[BUGFIX] Check existing Content-Type is in <head>

If a valid `Content-Type` `<meta>` element is present, DOM conversion will create a `<head>` element for it even without an explicit `<head>` tag in the HTML. However, to be valid, it must not be in the `<body>` element. As well as with an explicit `<body>` start tag, the `<body>` element also begins whenever a start tag for an element which cannot be in the `<head>` is encountered. This is now checked. Fixes #923.
MyIntervals · Mar 23, 2021 · 57aa616 · 57aa616
1 parent 1efffbb
commit 57aa616
Show file tree

Hide file tree

Showing 4 changed files with 160 additions and 9 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -16,6 +16,8 @@ This project adheres to [Semantic Versioning](https://semver.org/).
 ### Removed
 
 ### Fixed
+- Ignore `http-equiv` `Content-Type` in `<body>`
+  ([#961](https://github.com/MyIntervals/emogrifier/pull/961))
 - Allow "Content-Type" in content
   ([#959](https://github.com/MyIntervals/emogrifier/pull/959))
 

diff --git a/psalm.baseline.xml b/psalm.baseline.xml
@@ -65,12 +65,13 @@
     </PossiblyNullReference>
   </file>
   <file src="src/HtmlProcessor/AbstractHtmlProcessor.php">
-    <MixedOperand occurrences="6">
+    <MixedOperand occurrences="7">
       <code>static::PHP_UNRECOGNIZED_VOID_TAGNAME_MATCHER</code>
       <code>static::DEFAULT_DOCUMENT_TYPE</code>
       <code>static::CONTENT_TYPE_META_TAG</code>
       <code>static::CONTENT_TYPE_META_TAG</code>
       <code>static::CONTENT_TYPE_META_TAG</code>
+      <code>static::TAGNAME_ALLOWED_BEFORE_BODY_MATCHER</code>
       <code>static::PHP_UNRECOGNIZED_VOID_TAGNAME_MATCHER</code>
     </MixedOperand>
     <PossiblyNullPropertyAssignmentValue occurrences="1">

diff --git a/src/HtmlProcessor/AbstractHtmlProcessor.php b/src/HtmlProcessor/AbstractHtmlProcessor.php
@@ -32,6 +32,15 @@ abstract class AbstractHtmlProcessor
      */
     protected const PHP_UNRECOGNIZED_VOID_TAGNAME_MATCHER = '(?:command|embed|keygen|source|track|wbr)';
 
+    /**
+     * Regular expression part to match tag names that may appear before the start of the `<body>` element.  A start tag
+     * for any other element would implicitly start the `<body>` element due to tag omission rules.
+     *
+     * @var string
+     */
+    protected const TAGNAME_ALLOWED_BEFORE_BODY_MATCHER
+        = '(?:html|head|base|command|link|meta|noscript|script|style|template|title)';
+
     /**
      * @var \DOMDocument|null
      */
@@ -271,10 +280,7 @@ private function normalizeDocumentType(string $html): string
      */
     private function addContentTypeMetaTag(string $html): string
     {
-        $contentTypeMetaTagMatchCount
-            = \preg_match('%<meta(?=\\s)[^>]*\\shttp-equiv=(["\']?+)Content-Type\\g{-1}[\\s/>]%i', $html);
-        $hasContentTypeMetaTag = \is_int($contentTypeMetaTagMatchCount) && $contentTypeMetaTagMatchCount > 0;
-        if ($hasContentTypeMetaTag) {
+        if ($this->hasContentTypeMetaTagInHead($html)) {
             return $html;
         }
 
@@ -302,6 +308,110 @@ private function addContentTypeMetaTag(string $html): string
         return $reworkedHtml;
     }
 
