run name/content dedupe before munging HTML

fix for #53
aaronpk · Jan 12, 2018 · 66adfbe · 66adfbe
1 parent 4477039
commit 66adfbe
Show file tree

Hide file tree

Showing 3 changed files with 39 additions and 8 deletions.
diff --git a/lib/XRay/Formats/Format.php b/lib/XRay/Formats/Format.php
@@ -84,8 +84,16 @@ protected static function sanitizeHTML($html, $allowImg=true) {
     return trim($sanitized);
   }
 
+  // Return a plaintext version of the input HTML
   protected static function stripHTML($html) {
-    return trim(strip_tags($html));
+    $config = HTMLPurifier_Config::createDefault();
+    $config->set('Cache.DefinitionImpl', null);
+    $config->set('HTML.AllowedElements', ['br']);
+    $purifier = new HTMLPurifier($config);
+    $sanitized = $purifier->purify($html);
+    $sanitized = str_replace("&#xD;","\r",$sanitized);
+    $sanitized = html_entity_decode($sanitized);
+    return trim(str_replace('<br>',"\n", $sanitized));
   }
 
 

diff --git a/lib/XRay/Formats/Mf2.php b/lib/XRay/Formats/Mf2.php
@@ -227,7 +227,8 @@ private static function parseHTMLValue($property, $item) {
           $allowImg = true;
 
         $htmlContent = trim(self::sanitizeHTML($content['html'], $allowImg));
-        $textContent = trim(str_replace("&#xD;","\r",$content['value']));
+        #$textContent = trim(str_replace("&#xD;","\r",$content['value']));
+        $textContent = trim(self::stripHTML($htmlContent));
       } else {
         $textContent = trim($content['value']);
       }
@@ -339,10 +340,13 @@ private static function determineNameAndContent($item, &$data) {
     $textContent = null;
     $htmlContent = null;
 
-    $content = self::parseHTMLValue('content', $item);
-    if($content) {
+    $content = self::getHTMLValue($item, 'content');
+
+    if(is_string($content)) {
+      $textContent = $content;
+    } elseif($content) {
       $htmlContent = array_key_exists('html', $content) ? $content['html'] : null;
-      $textContent = array_key_exists('text', $content) ? $content['text'] : null;
+      $textContent = array_key_exists('value', $content) ? $content['value'] : null;
     }
 
     if($content) {
@@ -365,8 +369,9 @@ private static function determineNameAndContent($item, &$data) {
 
     // If there is content, always return the plaintext content, and return HTML content if it's different
     if($content) {
+      $content = self::parseHTMLValue('content', $item);
       $data['content']['text'] = $content['text'];
-      if(array_key_exists('html', $content))
+      if(isset($content['html']))
         $data['content']['html'] = $content['html'];
     }    
   }
@@ -762,6 +767,20 @@ private static function getPlaintext($mf2, $k, $fallback=null) {
     return $fallback;
   }
 
+  private static function getHTMLValue($mf2, $k, $fallback=null) {
+    // Return an array with html and value if the value is html, otherwise return a string
+    if(!empty($mf2['properties'][$k]) and is_array($mf2['properties'][$k])) {
+      // $mf2['properties'][$v] will always be an array since the input was from the mf2 parser
+      $value = $mf2['properties'][$k][0];
+      if(is_string($value)) {
+        return $value;
+      } elseif(isset($value['html'])) {
+        return $value;
+      }
+    }
+    return $fallback;
+  }
+
   private static function getPlaintextValues($mf2, $k, $values=[]) {
     if(!empty($mf2['properties'][$k]) and is_array($mf2['properties'][$k])) {
       foreach($mf2['properties'][$k] as $value) {

diff --git a/tests/SanitizeTest.php b/tests/SanitizeTest.php
@@ -151,7 +151,7 @@ public function testSanitizeEmailAuthorURL() {
     $this->assertEquals('http://sanitize.example/photo.jpg', $data->data->author->photo);
   }
 
-  public function testPhotoInContent() {
+  public function testPhotoInContentNoAlt() {
     // https://github.com/aaronpk/XRay/issues/52
 
     $url = 'http://sanitize.example/photo-in-content';
@@ -161,7 +161,11 @@ public function testPhotoInContent() {
     $this->assertEquals(200, $response->getStatusCode());
     $data = json_decode($body);
 
-    #print_r($data->data);
+    $this->assertObjectNotHasAttribute('name', $data->data);
+    $this->assertEquals('http://target.example.com/photo.jpg', $data->data->photo[0]);
+    $this->assertEquals('This is a photo post with an img tag inside the content.', $data->data->content->text);
+    $this->assertEquals('This is a photo post with an <code>img</code> tag inside the content.', $data->data->content->html);
+  }
 
     $this->assertObjectNotHasAttribute('name', $data->data);
     $this->assertEquals('http://target.example.com/photo.jpg', $data->data->photo[0]);