IBX-7987: Node filter for extraction text

src/bundle/Resources/config/settings/fieldtype_services.yaml

@@ -70,3 +70,22 @@ services:
 
     Ibexa\FieldTypeRichText\RichText\TextExtractor\FullTextExtractor: ~
 
+    Ibexa\Contracts\FieldTypeRichText\RichText\TextExtractor\NodeFilterInterface:
+        alias: Ibexa\FieldTypeRichText\RichText\TextExtractor\NodeFilter\AggregateFilter
+
+    Ibexa\Contracts\FieldTypeRichText\RichText\TextExtractor\NodeFilterFactoryInterface:
+        alias: Ibexa\FieldTypeRichText\RichText\TextExtractor\NodeFilter\NodeFilterFactory
+
+    Ibexa\FieldTypeRichText\RichText\TextExtractor\NodeFilter\NodeFilterFactory: ~
+
+    Ibexa\FieldTypeRichText\RichText\TextExtractor\NodeFilter\AggregateFilter:
+        arguments:
+            $filters: !tagged ibexa.field_type.richtext.text_extractor.node_filter
+
+    ibexa.field_type.richtext.text_extractor.node_filter.template:
+        class: Ibexa\Contracts\FieldTypeRichText\RichText\TextExtractor\NodeFilterInterface
+        factory: ['@Ibexa\Contracts\FieldTypeRichText\RichText\TextExtractor\NodeFilterFactoryInterface', 'createPathFilter']
+        arguments: ['eztemplate', 'ezconfig']
+        tags:
+            - { name: ibexa.field_type.richtext.text_extractor.node_filter }
+

src/contracts/RichText/TextExtractor/NodeFilterFactoryInterface.php

@@ -0,0 +1,14 @@
+<?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\Contracts\FieldTypeRichText\RichText\TextExtractor;
+
+interface NodeFilterFactoryInterface
+{
+    public function createPathFilter(string ...$path): NodeFilterInterface;
+}

src/contracts/RichText/TextExtractor/NodeFilterInterface.php

@@ -0,0 +1,22 @@
+<?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\Contracts\FieldTypeRichText\RichText\TextExtractor;
+
+use DOMNode;
+
+/**
+ * Filters nodes for text extraction.
+ */
+interface NodeFilterInterface
+{
+    /**
+     * Return false to preserve the node, true to remove it.
+     */
+    public function filter(DOMNode $node): bool;
+}

src/lib/RichText/TextExtractor/FullTextExtractor.php

@@ -10,6 +10,7 @@
 
 use DOMDocument;
 use DOMNode;
+use Ibexa\Contracts\FieldTypeRichText\RichText\TextExtractor\NodeFilterInterface;
 use Ibexa\Contracts\FieldTypeRichText\RichText\TextExtractorInterface;
 
 /**
@@ -19,6 +20,13 @@
  */
 final class FullTextExtractor implements TextExtractorInterface
 {
+    private NodeFilterInterface $filter;
+
+    public function __construct(NodeFilterInterface $filter)
+    {
+        $this->filter = $filter;
+    }
+
     public function extractText(DOMDocument $document): string
     {
         return null !== $document->documentElement
@@ -28,8 +36,12 @@ public function extractText(DOMDocument $document): string
 
     private function extractTextFromNode(DOMNode $node): string
     {
-        $text = '';
+        if ($this->filter->filter($node) === true) {
+            // Node is excluded
+            return '';
+        }
 
+        $text = '';
         if ($node->childNodes !== null && $node->childNodes->count() > 0) {
             foreach ($node->childNodes as $child) {
                 $text .= $this->extractTextFromNode($child);

src/lib/RichText/TextExtractor/NodeFilter/AggregateFilter.php

@@ -0,0 +1,37 @@
+<?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\FieldTypeRichText\RichText\TextExtractor\NodeFilter;
+
+use DOMNode;
+use Ibexa\Contracts\FieldTypeRichText\RichText\TextExtractor\NodeFilterInterface;
+
+final class AggregateFilter implements NodeFilterInterface
+{
+    /** @var \Ibexa\Contracts\FieldTypeRichText\RichText\TextExtractor\NodeFilterInterface[] */
+    private iterable $filters;
+
+    /**
+     * @param \Ibexa\Contracts\FieldTypeRichText\RichText\TextExtractor\NodeFilterInterface[]|iterable $filters
+     */
+    public function __construct(iterable $filters)
+    {
+        $this->filters = $filters;
+    }
+
+    public function filter(DOMNode $node): bool
+    {
+        foreach ($this->filters as $filter) {
+            if ($filter->filter($node)) {
+                return true;
+            }
+        }
+
+        return false;
+    }
+}

