IBX-10425: Translated string is returned as XML (#29)

* IBX-10425: Translated string is returned as XML

* Corrected phpstan for 7.4 and higher

* Corrected CS

* Added tests

* Corrected CS

* Corrected CS

* Registered TextFieldCdataCleaner as service

---------

Co-authored-by: Mateusz Dębiński <mateusz.debinski@ibexa.co>

Author: mateuszdebinski

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",

src/bundle/Resources/config/services.yaml

@@ -62,3 +62,5 @@ services:
             - { name: form.type_extension, extended_type: Ibexa\AdminUi\Form\Type\Language\LanguageCreateType }
 
     Ibexa\Bundle\AutomatedTranslation\EventListener\ContentProxyTranslateListener: ~
+
+    Ibexa\AutomatedTranslation\TextFieldCdataCleaner: ~

src/lib/Encoder.php

@@ -79,14 +79,18 @@ class Encoder
 
     private FieldEncoderManager $fieldEncoderManager;
 
+    private TextFieldCdataCleaner $textFieldCdataCleaner;
+
     public function __construct(
         ContentTypeService $contentTypeService,
         EventDispatcherInterface $eventDispatcher,
-        FieldEncoderManager $fieldEncoderManager
+        FieldEncoderManager $fieldEncoderManager,
+        TextFieldCdataCleaner $textFieldCdataCleaner
     ) {
         $this->contentTypeService = $contentTypeService;
         $this->eventDispatcher = $eventDispatcher;
         $this->fieldEncoderManager = $fieldEncoderManager;
+        $this->textFieldCdataCleaner = $textFieldCdataCleaner;
     }
 
     public function encode(Content $content): string
@@ -118,6 +122,8 @@ public function encode(Content $content): string
 
         $encoder = new XmlEncoder();
         $payload = $encoder->encode($results, XmlEncoder::FORMAT);
+        $payload = $this->textFieldCdataCleaner->clear($payload);
+
         // here Encoder has  decorated with CDATA, we don't want the CDATA
         return str_replace(
             ['<![CDATA[', ']]>'],

src/lib/Encoder/Field/PageBuilderFieldEncoder.php

@@ -10,6 +10,7 @@
 
 use Ibexa\AutomatedTranslation\Encoder\BlockAttribute\BlockAttributeEncoderManager;
 use Ibexa\AutomatedTranslation\Exception\EmptyTranslatedAttributeException;
+use Ibexa\AutomatedTranslation\TextFieldCdataCleaner;
 use Ibexa\Contracts\AutomatedTranslation\Encoder\Field\FieldEncoderInterface;
 use Ibexa\Contracts\Core\Repository\Values\Content\Field;
 use Ibexa\Core\FieldType\Value as APIValue;
@@ -26,12 +27,16 @@ final class PageBuilderFieldEncoder implements FieldEncoderInterface
 
     private BlockDefinitionFactoryInterface $blockDefinitionFactory;
 
+    private TextFieldCdataCleaner $textFieldCdataCleaner;
+
     public function __construct(
         BlockAttributeEncoderManager $blockAttributeEncoderManager,
-        BlockDefinitionFactoryInterface $blockDefinitionFactory
+        BlockDefinitionFactoryInterface $blockDefinitionFactory,
+        TextFieldCdataCleaner $textFieldCdataCleaner
     ) {
         $this->blockAttributeEncoderManager = $blockAttributeEncoderManager;
         $this->blockDefinitionFactory = $blockDefinitionFactory;
+        $this->textFieldCdataCleaner = $textFieldCdataCleaner;
     }
 
     public function canEncode(Field $field): bool
@@ -85,12 +90,11 @@ public function encode(Field $field): string
         $payload = $encoder->encode($blocks, XmlEncoder::FORMAT, [
             XmlEncoder::ROOT_NODE_NAME => 'blocks',
         ]);
-
-        $payload = str_replace('<?xml version="1.0"?>' . "\n", '', $payload);
+        $payload = $this->textFieldCdataCleaner->clear($payload);
 
