feat: import post content records

2026-04-28 18:22:01 -05:00
parent c66501d0e5
commit 6d11934fcc
4 changed files with 447 additions and 4 deletions
@@ -0,0 +1,215 @@
+<?php
+/**
+ * Imports post content records.
+ *
+ * @package WPContentSync
+ */
+
+namespace WPContentSync\Content;
+
+use WPContentSync\Logging\LoggerInterface;
+use WPContentSync\Sync\SyncContext;
+use WPContentSync\Sync\SyncResult;
+use WPContentSync\Url\MetadataUrlTransformer;
+use WPContentSync\Url\UrlMapping;
+use WPContentSync\Url\UrlMappingCollection;
+use WPContentSync\Url\UrlTransformer;
+
+final class PostContentHandler implements ContentHandlerInterface {
+	private ContentRecordNormalizer $normalizer;
+	private UrlTransformer $url_transformer;
+	private MetadataUrlTransformer $metadata_transformer;
+	private LoggerInterface $logger;
+
+	public function __construct(
+		ContentRecordNormalizer $normalizer,
+		UrlTransformer $url_transformer,
+		MetadataUrlTransformer $metadata_transformer,
+		LoggerInterface $logger
+	) {
+		$this->normalizer           = $normalizer;
+		$this->url_transformer      = $url_transformer;
+		$this->metadata_transformer = $metadata_transformer;
+		$this->logger               = $logger;
+	}
+
+	public function bucket(): string {
+		return 'posts';
+	}
+
+	/**
+	 * @param array<int, array<string, mixed>> $records Package records.
+	 */
+	public function importRecords( array $records, SyncContext $context ): SyncResult {
+		$created   = 0;
+		$updated   = 0;
+		$skipped   = 0;
+		$conflicts = 0;
+		$mappings  = $this->mappings( $context );
+		$errors    = array();
+
+		foreach ( $records as $record ) {
+			$normalized = $this->normalizer->post( $record );
+			$existing   = $this->findExistingPostId( (int) $normalized['id'] );
+
+			if ( $existing > 0 && 'manual_review' === $context->conflictStrategy() ) {
+				++$skipped;
+				++$conflicts;
+				$this->logger->warning(
+					'Skipped post import because manual review is required.',
+					array(
+						'source_id' => $normalized['id'],
+						'post_id'   => $existing,
+					)
+				);
+				continue;
+			}
+
+			try {
+				$post_id = $this->savePost( $normalized, $existing, $mappings );
+			} catch ( ContentImportException $exception ) {
+				$errors[] = $exception->getMessage();
+				$this->logger->error(
+					$exception->getMessage(),
+					array(
+						'bucket' => $exception->bucket(),
+						'record' => $exception->record(),
+					)
+				);
+				continue;
+			}
+
+			if ( $existing > 0 ) {
+				++$updated;
+			} else {
+				++$created;
+			}
+
+			$this->saveMeta( $post_id, $normalized, $context, $mappings );
+		}
+
+		if ( array() !== $errors ) {
+			return SyncResult::merge(
+				array(
+					SyncResult::success(
+						array(
+							'created'   => $created,
+							'updated'   => $updated,
+							'skipped'   => $skipped,
+							'conflicts' => $conflicts,
+						)
+					),
+					SyncResult::failure( $errors ),
+				)
+			);
+		}
+
+		$this->logger->info(
+			'Imported post content records.',
+			array(
+				'created'   => $created,
+				'updated'   => $updated,
+				'skipped'   => $skipped,
+				'conflicts' => $conflicts,
+			)
+		);
+
+		return SyncResult::success(
+			array(
+				'created'   => $created,
+				'updated'   => $updated,
+				'skipped'   => $skipped,
+				'conflicts' => $conflicts,
+			)
+		);
+	}
+
+	private function findExistingPostId( int $source_id ): int {
+		if ( $source_id <= 0 ) {
+			return 0;
+		}
+
+		$posts = get_posts(
+			array(
+				'post_type'  => 'any',
+				// phpcs:disable WordPress.DB.SlowDBQuery.slow_db_query_meta_key, WordPress.DB.SlowDBQuery.slow_db_query_meta_value -- Source ID lookup is the handler's stable import identity.
