<?php
/**
 * Crawlability & AI-traffic module (spec module 4–5):
 *  - Serves generated /llms.txt (index), /llms-full.txt (full content) and /ai.txt (usage policy).
 *  - Injects AI-crawler allow/block rules into robots.txt from the bot toggles.
 *  - Logs real AI-crawler hits and human referrals arriving from AI assistants.
 *  - Aggregates that traffic for the Bot Activity view.
 *
 * @package AICitare
 */

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

class AICitare_LLMS {

	public function hooks() {
		add_action( 'init', array( $this, 'add_rewrite' ) );
		add_filter( 'query_vars', array( $this, 'query_vars' ) );
		add_action( 'template_redirect', array( $this, 'maybe_serve_file' ) );
		add_filter( 'robots_txt', array( $this, 'robots_rules' ), 10, 2 );
		add_action( 'wp', array( $this, 'log_traffic' ) );
	}

	public function add_rewrite() {
		add_rewrite_rule( '^llms\.txt$', 'index.php?aicitare_file=llms', 'top' );
		add_rewrite_rule( '^llms-full\.txt$', 'index.php?aicitare_file=llms-full', 'top' );
		add_rewrite_rule( '^ai\.txt$', 'index.php?aicitare_file=ai', 'top' );
	}

	public function query_vars( $vars ) {
		$vars[] = 'aicitare_file';
		return $vars;
	}

	/** Output the requested generated text file. */
	public function maybe_serve_file() {
		$which = get_query_var( 'aicitare_file' );
		if ( ! $which ) {
			return;
		}
		header( 'Content-Type: text/plain; charset=utf-8' );
		header( 'X-Robots-Tag: index, follow', true );
		if ( 'llms-full' === $which ) {
			echo self::build_llms_full_txt(); // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped
		} elseif ( 'ai' === $which ) {
			echo self::build_ai_txt(); // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped
		} else {
			echo self::build_llms_txt(); // phpcs:ignore WordPress.Security.EscapeOutput.OutputNotEscaped
		}
		exit;
	}

	/** Published post/page objects for the generated files (respects noindex). */
	private static function published_docs( $limit = 100 ) {
		$q = new WP_Query(
			array(
				'post_type'      => array( 'post', 'page' ),
				'post_status'    => 'publish',
				'posts_per_page' => $limit,
				'no_found_rows'  => true,
				'orderby'        => 'modified',
				'order'          => 'DESC',
			)
		);
		$out = array();
		if ( $q->have_posts() ) {
			foreach ( $q->posts as $post ) {
				if ( 'noindex' === get_post_meta( $post->ID, '_yoast_wpseo_meta-robots-noindex', true ) ) {
					continue;
				}
				$out[] = $post;
			}
		}
		wp_reset_postdata();
		return $out;
	}

	/** Build a curated llms.txt index from published content. */
	public static function build_llms_txt() {
		$name = get_bloginfo( 'name' );
		$desc = get_bloginfo( 'description' );
		$out  = "# {$name}\n\n";
		if ( $desc ) {
			$out .= "> {$desc}\n\n";
		}
		$out .= '# Full text: ' . home_url( '/llms-full.txt' ) . "\n\n";
		$out .= "## Pages\n\n";

		$docs = self::published_docs( 100 );
		if ( $docs ) {
			foreach ( $docs as $post ) {
				$title   = get_the_title( $post );
				$url     = get_permalink( $post );
				$excerpt = wp_strip_all_tags( get_the_excerpt( $post ) );
				$excerpt = $excerpt ? ': ' . wp_trim_words( $excerpt, 20, '' ) : '';
				$out    .= "- [{$title}]({$url}){$excerpt}\n";
			}
		} else {
			// Demo fallback so the file is never empty on a fresh install.
			foreach ( AICitare_Data::seed_pages() as $p ) {
				$out .= "- [{$p['title']}]( " . home_url( $p['url'] ) . " )\n";
			}
		}
		return $out;
	}

	/**
	 * Build llms-full.txt — the expanded index that inlines each page's text so an
	 * LLM can ingest the whole site in one fetch (the emerging "full" companion to llms.txt).
	 */
	public static function build_llms_full_txt() {
		$name = get_bloginfo( 'name' );
		$desc = get_bloginfo( 'description' );
		$out  = "# {$name}\n\n";
		if ( $desc ) {
			$out .= "> {$desc}\n\n";
		}
		$docs = self::published_docs( 50 );
		if ( ! $docs ) {
			$out .= "_No published content yet._\n";
			return $out;
		}
		foreach ( $docs as $post ) {
			$title = get_the_title( $post );
			$url   = get_permalink( $post );
			// do_shortcode (not the_content) avoids third-party content-filter fatals in this context.
			$body  = wp_strip_all_tags( do_shortcode( $post->post_content ) );
			$body  = preg_replace( "/\n{3,}/", "\n\n", trim( $body ) );
			if ( strlen( $body ) > 6000 ) {
				$body = substr( $body, 0, 6000 ) . "…";
			}
			$out .= "\n---\n\n";
			$out .= "## {$title}\n";
			$out .= "URL: {$url}\n\n";
			$out .= $body . "\n";
		}
		return $out;
	}

