<?php
/**
 * Document Processing Orchestrator.
 *
 * @package AISupportAssistant\Documents
 */

// phpcs:disable WordPress.DB.DirectDatabaseQuery

namespace AISupportAssistant\Documents;

use AISupportAssistant\Database\DatabaseManager;
use AISupportAssistant\Admin\Settings;
use AISupportAssistant\Embeddings\EmbeddingService;
use AISupportAssistant\Helpers\Logger;
use AISupportAssistant\Helpers\TextHelper;

if ( ! defined( 'ABSPATH' ) ) {
	exit;
}

/**
 * Class DocumentManager
 */
class DocumentManager {

	/**
	 * Upload and index a WordPress Media Library attachment.
	 *
	 * @param int $attachment_id Media attachment ID.
	 * @return int Document ID on success, 0 on failure.
	 */
	public static function upload_attachment( int $attachment_id ): int {
		global $wpdb;
		$doc_table = DatabaseManager::get_documents_table();

		$file_path = get_attached_file( $attachment_id );
		if ( ! $file_path || ! file_exists( $file_path ) ) {
			Logger::error( "Attachment file not found for ID $attachment_id" );
			return 0;
		}

		$filename  = basename( $file_path );
		$mime_type = get_post_mime_type( $attachment_id );
		$title     = get_the_title( $attachment_id );

		// Check if already registered.
		$existing_id = (int) $wpdb->get_var(
			$wpdb->prepare(
				"SELECT id FROM %i WHERE file_path = %s",
				$doc_table,
				$file_path
			)
		);

		$data = array(
			'title'      => $title,
			'filename'   => $filename,
			'mime_type'  => $mime_type,
			'file_path'  => $file_path,
			'status'     => 'pending',
			'created_at' => current_time( 'mysql' ),
			'updated_at' => current_time( 'mysql' ),
		);

		if ( $existing_id > 0 ) {
			$wpdb->update( $doc_table, array( 'status' => 'pending', 'updated_at' => current_time( 'mysql' ) ), array( 'id' => $existing_id ) );
			$document_id = $existing_id;
		} else {
			$wpdb->insert( $doc_table, $data );
			$document_id = (int) $wpdb->insert_id;
		}

		if ( $document_id > 0 ) {
			Logger::info( "Document registered successfully. Starting indexing.", array( 'document_id' => $document_id ) );
			self::index_document( $document_id );
			return $document_id;
		}

		return 0;
	}

	/**
	 * Process text extraction, chunking, and vector embedding uploads.
	 *
	 * @param int $document_id Document database ID.
	 * @return bool True on success, false on failure.
	 */
	public static function index_document( int $document_id ): bool {
		global $wpdb;
		$doc_table   = DatabaseManager::get_documents_table();
		$chunks_table = DatabaseManager::get_chunks_table();

		// Get document record.
		$doc = $wpdb->get_row( $wpdb->prepare( "SELECT * FROM %i WHERE id = %d", $doc_table, $document_id ) );
		if ( ! $doc ) {
			Logger::error( "Document ID $document_id not found during indexing." );
			return false;
		}

		// Update status to indexing.
		$wpdb->update( $doc_table, array( 'status' => 'indexing', 'updated_at' => current_time( 'mysql' ) ), array( 'id' => $document_id ) );

		try {
			// Get settings.
			$settings      = Settings::get_instance();
			$chunk_size    = $settings->get( 'chunk_size', 700 );
			$chunk_overlap = $settings->get( 'chunk_overlap', 100 );
			$vector_store  = EmbeddingService::get_vector_store();

			// 1. Clean up existing chunks in MySQL and Vector Store.
			$vector_store->delete_document( $document_id );
			$wpdb->delete( $chunks_table, array( 'document_id' => $document_id ) );

			// 2. Extract Text.
			$extractor = ExtractorFactory::get_extractor( $doc->mime_type, $doc->filename );
			if ( ! $extractor ) {
				throw new \Exception( 'Unsupported file type format: ' . $doc->mime_type );
			}