+				'meta_key'   => '_wpcs_source_id',
+				'meta_value' => (string) $source_id,
+				// phpcs:enable WordPress.DB.SlowDBQuery.slow_db_query_meta_key, WordPress.DB.SlowDBQuery.slow_db_query_meta_value
+			)
+		);
+
+		if ( array() === $posts ) {
+			return 0;
+		}
+
+		return (int) $posts[0]->ID;
+	}
+
+	/**
+	 * @param array<string, mixed> $record   Normalized post record.
+	 * @param int                  $existing Existing post ID.
+	 */
+	private function savePost( array $record, int $existing, UrlMappingCollection $mappings ): int {
+		$post_data = array(
+			'post_type'    => $record['post_type'],
+			'post_title'   => $record['post_title'],
+			'post_content' => $this->url_transformer->transformString( (string) $record['post_content'], $mappings ),
+			'post_excerpt' => $this->url_transformer->transformString( (string) $record['post_excerpt'], $mappings ),
+			'post_status'  => $record['post_status'],
+			'post_name'    => $record['post_name'],
+			'post_parent'  => $record['post_parent'],
+			'menu_order'   => $record['menu_order'],
+		);
+
+		if ( $existing > 0 ) {
+			$post_data['ID'] = $existing;
+
+			return $this->postIdFromResult( wp_update_post( $post_data, true ), $record );
+		}
+
+		return $this->postIdFromResult( wp_insert_post( $post_data, true ), $record );
+	}
+
+	/**
+	 * @param int|\WP_Error        $result Post save result.
+	 * @param array<string, mixed> $record Normalized post record.
+	 */
+	private function postIdFromResult( $result, array $record ): int {
+		if ( is_wp_error( $result ) || (int) $result <= 0 ) {
+			throw new ContentImportException(
+				$this->bucket(),
+				$record,
+				sprintf( 'Post import failed for source ID %d.', (int) $record['id'] )
+			);
+		}
+
+		return (int) $result;
+	}
+
+	/**
+	 * @param array<string, mixed> $record Normalized post record.
+	 */
+	private function saveMeta( int $post_id, array $record, SyncContext $context, UrlMappingCollection $mappings ): void {
+		update_post_meta( $post_id, '_wpcs_source_id', (int) $record['id'] );
+		update_post_meta( $post_id, '_wpcs_source_site', $context->sourceUrl() );
+
+		foreach ( $record['meta'] as $key => $value ) {
+			update_post_meta(
+				$post_id,
+				(string) $key,
+				$this->metadata_transformer->transformValue( $value, $mappings )
+			);
+		}
+	}
+
+	private function mappings( SyncContext $context ): UrlMappingCollection {
+		$mappings = array();
+
+		foreach ( $context->urlMappings() as $source => $destination ) {
+			$mappings[] = new UrlMapping( $source, $destination );
+		}
+
+		return new UrlMappingCollection( $mappings );
+	}
+}
@@ -0,0 +1,218 @@
+<?php
+/**
+ * Tests for post content imports.
