternate-priority
+++ b/tests/data/source.example.com/rel-alternate-priority
@@ -13,7 +13,7 @@ Connection: keep-alive
   </head>
   <body>
     <div class="h-entry">
-      <p class="p-content">This should not be the content from XRay</p>
+      <p class="p-content">This is the content in the HTML page</p>
     </div>
   </body>
 </html>
diff --git a/tests/data/source.example.com/rel-alternate-priority.json b/tests/data/source.example.com/rel-alternate-priority.json
index a3839ba..27e9dca 100644
--- a/tests/data/source.example.com/rel-alternate-priority.json
+++ b/tests/data/source.example.com/rel-alternate-priority.json
@@ -12,7 +12,7 @@ Connection: keep-alive
             ],
             "properties": {
                 "content": [
-                    "This should be the content from XRay"
+                    "This is the content in the MF2 JSON file"
                 ]
             }
         }