Merge pull request #64 from aaronpk/mf2-dev

Update to php-mf2 0.4.0

composer.json

@@ -5,7 +5,7 @@
   "homepage": "https://github.com/aaronpk/XRay",
   "description": "X-Ray returns structured data from any URL",
   "require": {
-    "mf2/mf2": "^0.3.2",
+    "mf2/mf2": ">=0.4.0",
     "ezyang/htmlpurifier": "4.10.*",
     "indieweb/link-rel-parser": "0.1.*",
     "dg/twitter-php": "3.6.*",

composer.lock

@@ -4,7 +4,7 @@
         "Read more about it at https://getcomposer.org/doc/01-basic-usage.md#composer-lock-the-lock-file",
         "This file is @generated automatically"
     ],
-    "content-hash": "0a104ee89f03db919f34d3a9d387380c",
+    "content-hash": "60f545028d44eb8e15c499fbe7ecf4d0",
     "packages": [
         {
             "name": "cebe/markdown",
@@ -261,16 +261,16 @@
         },
         {
             "name": "mf2/mf2",
-            "version": "v0.3.2",
+            "version": "v0.4.0",
             "source": {
                 "type": "git",
                 "url": "https://github.com/indieweb/php-mf2.git",
-                "reference": "dc0d90d4ee30864bcf37cd3a8fc8db94f9134cc4"
+                "reference": "42ef6eb9777bffe654a70cbbc1dbd777a61c1445"
             },
             "dist": {
                 "type": "zip",
-                "url": "https://api.github.com/repos/indieweb/php-mf2/zipball/dc0d90d4ee30864bcf37cd3a8fc8db94f9134cc4",
-                "reference": "dc0d90d4ee30864bcf37cd3a8fc8db94f9134cc4",
+                "url": "https://api.github.com/repos/indieweb/php-mf2/zipball/42ef6eb9777bffe654a70cbbc1dbd777a61c1445",
+                "reference": "42ef6eb9777bffe654a70cbbc1dbd777a61c1445",
                 "shasum": ""
             },
             "require": {
@@ -296,7 +296,7 @@
             },
             "notification-url": "https://packagist.org/downloads/",
             "license": [
-                "CC0"
+                "CC0-1.0"
             ],
             "authors": [
                 {
@@ -312,7 +312,7 @@
                 "parser",
                 "semantic"
             ],
-            "time": "2017-05-27T15:27:47+00:00"
+            "time": "2018-03-13T23:33:15+00:00"
         },
         {
             "name": "p3k/http",

lib/XRay/Feeds.php

@@ -76,26 +76,28 @@ class Feeds {
       // Some other document was returned, parse the HTML and look for rel alternates and Microformats
 
       $mf2 = \mf2\Parse($body, $result['url']);
-      if(isset($mf2['alternates'])) {
-        foreach($mf2['alternates'] as $alt) {
-          if(isset($alt['type'])) {
-            if(strpos($alt['type'], 'application/json') !== false) {
-              $feeds[] = [
-                'url' => $alt['url'],
-                'type' => 'jsonfeed'
-              ];
-            }
-            if(strpos($alt['type'], 'application/atom+xml') !== false) {
-              $feeds[] = [
-                'url' => $alt['url'],
-                'type' => 'atom'
-              ];
-            }
-            if(strpos($alt['type'], 'application/rss+xml') !== false) {
-              $feeds[] = [
-                'url' => $alt['url'],
-                'type' => 'rss'
-              ];
+      if(isset($mf2['rel-urls'])) {
+        foreach($mf2['rel-urls'] as $rel=>$info) {
+          if(isset($info['rels']) && in_array('alternate', $info['rels'])) {
+            if(isset($info['type'])) {
+              if(strpos($info['type'], 'application/json') !== false) {
+                $feeds[] = [
+                  'url' => $rel,
+                  'type' => 'jsonfeed'
+                ];
+              }
+              if(strpos($info['type'], 'application/atom+xml') !== false) {
+                $feeds[] = [
+                  'url' => $rel,
+                  'type' => 'atom'
+                ];
+              }
+              if(strpos($info['type'], 'application/rss+xml') !== false) {
+                $feeds[] = [
+                  'url' => $rel,
+                  'type' => 'rss'
+                ];
+              }
             }
           }
         }

