<?php
/**
 * Sitewide Keyword Auto-Linker Engine with Tag-Safe Text Tokenizer.
 *
 * @package AffiStyle
 */

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

/**
 * Class AffiStyle_Auto_Linker
 *
 * Automatically detects target product keywords within post content on render
 * and safely wraps them in cloaked affiliate links without corrupting HTML.
 */
class AffiStyle_Auto_Linker {

	/**
	 * Transient cache key for the compiled keyword index.
	 *
	 * @var string
	 */
	const CACHE_KEY = 'affistyle_autolink_keyword_index';

	/**
	 * Constructor. Registers WordPress action and filter hooks.
	 */
	public function __construct() {
		// Hook into the_content filter at priority 20 (after shortcodes and embeds).
		add_filter( 'the_content', array( $this, 'filter_content' ), 20 );

		// Invalidate compiled keyword index on post save, update, trash, and delete.
		add_action( 'save_post_affistyle_box', array( __CLASS__, 'invalidate_cache' ) );
		add_action( 'wp_trash_post', array( __CLASS__, 'invalidate_cache' ) );
		add_action( 'before_delete_post', array( __CLASS__, 'invalidate_cache' ) );
		add_action( 'affistyle_clear_autolink_cache', array( __CLASS__, 'invalidate_cache' ) );
	}

	/**
	 * Invalidate the keyword index transient cache.
	 */
	public static function invalidate_cache() {
		delete_transient( self::CACHE_KEY );
	}

	/**
	 * Retrieve or compile the fast in-memory keyword index.
	 *
	 * @return array Associative array of keyword => card metadata.
	 */
	public static function get_keyword_index() {
		$cached_index = get_transient( self::CACHE_KEY );
		if ( false !== $cached_index && is_array( $cached_index ) ) {
			return $cached_index;
		}

		$index = array();

		$boxes = get_posts(
			array(
				'post_type'      => 'affistyle_box',
				'post_status'    => 'publish',
				'posts_per_page' => -1,
				'fields'         => 'ids',
				'no_found_rows'  => true,
			)
		);

		if ( empty( $boxes ) || ! is_array( $boxes ) ) {
			set_transient( self::CACHE_KEY, $index, 24 * HOUR_IN_SECONDS );
			return $index;
		}

		foreach ( $boxes as $box_id ) {
			$box_id   = absint( $box_id );
			$keywords = get_post_meta( $box_id, '_affistyle_keywords', true );

			if ( empty( $keywords ) ) {
				continue;
			}

			// Support both array and comma-separated string formats.
			if ( is_string( $keywords ) ) {
				$keywords = explode( ',', $keywords );
			}

			if ( ! is_array( $keywords ) ) {
				continue;
			}

			$title       = get_post_meta( $box_id, '_affistyle_title_1', true );
			$price       = get_post_meta( $box_id, '_affistyle_price_1', true );
			$rating      = get_post_meta( $box_id, '_affistyle_rating_score', true );
			$image       = get_post_meta( $box_id, '_affistyle_image_1', true );
			$button_text = get_post_meta( $box_id, '_affistyle_button_text', true );

			if ( empty( $title ) ) {
				$title = get_the_title( $box_id );
			}
			if ( empty( $image ) ) {
				$image = get_post_meta( $box_id, '_affistyle_image_url', true );
			}
			if ( empty( $button_text ) ) {
				$button_text = __( 'Check Deal', 'affistyle' );
			}

			$cloaked_url = class_exists( 'AffiStyle_Cloaking' )
				? AffiStyle_Cloaking::get_cloaked_url( $box_id, 1 )
				: get_permalink( $box_id );

			// Append tracking source to the destination.
			$cloaked_url = add_query_arg(
				array(
					'affistyle_source' => 'keyword_auto_link',
					'aff_src'          => 'keyword_auto_link',
				),
				$cloaked_url
			);

			$card_data = array(
				'box_id'      => $box_id,
				'title'       => $title,
				'price'       => $price,
				'rating'      => $rating,
				'image'       => $image,
				'button_text' => $button_text,
				'cloaked_url' => $cloaked_url,
			);

			foreach ( $keywords as $raw_kw ) {
				$clean_kw = trim( (string) $raw_kw );
				if ( '' === $clean_kw ) {
					continue;
				}

				$normalized_kw = function_exists( 'mb_strtolower' )
					? mb_strtolower( $clean_kw, 'UTF-8' )
					: strtolower( $clean_kw );

				// Avoid overwriting a more specific keyword if already set.
				if ( ! isset( $index[ $normalized_kw ] ) ) {
					$index[ $normalized_kw ] = $card_data;
				}
			}
		}

		set_transient( self::CACHE_KEY, $index, 24 * HOUR_IN_SECONDS );
		return $index;
	}