+    /**
+     * Tests whether the given HTML has a valid `Content-Type` metadata element within the `<head>` element.  Due to tag
+     * omission rules, HTML parsers are expected to end the `<head>` element and start the `<body>` element upon
+     * encountering a start tag for any element which is permitted only within the `<body>`.
+     *
+     * @param string $html
+     *
+     * @return bool
+     */
+    private function hasContentTypeMetaTagInHead(string $html): bool
+    {
+        $contentTypeMetaTagMatchCount = \preg_match(
+            '%<meta(?=\\s)[^>]*\\shttp-equiv=(["\']?+)Content-Type\\g{-1}[\\s/>]%i',
+            $html,
+            $matches,
+            PREG_OFFSET_CAPTURE
+        );
+        if (\is_int($contentTypeMetaTagMatchCount) && $contentTypeMetaTagMatchCount > 0) {
+            /** @psalm-var array<int, array{0:string, 1:int}> $matches */
+            $matchPosition = $matches[0][1];
+            $htmlBefore = \substr($html, 0, $matchPosition);
+            $hasContentTypeMetaTagInHead = !$this->hasEndOfHeadElement($htmlBefore);
+        } else {
+            $hasContentTypeMetaTagInHead = false;
+        }
+
+        return $hasContentTypeMetaTagInHead;
+    }
+
+    /**
+     * Tests whether the `<head>` element ends within the given HTML.  Due to tag omission rules, HTML parsers are
+     * expected to end the `<head>` element and start the `<body>` element upon encountering a start tag for any element
+     * which is permitted only within the `<body>`.
+     *
+     * @param string $html
+     *
+     * @return bool
+     */
+    private function hasEndOfHeadElement(string $html): bool
+    {
+        $headEndTagMatchCount
+            = \preg_match('%<(?!' . static::TAGNAME_ALLOWED_BEFORE_BODY_MATCHER . '[\\s/>])\\w|</head>%i', $html);
+        if (\is_int($headEndTagMatchCount) && $headEndTagMatchCount > 0) {
+            // An exception to the implicit end of the `<head>` is any content within a `<template>` element, as well in
+            // comments.  As an optimization, this is only checked for if a potential `<head>` end tag is found.
+            try {
+                $htmlWithoutCommentsOrTemplates = $this->removeHtmlTemplateElements($this->removeHtmlComments($html));
+                if ($htmlWithoutCommentsOrTemplates === $html) {
+                    $hasEndOfHeadElement = true;
+                } else {
+                    $hasEndOfHeadElement = $this->hasEndOfHeadElement($htmlWithoutCommentsOrTemplates);
+                }
+            } catch (\RuntimeException $ex) {
+                // If something unexpected occurs, assume the tag(s) that would end the `<head>` element were within a
+                // `<template>` or comment and thus do not end it.
+                \trigger_error($ex->getMessage());
+                $hasEndOfHeadElement = false;
+            }
+        } else {
+            $hasEndOfHeadElement = false;
+        }
+
+        return $hasEndOfHeadElement;
+    }
+
+    /**
+     * Removes comments from the given HTML, including any unclosed, for which the remainder of the string is removed.
+     *
+     * @param string $html
+     *
+     * @return string
+     *
+     * @throws \RuntimeException
+     */
+    private function removeHtmlComments(string $html): string
+    {
+        $result = \preg_replace('/<!--[^-]*+(?:-(?!->)[^-]*+)*+(?:-->|$)/', '', $html);
+        if (!\is_string($result)) {
+            throw new \RuntimeException('Internal PCRE error', 1616521475);
+        }
+
+        return $result;
+    }
+
+    /**
+     * Removes `<template>` elements from the given HTML, including any without an end tag, for which the remainder of
+     * the string is removed.
+     *
+     * @param string $html
+     *
+     * @return string
+     *
+     * @throws \RuntimeException
+     */
+    private function removeHtmlTemplateElements(string $html): string
+    {
+        $result = \preg_replace('%<template[\\s>][^<]*+(?:<(?!/template>)[^<]*+)*+(?:</template>|$)%i', '', $html);
+        if (!\is_string($result)) {
+            throw new \RuntimeException('Internal PCRE error', 1616519652);
+        }
+
+        return $result;
+    }
+
     /**
      * Makes sure that any self-closing tags not recognized as such by PHP's DOMDocument implementation have a
      * self-closing slash.

diff --git a/tests/Unit/HtmlProcessor/AbstractHtmlProcessorTest.php b/tests/Unit/HtmlProcessor/AbstractHtmlProcessorTest.php
@@ -442,12 +442,16 @@ public function provideContentWithoutHeadTag(): array
             'META element with Content-Type as a value' => ['<meta name="description" content="Content-Type">'],
             'BODY element with Content-Type in text' => ['<body>Content-Type</body>'],
             'body content only with Content-Type in text' => ['<p>Content-Type</p>'],
-            // broken: BODY element containing Content-Type META tag
-            // broken: body content only with Content-Type META tag
+            'http-equiv META element within BODY (not allowed)'
+                => ['<body><meta http-equiv="Content-Type" content="text/html; charset=utf-8"></body>'],
+            'http-equiv META element after P (implicit BODY, not allowed)'
+                => ['<p>hello</p><meta http-equiv="Content-Type" content="text/html; charset=utf-8">'],
             'http-equiv META element within P (not allowed)'
                 => ['<p><meta http-equiv="Content-Type" content="text/html; charset=utf-8"></p>'],
             'viewport META element within P (allowed)'
                 => ['<p><meta name="viewport" content="width=device-width, initial-scale=1.0"></p>'],
+            'http-equiv META element within HEADER (not allowed)'
+                => ['<header><meta http-equiv="Content-Type" content="text/html; charset=utf-8"></header>'],
         ];
     }
 