lib/XRay/Formats/Mf2.php

@@ -65,7 +65,7 @@ class Mf2 extends Format {
       }
     }
 
-    // Check the list of items on the page to see if one matches the URL of the page, 
+    // Check the list of items on the page to see if one matches the URL of the page,
     // and treat as a permalink for that object if so.
     foreach($mf2['items'] as $item) {
       if(array_key_exists('url', $item['properties'])) {
@@ -323,16 +323,20 @@ class Mf2 extends Format {
   }
 
   private static function collectArrayURLValues($properties, $item, &$data, &$refs, &$http) {
+    $keys = [];
+
     foreach($properties as $p) {
       if(array_key_exists($p, $item['properties'])) {
         foreach($item['properties'][$p] as $v) {
           if(is_string($v) && self::isURL($v)) {
             if(!array_key_exists($p, $data)) $data[$p] = [];
             $data[$p][] = $v;
+            $keys[] = $p;
           }
           elseif(self::isMicroformat($v) && ($u=self::getPlaintext($v, 'url')) && self::isURL($u)) {
             if(!array_key_exists($p, $data)) $data[$p] = [];
             $data[$p][] = $u;
+            $keys[] = $p;
             // parse the object and put the result in the "refs" object
             $ref = self::parse(['items'=>[$v]], $u, $http);
             if($ref) {
@@ -340,7 +344,12 @@ class Mf2 extends Format {
             }
           }
         }
-      }      
+      }
+    }
+
+    // Remove duplicate values
+    foreach(array_unique($keys) as $key) {
+      $data[$key] = array_unique($data[$key]);
     }
   }
 
@@ -394,7 +403,7 @@ class Mf2 extends Format {
           $data['name'] = $name;
         }
       }
-    }    
+    }
   }
 
   private static function parseAsHEntry($mf2, $item, $http, $url) {
@@ -712,7 +721,7 @@ class Mf2 extends Format {
               and in_array(\p3k\XRay\normalize_url($authorPage), \p3k\XRay\normalize_urls($i['properties']['url']))
               and array_key_exists('uid', $i['properties'])
               and in_array(\p3k\XRay\normalize_url($authorPage), \p3k\XRay\normalize_urls($i['properties']['uid']))
-            ) { 
+            ) {
               return self::parseAsHCard($i, $http, $url, $authorPage)['data'];
             }
 
@@ -740,7 +749,7 @@ class Mf2 extends Format {
           }
 
         }
-        // Also check the "author" property 
+        // Also check the "author" property
         // (for finding the author of an h-feed's children when the author is the p-author property of the h-feed)
         if(isset($i['properties']['author'])) {
           foreach($i['properties']['author'] as $ic) {
@@ -766,16 +775,16 @@ class Mf2 extends Format {
   }
 
   private static function hasNumericKeys(array $arr) {
-    foreach($arr as $key=>$val) 
-      if (is_numeric($key)) 
+    foreach($arr as $key=>$val)
+      if (is_numeric($key))
         return true;
     return false;
   }
 
   private static function isMicroformat($mf) {
-    return is_array($mf) 
-      and !self::hasNumericKeys($mf) 
-      and !empty($mf['type']) 
+    return is_array($mf)
+      and !self::hasNumericKeys($mf)
+      and !empty($mf['type'])
       and isset($mf['properties']);
   }
 

