Add StringUtil::getRawDecodedValue()

contao · May 19, 2021 · ee9eedf · ee9eedf
1 parent 6b28659
commit ee9eedf
Show file tree

Hide file tree

Showing 9 changed files with 51 additions and 28 deletions.
diff --git a/calendar-bundle/src/Resources/contao/modules/ModuleEventReader.php b/calendar-bundle/src/Resources/contao/modules/ModuleEventReader.php
@@ -140,16 +140,16 @@ protected function compile()
 			}
 			elseif ($objEvent->title)
 			{
-				$responseContext->setTitle(StringUtil::decodeEntities(Controller::replaceInsertTags($objEvent->title)));
+				$responseContext->setTitle(StringUtil::getRawDecodedValue($objEvent->title));
 			}
 
 			if ($objEvent->description)
 			{
-				$responseContext->setMetaDescription(StringUtil::decodeEntities(Controller::replaceInsertTags($objEvent->description)));
+				$responseContext->setMetaDescription(StringUtil::getRawDecodedValue($objEvent->description));
 			}
 			elseif ($objEvent->teaser)
 			{
-				$responseContext->setMetaDescription(StringUtil::decodeEntities(Controller::replaceInsertTags($objEvent->teaser)));
+				$responseContext->setMetaDescription(trim(preg_replace('/\s+/', ' ', StringUtil::getRawDecodedValue($objEvent->teaser))));
 			}
 
 			if ($objEvent->robots)

diff --git a/core-bundle/src/Resources/contao/library/Contao/StringUtil.php b/core-bundle/src/Resources/contao/library/Contao/StringUtil.php
@@ -971,6 +971,41 @@ public static function ampersand($strString, $blnEncode=true): string
 	{
 		return preg_replace('/&(amp;)?/i', ($blnEncode ? '&amp;' : '&'), $strString);
 	}
+
+	/**
+	 * Converts an input-encoded string back to its raw UTF-8 value it originated from.
+	 */
+	public static function getRawDecodedValue(string $strValue, bool $blnRemoveInsertTags = false): string
+	{
+		if ($blnRemoveInsertTags)
+		{
+			$strValue = static::stripInsertTags($strValue);
+		}
+		else
+		{
+			$strValue = Controller::replaceInsertTags($strValue, false);
+		}
+
+		$strValue = strip_tags($strValue);
+		$strValue = static::restoreBasicEntities($strValue);
+		$strValue = static::decodeEntities($strValue);
+
+		// Ensure valid UTF-8
+		if (preg_match('//u', $strValue) !== 1)
+		{
+			$substituteCharacter = mb_substitute_character();
+			mb_substitute_character(0xFFFD);
+
+			$strValue = mb_convert_encoding($strValue, 'UTF-8', 'UTF-8');
+
+			mb_substitute_character($substituteCharacter);
+		}
+
+		// TODO: We have to prevent insert tags from existing in this stage for security reasons, is using a ZWNJ the best we can do?
+		$strValue = str_replace('{', "\u{200C}{", $strValue);
+
+		return $strValue;
+	}
 }
 
 class_alias(StringUtil::class, 'StringUtil');
diff --git a/core-bundle/src/Resources/contao/models/PageModel.php b/core-bundle/src/Resources/contao/models/PageModel.php
@@ -314,11 +314,11 @@ public function __set($strKey, $varValue)
 				switch ($strKey)
 				{
 					case 'pageTitle':
-						$responseContext->setTitle(StringUtil::decodeEntities($varValue));
+						$responseContext->setTitle(StringUtil::getRawDecodedValue($varValue ?? ''));
 						break;
 
 					case 'description':
-						$responseContext->setMetaDescription(StringUtil::decodeEntities($varValue));
+						$responseContext->setMetaDescription(StringUtil::getRawDecodedValue($varValue ?? ''));
 						break;
 
 					case 'robots':

diff --git a/core-bundle/src/Resources/contao/modules/ModuleArticle.php b/core-bundle/src/Resources/contao/modules/ModuleArticle.php
@@ -171,11 +171,11 @@ protected function compile()
 
 			if ($responseContext instanceof WebpageResponseContext)
 			{
-				$responseContext->setTitle(StringUtil::decodeEntities($this->title));
+				$responseContext->setTitle(StringUtil::getRawDecodedValue($this->title ?? ''));
 
 				if ($this->teaser)
 				{
-					$responseContext->setMetaDescription(Controller::replaceInsertTags($this->teaser));
+					$responseContext->setMetaDescription(trim(preg_replace('/\s+/', ' ', StringUtil::getRawDecodedValue($this->teaser))));
 				}
 			}
 		}

diff --git a/core-bundle/src/Routing/ResponseContext/ContaoWebpageResponseContext.php b/core-bundle/src/Routing/ResponseContext/ContaoWebpageResponseContext.php
@@ -19,11 +19,11 @@ class ContaoWebpageResponseContext extends WebpageResponseContext
 {
     public function __construct(PageModel $pageModel)
     {
-        $title = $pageModel->pageTitle ?: StringUtil::decodeEntities($pageModel->title ?: '');
+        $title = $pageModel->pageTitle ?: StringUtil::getRawDecodedValue($pageModel->title ?: '');
 