         $payload = str_replace(
-            ['<![CDATA[', ']]>'],
-            ['<' . self::CDATA_FAKER_TAG . '>', '</' . self::CDATA_FAKER_TAG . '>'],
+            ['<?xml version="1.0"?>' . "\n", '<![CDATA[', ']]>'],
+            ['', '<' . self::CDATA_FAKER_TAG . '>', '</' . self::CDATA_FAKER_TAG . '>'],
             $payload
         );
 

src/lib/TextFieldCdataCleaner.php

@@ -0,0 +1,86 @@
+<?php
+
+/**
+ * @copyright Copyright (C) Ibexa AS. All rights reserved.
+ * @license For full copyright and license information view LICENSE file distributed with this source code.
+ */
+declare(strict_types=1);
+
+namespace Ibexa\AutomatedTranslation;
+
+use DOMCdataSection;
+use DOMDocument;
+use DOMElement;
+use DOMNode;
+use DOMXPath;
+use Ibexa\FieldTypeRichText\FieldType\RichText\Value as RichTextValue;
+use RuntimeException;
+
+final class TextFieldCdataCleaner
+{
+    public function clear(string $payload): string
+    {
+        $dom = $this->loadDocument($payload);
+        $this->processCdataNodes($dom);
+
+        return $this->saveDocument($dom);
+    }
+
+    private function loadDocument(string $payload): DOMDocument
+    {
+        $dom = new DOMDocument();
+        $dom->loadXML($payload);
+
+        return $dom;
+    }
+
+    private function processCdataNodes(DOMDocument $dom): void
+    {
+        $xpath = new DOMXPath($dom);
+        $textNodes = $xpath->query('//text()');
+
+        if ($textNodes === false) {
+            return;
+        }
+
+        foreach ($textNodes as $textNode) {
+            if (!$textNode instanceof DOMCdataSection) {
+                continue;
+            }
+
+            if ($this->shouldReplaceCdata($textNode)) {
+                $this->replaceWithTextNode($dom, $textNode);
+            }
+        }
+    }
+
+    private function shouldReplaceCdata(DOMNode $node): bool
+    {
+        $parent = $node->parentNode;
+        if (!$parent instanceof DOMElement) {
+            return false;
+        }
+
+        return $parent->getAttribute('type') !== RichTextValue::class;
+    }
+
+    private function replaceWithTextNode(DOMDocument $dom, DOMCdataSection $cdataNode): void
+    {
+        $newText = $dom->createTextNode($cdataNode->data);
+
+        if ($cdataNode->parentNode !== null) {
+            $cdataNode->parentNode->replaceChild($newText, $cdataNode);
+        }
+    }
+
+    private function saveDocument(DOMDocument $dom): string
+    {
+        $result = $dom->saveXML();
+
+        if ($result === false) {
+            throw new RuntimeException('Saving XML failed after removing CDATA.');
+        }
+
+        return $result;
+    }
+}

tests/lib/Encoder/Field/PageBuilderFieldEncoderTest.php

@@ -10,6 +10,7 @@
 
 use Ibexa\AutomatedTranslation\Encoder\BlockAttribute\BlockAttributeEncoderManager;
 use Ibexa\AutomatedTranslation\Encoder\Field\PageBuilderFieldEncoder;
+use Ibexa\AutomatedTranslation\TextFieldCdataCleaner;
 use Ibexa\Contracts\Core\Repository\Values\Content\Field;
 use Ibexa\Contracts\FieldTypePage\FieldType\LandingPage\Model\Attribute;
 use Ibexa\Contracts\FieldTypePage\FieldType\LandingPage\Model\BlockValue;
@@ -24,6 +25,7 @@
 final class PageBuilderFieldEncoderTest extends TestCase
 {
     public const ATTRIBUTE_VALUE = 'ibexa';
+    public const ATTRIBUTE_VALUE_CDATA = 'ibexa & ibexa';
 
