<?php
/**
 * HTML analysis / lightweight markdown conversion helpers.
 *
 * @package AiRankSuite
 */

declare(strict_types=1);

namespace AiRankSuite\Support;

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

final class Html {

	/**
	 * Rendered post content (blocks + shortcodes applied), filterable for builders.
	 */
	public static function rendered( \WP_Post $post ): string {
		/**
		 * Filter the raw content before rendering — the hook page builders use to inject their markup for analysis.
		 *
		 * @param string   $content Raw post_content.
		 * @param \WP_Post $post    The post being rendered.
		 */
		$content = (string) apply_filters( 'airanksuite/content/raw', $post->post_content, $post );
		$content = do_blocks( $content );
		$content = do_shortcode( $content );
		return $content;
	}

	/**
	 * Post content prepared for an AI provider.
	 *
	 * Deliberately not `rendered()`. That runs do_shortcode(), which executes
	 * arbitrary plugin code before the redactor ever sees the output — a
	 * [customer_list], a form plugin's prefill, a members-only block would
	 * each render their data straight into the prompt, and stripping addresses
	 * afterwards does not put that back. Here shortcodes are removed rather
	 * than run, and the private marker is honoured before anything else so a
	 * section marked private cannot be reached even by a filter.
	 *
	 * Blocks are still rendered: static blocks are the page's own markup, and
	 * without them a block-built page is empty text. Dynamic blocks can be
	 * excluded per site through the filter.
	 */
	public static function for_ai( \WP_Post $post ): string {
		$content = (string) $post->post_content;

		// Private sections first, before any filter or renderer can move them.
		$content = self::without_private( $content );

		/**
		 * Filters the raw content sent to an AI provider.
		 *
		 * Separate from `airanksuite/content/raw` on purpose: a page builder may want
		 * its markup analysed locally without also shipping it to a vendor.
		 *
		 * @param string   $content Raw post_content, private sections removed.
		 * @param \WP_Post $post    The post.
		 */
		$content = (string) apply_filters( 'airanksuite/ai/content/raw', $content, $post );

		// Remove, never execute.
		$content = strip_shortcodes( $content );

		/*
		 * Dynamic blocks are rendered by their plugin's PHP, which can print
		 * anything the page's viewer would see — a members-only notice, a
		 * personalised list, a private field. Static blocks are just the
		 * author's own saved markup and are safe to render.
		 *
		 * So dynamic blocks are dropped by default rather than run. A site that
		 * knows a particular dynamic block is safe (and wants its text analysed)
		 * can allow it by name through the filter.
		 */
		$blocks = parse_blocks( $content );
		$safe   = array();
		foreach ( $blocks as $block ) {
			$name = (string) ( $block['blockName'] ?? '' );
			if ( '' === $name ) {
				$safe[] = $block; // Freeform / classic content.
				continue;
			}
			$type = \WP_Block_Type_Registry::get_instance()->get_registered( $name );
			$dynamic = $type && is_callable( $type->render_callback );

			/**
			 * Filters whether a dynamic block may be rendered for an AI prompt.
			 *
			 * @param bool   $allow Default false for dynamic blocks.
			 * @param string $name  Block name, e.g. "core/latest-posts".
			 */
			if ( $dynamic && ! (bool) apply_filters( 'airanksuite/ai/allow_dynamic_block', false, $name ) ) {
				continue;
			}
			$safe[] = $block;
		}
		$content = implode( '', array_map( 'render_block', $safe ) );

		// render_block can re-introduce shortcodes from a block's saved content.
		$content = strip_shortcodes( $content );

		return $content;
	}

	/**
	 * Remove every [airanksuite-private]…[/airanksuite-private] (or the older [airs-private]) section.
	 *
	 * The marker is not a registered shortcode, so nothing in WordPress takes
	 * it out on its own — every path that publishes or transmits content has
	 * to ask. The regex lived in two places and now lives here, because a
	 * third copy is how one of them ends up a version behind.
	 *
	 * @param string $content Raw or rendered content.
	 */
	public static function without_private( string $content ): string {
		return (string) preg_replace( '#\[(airanksuite|airs)-private\].*?\[/\1-private\]#is', '', $content );
	}

