DEP Use masterminds/html5 for HTMLValue

silverstripe · Jan 16, 2023 · df04994 · df04994
1 parent 6d45425
commit df04994
Show file tree

Hide file tree

Showing 5 changed files with 54 additions and 64 deletions.
diff --git a/_config/html.yml b/_config/html.yml
@@ -2,10 +2,7 @@
 Name: corehtml
 ---
 SilverStripe\Core\Injector\Injector:
-  SilverStripe\View\Parsers\HTMLValue:
-    class: SilverStripe\View\Parsers\HTML4Value
   # Shorthand
-  HTMLValue: '%$SilverStripe\View\Parsers\HTMLValue'
   SilverStripe\Forms\HTMLEditor\HTMLEditorConfig:
     class: SilverStripe\Forms\HTMLEditor\TinyMCEConfig
   SilverStripe\Forms\HTMLEditor\TinyMCEScriptGenerator: '%$SilverStripe\Forms\HTMLEditor\TinyMCECombinedGenerator'

diff --git a/composer.json b/composer.json
@@ -29,6 +29,7 @@
         "embed/embed": "^4.4.7",
         "league/csv": "^9.8.0",
         "m1/env": "^2.2.0",
+        "masterminds/html5": "^2.7",
         "monolog/monolog": "^3.2.0",
         "nikic/php-parser": "^4.15.0",
         "psr/container": "^2.0",

diff --git a/src/View/Parsers/HTML4Value.php b/src/View/Parsers/HTML4Value.php
diff --git a/src/View/Parsers/HTMLValue.php b/src/View/Parsers/HTMLValue.php
@@ -4,6 +4,7 @@
 
 use SilverStripe\Core\Convert;
 use SilverStripe\View\ViewableData;
+use Masterminds\HTML5;
 use DOMNodeList;
 use DOMXPath;
 use DOMDocument;
@@ -12,14 +13,10 @@
  * This class handles the converting of HTML fragments between a string and a DOMDocument based
  * representation.
  *
- * It's designed to allow dependency injection to replace the standard HTML4 version with one that
- * handles XHTML or HTML5 instead
- *
  * @mixin DOMDocument
  */
-abstract class HTMLValue extends ViewableData
+class HTMLValue extends ViewableData
 {
-
     public function __construct($fragment = null)
     {
         if ($fragment) {
@@ -28,7 +25,25 @@ public function __construct($fragment = null)
         parent::__construct();
     }
 
-    abstract public function setContent($fragment);
+    /**
+     * @param string $content
+     * @return bool
+     */
+    public function setContent($content)
+    {
+        $content = preg_replace('#</?(html|head|body)[^>]*>#si', '', $content);
+        $html5 = new HTML5();
+        $document = $html5->loadHTML(
+            '<html><head><meta http-equiv="content-type" content="text/html; charset=utf-8"></head>' .
+            "<body>$content</body></html>"
+        );
+        if ($document) {
+            $this->setDocument($document);
+            return true;
+        }
+        $this->valid = false;
+        return false;
+    }
 
     /**
      * @return string
@@ -77,6 +92,9 @@ public function getContent()
         // Possible alternative solution: http://stackoverflow.com/questions/2142120/php-encoding-with-domdocument
         $from = mb_convert_encoding('&nbsp;', 'utf-8', 'html-entities');
         $res = str_replace($from ?? '', '&nbsp;', $res ?? '');
+
+        // remove closing </meta> tags when they're not required
+        $res = preg_replace('#(<meta[^>]*>)</meta>#', '$1', $res);
 
         return $res;
     }

diff --git a/tests/php/View/Parsers/HTML4ValueTest.php → tests/php/View/Parsers/HTMLValueTest.php b/tests/php/View/Parsers/HTML4ValueTest.php → tests/php/View/Parsers/HTMLValueTest.php
@@ -3,13 +3,13 @@
 namespace SilverStripe\View\Tests\Parsers;
 
 use SilverStripe\Dev\SapphireTest;
-use SilverStripe\View\Parsers\HTML4Value;
+use SilverStripe\View\Parsers\HTMLValue;
 
-class HTML4ValueTest extends SapphireTest
+class HTMLValueTest extends SapphireTest
 {
     public function testInvalidHTMLSaving()
     {
-        $value = new HTML4Value();
+        $value = new HTMLValue();
 
         $invalid = [
             '<p>Enclosed Value</p></p>'
@@ -32,35 +32,40 @@ public function testInvalidHTMLSaving()
 
     public function testUtf8Saving()
     {
-        $value = new HTML4Value();
+        $value = new HTMLValue();
 
         $value->setContent('<p>ö ß ā い 家</p>');
         $this->assertEquals('<p>ö ß ā い 家</p>', $value->getContent());
     }
 
-    public function testInvalidHTMLTagNames()
+    /**
+     * @dataProvider provideInvalidHTMLTagNames
+     */
+    public function testInvalidHTMLTagNames(string $input)
     {
-        $value = new HTML4Value();
+        $value = new HTMLValue();
+        $value->setContent($input);
+        $link = $value->getElementsByTagName('a')->item(0);
+        $this->assertNotNull($link);
+        $this->assertEquals(
+            'test-link',
+            $link->getAttribute('href'),
+            'Link data can be extracted from malformed HTML'
+        );
+    }
 
-        $invalid = [
-            '<p><div><a href="test-link"></p></div>',
-            '<html><div><a href="test-link"></a></a></html_>',
-            '""\'\'\'"""\'""<<<>/</<htmlbody><a href="test-link"<<>'
+    public function provideInvalidHTMLTagNames()
+    {
+        return [
+            ['<p><div><a href="test-link"></p></div>'],
+            ['<html><div><a href="test-link"></a></a></html_>'],
+            ['<x<b<c c=\'"\'><htmlbody><a href="test-link">']
         ];
-
-        foreach ($invalid as $input) {
-            $value->setContent($input);
-            $this->assertEquals(
-                'test-link',
-                $value->getElementsByTagName('a')->item(0)->getAttribute('href'),
-                'Link data can be extracted from malformed HTML'
-            );
-        }
     }
 
     public function testMixedNewlines()
     {
-        $value = new HTML4Value();
+        $value = new HTMLValue();
 
         $value->setContent("<p>paragraph</p>\n<ul><li>1</li>\r\n</ul>");
         $this->assertEquals(
@@ -72,18 +77,18 @@ public function testMixedNewlines()
 
     public function testAttributeEscaping()
     {
-        $value = new HTML4Value();
+        $value = new HTMLValue();
 
         $value->setContent('<a href="[]"></a>');
         $this->assertEquals('<a href="[]"></a>', $value->getContent(), "'[' character isn't escaped");
 
-        $value->setContent('<a href="&quot;"></a>');
-        $this->assertEquals('<a href="&quot;"></a>', $value->getContent(), "'\"' character is escaped");
+        $value->setContent('<a href="\'&quot;"></a>');
+        $this->assertEquals('<a href="\'&quot;"></a>', $value->getContent(), "'\"' character is escaped");
     }
 
     public function testGetContent()
     {
-        $value = new HTML4Value();
+        $value = new HTMLValue();
 
         $value->setContent('<p>This is valid</p>');
         $this->assertEquals('<p>This is valid</p>', $value->getContent(), "Valid content is returned");