Find and move any misplaced comment nodes

bigsweater · colinodell · commit 1cb16388bb0f · 2021-11-06T01:35:25.000-04:00
diff --git a/src/HtmlConverter.php b/src/HtmlConverter.php
@@ -121,13 +121,45 @@ private function createDOMDocument(string $html): \DOMDocument
         $document->loadHTML('<?xml encoding="UTF-8">' . $html);
         $document->encoding = 'UTF-8';
 
+        $this->replaceMisplacedComments($document);
+
         if ($this->getConfig()->getOption('suppress_errors')) {
             \libxml_clear_errors();
         }
 
         return $document;
     }
 
+    /**
+     * Finds any comment nodes outside <html> element and moves them into <body>.
+     *
+     * @see https://github.com/thephpleague/html-to-markdown/issues/212
+     * @see https://3v4l.org/7bC33
+     */
+    private function replaceMisplacedComments(\DOMDocument $document): void
+    {
+        // Find ny comment nodes at the root of the document.
+        $misplacedComments = (new \DOMXPath($document))->query('/comment()');
+        if ($misplacedComments === false) {
+            return;
+        }
+
+        $body = $document->getElementsByTagName('body')->item(0);
+        if ($body === null) {
+            return;
+        }
+
+        // Loop over comment nodes in reverse so we put them inside <body> in
+        // their original order.
+        for ($index = $misplacedComments->length - 1; $index >= 0; $index--) {
+            if ($body->firstChild === null) {
+                $body->insertBefore($misplacedComments[$index]);
+            } else {
+                $body->insertBefore($misplacedComments[$index], $body->firstChild);
+            }
+        }
+    }
+
     /**
      * Convert Children
      *
diff --git a/tests/HtmlConverterTest.php b/tests/HtmlConverterTest.php
@@ -327,14 +327,21 @@ public function testStripUnmarkdownable(): void
     public function testStripComments(): void
     {
         $this->assertHtmlGivesMarkdown('<p>Test</p><!-- Test comment -->', 'Test');
+        $this->assertHtmlGivesMarkdown('<!-- Test comment --><p>Test</p>', 'Test');
         $this->assertHtmlGivesMarkdown('<p>Test</p><!-- Test comment -->', 'Test', ['strip_tags' => true]);
+        $this->assertHtmlGivesMarkdown('<!-- Test comment --><p>Test</p>', 'Test', ['strip_tags' => true]);
     }
 
     public function testPreserveComments(): void
     {
-        $this->assertHtmlGivesMarkdown('<p>Test</p><!-- Test comment -->', "Test\n\n<!-- Test comment -->", ['preserve_comments' => true]);
         $this->assertHtmlGivesMarkdown('<p>Test</p><!-- more -->', "Test\n\n<!-- more -->", ['preserve_comments' => ['more']]);
         $this->assertHtmlGivesMarkdown('<p>Test</p><!-- Test comment --><!-- more -->', "Test\n\n<!-- more -->", ['preserve_comments' => ['more']]);
+        $this->assertHtmlGivesMarkdown('<!-- Test comment --><p>Test</p><!-- Test comment -->', "<!-- Test comment -->Test\n\n<!-- Test comment -->", ['preserve_comments' => true]);
+    }
+
+    public function testPreserveCommentOrder(): void
+    {
+        $this->assertHtmlGivesMarkdown('<!-- 1 --><!-- 2 --><p>Test</p><!-- 3 -->', "<!-- 1 --><!-- 2 -->Test\n\n<!-- 3 -->", ['preserve_comments' => true]);
     }
 
     public function testPreserveWhitespace(): void

Original file line number	Diff line number	Diff line change
`@@ -327,14 +327,21 @@ public function testStripUnmarkdownable(): void`
`327`	`327`	`public function testStripComments(): void`
`328`	`328`	`{`
`329`	`329`	`$this->assertHtmlGivesMarkdown('<p>Test</p><!-- Test comment -->', 'Test');`
	`330`	`+ $this->assertHtmlGivesMarkdown('<!-- Test comment --><p>Test</p>', 'Test');`
`330`	`331`	`$this->assertHtmlGivesMarkdown('<p>Test</p><!-- Test comment -->', 'Test', ['strip_tags' => true]);`
	`332`	`+ $this->assertHtmlGivesMarkdown('<!-- Test comment --><p>Test</p>', 'Test', ['strip_tags' => true]);`
`331`	`333`	`}`
`332`	`334`
`333`	`335`	`public function testPreserveComments(): void`
`334`	`336`	`{`
`335`		`- $this->assertHtmlGivesMarkdown('<p>Test</p><!-- Test comment -->', "Test\n\n<!-- Test comment -->", ['preserve_comments' => true]);`
`336`	`337`	`$this->assertHtmlGivesMarkdown('<p>Test</p><!-- more -->', "Test\n\n<!-- more -->", ['preserve_comments' => ['more']]);`
`337`	`338`	`$this->assertHtmlGivesMarkdown('<p>Test</p><!-- Test comment --><!-- more -->', "Test\n\n<!-- more -->", ['preserve_comments' => ['more']]);`
	`339`	`+ $this->assertHtmlGivesMarkdown('<!-- Test comment --><p>Test</p><!-- Test comment -->', "<!-- Test comment -->Test\n\n<!-- Test comment -->", ['preserve_comments' => true]);`
	`340`	`+ }`
	`341`	`+`
	`342`	`+ public function testPreserveCommentOrder(): void`
	`343`	`+ {`
	`344`	`+ $this->assertHtmlGivesMarkdown('<!-- 1 --><!-- 2 --><p>Test</p><!-- 3 -->', "<!-- 1 --><!-- 2 -->Test\n\n<!-- 3 -->", ['preserve_comments' => true]);`
`338`	`345`	`}`
`339`	`346`
`340`	`347`	`public function testPreserveWhitespace(): void`