	/**
	 * Plain text with entities decoded and whitespace normalized.
	 */
	public static function text( string $html ): string {
		$text = wp_strip_all_tags( $html, false );
		$text = html_entity_decode( $text, ENT_QUOTES | ENT_HTML5, 'UTF-8' );
		return trim( (string) preg_replace( '/[ \t]+/', ' ', $text ) );
	}

	public static function word_count( string $html ): int {
		$text = self::text( $html );
		return '' === $text ? 0 : count( preg_split( '/\s+/u', $text ) ?: array() );
	}

	/**
	 * All headings: [ [level, text], … ].
	 *
	 * @return array<int, array{0:int, 1:string}>
	 */
	public static function headings( string $html ): array {
		$out = array();
		if ( preg_match_all( '/<h([1-6])[^>]*>(.*?)<\/h\1>/is', $html, $m, PREG_SET_ORDER ) ) {
			foreach ( $m as $hit ) {
				$out[] = array( (int) $hit[1], self::text( $hit[2] ) );
			}
		}
		return $out;
	}

	/**
	 * Paragraph texts (non-empty).
	 *
	 * @return array<int, string>
	 */
	public static function paragraphs( string $html ): array {
		$out = array();
		if ( preg_match_all( '/<p[^>]*>(.*?)<\/p>/is', $html, $m ) ) {
			foreach ( $m[1] as $p ) {
				$text = self::text( $p );
				if ( '' !== $text ) {
					$out[] = $text;
				}
			}
		}
		if ( empty( $out ) ) {
			// Fallback for content without <p> tags (some builders): split on double newlines.
			$out = array_values( array_filter( array_map( 'trim', preg_split( '/\n{2,}/', self::text( $html ) ) ?: array() ) ) );
		}
		return $out;
	}

	/**
	 * Images with/without alt: [total, missing_alt].
	 *
	 * @return array{0:int, 1:int}
	 */
	public static function images_alt( string $html ): array {
		$total   = 0;
		$missing = 0;
		if ( preg_match_all( '/<img\b[^>]*>/i', $html, $m ) ) {
			foreach ( $m[0] as $img ) {
				++$total;
				if ( ! preg_match( '/\balt\s*=\s*("([^"]+)"|\'([^\']+)\')/i', $img ) ) {
					++$missing;
				}
			}
		}
		return array( $total, $missing );
	}

	/**
	 * Links split into [internal, external] counts relative to home host.
	 *
	 * @return array{0:int, 1:int}
	 */
	public static function links( string $html ): array {
		$internal = 0;
		$external = 0;
		$home     = (string) wp_parse_url( home_url(), PHP_URL_HOST );
		if ( preg_match_all( '/<a\b[^>]*href\s*=\s*("([^"]+)"|\'([^\']+)\')/i', $html, $m ) ) {
			foreach ( $m[2] as $i => $href ) {
				$href = '' !== $href ? $href : $m[3][ $i ];
				if ( str_starts_with( $href, '#' ) || str_starts_with( $href, 'mailto:' ) || str_starts_with( $href, 'tel:' ) ) {
					continue;
				}
				$host = (string) wp_parse_url( $href, PHP_URL_HOST );
				if ( '' === $host || $host === $home ) {
					++$internal;
				} else {
					++$external;
				}
			}
		}
		return array( $internal, $external );
	}

	/**
	 * Lightweight HTML → markdown for llms-full.txt (headings, lists, paragraphs).
	 */
	public static function to_markdown( string $html ): string {
		$html = preg_replace( '/<(script|style)\b.*?<\/\1>/is', '', $html ) ?? '';
		$html = preg_replace_callback(
			'/<h([1-6])[^>]*>(.*?)<\/h\1>/is',
			static fn( $m ) => "\n\n" . str_repeat( '#', min( 6, (int) $m[1] + 1 ) ) . ' ' . self::text( $m[2] ) . "\n\n",
			$html
		) ?? '';
		$html = preg_replace_callback( '/<li[^>]*>(.*?)<\/li>/is', static fn( $m ) => "\n- " . self::text( $m[1] ), $html ) ?? '';
		$html = preg_replace( '/<\/(p|div|ul|ol|table|blockquote)>/i', "\n\n", $html ) ?? '';
		$html = preg_replace( '/<br\s*\/?>/i', "\n", $html ) ?? '';
		$text = self::text( $html );
		// Re-apply markdown newlines lost by text(): text() keeps \n, only collapses spaces.
		return trim( (string) preg_replace( "/\n{3,}/", "\n\n", $text ) );
	}
}