adds a bunch of broken tests for #52

lib/XRay/Formats/Format.php

@@ -34,10 +34,8 @@ abstract class Format implements iFormat {
     return [$doc, $xpath];
   }
 
-  protected static function sanitizeHTML($html) {
-    $config = HTMLPurifier_Config::createDefault();
-    $config->set('Cache.DefinitionImpl', null);
-    $config->set('HTML.AllowedElements', [
+  protected static function sanitizeHTML($html, $allowImg=true) {
+    $allowed = [
       'a',
       'abbr',
       'b',
@@ -45,7 +43,6 @@ abstract class Format implements iFormat {
       'del',
       'em',
       'i',
-      'img',
       'q',
       'strike',
       'strong',
@@ -62,7 +59,13 @@ abstract class Format implements iFormat {
       'ul',
       'li',
       'ol'
-    ]);
+    ];
+    if($allowImg)
+      $allowed[] = 'img';
+
+    $config = HTMLPurifier_Config::createDefault();
+    $config->set('Cache.DefinitionImpl', null);
+    $config->set('HTML.AllowedElements', $allowed);
     $def = $config->getHTMLDefinition(true);
     $def->addElement(
       'time',

lib/XRay/Formats/Mf2.php

@@ -220,21 +220,39 @@ class Mf2 extends Format {
       $textContent = $content;
     } elseif(!is_string($content) && is_array($content) && array_key_exists('value', $content)) {
       if(array_key_exists('html', $content)) {
-        $htmlContent = trim(self::sanitizeHTML($content['html']));
-        #$textContent = trim(str_replace("
","\r",strip_tags($htmlContent)));
+        // Only allow images in the content if there is no photo property set
+        if(isset($item['properties']['photo']))
+          $allowImg = false;
+        else
+          $allowImg = true;
+
+        $htmlContent = trim(self::sanitizeHTML($content['html'], $allowImg));
         $textContent = trim(str_replace("
","\r",$content['value']));
       } else {
         $textContent = trim($content['value']);
       }
     }
 
-    $data = [
-      'text' => $textContent
-    ];
-    if($htmlContent && $textContent != $htmlContent) {
-      $data['html'] = $htmlContent;
+    if($textContent || $htmlContent) {
+      $data = [
+        'text' => $textContent
+      ];
+      // Only add HTML content if there is actual content.
+      // If the text content ends up empty, then the HTML should be too
+      // e.g. <div class="e-content"><a href=""><img src="" class="u-photo"></a></div>
+      // should not return content of <a href=""></a>
+      // TODO: still need to remove empty <a> tags when there is other text in the content
+      if($htmlContent && $textContent && $textContent != $htmlContent) {
+        $data['html'] = $htmlContent;
+      }
+
+      if(!$data['text'])
+        return null;
+
+      return $data;
+    } else {
+      return null;
     }
-    return $data;
   }
 
   // Always return arrays, and may contain plaintext content

tests/SanitizeTest.php

@@ -151,4 +151,104 @@ class SanitizeTest extends PHPUnit_Framework_TestCase {
     $this->assertEquals('http://sanitize.example/photo.jpg', $data->data->author->photo);
   }
 