tests/ParseTest.php

@@ -215,7 +215,7 @@ class ParseTest extends PHPUnit_Framework_TestCase {
     $body = $response->getContent();
     $this->assertEquals(200, $response->getStatusCode());
     $data = json_decode($body, true);
-    $this->assertEquals('entry', $data['data']['type']);    
+    $this->assertEquals('entry', $data['data']['type']);
     $this->assertEquals('http://example.com/100', $data['data']['in-reply-to'][0]);
     $this->assertArrayHasKey('http://example.com/100', $data['data']['refs']);
     $this->assertEquals('Example Post', $data['data']['refs']['http://example.com/100']['name']);
@@ -289,7 +289,7 @@ class ParseTest extends PHPUnit_Framework_TestCase {
     $this->assertEquals(200, $response->getStatusCode());
     $data = json_decode($body, true);
     $this->assertEquals('entry', $data['data']['type']);
-    $this->assertEquals('I\'ll be there!', $data['data']['name']);
+    $this->assertEquals('I\'ll be there!', $data['data']['content']['text']);
     $this->assertEquals('yes', $data['data']['rsvp']);
   }
 
@@ -454,7 +454,7 @@ class ParseTest extends PHPUnit_Framework_TestCase {
     $data = json_decode($body, true);
 
     $this->assertEquals('review', $data['data']['type']);
-    $this->assertEquals('Not great', $data['data']['summary']);
+    $this->assertEquals('Not great', $data['data']['name']);
     $this->assertEquals('3', $data['data']['rating']);
     $this->assertEquals('5', $data['data']['best']);
     $this->assertEquals('This is the full text of the review', $data['data']['content']['text']);
@@ -660,4 +660,28 @@ class ParseTest extends PHPUnit_Framework_TestCase {
     $this->assertObjectNotHasAttribute('photo', $data->data);
   }
 
+  public function testDuplicateReplyURLValues() {
+    $url = 'http://source.example.com/duplicate-in-reply-to-urls';
+    $response = $this->parse(['url' => $url]);
+
+    $body = $response->getContent();
+    $this->assertEquals(200, $response->getStatusCode());
+    $data = json_decode($body, true);
+
+    $this->assertEquals('http://example.com/100', $data['data']['in-reply-to'][0]);
+    $this->assertEquals(1, count($data['data']['in-reply-to']));
+  }
+
+  public function testDuplicateLikeOfURLValues() {
+    $url = 'http://source.example.com/duplicate-like-of-urls';
+    $response = $this->parse(['url' => $url]);
+
+    $body = $response->getContent();
+    $this->assertEquals(200, $response->getStatusCode());
+    $data = json_decode($body, true);
+
+    $this->assertEquals('http://example.com/100', $data['data']['like-of'][0]);
+    $this->assertEquals(1, count($data['data']['like-of']));
+  }
+
 }

tests/data/source.example.com/duplicate-in-reply-to-urls

@@ -0,0 +1,19 @@
+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 class="h-entry">
+    <a href="http://example.com/100" class="u-in-reply-to">in reply to</a>
+    <div class="u-in-reply-to h-cite">
+      <a href="http://example.com/100" class="u-url">this post</a>
+    </div>
+    <p class="e-content">This page has duplicate in-reply-to values.</p>
+    <a href="/duplicate-in-reply-to-urls" class="u-url">permalink</a>
+  </body>
+</html>

tests/data/source.example.com/duplicate-like-of-urls

@@ -0,0 +1,19 @@
+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 class="h-entry">
+    liked
+    <a href="http://example.com/100" class="u-like-of">a post</a>
+    <div class="u-like-of h-cite">
+      <a href="http://example.com/100" class="u-url">this post</a>
+    </div>
+    <a href="/duplicate-like-of-urls" class="u-url">permalink</a>
+  </body>
+</html>

tests/data/source.example.com/h-entry-rsvp

@@ -10,7 +10,7 @@ Connection: keep-alive
   </head>
   <body>
     <div class="h-entry">
-      <data class="p-rsvp" value="yes">I'll be there!</data>
+      <data class="p-rsvp" value="yes"><span class="p-content">I'll be there!</span></data>
     </div>
   </body>
 </html>