Merge pull request #524 from ampproject/fix/522-uncaught-domexception

ediamin · web-flow · commit 517f67b65fc6 · 2022-05-18T10:16:44.000+06:00
Fix error when source is missing closing head tag
diff --git a/src/Dom/Document.php b/src/Dom/Document.php
@@ -566,7 +566,13 @@ private function normalizeDocumentStructure($content)
         } elseif (! preg_match(self::HTML_STRUCTURE_BODY_END_TAG, $content, $matches)) {
             // Only <body> missing.
             // @todo This is an expensive regex operation, look into further optimization.
-            $content = preg_replace(self::HTML_STRUCTURE_HEAD_TAG, '$0<body>', $content, 1);
+            $content = preg_replace(self::HTML_STRUCTURE_HEAD_TAG, '$0<body>', $content, 1, $count);
+
+            // Closing </head> tag is missing.
+            if (! $count) {
+                $content = $content . '</head><body>';
+            }
+
             $content .= '</body>';
         }
 
diff --git a/tests/Dom/DocumentTest.php b/tests/Dom/DocumentTest.php
@@ -11,6 +11,7 @@
 use AmpProject\Html\Tag;
 use AmpProject\Tests\MarkupComparison;
 use AmpProject\Tests\TestCase;
+use AmpProject\Tests\TestMarkup;
 use AmpProject\Validator\Spec\CssRuleset\AmpNoTransformed;
 use AmpProject\Validator\Spec\SpecRule;
 use DOMNode;
@@ -752,6 +753,17 @@ public function testInvalidHeadNodes()
         $this->assertEquals('meta', $dom->head->firstChild->tagName);
         $this->assertNull($dom->head->firstChild->nextSibling);
         $this->assertEquals(6, $dom->body->childNodes->length);
+
+        // Test with missing closing head tag.
+        $expected = TestMarkup::DOCTYPE . '<html>'
+            . '<head>' . TestMarkup::META_CHARSET . '<meta name="foo" content="bar"></head>'
+            . '<body><foo></foo><bar></bar></body></html>';
+
+        $html     = '<html><head><meta name="foo" content="bar"><foo></foo><bar></bar>';
+        $document = Document::fromHtml($html);
+        $output   = $document->saveHTML();
+
+        $this->assertEqualMarkup($expected, $output);
     }
 
     /**
