<?php
/**
 * Builds the markdown document (frontmatter + body) for a post.
 *
 * @package AEL42A
 */

namespace AEL42A;

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

/**
 * Rendered-HTML -> markdown document builder.
 *
 * Frontmatter mirrors the 42A mirror format (convex/mirroring/readability.ts)
 * adapted for first-party serving:
 *
 *   ---
 *   source_url: https://example.com/hello-world/
 *   title: "Hello: World"
 *   generated_at: 2026-07-06T12:00:00Z
 *   host: example.com
 *   type: post
 *   generator: ael-42a/1.0.0
 *   ---
 */
class Markdown_Converter {

	/**
	 * Build (or fetch from cache) the full markdown document for a post.
	 *
	 * @param \WP_Post $post Post object.
	 * @return string Markdown document.
	 */
	public function document( $post ) {
		$cached = Cache::get_markdown( $post );
		if ( null !== $cached ) {
			return $cached;
		}

		$markdown = $this->build_document( $post );
		Cache::set_markdown( $post, $markdown );

		return $markdown;
	}

	/**
	 * Build the document without touching the cache.
	 *
	 * @param \WP_Post $post Post object.
	 * @return string
	 */
	public function build_document( $post ) {
		$permalink = get_permalink( $post );
		$title     = get_the_title( $post );
		$host      = (string) wp_parse_url( home_url(), PHP_URL_HOST );

		$frontmatter = array(
			'---',
			'source_url: ' . $this->yaml_quote_if_needed( $permalink ),
			'title: ' . $this->yaml_quote( $title ),
			'generated_at: ' . gmdate( 'Y-m-d\TH:i:s\Z' ),
			'host: ' . $host,
			'type: ' . $post->post_type,
			'generator: ael-42a/' . AEL42A_VERSION,
			'---',
		);

		$body = $this->body( $post );

		$document = implode( "\n", $frontmatter ) . "\n\n# " . $title . "\n\n" . $body;

		if ( 'product' === $post->post_type && function_exists( 'wc_get_product' ) ) {
			$details = $this->product_details( $post );
			if ( '' !== $details ) {
				$document = rtrim( $document ) . "\n\n" . $details;
			}
		}

		return rtrim( $document ) . "\n";
	}

	/**
	 * Convert the rendered post content to markdown.
	 *
	 * @param \WP_Post $post Post object.
	 * @return string
	 */
	public function body( $post ) {
		$html = apply_filters( 'the_content', $post->post_content ); // phpcs:ignore WordPress.NamingConventions.PrefixAllGlobals.NonPrefixedHooknameFound -- applying the WordPress core content filter, not defining a plugin hook.
		$html = (string) $html;

		// Strauss-prefixed vendor build takes precedence when available.
		$league_class = '\\AEL42A\\Vendor\\League\\HTMLToMarkdown\\HtmlConverter';
		if ( class_exists( $league_class ) ) {
			try {
				$converter = new $league_class(
					array(
						'strip_tags'   => true,
						'header_style' => 'atx',
						'remove_nodes' => 'script style',
					)
				);
				return trim( $converter->convert( $html ) ) . "\n";
			} catch ( \Throwable $e ) {
				// Fall through to the built-in converter.
			}
		}

		$fallback = new Html_To_Markdown_Fallback();
		return $fallback->convert( $html );
	}

	/**
	 * Structured product details section for WooCommerce products.
	 *
	 * @param \WP_Post $post Product post.
	 * @return string Markdown section, or '' when unavailable.
	 */
	private function product_details( $post ) {
		$product = wc_get_product( $post->ID );
		if ( ! $product ) {
			return '';
		}

		$categories = array();
		$terms      = get_the_terms( $post->ID, 'product_cat' );
		if ( is_array( $terms ) ) {
			foreach ( $terms as $term ) {
				$categories[] = $term->name;
			}
		}

		$rows   = array();
		$rows[] = '## ' . __( 'Product details', 'ael-42a' );
		$rows[] = '';
		$rows[] = '- ' . __( 'Price', 'ael-42a' ) . ': ' . ( '' !== (string) $product->get_price() ? (string) $product->get_price() : __( 'n/a', 'ael-42a' ) );
		$rows[] = '- ' . __( 'Currency', 'ael-42a' ) . ': ' . get_woocommerce_currency();
		if ( $product->get_sku() ) {
			$rows[] = '- ' . __( 'SKU', 'ael-42a' ) . ': ' . $product->get_sku();
		}
		$rows[] = '- ' . __( 'Stock status', 'ael-42a' ) . ': ' . $product->get_stock_status();
		if ( ! empty( $categories ) ) {
			$rows[] = '- ' . __( 'Categories', 'ael-42a' ) . ': ' . implode( ', ', $categories );
		}

		return implode( "\n", $rows );
	}

	/**
	 * Always double-quote a YAML value (used for title).
	 *
	 * @param string $value Raw value.
	 * @return string
	 */
	private function yaml_quote( $value ) {
		$escaped = str_replace( array( '\\', '"' ), array( '\\\\', '\\"' ), (string) $value );
		$escaped = str_replace( array( "\r", "\n" ), ' ', $escaped );
		return '"' . $escaped . '"';
	}

	/**
	 * Quote a YAML scalar only when it contains special characters.
	 *
	 * @param string $value Raw value.
	 * @return string
	 */
	private function yaml_quote_if_needed( $value ) {
		$value = (string) $value;
		if ( preg_match( '/[:#\[\]{},&*?|<>=!%@`\'"\\\\]/', $value ) || '' === $value ) {
			return $this->yaml_quote( $value );
		}
		return $value;
	}
}