IBX-10425: Translated string is returned as XML

Author: Mateusz Dębiński

composer.json

@@ -32,6 +32,7 @@
     "require": {
         "php": "^7.4 || ^8.0",
         "ext-libxml": "*",
+        "ext-dom": "*",
         "symfony/yaml": "^5.0",
         "symfony/routing": "^5.0",
         "symfony/console": "^5.0",

phpstan-baseline.neon

@@ -30,6 +30,18 @@ parameters:
 			count: 1
 			path: src/bundle/Form/TranslationAddDataTransformer.php
 
+		-
+			message: '#^Call to an undefined method DOMNode\:\:getAttribute\(\)\.$#'
+			identifier: method.notFound
+			count: 1
+			path: src/lib/Encoder.php
+
+		-
+			message: '#^Call to an undefined method DOMNode\:\:getAttribute\(\)\.$#'
+			identifier: method.notFound
+			count: 1
+			path: src/lib/Encoder/Field/PageBuilderFieldEncoder.php
+
 		-
 			message: '#^Cannot call method setValue\(\) on Ibexa\\Contracts\\FieldTypePage\\FieldType\\LandingPage\\Model\\Attribute\|null\.$#'
 			identifier: method.nonObject

src/lib/Encoder.php

@@ -8,6 +8,9 @@
 
 namespace Ibexa\AutomatedTranslation;
 
+use DOMCdataSection;
+use DOMDocument;
+use DOMXPath;
 use Ibexa\AutomatedTranslation\Encoder\Field\FieldEncoderManager;
 use Ibexa\AutomatedTranslation\Exception\EmptyTranslatedFieldException;
 use Ibexa\Bundle\AutomatedTranslation\Event\FieldDecodeEvent;
@@ -18,7 +21,9 @@
 use Ibexa\Contracts\Core\Repository\Values\Content\Content;
 use Ibexa\Contracts\Core\Repository\Values\Content\Field;
 use Ibexa\Core\FieldType\Value;
+use Ibexa\FieldTypeRichText\FieldType\RichText\Value as RichTextValue;
 use InvalidArgumentException;
+use RuntimeException;
 use Symfony\Component\Serializer\Encoder\XmlEncoder;
 use Symfony\Contracts\EventDispatcher\EventDispatcherInterface;
 
@@ -118,6 +123,36 @@ public function encode(Content $content): string
 
         $encoder = new XmlEncoder();
         $payload = $encoder->encode($results, XmlEncoder::FORMAT);
+
+        $dom = new DOMDocument();
+        $dom->loadXML($payload);
+        $xpath = new DOMXPath($dom);
+        $textNodes = $xpath->query('//text()');
+        if ($textNodes !== false) {
+            foreach ($textNodes as $textNode) {
+                if ($textNode instanceof DOMCdataSection) {
+                    $parent = $textNode->parentNode;
+                    if ($parent === null) {
+                        continue;
+                    }
+
+                    $type = $parent->getAttribute('type');
+
+                    if ($type !== RichTextValue::class) {
+                        $newText = $dom->createTextNode($textNode->data);
+                        $parent->replaceChild($newText, $textNode);
+                    }
+                }
+            }
+            $payload = $dom->saveXML();
+
+            if (!$payload) {
+                throw new RuntimeException(
+                    sprintf('Saving XML failed after removing CDATA, error: %s', preg_last_error_msg())
+                );
+            }
+        }
+
         // here Encoder has  decorated with CDATA, we don't want the CDATA
         return str_replace(
             ['<![CDATA[', ']]>'],

src/lib/Encoder/Field/PageBuilderFieldEncoder.php

@@ -8,14 +8,19 @@
 
 namespace Ibexa\AutomatedTranslation\Encoder\Field;
 
+use DOMCdataSection;
+use DOMDocument;
+use DOMXPath;
 use Ibexa\AutomatedTranslation\Encoder\BlockAttribute\BlockAttributeEncoderManager;
 use Ibexa\AutomatedTranslation\Exception\EmptyTranslatedAttributeException;
 use Ibexa\Contracts\AutomatedTranslation\Encoder\Field\FieldEncoderInterface;
 use Ibexa\Contracts\Core\Repository\Values\Content\Field;
 use Ibexa\Core\FieldType\Value as APIValue;
 use Ibexa\FieldTypePage\FieldType\LandingPage\Value;
 use Ibexa\FieldTypePage\FieldType\Page\Block\Definition\BlockDefinitionFactoryInterface;
+use Ibexa\FieldTypeRichText\FieldType\RichText\Value as RichTextValue;
 use InvalidArgumentException;
+use RuntimeException;
 use Symfony\Component\Serializer\Encoder\XmlEncoder;
 
 final class PageBuilderFieldEncoder implements FieldEncoderInterface
@@ -84,6 +89,35 @@ public function encode(Field $field): string
             XmlEncoder::ROOT_NODE_NAME => 'blocks',
         ]);
 
+        $dom = new DOMDocument();
+        $dom->loadXML($payload);
+        $xpath = new DOMXPath($dom);
+        $textNodes = $xpath->query('//text()');
+        if ($textNodes !== false) {
+            foreach ($textNodes as $textNode) {
+                if ($textNode instanceof DOMCdataSection) {
+                    $parent = $textNode->parentNode;
+                    if ($parent === null) {
+                        continue;
+                    }
+
+                    $type = $parent->getAttribute('type');
+
+                    if ($type !== RichTextValue::class) {
+                        $newText = $dom->createTextNode($textNode->data);
+                        $parent->replaceChild($newText, $textNode);
+                    }
+                }
+            }
+            $payload = $dom->saveXML();
+
+            if (!$payload) {
+                throw new RuntimeException(
+                    sprintf('Saving XML failed after removing CDATA, error: %s', preg_last_error_msg())
+                );
+            }
+        }
+
         $payload = str_replace('<?xml version="1.0"?>' . "\n", '', $payload);
 
         $payload = str_replace(