check for meta-equiv HTTP deleted

closes #16
8 years ago · 227311faa9
--- a/controllers/Parse.php
+++ b/controllers/Parse.php
@ -226,6 +226,16 @@ class Parse {
      }
    }

    // Check for meta http equiv and replace the status code if present
    foreach($xpath->query('//meta[@http-equiv=\'status\']') as $el) {
      $equivStatus = ''.$el->getAttribute('content');
      if($equivStatus && is_string($equivStatus)) {
        if(preg_match('/^(\d+)/', $equivStatus, $match)) {
          $result['code'] = (int)$match[1];
        }
      }
    }

    // If the URL has a fragment ID, find the DOM starting at that node and parse it instead
    $html = $result['body'];

--- a/tests/FetchTest.php
+++ b/tests/FetchTest.php
@ -103,4 +103,17 @@ class FetchTest extends PHPUnit_Framework_TestCase {
    $this->assertEquals(401, $data->code);
  }

  public function testMetaEquivDeleted() {
    $url = 'http://source.example.com/deleted';
    $response = $this->parse([
      'url' => $url
    ]);

    $body = $response->getContent();
    $this->assertEquals(200, $response->getStatusCode());
    $data = json_decode($body);
    $this->assertObjectNotHasAttribute('error', $data);
    $this->assertEquals(410, $data->code);
    $this->assertEquals('This post has been deleted.', $data->data->content->text);
  }
 }
--- a/tests/data/source.example.com/deleted
+++ b/tests/data/source.example.com/deleted
@ -0,0 +1,15 @@
 HTTP/1.1 200 OK
 Server: Apache
 Date: Wed, 09 Dec 2015 03:29:14 GMT
 Content-Type: text/html; charset=utf-8
 Connection: keep-alive

 <html>
  <head>
    <title>Test</title>
    <meta http-equiv="status" content="410 Gone" />
  </head>
  <body class="h-entry">
    <p class="e-content">This post has been deleted.</p>
  </body>
 </html>