			$text = $extractor->extract( $doc->file_path );
			if ( empty( trim( $text ) ) ) {
				if ( 'application/pdf' === $doc->mime_type || 'pdf' === strtolower( pathinfo( $doc->filename, PATHINFO_EXTENSION ) ) ) {
					throw new \Exception( 'Extracted PDF text content is empty. This document may be scanned or image-only; please ensure it contains selectable text.' );
				}
				throw new \Exception( 'Extracted document text content is empty.' );
			}

			// 3. Chunk Text.
			$chunks = ChunkingService::chunk( $text, $chunk_size, $chunk_overlap );
			if ( empty( $chunks ) ) {
				throw new \Exception( 'Failed to partition document text into chunks.' );
			}

			Logger::info( sprintf( 'Document chunked into %d elements.', count( $chunks ) ), array( 'document_id' => $document_id ) );

			// 4. Batch Embedding Generation and Storage.
			// Group chunks for batch processing to minimize API requests.
			$batch_size = 16;
			$chunks_count = count( $chunks );
			
			for ( $offset = 0; $offset < $chunks_count; $offset += $batch_size ) {
				$batch = array_slice( $chunks, $offset, $batch_size );
				
				foreach ( $batch as $chunk_key => $chunk ) {
					$chunk_text = $chunk['chunk_text'];
					
					// Generate vector embedding.
					$vector = EmbeddingService::generate( $chunk_text );
					if ( empty( $vector ) ) {
						throw new \Exception( "Failed to generate vector embedding for chunk #{$chunk['chunk_number']}." );
					}

					// Upsert to local MySQL cache AND selected vector database.
					$meta = array(
						'page_number' => $chunk['page_number'],
						'token_count' => $chunk['token_count'],
						'filename'    => $doc->filename,
					);

					// Local database cache insert.
					$local_saved = $vector_store->upsert( $document_id, $chunk['chunk_number'], $vector, $chunk_text, $meta );
					if ( ! $local_saved ) {
						throw new \Exception( "Failed to save vector embedding in store for chunk #{$chunk['chunk_number']}." );
					}
				}
			}

			// Complete index status.
			$wpdb->update( $doc_table, array( 'status' => 'completed', 'error_message' => null, 'updated_at' => current_time( 'mysql' ) ), array( 'id' => $document_id ) );
			delete_transient( 'ai_support_assistant_statistics' ); // Invalidate statistics cache.
			Logger::info( "Document ID $document_id indexed successfully." );
			return true;

		} catch ( \Exception $e ) {
			Logger::error( "Indexing failed for Document ID $document_id: " . $e->getMessage() );
			$wpdb->update(
				$doc_table,
				array(
					'status'        => 'failed',
					'error_message' => $e->getMessage(),
					'updated_at'    => current_time( 'mysql' ),
				),
				array( 'id' => $document_id )
			);
			throw $e;
		}
	}

	/**
	 * Crawl a webpage URL and index it.
	 *
	 * @param string $url Webpage URL.
	 * @return int Document ID on success, 0 on failure.
	 */
	public static function crawl_and_index_url( string $url ): int {
		global $wpdb;
		$doc_table   = DatabaseManager::get_documents_table();
		$chunks_table = DatabaseManager::get_chunks_table();

		// Check if already registered.
		$existing_id = (int) $wpdb->get_var(
			$wpdb->prepare(
				"SELECT id FROM %i WHERE file_path = %s",
				$doc_table,
				$url
			)
		);

		// Crawl the webpage.
		try {
			$crawled = UrlCrawler::crawl( $url );
		} catch ( \Exception $e ) {
			Logger::error( "Crawl failed for URL $url: " . $e->getMessage() );
			
			// If already registered, set status to failed.
			if ( $existing_id > 0 ) {
				$wpdb->update(
					$doc_table,
					array(
						'status'        => 'failed',
						'error_message' => $e->getMessage(),
						'updated_at'    => current_time( 'mysql' ),
					),
					array( 'id' => $existing_id )
				);
			}
			throw $e;
		}

		$title    = $crawled['title'];
		$text     = $crawled['text'];
		$filename = wp_parse_url( $url, PHP_URL_HOST ) . wp_parse_url( $url, PHP_URL_PATH );
		$filename = trim( $filename, '/' );

