use effective URL after following redirects when comparing URLs

8 years ago · 8dc0caa4d0
--- a/controllers/Parse.php
+++ b/controllers/Parse.php
@ -124,10 +124,10 @@ class Parse {
    }

    // Now start pulling in the data from the page. Start by looking for microformats2
    $mf2 = mf2\Parse($result['body'], $url);
    $mf2 = mf2\Parse($result['body'], $result['url']);

    if($mf2 && count($mf2['items']) > 0) {
      $data = Formats\Mf2::parse($mf2, $url, $this->http);
      $data = Formats\Mf2::parse($mf2, $result['url'], $this->http);
      if($data) {
        return $this->respond($response, 200, $data);
      }
--- a/lib/HTTPCurl.php
+++ b/lib/HTTPCurl.php
@ -18,6 +18,7 @@ class HTTPCurl {
      'error' => self::error_string_from_code(curl_errno($ch)),
      'error_description' => curl_error($ch),
      'error_code' => curl_errno($ch),
      'url' => curl_getinfo($ch, CURLINFO_EFFECTIVE_URL),
    );
  }

@ -36,6 +37,7 @@ class HTTPCurl {
      'error' => self::error_string_from_code(curl_errno($ch)),
      'error_description' => curl_error($ch),
      'error_code' => curl_errno($ch),
      'url' => curl_getinfo($ch, CURLINFO_EFFECTIVE_URL),
    );
  }

@ -50,6 +52,7 @@ class HTTPCurl {
      'error' => self::error_string_from_code(curl_errno($ch)),
      'error_description' => curl_error($ch),
      'error_code' => curl_errno($ch),
      'url' => curl_getinfo($ch, CURLINFO_EFFECTIVE_URL),
    );
  }

--- a/lib/HTTPStream.php
+++ b/lib/HTTPStream.php
@ -68,6 +68,7 @@ class HTTPStream {
      'body' => $body,
      'error' => $error ? $error['code'] : false,
      'error_description' => $error ? $error['description'] : false,
      'url' => $url,
    );
  }

--- a/lib/HTTPTest.php
+++ b/lib/HTTPTest.php
@ -23,7 +23,8 @@ class HTTPTest extends HTTPCurl {
      'code' => $response['code'],
      'headers' => $response['headers'],
      'error' => '',
      'error_description' => ''
      'error_description' => '',
      'url' => $response['url']
    );
  }

@ -46,13 +47,21 @@ class HTTPTest extends HTTPCurl {
    }

    $headers = preg_replace('/HTTP\/1\.1 \d+ .+/', '', $headers);
    $parsedHeaders = self::parse_headers($headers);

    if(array_key_exists('Location', $parsedHeaders)) {
      $effectiveUrl = \mf2\resolveUrl($url, $parsedHeaders['Location']);
    } else {
      $effectiveUrl = $url;
    }

    return array(
      'code' => $code,
      'headers' => self::parse_headers($headers),
      'headers' => $parsedHeaders,
      'body' => $body,
      'error' => '',
      'error_description' => ''
      'error_description' => '',
      'url' => $effectiveUrl
    );
  }

--- a/tests/ParseTest.php
+++ b/tests/ParseTest.php
@ -213,4 +213,26 @@ class ParseTest extends PHPUnit_Framework_TestCase {
    $this->assertEquals('Primary Post', $data['data']['name']);
  }

  public function testHEntryWithHCardSibling() {
    $url = 'http://source.example.com/h-entry-with-h-card-sibling';
    $response = $this->parse(['url' => $url]);

    $body = $response->getContent();
    $this->assertEquals(200, $response->getStatusCode());
    $data = json_decode($body, true);
    $this->assertEquals('entry', $data['data']['type']);
    $this->assertEquals('Hello World', $data['data']['content']['text']);
  }

  public function testHEntryRedirectWithHCardSibling() {
    $url = 'http://source.example.com/h-entry-redirect-with-h-card-sibling';
    $response = $this->parse(['url' => $url]);

    $body = $response->getContent();
    $this->assertEquals(200, $response->getStatusCode());
    $data = json_decode($body, true);
    $this->assertEquals('entry', $data['data']['type']);
    $this->assertEquals('Hello World', $data['data']['content']['text']);
  }

 }
--- a/tests/data/source.example.com/h-entry-redirect-with-h-card-sibling
+++ b/tests/data/source.example.com/h-entry-redirect-with-h-card-sibling
@ -0,0 +1,20 @@
 HTTP/1.1 301 Moved Permanently
 Server: Apache
 Date: Wed, 09 Dec 2015 03:29:14 GMT
 Content-Type: text/html; charset=utf-8
 Connection: keep-alive
 Location: /h-entry-with-h-card-sibling

 <html>
  <head>
    <title>Test</title>
  </head>
  <body>
    <div class="h-entry">
      <p class="p-content">Hello World</p>
      <a href="/h-entry-with-h-card-sibling">permalink</a>
      <!-- testing when the URL to this page is a redirect to the actual URL, so that the reported URL is different from what was fetched -->
    </div>
    <a href="/" class="h-card">Author Name</a>
  </body>
 </html>
--- a/tests/data/source.example.com/h-entry-with-h-card-sibling
+++ b/tests/data/source.example.com/h-entry-with-h-card-sibling
@ -0,0 +1,18 @@
 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>
  </head>
  <body>
    <div class="h-entry">
      <p class="p-content">Hello World</p>
      <a href="/h-entry-with-h-card-sibling">permalink</a>
    </div>
    <a href="/" class="h-card">Author Name</a>
  </body>
 </html>