+  public function testPhotoInContent() {
+    // https://github.com/aaronpk/XRay/issues/52
+
+    $url = 'http://sanitize.example/photo-in-content';
+    $response = $this->parse(['url' => $url]);
+
+    $body = $response->getContent();
+    $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);
+  }
+
+  public function testPhotoInContentEmptyAltAttribute() {
+    // https://github.com/aaronpk/XRay/issues/52
+    
+    $url = 'http://sanitize.example/photo-in-content-empty-alt';
+    $response = $this->parse(['url' => $url]);
+
+    $body = $response->getContent();
+    $this->assertEquals(200, $response->getStatusCode());
+    $data = json_decode($body);
+
+    $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);
+  }
+
+  public function testPhotoInContentWithAlt() {
+    // https://github.com/aaronpk/XRay/issues/52
+    
+    $url = 'http://sanitize.example/photo-in-content-with-alt';
+    $response = $this->parse(['url' => $url]);
+
+    $body = $response->getContent();
+    $this->assertEquals(200, $response->getStatusCode());
+    $data = json_decode($body);
+
+    $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);    
+  }
+
+  public function testPhotoInContentWithNoText() {
+    $url = 'http://sanitize.example/cleverdevil';
+    $response = $this->parse(['url' => $url]);
+
+    $body = $response->getContent();
+    $this->assertEquals(200, $response->getStatusCode());
+    $data = json_decode($body);
+
+    $this->assertObjectHasAttribute('name', $data->data);
+    $this->assertEquals('Oh, how well they know me! 🥃', $data->data->name);
+    $this->assertObjectNotHasAttribute('content', $data->data);
+    $this->assertEquals('https://cleverdevil.io/file/5bf2fa91c3d4c592f9978200923cb56e/thumb.jpg', $data->data->photo[0]);
+  }
+
+  public function testPhotosWithAlt() {
+    // https://github.com/microformats/microformats2-parsing/issues/16
+
+    $url = 'http://sanitize.example/photos-with-alt';
+    $response = $this->parse(['url' => $url]);
+
+    $body = $response->getContent();
+    $this->assertEquals(200, $response->getStatusCode());
+    $data = json_decode($body);
+
+    #print_r($data->data);
+
+    $this->assertEquals('🌆 Made it to the first #NPSF #earlygang of the year, did in-betweeners abs, and 6:30 workout with a brutal burnout that was really its own workout. But wow pretty sunrise. Plus 50+ deg F? I’ll take it. #100PDPD#justshowup #darknesstodawn #wakeupthesun #fromwhereirun #NovemberProject #sunrise #latergram #nofilter', $data->data->content->text);
+    $this->assertObjectNotHasAttribute('name', $data->data);
+    $this->assertEquals('https://igx.4sqi.net/img/general/original/476_g7yruXflacsGr7PyVmECefyTBMB_R99zmPQxW7pftzA.jpg', $data->data->photo[0]);
+    $this->assertEquals('https://igx.4sqi.net/img/general/original/476_zM3UgU9JHNhom907Ac_1WCEcUhGOJZaNWGlRmev86YA.jpg', $data->data->photo[1]);
+  }
+
+  public function testEntryWithImgNoImpliedPhoto() {
+    // See https://github.com/microformats/microformats2-parsing/issues/6#issuecomment-357286985
+    // and https://github.com/aaronpk/XRay/issues/52#issuecomment-357269683
+    // and https://github.com/microformats/microformats2-parsing/issues/16
+    $url = 'http://sanitize.example/entry-with-img-no-implied-photo';
+    $response = $this->parse(['url' => $url]);
+
+    $body = $response->getContent();
+    $this->assertEquals(200, $response->getStatusCode());
+    $data = json_decode($body);
+
+    $this->assertObjectNotHasAttribute('photo', $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, which does not have a u-photo class so should not be removed.', $data->data->content->text);
+    $this->assertEquals('This is a photo post with an <code>img</code> tag inside the content, which does not have a u-photo class so should not be removed. <img src="http://target.example.com/photo.jpg" alt="a photo">', $data->data->content->html);
+  }
+
 }

tests/data/sanitize.example/cleverdevil

@@ -0,0 +1,47 @@
+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="col-md-8 col-md-offset-2 h-entry idno-photos idno-object idno-content">
+                    <div>
+                        <div class="p-author author h-card vcard">
+                            <a href="https://cleverdevil.io/profile/cleverdevil" class="u-url icon-container"><img class="u-photo"
+                                                                                                src="https://cleverdevil.io/file/2fa19f964fb8970faaf20b909c69d6cb/thumb.png"/></a>
+                            <a class="p-name fn u-url url"
+                               href="https://cleverdevil.io/profile/cleverdevil">Jonathan LaCour</a>
+                            <a class="u-url" href="https://cleverdevil.io/profile/cleverdevil">
+                                <!-- This is here to force the hand of your MF2 parser --></a>
+                        </div>
+                        <div class="break">&nbsp;</div>
+                    </div>
+                    <div class="datestamp">
+                        <p>
+                            <a class="u-url url" href="https://cleverdevil.io/2018/oh-how-well-they-know-me" rel="permalink">
+                                <time class="dt-published"
+                                      datetime="2018-01-11T23:03:32+00:00">January 11, 2018</time>
+                            </a>
+                        </p>
+                    </div>
+                                        <div class="idno-body">
+                              <h2 class="photo-title p-name"><a
+                href="https://cleverdevil.io/2018/oh-how-well-they-know-me">Oh, how well they know me! 🥃</a>
+        </h2>
+    
+<div class="e-content entry-content">
+
+                <div class="photo-view">
+                <a href="https://cleverdevil.io/file/8ff3c164f54ff736bb5fb64d50e1276c/C66353A6-9AAE-46F2-B9BC-1BA7F5574D54.jpeg" 
+                   data-original-img="https://cleverdevil.io/file/8ff3c164f54ff736bb5fb64d50e1276c/C66353A6-9AAE-46F2-B9BC-1BA7F5574D54.jpeg"
+                   data-title="Oh, how well they know me! 🥃" 
+                   data-footer=""><img src="https://cleverdevil.io/file/5bf2fa91c3d4c592f9978200923cb56e/thumb.jpg" class="u-photo" alt="Oh, how well they know me! 🥃" /></a>
+            </div>
+    </body>
+</html>