+ *
+ * @package WPContentSync
+ */
+
+namespace WPContentSync\Tests\Unit\Content;
+
+use PHPUnit\Framework\TestCase;
+use WPContentSync\Content\ContentRecordNormalizer;
+use WPContentSync\Content\PostContentHandler;
+use WPContentSync\Logging\LoggerInterface;
+use WPContentSync\Sync\SyncContext;
+use WPContentSync\Url\MetadataUrlTransformer;
+use WPContentSync\Url\UrlTransformer;
+
+class PostContentHandlerTest extends TestCase {
+	/** @var array<int, array<string, mixed>> */
+	private array $logs = array();
+
+	protected function tearDown(): void {
+		unset(
+			$GLOBALS['wpcs_test_posts'],
+			$GLOBALS['wpcs_test_next_post_id'],
+			$GLOBALS['wpcs_test_post_meta']
+		);
+
+		$this->logs = array();
+
+		parent::tearDown();
+	}
+
+	public function test_it_creates_new_post_records(): void {
+		$result = $this->handler()->importRecords(
+			array(
+				$this->postRecord(),
+			),
+			$this->context( 'last_write_wins' )
+		);
+
+		self::assertTrue( $result->isSuccessful() );
+		self::assertSame( 1, $result->created() );
+		self::assertSame( 'Imported Title', get_post( 1 )['post_title'] );
+		self::assertSame( 42, get_post_meta( 1, '_wpcs_source_id', true ) );
+		self::assertSame( 'https://source.test', get_post_meta( 1, '_wpcs_source_site', true ) );
+	}
+
+	public function test_it_updates_existing_posts_with_last_write_wins(): void {
+		$post_id = wp_insert_post(
+			array(
+				'post_title' => 'Old Title',
+				'post_type'  => 'post',
+			),
+			true
+		);
+		update_post_meta( $post_id, '_wpcs_source_id', 42 );
+
+		$result = $this->handler()->importRecords(
+			array(
+				$this->postRecord( array( 'post_title' => 'New Title' ) ),
+			),
+			$this->context( 'last_write_wins' )
+		);
+
+		self::assertTrue( $result->isSuccessful() );
+		self::assertSame( 1, $result->updated() );
+		self::assertSame( 'New Title', get_post( $post_id )['post_title'] );
+	}
+
+	public function test_it_skips_existing_posts_with_manual_review_conflict(): void {
+		$post_id = wp_insert_post(
+			array(
+				'post_title' => 'Old Title',
+				'post_type'  => 'post',
+			),
+			true
+		);
+		update_post_meta( $post_id, '_wpcs_source_id', 42 );
+
+		$result = $this->handler()->importRecords(
+			array(
+				$this->postRecord( array( 'post_title' => 'New Title' ) ),
+			),
+			$this->context( 'manual_review' )
+		);
+
+		self::assertTrue( $result->isSuccessful() );
+		self::assertSame( 1, $result->skipped() );
+		self::assertSame( 1, $result->conflicts() );
+		self::assertSame( 'Old Title', get_post( $post_id )['post_title'] );
+		self::assertSame( 'Skipped post import because manual review is required.', $this->logs[0]['message'] );
+	}
+
+	public function test_it_rewrites_post_content_excerpt_and_meta_urls(): void {
+		$result = $this->handler()->importRecords(
+			array(
+				$this->postRecord(
+					array(
+						'post_content' => '<a href="https://source.test/page">Page</a>',
+						'post_excerpt' => 'Read https://source.test/page',
+						'meta'         => array(
+							'_source_url' => 'https://source.test/page',
+							'_json_links' => '{"url":"https://source.test/page"}',
+						),
+					)
+				),
+			),
+			$this->context( 'last_write_wins' )
+		);
+
+		self::assertTrue( $result->isSuccessful() );
+		self::assertSame( '<a href="https://destination.test/page">Page</a>', get_post( 1 )['post_content'] );
+		self::assertSame( 'Read https://destination.test/page', get_post( 1 )['post_excerpt'] );
+		self::assertSame( 'https://destination.test/page', get_post_meta( 1, '_source_url', true ) );
+		self::assertSame( '{"url":"https:\/\/destination.test\/page"}', get_post_meta( 1, '_json_links', true ) );
+	}
+
+	public function test_it_returns_failure_when_wordpress_rejects_post_save(): void {
+		$result = $this->handler()->importRecords(
+			array(
+				$this->postRecord(
+					array(
+						'id'        => 0,
+						'post_type' => '',
+					)
+				),
+			),
+			$this->context( 'last_write_wins' )
+		);
+
+		self::assertFalse( $result->isSuccessful() );
+		self::assertSame( array( 'Post import failed for source ID 0.' ), $result->errors() );
+		self::assertSame( array(), get_post_meta( 0, '_wpcs_source_id', false ) );
+	}
+
+
+	private function handler(): PostContentHandler {
+		return new PostContentHandler(
+			new ContentRecordNormalizer(),
+			new UrlTransformer(),
+			new MetadataUrlTransformer( new UrlTransformer() ),
+			$this->logger()
+		);
+	}
+
+	private function context( string $conflict_strategy ): SyncContext {
+		return SyncContext::forImport(
+			array( 'site_url' => 'https://source.test' ),
+			array( 'site_url' => 'https://destination.test' ),
+			$conflict_strategy,
+			'operation-1'
+		);
+	}
+
+	/**
+	 * @param array<string, mixed> $overrides Record overrides.
