feat: normalize content records

src/Content/ContentRecordNormalizer.php

@@ -0,0 +1,68 @@
+<?php
+/**
+ * Normalizes package content records for handlers.
+ *
+ * @package WPContentSync
+ */
+
+namespace WPContentSync\Content;
+
+final class ContentRecordNormalizer {
+	/**
+	 * @param array<string, mixed> $record Raw post record.
+	 * @return array<string, mixed>
+	 */
+	public function post( array $record ): array {
+		return array(
+			'id'           => (int) ( $record['id'] ?? 0 ),
+			'post_type'    => sanitize_text_field( (string) ( $record['post_type'] ?? 'post' ) ),
+			'post_title'   => sanitize_text_field( (string) ( $record['post_title'] ?? '' ) ),
+			'post_content' => (string) ( $record['post_content'] ?? '' ),
+			'post_excerpt' => sanitize_text_field( (string) ( $record['post_excerpt'] ?? '' ) ),
+			'post_status'  => sanitize_text_field( (string) ( $record['post_status'] ?? 'draft' ) ),
+			'post_name'    => sanitize_text_field( (string) ( $record['post_name'] ?? '' ) ),
+			'post_parent'  => (int) ( $record['post_parent'] ?? 0 ),
+			'menu_order'   => (int) ( $record['menu_order'] ?? 0 ),
+			'meta'         => $this->arrayValue( $record['meta'] ?? array() ),
+		);
+	}
+
+	/**
+	 * @param array<string, mixed> $record Raw term record.
+	 * @return array<string, mixed>
+	 */
+	public function term( array $record ): array {
+		return array(
+			'id'          => (int) ( $record['id'] ?? 0 ),
+			'taxonomy'    => sanitize_text_field( (string) ( $record['taxonomy'] ?? '' ) ),
+			'name'        => sanitize_text_field( (string) ( $record['name'] ?? '' ) ),
+			'slug'        => sanitize_text_field( (string) ( $record['slug'] ?? '' ) ),
+			'description' => (string) ( $record['description'] ?? '' ),
+			'parent'      => (int) ( $record['parent'] ?? 0 ),
+			'meta'        => $this->arrayValue( $record['meta'] ?? array() ),
+		);
+	}
+
+	/**
+	 * @param array<string, mixed> $record Raw media record.
+	 * @return array<string, mixed>
+	 */
+	public function media( array $record ): array {
+		return array(
+			'id'             => (int) ( $record['id'] ?? 0 ),
+			'post_title'     => sanitize_text_field( (string) ( $record['post_title'] ?? '' ) ),
+			'post_mime_type' => sanitize_text_field( (string) ( $record['post_mime_type'] ?? '' ) ),
+			'source_url'     => esc_url_raw( (string) ( $record['source_url'] ?? '' ) ),
+			'metadata'       => $this->arrayValue( $record['metadata'] ?? array() ),
+			'meta'           => $this->arrayValue( $record['meta'] ?? array() ),
+		);
+	}
+
+	/**
+	 * @param mixed $value Value to normalize.
+	 * @return array<string, mixed>
+	 */
+	private function arrayValue( $value ): array {
+		return is_array( $value ) ? $value : array();
+	}
+}

tests/Unit/Content/ContentRecordNormalizerTest.php

@@ -0,0 +1,120 @@
+<?php
+/**
+ * Tests for content record normalization.
+ *
+ * @package WPContentSync
+ */
+
+namespace WPContentSync\Tests\Unit\Content;
+
+use PHPUnit\Framework\TestCase;
+use WPContentSync\Content\ContentRecordNormalizer;
+
+class ContentRecordNormalizerTest extends TestCase {
+	public function test_it_normalizes_post_records(): void {
+		$normalizer = new ContentRecordNormalizer();
+
+		$record = $normalizer->post(
+			array(
+				'id'           => '42',
+				'post_type'    => '<b>post</b>',
+				'post_title'   => "Hello\nWorld",
+				'post_content' => '<p>Keep HTML</p>',
+				'post_excerpt' => '<em>Excerpt</em>',
+				'post_status'  => 'publish',
+				'post_name'    => 'hello-world',
+				'post_parent'  => '7',
+				'menu_order'   => '3',
+				'meta'         => array(
+					'_source_url' => 'https://source.test/page',
+				),
+			)
+		);
+
+		self::assertSame(
+			array(
+				'id'           => 42,
+				'post_type'    => 'post',
+				'post_title'   => 'Hello World',
+				'post_content' => '<p>Keep HTML</p>',
+				'post_excerpt' => 'Excerpt',
+				'post_status'  => 'publish',
+				'post_name'    => 'hello-world',
+				'post_parent'  => 7,
+				'menu_order'   => 3,
+				'meta'         => array(
+					'_source_url' => 'https://source.test/page',
+				),
+			),
+			$record
+		);
+	}
+
+	public function test_it_normalizes_term_records(): void {
+		$normalizer = new ContentRecordNormalizer();
+
+		$record = $normalizer->term(
+			array(
+				'id'          => '9',
+				'taxonomy'    => '<b>category</b>',
+				'name'        => "News\nUpdates",
+				'slug'        => 'news-updates',
+				'description' => '<p>Keep description HTML</p>',
+				'parent'      => '2',
+				'meta'        => array(
+					'landing_url' => 'https://source.test/news',
+				),
+			)
+		);
+
+		self::assertSame(
+			array(
+				'id'          => 9,
+				'taxonomy'    => 'category',
+				'name'        => 'News Updates',
+				'slug'        => 'news-updates',
+				'description' => '<p>Keep description HTML</p>',
+				'parent'      => 2,
+				'meta'        => array(
+					'landing_url' => 'https://source.test/news',
+				),
+			),
+			$record
+		);
+	}
+
+	public function test_it_normalizes_media_records(): void {
+		$normalizer = new ContentRecordNormalizer();
+
+		$record = $normalizer->media(
+			array(
+				'id'             => '12',
+				'post_title'     => "Hero\nImage",
+				'post_mime_type' => 'image/jpeg',
+				'source_url'     => 'https://source.test/uploads/hero.jpg',
+				'metadata'       => array(
+					'width' => 1200,
+				),
+				'meta'           => array(
+					'_wp_attachment_image_alt' => 'Hero',
+				),
+			)
+		);
+
+		self::assertSame(
+			array(
+				'id'             => 12,
+				'post_title'     => 'Hero Image',
+				'post_mime_type' => 'image/jpeg',
+				'source_url'     => 'https://source.test/uploads/hero.jpg',
+				'metadata'       => array(
+					'width' => 1200,
+				),
+				'meta'           => array(
+					'_wp_attachment_image_alt' => 'Hero',
+				),
+			),
+			$record
+		);
+	}
+}