<?php
/**
 * Block-safe translation of strings, block markup and posts.
 *
 * Only text-bearing HTML chunks and whitelisted string attributes are sent to
 * the provider (BlockTexts); identical strings are sent once; whitespace
 * around each string is kept; page context and per-text notes go to AI
 * providers so short labels are translated in context.
 */

declare( strict_types = 1 );

namespace AddBlack\Lang\MachineTranslation;

defined( 'ABSPATH' ) || exit;

use AddBlack\Lang\Settings\SettingsRepository;
use WP_Error;
use WP_Post;

final class Translator {

	public function __construct(
		private readonly SettingsRepository $settings,
		private readonly ProviderFactory $providers
	) {}

	/**
	 * Translate a list of strings. Kind per item: "html" (default) or "text".
	 *
	 * @param array<string|int, string>                                                 $strings
	 * @param array<string|int, string>                                                 $kinds   Same keys as $strings.
	 * @param array{hints?: array<string|int, string>, context?: array<string, string>} $options
	 *        hints: one note per string (same keys); context: title, type, outline, style.
	 * @return array<string|int, string>|WP_Error Same keys as $strings.
	 */
	public function translate_strings( array $strings, array $kinds, string $from, string $to, ?Provider $provider = null, array $options = array() ): array|WP_Error {
		$settings  = $this->settings->get();
		$from_lang = $settings->language( $from );
		$to_lang   = $settings->language( $to );
		if ( ! $from_lang || ! $to_lang ) {
			return new WP_Error( 'blocklingo_bad_lang', __( 'Unknown language.', 'blocklingo' ) );
		}
		$provider = $provider ?? $this->providers->make();
		if ( is_wp_error( $provider ) ) {
			return $provider;
		}

		$unique = array(); // Payload => index of the first occurrence.
		$plan   = array(); // Key => [ lead, payload|null, trail, kind, original ].
		$hints  = (array) ( $options['hints'] ?? array() );
		$notes  = array(); // One note per unique payload.

		foreach ( $strings as $k => $s ) {
			$s    = (string) $s;
			$kind = $kinds[ $k ] ?? 'html';
			if ( ! BlockTexts::has_text( $s ) ) {
				$plan[ $k ] = array( '', null, '', $kind, $s );
				continue;
			}
			preg_match( '/^(\s*)(.*?)(\s*)$/su', $s, $m );
			$payload = 'text' === $kind ? esc_html( $m[2] ) : $m[2];
			if ( ! isset( $unique[ $payload ] ) ) {
				$notes[]            = (string) ( $hints[ $k ] ?? '' );
				$unique[ $payload ] = count( $unique );
			}
			$plan[ $k ] = array( $m[1], $payload, $m[3], $kind, $s );
		}

		$translated = array();
		if ( $unique ) {
			$context_in = (array) ( $options['context'] ?? array() );
			if ( ! isset( $context_in['style'] ) ) {
				$context_in['style'] = $settings->style_for( $to );
			}
			$context    = $this->build_context( $context_in )->with_notes( $notes );
			$translated = $provider->translate( array_map( 'strval', array_keys( $unique ) ), $from_lang, $to_lang, $context );
			if ( is_wp_error( $translated ) ) {
				return $translated;
			}
		}

		$out = array();
		foreach ( $plan as $k => $p ) {
			list( $lead, $payload, $trail, $kind, $orig ) = $p;
			if ( null === $payload ) {
				$out[ $k ] = $orig;
				continue;
			}
			$t = $translated[ $unique[ $payload ] ];
			if ( 'text' === $kind ) {
				$t = html_entity_decode( $t, ENT_QUOTES | ENT_HTML5, 'UTF-8' );
			}
			$out[ $k ] = $lead . MarkupGuard::clean( $t, $orig ) . $trail;
		}
		return $out;
	}