	/**
	 * Main the_content filter callback. Parses and injects auto-links safely.
	 *
	 * @param string $content Post content HTML.
	 * @return string Filtered post content.
	 */
	public function filter_content( $content ) {
		// Guard clause 1: Verify string input.
		if ( empty( $content ) || ! is_string( $content ) ) {
			return $content;
		}

		// Guard clause 2: Skip in admin, feeds, XML-RPC, and REST API requests.
		if ( is_admin() || is_feed() || wp_is_json_request() ) {
			return $content;
		}

		if ( defined( 'REST_REQUEST' ) && REST_REQUEST ) {
			return $content;
		}

		// Guard clause 3: Allow external filters to bypass auto-linking (e.g. Pro post opt-out).
		$can_process = apply_filters( 'affistyle_autolink_can_process', true, $content, get_the_ID() );
		if ( true !== $can_process ) {
			return $content;
		}

		// Retrieve active keyword dictionary.
		$keyword_index = static::get_keyword_index();
		if ( empty( $keyword_index ) || ! is_array( $keyword_index ) ) {
			return $content;
		}

		// Sort keywords by string length in descending order to avoid prefix collision.
		uksort(
			$keyword_index,
			array( $this, 'sort_by_length_desc' )
		);

		// Link limits.
		$max_total_links  = apply_filters( 'affistyle_autolink_max_links_per_post', 2, get_the_ID() );
		$max_per_keyword  = apply_filters( 'affistyle_autolink_max_per_keyword', 1, get_the_ID() );
		$total_links_done = 0;
		$keyword_counts   = array();

		// Compile the master alternation regex pattern.
		$escaped_keywords = array();
		foreach ( array_keys( $keyword_index ) as $kw ) {
			$escaped_keywords[] = preg_quote( $kw, '/' );
		}
		$master_pattern = '/(?<!\p{L})(' . implode( '|', $escaped_keywords ) . ')(?!\p{L})/iu';

		// Fast bail: If none of the active keywords exist in the content, bypass tokenization completely.
		if ( ! preg_match( $master_pattern, $content ) ) {
			return $content;
		}

		// Split content into alternating HTML tags and text tokens.
		$tokens = preg_split( '/(<[^>]+>)/u', $content, -1, PREG_SPLIT_DELIM_CAPTURE );
		if ( empty( $tokens ) || ! is_array( $tokens ) ) {
			return $content;
		}

		$output_buffer = '';
		$tag_stack     = array();

		// List of HTML tags inside which text must never be linked.
		$protected_tags = array(
			'a',
			'h1',
			'h2',
			'h3',
			'h4',
			'h5',
			'h6',
			'script',
			'style',
			'code',
			'pre',
			'kbd',
			'button',
			'input',
			'select',
			'textarea',
			'svg',
			'canvas',
		);

		foreach ( $tokens as $token_idx => $token ) {
			// Fast exit: If maximum links reached for this post, append the remaining tokens and stop.
			if ( $total_links_done >= $max_total_links ) {
				$output_buffer .= implode( '', array_slice( $tokens, $token_idx ) );
				break;
			}

			// Check if token is an HTML tag.
			if ( isset( $token[0] ) && '<' === $token[0] && '>' === substr( $token, -1 ) ) {
				// 1. Closing tag: </tag>
				if ( preg_match( '#^</([a-z0-9_-]+)>#i', $token, $close_match ) ) {
					$closing_tag = strtolower( $close_match[1] );
					// Pop from tag stack.
					$pos = array_search( $closing_tag, $tag_stack, true );
					if ( false !== $pos ) {
						array_splice( $tag_stack, $pos, 1 );
					}
				} elseif ( ! preg_match( '#/>$#', $token ) ) {
					// 2. Opening tag (excluding self-closing tags like <img ... />, <br>, <hr>).
					if ( preg_match( '#^<([a-z0-9_-]+)(\s[^>]*)?>#i', $token, $open_match ) ) {
						$opening_tag = strtolower( $open_match[1] );
						$tag_attrs   = isset( $open_match[2] ) ? $open_match[2] : '';

						// Void tags in HTML5 that never have a closing tag.
						$void_tags = array( 'img', 'br', 'hr', 'input', 'meta', 'link', 'area', 'base', 'col', 'embed', 'param', 'source', 'track', 'wbr' );

						if ( ! in_array( $opening_tag, $void_tags, true ) ) {
							// Check if tag is explicitly protected or has no-autolink classes.
							$is_protected = in_array( $opening_tag, $protected_tags, true );

							if ( ! $is_protected && ! empty( $tag_attrs ) ) {
								if ( preg_match( '/class=["\'][^"\']*(affistyle-card|affistyle-box-container|affistyle-no-autolink)[^"\']*["\']/i', $tag_attrs ) ) {
									$is_protected = true;
								}
								if ( false !== stripos( $tag_attrs, 'data-no-autolink' ) ) {
									$is_protected = true;
								}
							}

							if ( $is_protected ) {
								$tag_stack[] = $opening_tag;
							}
						}
					}
				}

