handle the case where the server returns multiple content-type headers

6 years ago · 43db6098fc
--- a/lib/XRay/Fetcher.php
+++ b/lib/XRay/Fetcher.php
@ -89,9 +89,17 @@ class Fetcher {
    // Show an error if the content type returned is not a recognized type
    $format = null;
    if(isset($result['headers']['Content-Type']) && is_string($result['headers']['Content-Type'])) {
      $type = new MediaType($result['headers']['Content-Type']);
      $format = $type->format;
    if(isset($result['headers']['Content-Type'])) {
      $contentType = null;
      if(is_array($result['headers']['Content-Type'])) {
        $contentType = $result['headers']['Content-Type'][0];
      } elseif(is_string($result['headers']['Content-Type'])) {
        $contentType = $result['headers']['Content-Type'];
      }
      if($contentType) {
        $type = new MediaType($contentType);
        $format = $type->format;
      }
    }
    if(!$format ||
--- a/tests/ParseTest.php
+++ b/tests/ParseTest.php
@ -940,4 +940,15 @@ class ParseTest extends PHPUnit_Framework_TestCase {
    $this->assertEquals('XRay should use this content since the JSON in the rel-alternate is invalid', $data['data']['content']['text']);
  }
  public function testMultipleContentTypeHeaders() {
    $url = 'http://source.example.com/multiple-content-type';
    $response = $this->parse(['url' => $url]);
    $body = $response->getContent();
    $this->assertEquals(200, $response->getStatusCode());
    $data = json_decode($body, true);
    $this->assertEquals('mf2+html', $data['source-format']);
  }
 }
--- a/tests/data/source.example.com/multiple-content-type
+++ b/tests/data/source.example.com/multiple-content-type
@ -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
 Content-Type: text/html
 Connection: keep-alive
 <html>
  <head>
    <title>Test</title>
  </head>
  <body class="h-entry">
    <p class="p-content">This page has a link to <a href="http://target.example.com">target.example.com</a> and some <b>formatted text</b> but is in a p-content element so is plaintext.</p>
  </body>
 </html>