		$data = array(
			'title'      => $title,
			'filename'   => $filename,
			'mime_type'  => 'text/html',
			'file_path'  => $url,
			'status'     => 'indexing',
			'created_at' => current_time( 'mysql' ),
			'updated_at' => current_time( 'mysql' ),
		);

		if ( $existing_id > 0 ) {
			$wpdb->update(
				$doc_table,
				array(
					'title'         => $title,
					'status'        => 'indexing',
					'error_message' => null,
					'updated_at'    => current_time( 'mysql' ),
				),
				array( 'id' => $existing_id )
			);
			$document_id = $existing_id;
		} else {
			$wpdb->insert( $doc_table, $data );
			$document_id = (int) $wpdb->insert_id;
		}

		if ( $document_id <= 0 ) {
			return 0;
		}

		try {
			// Get settings.
			$settings      = Settings::get_instance();
			$chunk_size    = $settings->get( 'chunk_size', 700 );
			$chunk_overlap = $settings->get( 'chunk_overlap', 100 );
			$vector_store  = EmbeddingService::get_vector_store();

			// 1. Clean up existing chunks in MySQL and Vector Store.
			$vector_store->delete_document( $document_id );
			$wpdb->delete( $chunks_table, array( 'document_id' => $document_id ) );

			// 2. Chunk text content.
			$chunks = ChunkingService::chunk( $text, $chunk_size, $chunk_overlap );
			if ( empty( $chunks ) ) {
				throw new \Exception( 'Failed to partition webpage text into chunks.' );
			}

			Logger::info( sprintf( 'URL %s chunked into %d elements.', $url, count( $chunks ) ), array( 'document_id' => $document_id ) );

			// 3. Batch Embedding Generation and Storage.
			$batch_size = 16;
			$chunks_count = count( $chunks );
			
			for ( $offset = 0; $offset < $chunks_count; $offset += $batch_size ) {
				$batch = array_slice( $chunks, $offset, $batch_size );
				
				foreach ( $batch as $chunk ) {
					$chunk_text = $chunk['chunk_text'];
					
					// Generate vector embedding.
					$vector = EmbeddingService::generate( $chunk_text );
					if ( empty( $vector ) ) {
						throw new \Exception( "Failed to generate vector embedding for chunk #{$chunk['chunk_number']}." );
					}

					$meta = array(
						'page_number' => $chunk['page_number'],
						'token_count' => $chunk['token_count'],
						'filename'    => $filename,
					);

					// Local database cache insert.
					$local_saved = $vector_store->upsert( $document_id, $chunk['chunk_number'], $vector, $chunk_text, $meta );
					if ( ! $local_saved ) {
						throw new \Exception( "Failed to save vector embedding in store for chunk #{$chunk['chunk_number']}." );
					}
				}
			}

			// Complete index status.
			$wpdb->update(
				$doc_table,
				array(
					'status'        => 'completed',
					'error_message' => null,
					'updated_at'    => current_time( 'mysql' ),
				),
				array( 'id' => $document_id )
			);
			delete_transient( 'ai_support_assistant_statistics' ); // Invalidate statistics cache.
			Logger::info( "URL ID $document_id indexed successfully." );
			return $document_id;
		} catch ( \Exception $e ) {
			Logger::error( "Indexing failed for URL ID $document_id: " . $e->getMessage() );
			$wpdb->update(
				$doc_table,
				array(
					'status'        => 'failed',
					'error_message' => $e->getMessage(),
					'updated_at'    => current_time( 'mysql' ),
				),
				array( 'id' => $document_id )
			);
			throw $e;
		}
	}

	/**
	 * Delete a document and all associated text chunks and vector embeddings.
	 *
	 * @param int $document_id Document database ID.
	 * @return bool True on success.
	 */
	public static function delete_document( int $document_id ): bool {
		global $wpdb;
		$doc_table = DatabaseManager::get_documents_table();

		// Delete vectors from active store.
		$vector_store = EmbeddingService::get_vector_store();
		$vector_store->delete_document( $document_id );

		// Delete from local cache tables.
		$wpdb->delete( DatabaseManager::get_chunks_table(), array( 'document_id' => $document_id ) );
		$result = $wpdb->delete( $doc_table, array( 'id' => $document_id ) );

		delete_transient( 'ai_support_assistant_statistics' );
		Logger::info( "Document ID $document_id and all its chunks deleted." );

		return false !== $result;
	}
}