1.2.5, check changelog

heimrichhannot · Oct 22, 2018 · a2e171a · a2e171a
1 parent 7830c7a
commit a2e171a
Show file tree

Hide file tree

Showing 3 changed files with 28 additions and 11 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -1,6 +1,11 @@
 # Changelog
 All notable changes to this project will be documented in this file.
 
+## [1.2.5] - 2018-10-22
+
+### Fixed
+- prevent unescape unicode html entities (email obfuscation)
+
 ## [1.2.4] - 2018-07-12
 
 ### Fixed

diff --git a/src/Hyphenator/FrontendHyphenator.php b/src/Hyphenator/FrontendHyphenator.php
@@ -37,7 +37,7 @@ public function hyphenate($strBuffer)
 
         $arrSkipPages = Config::get('hyphenator_skipPages');
 
-        if (null === $objPage || is_array($arrSkipPages) && in_array($objPage->id, $arrSkipPages, true)) {
+        if (null === $objPage || \is_array($arrSkipPages) && \in_array($objPage->id, $arrSkipPages, true)) {
             return $strBuffer;
         }
 
@@ -56,14 +56,21 @@ function ($matches) {
             $strBuffer
         );
 
+        // prevent unescape unicode html entities (email obfuscation)
+        $strBuffer = preg_replace('/&(#+[x0-9a-fA-F]+);/', '&_$1;', $strBuffer);
+
         $doc = HtmlPageCrawler::create($strBuffer);
+        $isHtmlDocument = $doc->isHtmlDocument();
+
+        if (false === $isHtmlDocument) {
+            $doc = HtmlPageCrawler::create(sprintf('<div id="crawler-root">%s</div>', $strBuffer));
+        }
 
         $cacheEnabled = (bool) Config::get('hyphenator_enableCache');
         $cache = [];
 
         $doc->filter(Config::get('hyphenator_tags'))->each(
-            function ($node, $i) use ($h, &$cache, $cacheEnabled) {
-                /** @var $node HtmlPageCrawler */
+            function (HtmlPageCrawler $node, $i) use ($h, &$cache, $cacheEnabled) {
                 $clone = $node->makeClone(); // make a clone to prevent `Couldn't fetch DOMElement. Node no longer exists`
                 $html = $clone->html(); // restore nested inserttags that were replaced with %7B or %7D
                 $cacheKey = $html;
@@ -89,15 +96,18 @@ function ($node, $i) use ($h, &$cache, $cacheEnabled) {
 
                 $html = $matches['content'];
                 $clone->html(StringUtil::decodeEntities($html));
-                $node->replaceWith($clone);
+                $node->replaceWith($clone->saveHTML());
 
                 $cache[$cacheKey] = $html;
 
                 return $node;
             }
         );
 
-        $strBuffer = $doc->saveHTML();
+        $strBuffer = false === $isHtmlDocument ? $doc->filter('#crawler-root')->getInnerHtml() : $doc->saveHTML();
+
+        // prevent unescape unicode html entities (email obfuscation)
+        $strBuffer = preg_replace('/&amp;_(#+[x0-9a-fA-F]+);/', '&$1;', $strBuffer);
 
         $strBuffer = preg_replace_callback(
             '/####esi:open####(.*)####esi:close####/',

diff --git a/tests/Hyphenator/FrontendHyphenatorTest.php b/tests/Hyphenator/FrontendHyphenatorTest.php
@@ -28,7 +28,7 @@ protected function setUp()
     {
         parent::setUp();
 
-        if (!defined('TL_ROOT')) {
+        if (!\defined('TL_ROOT')) {
             \define('TL_ROOT', $this->getFixturesDir());
         }
 
@@ -86,13 +86,9 @@ public function hyphenationProvider()
     {
         $GLOBALS['TL_CONFIG']['hyphenator_tags'] = 'h1, h1> a, h2, h2 > a, h3, h3 > a, h4, h4 > a, h5, h5 > a, h6, h6 > a, p';
         $GLOBALS['TL_CONFIG']['hyphenator_wordMin'] = 10;
-        $GLOBALS['TL_CONFIG']['hyphenator_leftMin'] = 5;
-        $GLOBALS['TL_CONFIG']['hyphenator_rightMin'] = 5;
-        $GLOBALS['TL_CONFIG']['hyphenator_quality'] = 9;
         $GLOBALS['TL_CONFIG']['hyphenator_hyphen'] = '&shy;';
-        $GLOBALS['TL_CONFIG']['hyphenator_filter'] = 'Simple';
-        $GLOBALS['TL_CONFIG']['hyphenator_tokenizers'] = ['Whitespace', 'Punctuation'];
         $GLOBALS['TL_CONFIG']['hyphenator_skipPages'] = [];
+        $GLOBALS['TL_CONFIG']['hyphenator_enableCache'] = true;
 
         return [
             [
@@ -143,6 +139,12 @@ public function hyphenationProvider()
                 $this->getConfig(),
                 '<p><br> Tel: +49 40 123 45 67<br> Fax: +49 40 123 45 68 <br> {{email::test@test.de} <br><br><a class="more" href="mailto:%7B%7Bemail_url::test@test.de%7D%7D">Kon&shy;tak&shy;tie&shy;ren Sie uns</a></p>',
             ],
+            [
+                '<h1><a href="&#109;&#97;&#105;&#108;&#116;&#111;&#58;&#116;&#x65;&#x73;&#x74;&#64;&#x74;&#101;&#115;&#x74;&#46;&#99;&#x6F;&#109;" title="email obfuscation test for test@test.com">&#116;&#x65;&#x73;&#x74;&#64;&#x74;&#101;&#115;&#x74;&#46;&#99;&#x6F;&#109;</a></h1>',
+                $this->getPage(),
+                $this->getConfig(),
+                '<h1><a href="&#109;&#97;&#105;&#108;&#116;&#111;&#58;&#116;&#x65;&#x73;&#x74;&#64;&#x74;&#101;&#115;&#x74;&#46;&#99;&#x6F;&#109;" title="email obfuscation test for test@test.com">&#116;&#x65;&#x73;&#x74;&#64;&#x74;&#101;&#115;&#x74;&#46;&#99;&#x6F;&#109;</a></h1>',
+            ],
         ];
     }