+	 * @return array<string, mixed>
+	 */
+	private function postRecord( array $overrides = array() ): array {
+		return array_merge(
+			array(
+				'id'           => 42,
+				'post_type'    => 'post',
+				'post_title'   => 'Imported Title',
+				'post_content' => 'Imported content',
+				'post_excerpt' => 'Imported excerpt',
+				'post_status'  => 'publish',
+				'post_name'    => 'imported-title',
+				'post_parent'  => 0,
+				'menu_order'   => 0,
+				'meta'         => array(),
+			),
+			$overrides
+		);
+	}
+
+	private function logger(): LoggerInterface {
+		return new class( $this->logs ) implements LoggerInterface {
+			/** @var array<int, array<string, mixed>> */
+			private array $logs;
+
+			/**
+			 * @param array<int, array<string, mixed>> $logs Logs.
+			 */
+			public function __construct( array &$logs ) {
+				$this->logs = &$logs;
+			}
+
+			public function error( string $message, array $context = array() ): void {
+				$this->record( 'error', $message, $context );
+			}
+
+			public function warning( string $message, array $context = array() ): void {
+				$this->record( 'warning', $message, $context );
+			}
+
+			public function info( string $message, array $context = array() ): void {
+				$this->record( 'info', $message, $context );
+			}
+
+			public function debug( string $message, array $context = array() ): void {
+				$this->record( 'debug', $message, $context );
+			}
+
+			/**
+			 * @param array<string, mixed> $context Context.
+			 */
+			private function record( string $level, string $message, array $context ): void {
+				$this->logs[] = array(
+					'level'   => $level,
+					'message' => $message,
+					'context' => $context,
+				);
+			}
+		};
+	}
+}
@@ -119,7 +119,7 @@ class WordPressContentStubTest extends TestCase {
 		);

 		self::assertCount( 1, $posts );
-		self::assertSame( $first_post_id, $posts[0]['ID'] );
+		self::assertSame( $first_post_id, $posts[0]->ID );
 		self::assertSame( array(), get_post_meta( $second_post_id, '_wpcs_source_id', false ) );
 		self::assertNull( get_post( $second_post_id ) );
 		self::assertTrue( $GLOBALS['wpcs_test_force_delete'][ $second_post_id ] );
@@ -545,6 +545,10 @@ if ( ! function_exists( 'wp_insert_post' ) ) {
 	 * @return int|\WP_Error
 	 */
 	function wp_insert_post( array $postarr, $wp_error = false ) {
+		if ( empty( $postarr['post_type'] ) ) {
+			return $wp_error ? new WP_Error( 'invalid_post_type', 'Post type is required.' ) : 0;
+		}
+
 		if ( isset( $postarr['ID'] ) && (int) $postarr['ID'] > 0 ) {
 			$post_id = (int) $postarr['ID'];
 		} else {
@@ -629,7 +633,7 @@ if ( ! function_exists( 'get_posts' ) ) {
 	 * Minimal posts query for unit tests.
 	 *
 	 * @param array<string, mixed> $args Query args.
-	 * @return array<int, array<string, mixed>>
+	 * @return array<int, object>
 	 */
 	function get_posts( array $args = array() ) {
 		$posts = array_values( $GLOBALS['wpcs_test_posts'] ?? array() );
@@ -650,12 +654,18 @@ if ( ! function_exists( 'get_posts' ) ) {
 				static function ( array $post ) use ( $args ): bool {
 					$values = $GLOBALS['wpcs_test_post_meta'][ (int) $post['ID'] ][ (string) $args['meta_key'] ] ?? array();

-					return in_array( $args['meta_value'], $values, true );
+					foreach ( $values as $value ) {
+						if ( (string) $args['meta_value'] === (string) $value ) {
+							return true;
+						}
+					}
+
+					return false;
 				}
 			);
 		}

-		return array_values( $posts );
+		return array_values( array_map( static fn( array $post ): object => (object) $post, $posts ) );
 	}
 }