normalize relative URLs in JSONFeed items

closes #77
aaronpk · Nov 9, 2018 · 9163341 · 9163341
1 parent b618f9a
commit 9163341
Show file tree

Hide file tree

Showing 7 changed files with 106 additions and 14 deletions.
diff --git a/lib/XRay/Formats/Format.php b/lib/XRay/Formats/Format.php
@@ -34,7 +34,7 @@ protected static function _loadHTML($html) {
     return [$doc, $xpath];
   }
 
-  protected static function sanitizeHTML($html, $allowImg=true) {
+  protected static function sanitizeHTML($html, $allowImg=true, $baseURL=false) {
     $allowed = [
       'a',
       'abbr',
@@ -68,6 +68,12 @@ protected static function sanitizeHTML($html, $allowImg=true) {
     $config = HTMLPurifier_Config::createDefault();
     $config->set('Cache.DefinitionImpl', null);
     $config->set('HTML.AllowedElements', $allowed);
+
+    if($baseURL) {
+      $config->set('URI.MakeAbsolute', true);
+      $config->set('URI.Base', $baseURL);
+    }
+
     $def = $config->getHTMLDefinition(true);
     $def->addElement(
       'time',

diff --git a/lib/XRay/Formats/JSONFeed.php b/lib/XRay/Formats/JSONFeed.php
@@ -23,14 +23,14 @@ public static function parse($feed, $url) {
       $result['data']['type'] = 'feed';
 
       foreach($feed['items'] as $item) {
-        $result['data']['items'][] = self::_hEntryFromFeedItem($item, $feed);
+        $result['data']['items'][] = self::_hEntryFromFeedItem($item, $feed, $url);
       }
     }
 
     return $result;
   }
 
-  private static function _hEntryFromFeedItem($item, $feed) {
+  private static function _hEntryFromFeedItem($item, $feed, $feedurl) {
     $entry = [
       'type' => 'entry',
       'author' => [
@@ -64,14 +64,16 @@ private static function _hEntryFromFeedItem($item, $feed) {
       $entry['name'] = trim($item['title']);
     }
 
+    $baseURL = isset($entry['url']) ? $entry['url'] : $feedurl;
+
     if(isset($item['content_html']) && isset($item['content_text'])) {
       $entry['content'] = [
-        'html' => self::sanitizeHTML($item['content_html']),
+        'html' => self::sanitizeHTML($item['content_html'], true, $baseURL),
         'text' => trim($item['content_text'])
       ];
     } elseif(isset($item['content_html'])) {
       $entry['content'] = [
-        'html' => self::sanitizeHTML($item['content_html']),
+        'html' => self::sanitizeHTML($item['content_html'], true, $baseURL),
         'text' => self::stripHTML($item['content_html'])
       ];
     } elseif(isset($item['content_text'])) {
@@ -93,7 +95,7 @@ private static function _hEntryFromFeedItem($item, $feed) {
     }
 
     if(isset($item['image'])) {
-      $entry['photo'] = $item['image'];
+      $entry['photo'] = \Mf2\resolveUrl($baseURL, $item['image']);
     }
 
     if(isset($item['tags'])) {

diff --git a/tests/FeedTest.php b/tests/FeedTest.php
@@ -199,7 +199,7 @@ public function testJSONFeed() {
     $this->assertEquals('feed+json', $result->{'source-format'});
     $data = $result->data;
 
-    $this->assertEquals(10, count($data->items));
+    $this->assertEquals(11, count($data->items));
     for($i=0; $i<8; $i++) {
       $this->assertEquals('entry', $data->items[$i]->type);
       $this->assertEquals('manton', $data->items[$i]->author->name);
@@ -213,15 +213,35 @@ public function testJSONFeed() {
     $this->assertEquals('note', $data->items[0]->{'post-type'});
     $this->assertEquals('article', $data->items[4]->{'post-type'});
 
-    $this->assertEquals('<p>Lots of good feedback on <a href="http://help.micro.blog/2017/wordpress-import/">the WordPress import</a>. Made a couple improvements this morning. Overall, pretty good.</p>', $data->items[9]->content->html);
-    $this->assertEquals('Lots of good feedback on the WordPress import. Made a couple improvements this morning. Overall, pretty good.', $data->items[9]->content->text);
-    $this->assertEquals('http://www.manton.org/2017/11/5975.html', $data->items[9]->url);
-    $this->assertEquals('http://www.manton.org/2017/11/5975.html', $data->items[9]->uid);
-    $this->assertEquals('2017-11-07T15:04:01+00:00', $data->items[9]->published);
+    $this->assertEquals('<p>Coming up on a year since I wrote about how <a href="http://www.manton.org/2016/11/todays-social-networks-are-broken.html">today’s social networks are broken</a>. Still what I believe.</p>', $data->items[7]->content->html);
+    $this->assertEquals('Coming up on a year since I wrote about how today’s social networks are broken. Still what I believe.', $data->items[7]->content->text);
+    $this->assertEquals('http://www.manton.org/2017/11/5979.html', $data->items[7]->url);
+    $this->assertEquals('http://www.manton.org/2017/11/5979.html', $data->items[7]->uid);
+    $this->assertEquals('2017-11-07T21:00:42+00:00', $data->items[7]->published);
 