	/**
	 * Translate block markup that is not a post (template parts, patterns).
	 *
	 * @param array<string, string> $context title, type.
	 */
	public function translate_content( string $content, string $from, string $to, array $context = array() ): string|WP_Error {
		$blocks = parse_blocks( $content );
		$found  = BlockTexts::collect( $blocks );
		if ( ! $found->refs ) {
			return $content;
		}
		$strings = array();
		$hints   = array();
		$kinds   = array();
		foreach ( $found->texts() as $i => $text ) {
			$strings[ 'c' . $i ] = $text;
			$hints[ 'c' . $i ]   = $found->where[ $i ];
			$kinds[ 'c' . $i ]   = $found->kinds[ $i ];
		}
		$context['outline'] = implode( "\n", $found->texts() );

		$result = $this->translate_strings(
			$strings,
			$kinds,
			$from,
			$to,
			null,
			array(
				'hints'   => $hints,
				'context' => $context,
			)
		);
		if ( is_wp_error( $result ) ) {
			return $result;
		}
		foreach ( array_keys( $found->refs ) as $i ) {
			$found->refs[ $i ] = $result[ 'c' . $i ];
		}
		return serialize_blocks( $blocks );
	}

	/**
	 * Title, excerpt and block content of a post, in one provider pass.
	 *
	 * @return array{title: string, excerpt: string, content: string}|WP_Error
	 */
	public function translate_post( WP_Post $post, string $from, string $to ): array|WP_Error {
		$blocks = parse_blocks( $post->post_content );
		$found  = BlockTexts::collect( $blocks );

		$strings = array();
		$hints   = array(
			'title'   => 'Page title',
			'excerpt' => 'Excerpt',
		);
		$kinds   = array(
			'title'   => 'text',
			'excerpt' => 'text',
		);
		foreach ( $found->texts() as $i => $text ) {
			$strings[ 'c' . $i ] = $text;
			$hints[ 'c' . $i ]   = $found->where[ $i ];
			$kinds[ 'c' . $i ]   = $found->kinds[ $i ];
		}
		$strings['title']   = $post->post_title;
		$strings['excerpt'] = $post->post_excerpt;

		$type   = get_post_type_object( $post->post_type );
		$result = $this->translate_strings(
			$strings,
			$kinds,
			$from,
			$to,
			null,
			array(
				'hints'   => $hints,
				'context' => array(
					'title'   => $post->post_title,
					'type'    => $type ? (string) $type->labels->singular_name : $post->post_type,
					'outline' => implode( "\n", $found->texts() ),
				),
			)
		);
		if ( is_wp_error( $result ) ) {
			return $result;
		}
		foreach ( array_keys( $found->refs ) as $i ) {
			$found->refs[ $i ] = $result[ 'c' . $i ];
		}

		return array(
			'title'   => $result['title'],
			'excerpt' => $result['excerpt'],
			'content' => serialize_blocks( $blocks ),
		);
	}

	/**
	 * Page context for AI providers: the site name added, the outline as plain
	 * text capped at blocklingo_context_max_chars (6000).
	 *
	 * @param array<string, mixed> $c title, type, outline, style.
	 */
	public function build_context( array $c ): Context {
		$site    = trim( wp_specialchars_decode( (string) get_bloginfo( 'name' ), ENT_QUOTES ) );
		$desc    = trim( wp_specialchars_decode( (string) get_bloginfo( 'description' ), ENT_QUOTES ) );
		$outline = '';
		if ( ! empty( $c['outline'] ) ) {
			$plain   = html_entity_decode( wp_strip_all_tags( (string) $c['outline'] ), ENT_QUOTES | ENT_HTML5, 'UTF-8' );
			$lines   = array_filter(
				array_map(
					static fn( string $l ): string => trim( (string) preg_replace( '/\s+/u', ' ', $l ) ),
					(array) preg_split( '/\r\n|\r|\n/', $plain )
				),
				static fn( string $s ): bool => '' !== $s
			);
			$outline = implode( "\n", $lines );
			$max     = (int) apply_filters( 'blocklingo_context_max_chars', 6000 );
			if ( mb_strlen( $outline ) > $max ) {
				$outline = mb_substr( $outline, 0, $max ) . ' …';
			}
		}
		return new Context(
			$desc ? $site . ' - ' . $desc : $site,
			sanitize_text_field( (string) ( $c['title'] ?? '' ) ),
			sanitize_text_field( (string) ( $c['type'] ?? '' ) ),
			$outline,
			sanitize_textarea_field( (string) ( $c['style'] ?? '' ) )
		);
	}
}