tests/data/sanitize.example/entry-with-img-no-implied-photo

@@ -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 p-name">This is a photo post with an <code>img</code> tag inside the content, which does not have a u-photo class so should not be removed. <img src="http://target.example.com/photo.jpg" alt="a photo"></p>
+  </body>
+</html>

tests/data/sanitize.example/photo-in-content

@@ -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 p-name">This is a photo post with an <code>img</code> tag inside the content. <img class="u-photo" src="http://target.example.com/photo.jpg"></p>
+  </body>
+</html>

tests/data/sanitize.example/photo-in-content-empty-alt

@@ -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 p-name">This is a photo post with an <code>img</code> tag inside the content. <img class="u-photo" src="http://target.example.com/photo.jpg" alt=""></p>
+  </body>
+</html>

tests/data/sanitize.example/photo-in-content-with-alt

@@ -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 p-name">This is a photo post with an <code>img</code> tag inside the content. <img class="u-photo" src="http://target.example.com/photo.jpg" alt="a photo"></p>
+  </body>
+</html>

tests/data/sanitize.example/photos-with-alt

@@ -0,0 +1,73 @@
+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>
+<body class="post hentry h-entry as-note">
+<header><div class="header">
+<h1><a href="../" rel="author home">tantek.com</a></h1>
+
+<form class="search" action="https://duckduckgo.com/html/" method="get">
+<input type="hidden" name="type" value="list" />
+<input type="search" value="site:tantek.com " name="q" />
+<button type="submit">Search</button>
+</form>
+
+</div></header>
+
+<div class="sidestuff">
+<ul class="snav">
+<li><a href="003/t2/first-yoga-class-this-year" id="previtem" title="View the previous (older) item in the stream." rel="prev"><abbr>&#x2190;</abbr></a></li><li><a href="005/t1/yesterday-rest-homework-day" id="nextitem" title="View the next (newer) item in the stream" rel="next"><abbr>&#x2192;</abbr></a></li>
+</ul>
+</div>
+<a href="../" class="p-author h-card author-icon" rel="author" title="Tantek Çelik"><img src="../logo.jpg" alt="Tantek Çelik" /></a>
+
+<p class="p-name entry-title e-content entry-content article"><a class="auto-link figure u-photo" href="https://igx.4sqi.net/img/general/original/476_g7yruXflacsGr7PyVmECefyTBMB_R99zmPQxW7pftzA.jpg"><img class="auto-embed" alt="a jpg. " src="https://igx.4sqi.net/img/general/width960/476_g7yruXflacsGr7PyVmECefyTBMB_R99zmPQxW7pftzA.jpg"/></a><a class="auto-link figure u-photo" href="https://igx.4sqi.net/img/general/original/476_zM3UgU9JHNhom907Ac_1WCEcUhGOJZaNWGlRmev86YA.jpg"><img class="auto-embed" alt="a jpg. " src="https://igx.4sqi.net/img/general/width960/476_zM3UgU9JHNhom907Ac_1WCEcUhGOJZaNWGlRmev86YA.jpg"/></a>🌆 Made it to the first #<span class="p-category auto-tag">NPSF</span> #<span class="p-category auto-tag">earlygang</span> of the year, did in-betweeners abs, and 6:30 workout with a brutal burnout that was really its own workout. But wow pretty sunrise. Plus 50+ deg F? I’ll take it. #<span class="p-category auto-tag">100PDPD</span><br class="auto-break"/><br class="auto-break"/>#<span class="p-category auto-tag">justshowup</span> #<span class="p-category auto-tag">darknesstodawn</span> #<span class="p-category auto-tag">wakeupthesun</span> #<span class="p-category auto-tag">fromwhereirun</span> #<span class="p-category auto-tag">NovemberProject</span> #<span class="p-category auto-tag">sunrise</span> #<span class="p-category auto-tag">latergram</span> #<span class="p-category auto-tag">nofilter</span></p><span class="info footer">
