<?php
/**
 * Block-aware content translator.
 *
 * Walks the parsed block tree from `parse_blocks()`, translates each leaf
 * block's `innerHTML` plus a curated set of text attributes for known core
 * blocks, then reserializes via `serialize_blocks()`. Keeps Gutenberg block
 * structure intact across complex/dynamic blocks where the v0.1 "translate
 * the whole post_content as HTML" path could mangle attribute JSON.
 *
 * @package AdaraTranslate
 */

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

final class Adara_Translate_Block_Translator {

	/**
	 * Block names whose content must NOT be translated. Their innerHTML is
	 * code, raw HTML, or a shortcode literal — translation would corrupt it.
	 * Curated attrs are still translated for these blocks (none are listed
	 * below, so in practice nothing inside them changes).
	 */
	private const SKIP_CONTENT = [
		'core/code',
		'core/html',
		'core/shortcode',
		'core/freeform',
	];

	/**
	 * Per-block-type list of attribute keys whose string value is user-facing
	 * text and must be translated alongside innerHTML. Anything not listed
	 * here is preserved unchanged.
	 *
	 * Add to this list as users hit cases — keep it conservative; translating
	 * a key that holds a URL, ID, or technical token will silently corrupt
	 * the block on the next editor open.
	 */
	private const TRANSLATABLE_ATTRS = [
		'core/button'      => [ 'text' ],
		'core/image'       => [ 'alt', 'caption' ],
		'core/cover'       => [ 'alt' ],
		'core/quote'       => [ 'citation' ],
		'core/pullquote'   => [ 'citation' ],
		'core/audio'       => [ 'caption' ],
		'core/video'       => [ 'caption' ],
		'core/embed'       => [ 'caption' ],
		'core/file'        => [ 'downloadButtonText' ],
		'core/search'      => [ 'label', 'buttonText', 'placeholder' ],
		'core/post-title'  => [ 'placeholder' ],
		'core/site-title'  => [ 'placeholder' ],
		'core/site-tagline' => [ 'placeholder' ],
	];

	/**
	 * Translate a post_content string.
	 *
	 * @return string|WP_Error Translated HTML/block markup on success.
	 */
	public static function translate_content( string $content, string $target_lang, ?string $source_lang ) {
		if ( '' === trim( $content ) ) {
			return '';
		}

		if ( ! has_blocks( $content ) ) {
			$result = Adara_Translate_Client::translate( $content, $target_lang, $source_lang );
			if ( is_wp_error( $result ) ) {
				return $result;
			}
			return (string) $result['translatedText'];
		}

		$blocks = parse_blocks( $content );
		$out    = [];
		foreach ( $blocks as $block ) {
			$translated = self::translate_block( $block, $target_lang, $source_lang );
			if ( is_wp_error( $translated ) ) {
				return $translated;
			}
			$out[] = $translated;
		}

		return serialize_blocks( $out );
	}

	/**
	 * Translate a single block (recursively for innerBlocks).
	 *
	 * @return array|WP_Error
	 */
	private static function translate_block( array $block, string $target_lang, ?string $source_lang ) {
		$name = isset( $block['blockName'] ) ? (string) $block['blockName'] : '';

		// Translate curated text attributes first. These exist regardless of
		// whether the block has innerBlocks.
		if ( '' !== $name && isset( self::TRANSLATABLE_ATTRS[ $name ] ) && ! empty( $block['attrs'] ) && is_array( $block['attrs'] ) ) {
			foreach ( self::TRANSLATABLE_ATTRS[ $name ] as $attr_key ) {
				if ( ! isset( $block['attrs'][ $attr_key ] ) || ! is_string( $block['attrs'][ $attr_key ] ) ) {
					continue;
				}
				if ( '' === trim( $block['attrs'][ $attr_key ] ) ) {
					continue;
				}
				$attr_result = Adara_Translate_Client::translate(
					$block['attrs'][ $attr_key ],
					$target_lang,
					$source_lang
				);
				if ( is_wp_error( $attr_result ) ) {
					return $attr_result;
				}
				$block['attrs'][ $attr_key ] = (string) $attr_result['translatedText'];
			}
		}

		$content_locked = in_array( $name, self::SKIP_CONTENT, true );

		$has_children = ! empty( $block['innerBlocks'] ) && is_array( $block['innerBlocks'] );

		if ( $has_children ) {
			$translated_children = [];
			foreach ( $block['innerBlocks'] as $child ) {
				$result = self::translate_block( $child, $target_lang, $source_lang );
				if ( is_wp_error( $result ) ) {
					return $result;
				}
				$translated_children[] = $result;
			}
			$block['innerBlocks'] = $translated_children;

			// Parent blocks store wrapping markup in innerContent[] interleaved
			// with null placeholders for child positions. Translate only
			// fragments that contain visible text — most wrappers are pure
			// markup (e.g. "<div class=\"wp-block-columns\">") and don't need
			// a translation API call.
			if ( ! $content_locked && ! empty( $block['innerContent'] ) && is_array( $block['innerContent'] ) ) {
				foreach ( $block['innerContent'] as $i => $piece ) {
					if ( null === $piece || ! is_string( $piece ) ) {
						continue;
					}
					if ( '' === trim( wp_strip_all_tags( $piece ) ) ) {
						continue;
					}
					$piece_result = Adara_Translate_Client::translate( $piece, $target_lang, $source_lang );
					if ( is_wp_error( $piece_result ) ) {
						return $piece_result;
					}
					$block['innerContent'][ $i ] = (string) $piece_result['translatedText'];
				}
			}
		} else {
			// Leaf block: translate innerHTML, then keep innerContent in sync.
			// serialize_block() prefers innerContent over innerHTML for the
			// final output, so both must hold the translated string.
			if ( ! $content_locked ) {
				$html = isset( $block['innerHTML'] ) ? (string) $block['innerHTML'] : '';
				if ( '' !== trim( wp_strip_all_tags( $html ) ) ) {
					$leaf_result = Adara_Translate_Client::translate( $html, $target_lang, $source_lang );
					if ( is_wp_error( $leaf_result ) ) {
						return $leaf_result;
					}
					$translated_html        = (string) $leaf_result['translatedText'];
					$block['innerHTML']     = $translated_html;
					$block['innerContent']  = [ $translated_html ];
				}
			}
		}

		return $block;
	}
}