     $this->assertEquals('feed', $data->type);
   }
 
+  public function testJSONFeedRelativeImages() {
+    $url = 'http://feed.example.com/jsonfeed';
+    $response = $this->parse(['url' => $url, 'expect' => 'feed']);
+
+    $body = $response->getContent();
+    $this->assertEquals(200, $response->getStatusCode());
+    $result = json_decode($body);
+    $this->assertEquals('feed+json', $result->{'source-format'});
+    $data = $result->data;
+
+    // Relative image on an item that has a url
+    $this->assertEquals('http://www.manton.org/2017/11/image.jpg', $data->items[9]->photo);
+
+    // Relative image on an item that has no URL, fall back to feed URL
+    $this->assertEquals('http://feed.example.com/image.jpg', $data->items[10]->photo);
+
+    // Relative image inside the content html
+    $this->assertContains('http://www.manton.org/2017/11/img.jpg', $data->items[9]->content->html);
+  }
+
   public function testAtomFeed() {
     $url = 'http://feed.example.com/atom';
     $response = $this->parse(['url' => $url, 'expect' => 'feed']);

diff --git a/tests/SanitizeTest.php b/tests/SanitizeTest.php
@@ -186,6 +186,28 @@ public function testPhotoInTextContentNoAlt() {
   }
   */
 
+  public function testRelativePhotoInContent() {
+    $url = 'http://sanitize.example/photo-in-content-relative';
+    $response = $this->parse(['url' => $url]);
+
+    $body = $response->getContent();
+    $this->assertEquals(200, $response->getStatusCode());
+    $data = json_decode($body);
+
+    $this->assertContains('http://sanitize.example/photo1.jpg', $data->data->content->html);
+  }
+
+  public function testRelativePhotoProperty() {
+    $url = 'http://sanitize.example/photo-relative';
+    $response = $this->parse(['url' => $url]);
+
+    $body = $response->getContent();
+    $this->assertEquals(200, $response->getStatusCode());
+    $data = json_decode($body);
+
+    $this->assertEquals('http://sanitize.example/photo.jpg', $data->data->photo[0]);
+  }
+
   public function testPhotoInContentEmptyAltAttribute() {
     // https://github.com/aaronpk/XRay/issues/52
 

diff --git a/tests/data/feed.example.com/jsonfeed b/tests/data/feed.example.com/jsonfeed
@@ -119,12 +119,25 @@ Content-Type: application/json; charset=UTF-8
             "id": "http://www.manton.org/2017/11/5975.html",
             "url": "http://www.manton.org/2017/11/5975.html",
             "title": "",
-            "content_html": "<p>Lots of good feedback on <a href=\"http://help.micro.blog/2017/wordpress-import/\">the WordPress import</a>. Made a couple improvements this morning. Overall, pretty good.</p>\n",
+            "content_html": "<p><img src=\"img.jpg\"></p>\n",
+            "image": "image.jpg",
+            "banner_image": "banner_image.jpg",
+            "date_published": "2017-11-07T15:04:01+00:00",
+            "date_modified": "2017-11-07T15:04:01+00:00",
+            "author": {
+                "name": "manton"
+            }
+        },
+        {
+            "id": "http://www.manton.org/2017/11/5975.html",
+            "title": "",
+            "content_html": "<p><img src=\"img.jpg\"></p>\n",
+            "image": "image.jpg",
             "date_published": "2017-11-07T15:04:01+00:00",
             "date_modified": "2017-11-07T15:04:01+00:00",
             "author": {
                 "name": "manton"
             }
         }
     ]
-}
+}
diff --git a/tests/data/sanitize.example/photo-in-content-relative b/tests/data/sanitize.example/photo-in-content-relative
@@ -0,0 +1,14 @@
+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">
+    <p class="e-content">Test of <b>relative URL resolution</b> with two <img src="photo1.jpg"> images <img src="photo2.jpg"> inside the content</p>
+  </body>
+</html>
diff --git a/tests/data/sanitize.example/photo-relative b/tests/data/sanitize.example/photo-relative
@@ -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>
+  </head>
+  <body class="h-entry">
+    <p class="e-content">Test of <b>relative URL resolution</b> with an photo property</p>
+    <img class="u-photo" src="photo.jpg">
+  </body>
+</html>