+  <span class="dt-published published dt-updated updated">
+  <time class="value" datetime="21:08-0800">21:08</time> on <time class="value">2018-01-03</time>  </span>
+ <span class="lt">
+ (ttk.me t4sE3) </span>
+  using <span class="using">BBEdit</span> </span> 
+<div class="info footer">
+<form action="http://tantek.com/2018/003/t3/first-npsf-earlygang-year"><div>
+<label><span class="lt">URL:</span>
+<input class="u-url url u-uid uid bookmark" type="url" size="70" style="max-width:100%" value="http://tantek.com/2018/003/t3/first-npsf-earlygang-year" />
+</label>
+</div></form>
+</div>
+<div class="info footer">
+<a class="u-syndication" rel="syndication" style="float:right; clear:right; margin-left:1em; height:2em"
+   href="https://www.instagram.com/p/BdhBVjNFi35">
+   <img src="/icon/instagram.png" style="vertical-align:-30%;margin:-3px -5px -4px 0" alt=""/> 
+   View on Instagram</a>
+<a class="u-syndication" rel="syndication" style="float:right; clear:right; margin-left:1em; height:2em"
+   href="https://www.facebook.com/10103634146088903">
+   <img src="/icon/facebook.png" style="vertical-align:-30%;margin:-3px -10px -4px 0" alt=""/> 
+   View on Facebook</a>
+<a class="u-syndication" rel="syndication" style="float:right; clear:right; margin-left:1em; height:2em"
+   href="https://www.flickr.com/photos/tantek/24621899937/">
+   <span style="display:inline-block; margin-top:.1em;  font-size:1.8em;vertical-align:-5%">••</span>
+   View on Flickr</a>
+<a class="u-syndication" rel="syndication" style="float:right; clear:right; height:2em"
+   href="https://twitter.com/t/status/948789001854517248">
+   <img src="/icon/twitter.png" style="vertical-align:-30%;margin:-3px -6px -4px 0" alt=""/> 
+   View 
+      on Twitter
+</a><div style="line-height:2.6em; font-size:1em;">
+<indie-action do="like" with="http://tantek.com/2018/003/t3/first-npsf-earlygang-year">
+<a class="action favorite" target="_blank" href="https://twitter.com/intent/favorite?tweet_id=948789001854517248"><img src="https://g.twimg.com/dev/documentation/image/like-icon-16.png" style="vertical-align:-15%" alt=""/> Like</a>
+</indie-action>
+<indie-action do="repost" with="http://tantek.com/2018/003/t3/first-npsf-earlygang-year">
+<a class="action reply" target="_blank" href="https://twitter.com/intent/retweet?tweet_id=948789001854517248"><img src="https://g.twimg.com/dev/documentation/image/retweet-icon-16.png" style="vertical-align:0%;margin:0 -0.1em 0 0.5em" alt=""/> Repost</a>
+</indie-action>
+<indie-action do="reply" with="http://tantek.com/2018/003/t3/first-npsf-earlygang-year">
+<a class="action reply" target="_blank" href="https://twitter.com/intent/tweet?in_reply_to=948789001854517248"><img src="https://g.twimg.com/dev/documentation/image/reply-icon-16.png" style="vertical-align:0%;margin:0 -0.1em 0 0.5em" alt=""/> Reply</a>
+</indie-action>
+</div>
+
+</div>
+ </li>
+</body>
+</html>