         $this
             ->setTitle($title ?: '')
-            ->setMetaDescription($pageModel->description ?: '')
+            ->setMetaDescription(StringUtil::getRawDecodedValue($pageModel->description ?: ''))
         ;
 
         if ($pageModel->robots) {

diff --git a/core-bundle/src/Routing/ResponseContext/WebpageResponseContext.php b/core-bundle/src/Routing/ResponseContext/WebpageResponseContext.php
@@ -37,8 +37,6 @@ public function getTitle(): string
 
     public function setTitle(string $title): self
     {
-        $title = self::cleanString($title);
-
         $this->title = $title;
 
         return $this;
@@ -51,8 +49,6 @@ public function getMetaDescription(): string
 
     public function setMetaDescription(string $metaDescription): self
     {
-        $metaDescription = self::cleanString($metaDescription);
-
         $this->metaDescription = $metaDescription;
 
         return $this;
@@ -69,12 +65,4 @@ public function setMetaRobots(string $metaRobots): self
 
         return $this;
     }
-
-    protected static function cleanString(string $string): string
-    {
-        $string = strip_tags($string);
-        $string = str_replace("\n", ' ', $string);
-
-        return trim($string);
-    }
 }
diff --git a/faq-bundle/src/Resources/contao/modules/ModuleFaqReader.php b/faq-bundle/src/Resources/contao/modules/ModuleFaqReader.php
@@ -121,16 +121,16 @@ protected function compile()
 			}
 			elseif ($objFaq->question)
 			{
-				$responseContext->setTitle(StringUtil::decodeEntities(Controller::replaceInsertTags($objFaq->title)));
+				$responseContext->setTitle(StringUtil::getRawDecodedValue($objFaq->question));
 			}
 
 			if ($objFaq->description)
 			{
-				$responseContext->setMetaDescription(StringUtil::decodeEntities(Controller::replaceInsertTags($objFaq->description)));
+				$responseContext->setMetaDescription(StringUtil::getRawDecodedValue($objFaq->description));
 			}
 			elseif ($objFaq->question)
 			{
-				$responseContext->setMetaDescription(StringUtil::decodeEntities(Controller::replaceInsertTags($objFaq->question)));
+				$responseContext->setMetaDescription(StringUtil::getRawDecodedValue($objFaq->question));
 			}
 
 			if ($objFaq->robots)

diff --git a/news-bundle/src/Resources/contao/modules/ModuleNewsReader.php b/news-bundle/src/Resources/contao/modules/ModuleNewsReader.php
@@ -147,16 +147,16 @@ protected function compile()
 			}
 			elseif ($objArticle->headline)
 			{
-				$responseContext->setTitle(StringUtil::decodeEntities(Controller::replaceInsertTags($objArticle->headline)));
+				$responseContext->setTitle(StringUtil::getRawDecodedValue($objArticle->headline));
 			}
 
 			if ($objArticle->description)
 			{
-				$responseContext->setMetaDescription(StringUtil::decodeEntities(Controller::replaceInsertTags($objArticle->description)));
+				$responseContext->setMetaDescription(StringUtil::getRawDecodedValue($objArticle->description));
 			}
 			elseif ($objArticle->teaser)
 			{
-				$responseContext->setMetaDescription(StringUtil::decodeEntities(Controller::replaceInsertTags($objArticle->teaser)));
+				$responseContext->setMetaDescription(trim(preg_replace('/\s+/', ' ', StringUtil::getRawDecodedValue($objArticle->teaser))));
 			}
 
 			if ($objArticle->robots)

diff --git a/newsletter-bundle/src/Resources/contao/modules/ModuleNewsletterReader.php b/newsletter-bundle/src/Resources/contao/modules/ModuleNewsletterReader.php
@@ -113,7 +113,7 @@ protected function compile()
 
 			if ($responseContext instanceof WebpageResponseContext)
 			{
-				$responseContext->setTitle(Controller::replaceInsertTags($objNewsletter->subject)); // Already stored decoded
+				$responseContext->setTitle(StringUtil::getRawDecodedValue($objNewsletter->subject));
 			}
 		}