src/lib/RichText/TextExtractor/NodeFilter/NodeFilterFactory.php

@@ -0,0 +1,20 @@
+<?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\FieldTypeRichText\RichText\TextExtractor\NodeFilter;
+
+use Ibexa\Contracts\FieldTypeRichText\RichText\TextExtractor\NodeFilterFactoryInterface;
+use Ibexa\Contracts\FieldTypeRichText\RichText\TextExtractor\NodeFilterInterface;
+
+final class NodeFilterFactory implements NodeFilterFactoryInterface
+{
+    public function createPathFilter(string ...$path): NodeFilterInterface
+    {
+        return new NodePathFilter(...$path);
+    }
+}

src/lib/RichText/TextExtractor/NodeFilter/NodePathFilter.php

@@ -0,0 +1,40 @@
+<?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\FieldTypeRichText\RichText\TextExtractor\NodeFilter;
+
+use DOMNode;
+use Ibexa\Contracts\FieldTypeRichText\RichText\TextExtractor\NodeFilterInterface;
+
+final class NodePathFilter implements NodeFilterInterface
+{
+    /**
+     * Path in reverse order.
+     *
+     * @var string[]
+     */
+    private array $path;
+
+    public function __construct(string ...$path)
+    {
+        $this->path = array_reverse($path);
+    }
+
+    public function filter(DOMNode $node): bool
+    {
+        foreach ($this->path as $name) {
+            if ($node === null || $node->nodeName !== $name) {
+                return false;
+            }
+
+            $node = $node->parentNode;
+        }
+
+        return true;
+    }
+}

tests/lib/RichText/TextExtractor/FullTextExtractorTest.php

@@ -8,13 +8,17 @@
 
 namespace Ibexa\Tests\FieldTypeRichText\RichText\TextExtractor;
 
+use Ibexa\Contracts\FieldTypeRichText\RichText\TextExtractor\NodeFilterInterface;
 use Ibexa\FieldTypeRichText\RichText\TextExtractor\FullTextExtractor;
 
 final class FullTextExtractorTest extends BaseTest
 {
     protected function setUp(): void
     {
-        $this->textExtractor = new FullTextExtractor();
+        $filter = $this->createMock(NodeFilterInterface::class);
+        $filter->method('filter')->willReturn(false);
+
+        $this->textExtractor = new FullTextExtractor($filter);
     }
 
     public function providerForTestExtractText(): array

tests/lib/RichText/TextExtractor/NodeFilter/AggregateFilterTest.php

@@ -0,0 +1,33 @@
+<?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\Tests\FieldTypeRichText\RichText\TextExtractor\NodeFilter;
+
+use DOMNode;
+use Ibexa\Contracts\FieldTypeRichText\RichText\TextExtractor\NodeFilterInterface;
+use Ibexa\FieldTypeRichText\RichText\TextExtractor\NodeFilter\AggregateFilter;
+use PHPUnit\Framework\TestCase;
+
+final class AggregateFilterTest extends TestCase
+{
+    public function testFilter(): void
+    {
+        $node = $this->createMock(DOMNode::class);
+
+        $filterA = $this->createMock(NodeFilterInterface::class);
+        $filterA->expects(self::once())->method('filter')->with($node)->willReturn(false);
+        $filterB = $this->createMock(NodeFilterInterface::class);
+        $filterB->expects(self::once())->method('filter')->with($node)->willReturn(true);
+        $filterC = $this->createMock(NodeFilterInterface::class);
+        $filterC->expects(self::never())->method('filter');
+
+        $aggregateFilter = new AggregateFilter([$filterA, $filterB, $filterC]);
+
+        self::assertTrue($aggregateFilter->filter($node));
+    }
+}

tests/lib/RichText/TextExtractor/NodeFilter/NodePathFilterTest.php

@@ -0,0 +1,47 @@
+<?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\Tests\FieldTypeRichText\RichText\TextExtractor\NodeFilter;
+
+use DOMDocument;
+use DOMNode;
+use DOMNodeList;
+use DOMXPath;
+use Ibexa\FieldTypeRichText\RichText\TextExtractor\NodeFilter\NodePathFilter;
+use PHPUnit\Framework\TestCase;
+use RuntimeException;
+
+final class NodePathFilterTest extends TestCase
+{
+    public function testFilter(): void
+    {
+        $document = new DOMDocument();
+        $document->loadXML('<a><b><c></c></b></a>');
+
+        $nodeA = $this->getNode($document, '//a');
+        $nodeB = $this->getNode($document, '//b');
+        $nodeC = $this->getNode($document, '//c');
+
+        self::assertFalse((new NodePathFilter('b', 'c'))->filter($nodeB));
+        self::assertTrue((new NodePathFilter('b', 'c'))->filter($nodeC));
+        self::assertFalse((new NodePathFilter('a', 'b', 'c', 'd'))->filter($nodeA));
+    }
+
+    private function getNode(DOMDocument $document, string $expression): DOMNode
+    {
+        $xpath = new DOMXPath($document);
+
+        $results = $xpath->query($expression);
+        if ($results instanceof DOMNodeList) {
+            /** @var \DOMNode */
+            return $results->item(0);
+        }
+
+        throw new RuntimeException("Expression '$expression' did not return a node.");
+    }
+}