     /** @var \Ibexa\AutomatedTranslation\Encoder\BlockAttribute\BlockAttributeEncoderManager&\PHPUnit\Framework\MockObject\MockObject */
     private BlockAttributeEncoderManager $blockAttributeEncoderManagerMock;
@@ -52,7 +54,8 @@ public function testEncode(): void
         $field = $this->getLandingPageField();
         $subject = new PageBuilderFieldEncoder(
             $this->blockAttributeEncoderManagerMock,
-            $this->blockDefinitionFactoryMock
+            $this->blockDefinitionFactoryMock,
+            new TextFieldCdataCleaner()
         );
 
         $result = $subject->encode($field);
@@ -75,7 +78,8 @@ public function testEncodeMissingAttribute(): void
         $field = $this->getLandingPageField();
         $subject = new PageBuilderFieldEncoder(
             $this->blockAttributeEncoderManagerMock,
-            $this->blockDefinitionFactoryMock
+            $this->blockDefinitionFactoryMock,
+            new TextFieldCdataCleaner()
         );
 
         $result = $subject->encode($field);
@@ -85,12 +89,38 @@ public function testEncodeMissingAttribute(): void
         self::assertEquals($expectedResult, $result);
     }
 
+    public function testEncodeCDATAInTextField(): void
+    {
+        $this->blockDefinitionFactoryMock
+            ->method('getBlockDefinition')
+            ->withAnyParameters()
+            ->willReturn($this->getBlockDefinition());
+
+        $this->blockAttributeEncoderManagerMock
+            ->method('encode')
+            ->withAnyParameters()
+            ->willReturn(self::ATTRIBUTE_VALUE_CDATA);
+
+        $field = $this->getLandingPageField();
+        $subject = new PageBuilderFieldEncoder(
+            $this->blockAttributeEncoderManagerMock,
+            $this->blockDefinitionFactoryMock,
+            new TextFieldCdataCleaner()
+        );
+
+        $result = $subject->encode($field);
+        $encodedValue = htmlspecialchars(self::ATTRIBUTE_VALUE_CDATA);
+
+        self::assertEquals($this->getEncodeResult($encodedValue), $result);
+    }
+
     public function testCanEncode(): void
     {
         $field = $this->getLandingPageField();
         $subject = new PageBuilderFieldEncoder(
             $this->blockAttributeEncoderManagerMock,
-            $this->blockDefinitionFactoryMock
+            $this->blockDefinitionFactoryMock,
+            new TextFieldCdataCleaner()
         );
 
         self::assertTrue($subject->canEncode($field));
@@ -107,7 +137,8 @@ public function testDecode(): void
         $field = $this->getLandingPageField();
         $subject = new PageBuilderFieldEncoder(
             $this->blockAttributeEncoderManagerMock,
-            $this->blockDefinitionFactoryMock
+            $this->blockDefinitionFactoryMock,
+            new TextFieldCdataCleaner()
         );
 
         $result = $subject->decode(
@@ -124,7 +155,8 @@ public function testCanDecode(): void
         $field = $this->getLandingPageField();
         $subject = new PageBuilderFieldEncoder(
             $this->blockAttributeEncoderManagerMock,
-            $this->blockDefinitionFactoryMock
+            $this->blockDefinitionFactoryMock,
+            new TextFieldCdataCleaner()
         );
 
         self::assertTrue($subject->canDecode(get_class($field->value)));
@@ -202,10 +234,10 @@ private function createBlockDefinition(): BlockDefinition
         return $blockDefinition;
     }
 
-    private function getEncodeResult(): string
+    private function getEncodeResult(string $value = self::ATTRIBUTE_VALUE): string
     {
         return '<blocks><item key="1"><name>Code</name><attributes><content type="string">' .
-            self::ATTRIBUTE_VALUE . '</content></attributes></item></blocks>
+            $value . '</content></attributes></item></blocks>
 ';
     }
 }