				$output_buffer .= $token;
				continue;
			}

			// If inside any protected tag, pass text token untouched.
			if ( ! empty( $tag_stack ) ) {
				$output_buffer .= $token;
				continue;
			}

			// Skip token if it does not contain any of the target keywords.
			if ( ! preg_match( $master_pattern, $token ) ) {
				$output_buffer .= $token;
				continue;
			}

			// Process pure text token using safe preg_replace_callback.
			$processed_text = preg_replace_callback(
				$master_pattern,
				function ( $matches ) use ( &$total_links_done, &$keyword_counts, $max_total_links, $max_per_keyword, $keyword_index ) {
					$match_text = $matches[0];

					if ( $total_links_done >= $max_total_links ) {
						return $match_text;
					}

					$normalized_match = function_exists( 'mb_strtolower' )
						? mb_strtolower( $match_text, 'UTF-8' )
						: strtolower( $match_text );

					if ( ! isset( $keyword_index[ $normalized_match ] ) ) {
						return $match_text;
					}

					$current_kw_count = isset( $keyword_counts[ $normalized_match ] ) ? $keyword_counts[ $normalized_match ] : 0;
					if ( $current_kw_count >= $max_per_keyword ) {
						return $match_text;
					}

					$item_data   = $keyword_index[ $normalized_match ];
					$box_id      = absint( $item_data['box_id'] );
					$cloaked_url = esc_url( $item_data['cloaked_url'] );

					$rel_attr    = class_exists( 'AffiStyle_Compliance' )
						? AffiStyle_Compliance::get_rel_attributes( $box_id )
						: 'nofollow sponsored noopener';
					$target_attr = class_exists( 'AffiStyle_Compliance' )
						? AffiStyle_Compliance::get_link_target( $box_id )
						: '_blank';

					$link_classes = apply_filters(
						'affistyle_autolink_link_classes',
						array( 'affistyle-autolink' ),
						$box_id,
						$item_data
					);

					if ( is_array( $link_classes ) ) {
						$link_classes = implode( ' ', array_map( 'sanitize_html_class', $link_classes ) );
					} else {
						$link_classes = sanitize_text_field( (string) $link_classes );
					}

					$extra_attrs = apply_filters(
						'affistyle_autolink_anchor_attributes',
						'',
						$box_id,
						$normalized_match,
						$item_data
					);

					// Increment counters.
					$total_links_done++;
					$keyword_counts[ $normalized_match ] = $current_kw_count + 1;

					// Construct the HTML anchor.
					return sprintf(
						'<a href="%1$s" class="%2$s" target="%3$s" rel="%4$s" data-affistyle-box="%5$d"%6$s>%7$s</a>',
						$cloaked_url,
						esc_attr( $link_classes ),
						esc_attr( $target_attr ),
						esc_attr( $rel_attr ),
						$box_id,
						! empty( $extra_attrs ) ? ' ' . trim( $extra_attrs ) : '',
						esc_html( $match_text )
					);
				},
				$token
			);

			$output_buffer .= $processed_text;
		}

		// If auto-links were injected, ensure frontend styling is enqueued.
		if ( $total_links_done > 0 && function_exists( 'wp_enqueue_style' ) && defined( 'AFFISTYLE_URL' ) && defined( 'AFFISTYLE_VERSION' ) ) {
			if ( ! wp_style_is( 'affistyle-frontend-css', 'enqueued' ) && ! wp_style_is( 'affistyle-block-frontend-css', 'enqueued' ) ) {
				wp_enqueue_style(
					'affistyle-frontend-css',
					AFFISTYLE_URL . 'assets/css/affistyle-frontend.css',
					array(),
					AFFISTYLE_VERSION
				);
			}
		}

		return $output_buffer;
	}

	/**
	 * Helper comparison callback to sort keywords by length descending.
	 *
	 * @param string $a First keyword.
	 * @param string $b Second keyword.
	 * @return int Comparison result.
	 */
	private function sort_by_length_desc( $a, $b ) {
		$len_a = function_exists( 'mb_strlen' ) ? mb_strlen( $a, 'UTF-8' ) : strlen( $a );
		$len_b = function_exists( 'mb_strlen' ) ? mb_strlen( $b, 'UTF-8' ) : strlen( $b );

		if ( $len_a === $len_b ) {
			return 0;
		}
		return ( $len_a > $len_b ) ? -1 : 1;
	}
}