@@ -480,13 +484,22 @@ public function provideContentWithHeadTag(): array
             'HEAD element with attribute' => ['<head lang="en"></head>'],
             'HEAD element and HEADER element' => ['<head></head><header></header>'],
             'HEAD element with Content-Type in comment' => ['<head><!-- Content-Type --></head>'],
-            'HEAD element with Content-Type as META value' => ['<meta name="description" content="Content-Type">'],
+            'HEAD element with Content-Type as META value'
+                => ['<head><meta name="description" content="Content-Type"></head>'],
             'with BODY element with Content-Type in text' => ['<head></head><body>Content-Type</body>'],
             'with implicit body content with Content-Type in text' => ['<head></head><p>Content-Type</p>'],
             'with BODY element containing Content-Type META tag'
                 => ['<head></head><body><meta http-equiv="Content-Type" content="text/html; charset=utf-8"></body>'],
             'with implicit body content with Content-Type META tag'
                 => ['<head></head><p>hello</p><meta http-equiv="Content-Type" content="text/html; charset=utf-8">'],
+            'with end tag omitted and BODY element containing Content-Type META tag'
+                => ['<head><body><meta http-equiv="Content-Type" content="text/html; charset=utf-8"></body>'],
+            'with end tag omitted and implicit body content with Content-Type META tag'
+                => ['<head><p>hello</p><meta http-equiv="Content-Type" content="text/html; charset=utf-8">'],
+            'with Content-Type META tag after end tag'
+                => ['<head></head><meta http-equiv="Content-Type" content="text/html; charset=utf-8">'],
+            'with Content-Type META tag after uppercase end tag'
+                => ['<HEAD></HEAD><meta http-equiv="Content-Type" content="text/html; charset=utf-8">'],
         ];
     }
 
@@ -718,8 +731,16 @@ public function addsMissingContentTypeMetaTagOnlyOnce(string $html): void
 
         $result = $subject->render();
 
+        $headEndPosition = \stripos($result, '</head>');
+        $resultBeforeHeadEnd = $headEndPosition !== false ? \substr($result, 0, $headEndPosition) : $result;
+        // PHP DOM does not understand `<header>` element so does not know it would implicitly start `<body>`.
+        $headerStartPosition = \stripos($resultBeforeHeadEnd, '<header');
+        if ($headerStartPosition !== false) {
+            $resultBeforeHeadEnd = \substr($resultBeforeHeadEnd, 0, $headerStartPosition);
+        }
+
         $numberOfContentTypeMetaTags = \substr_count(
-            $result,
+            $resultBeforeHeadEnd,
             '<meta http-equiv="Content-Type" content="text/html; charset=utf-8">'
         );
         self::assertSame(1, $numberOfContentTypeMetaTags);
@@ -763,6 +784,23 @@ public function provideHtmlAroundContentType(): array
             'HEAD element with attribute' => ['<head lang="en">', '</head>'],
             'HTML, HEAD, and BODY with HEADER elements'
                 => ['<html><head>', '</head><body><header></header></body></html>'],
+            'HEAD element with comment' => ['<head><!--Test-->', '</head>'],
+            'HEAD element with commented-out BODY start tag' => ['<head><!--<body>-->', '</head>'],
+            'HEAD element with BASE element' => ['<head><base href="https://example.com"/>', '</head>'],
+            'HEAD element with COMMAND element' => ['<head><command type="command"/>', '</head>'],
+            'HEAD element with LINK element'
+                => ['<head><link rel="stylesheet" href="https://example.org/css.css"/>', '</head>'],
+            'HEAD element with another META element' => ['<head><meta name="title" content="Test"/>', '</head>'],
+            'HEAD element with NOSCRIPT element'
+                => ['<head><noscript><style>p{color:green}</style></noscript>', '</head>'],
+            'HEAD element with SCRIPT element' => ['<head><script>console.log("Test");</script>', '</head>'],
+            'HEAD element with STYLE element' => ['<head><style>p{color:green}</style>', '</head>'],
+            'HEAD element with TEMPLATE element'
+                => ['<head><template id="test"><p>Test</p></template></title>', '</head>'],
+            'HEAD element with TITLE element' => ['<head><title>Test</title>', '</head>'],
+            'HEAD element with uppercase TEMPLATE element'
+                => ['<head><TEMPLATE id="test"><p>Test</p></TEMPLATE></title>', '</head>'],
+            'HEAD element with uppercase TITLE element' => ['<head><TITLE>Test</TITLE>', '</head>'],
         ];
     }