	/**
	 * Build ai.txt — a plain-text AI usage policy that mirrors the crawler toggles.
	 * (Companion to robots.txt aimed specifically at AI/LLM agents.)
	 */
	public static function build_ai_txt() {
		$name = get_bloginfo( 'name' );
		$bots = AICitare_Data::get_bots();
		$meta = array();
		foreach ( AICitare_Data::bot_meta() as $b ) {
			$meta[ $b['id'] ] = $b;
		}
		$out  = "# ai.txt — AI usage policy for {$name}\n";
		$out .= "# Generated by AICitare. Declares which AI agents may access this site's content.\n";
		$out .= '# Index: ' . home_url( '/llms.txt' ) . "\n";
		$out .= '# Full text: ' . home_url( '/llms-full.txt' ) . "\n\n";
		foreach ( $bots as $bot => $allowed ) {
			$owner = isset( $meta[ $bot ]['owner'] ) ? $meta[ $bot ]['owner'] : '';
			$out  .= "# {$owner}\n";
			$out  .= "User-Agent: {$bot}\n";
			$out  .= ( $allowed ? "Allow: /\n\n" : "Disallow: /\n\n" );
		}
		return $out;
	}

	/** Add allow/block directives for each AI crawler to robots.txt. */
	public function robots_rules( $output, $public ) {
		$bots = AICitare_Data::get_bots();
		$output .= "\n# --- AICitare: AI crawler management ---\n";
		$output .= '# AI policy: ' . home_url( '/ai.txt' ) . "\n";
		$output .= '# LLM index: ' . home_url( '/llms.txt' ) . "\n";
		foreach ( $bots as $bot => $allowed ) {
			$output .= "\nUser-agent: {$bot}\n";
			$output .= $allowed ? "Allow: /\n" : "Disallow: /\n";
		}
		return $output;
	}

	/** On a front-end request, log an AI crawler hit or a human AI referral. */
	public function log_traffic() {
		if ( is_admin() ) {
			return;
		}
		$ua  = isset( $_SERVER['HTTP_USER_AGENT'] ) ? sanitize_text_field( wp_unslash( $_SERVER['HTTP_USER_AGENT'] ) ) : '';
		$uri = isset( $_SERVER['REQUEST_URI'] ) ? esc_url_raw( wp_unslash( $_SERVER['REQUEST_URI'] ) ) : '';

		// 1) AI crawler?
		if ( $ua ) {
			foreach ( AICitare_Data::bot_meta() as $b ) {
				if ( false !== stripos( $ua, $b['id'] ) ) {
					self::insert_hit( $b['id'], $uri, $ua );
					return;
				}
			}
		}

		// 2) Human visitor arriving from an AI assistant (by referrer)?
		$ref = isset( $_SERVER['HTTP_REFERER'] ) ? sanitize_text_field( wp_unslash( $_SERVER['HTTP_REFERER'] ) ) : '';
		if ( ! $ref ) {
			return;
		}
		$host = wp_parse_url( $ref, PHP_URL_HOST );
		if ( ! $host || false !== stripos( $host, wp_parse_url( home_url(), PHP_URL_HOST ) ) ) {
			return; // ignore internal referrals
		}
		foreach ( AICitare_Data::ai_referrers() as $source => $needles ) {
			foreach ( $needles as $needle ) {
				if ( false !== stripos( $host, $needle ) ) {
					self::insert_referral( $source, $uri );
					return;
				}
			}
		}
	}

	private static function insert_hit( $bot, $uri, $ua ) {
		global $wpdb;
		$wpdb->insert( // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching
			$wpdb->prefix . 'aicitare_bot_hits',
			array(
				'bot_name'   => $bot,
				'url'        => $uri,
				'user_agent' => $ua,
				'ip'         => isset( $_SERVER['REMOTE_ADDR'] ) ? sanitize_text_field( wp_unslash( $_SERVER['REMOTE_ADDR'] ) ) : '',
				'hit_at'     => current_time( 'mysql' ),
			),
			array( '%s', '%s', '%s', '%s', '%s' )
		);
	}

	private static function insert_referral( $source, $uri ) {
		global $wpdb;
		$wpdb->insert( // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching
			$wpdb->prefix . 'aicitare_ai_refs',
			array(
				'source' => $source,
				'url'    => $uri,
				'ref_at' => current_time( 'mysql' ),
			),
			array( '%s', '%s', '%s' )
		);
	}

	/** Recent bot hits for the dashboard; falls back to demo data if empty. */
	public static function recent_hits( $limit = 12 ) {
		global $wpdb;
		$table = $wpdb->prefix . 'aicitare_bot_hits';
		$rows  = $wpdb->get_results( $wpdb->prepare( "SELECT bot_name, url, hit_at FROM {$table} ORDER BY hit_at DESC LIMIT %d", $limit ), ARRAY_A ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared, PluginCheck.Security.DirectDB.UnescapedDBParameter -- table name is $wpdb->prefix + a hardcoded literal; the value is bound via prepare().
		if ( empty( $rows ) ) {
			return AICitare_Data::bot_log();
		}
		$bots = AICitare_Data::get_bots();
		$out  = array();
		foreach ( $rows as $r ) {
			$out[] = array(
				'bot'  => $r['bot_name'],
				'page' => $r['url'],
				'time' => human_time_diff( strtotime( $r['hit_at'] ), time() ) . ' ago',
				'ok'   => ! empty( $bots[ $r['bot_name'] ] ),
			);
		}
		return $out;
	}

	/**
	 * Aggregate AI-crawler + AI-referral activity over the last $days days.
	 * Falls back to a demo-derived summary when there is no real data yet and Demo mode is on.
	 */
	public static function bot_stats( $days = 30 ) {
		$days = max( 1, min( 365, (int) $days ) );

		// Demo mode shows the sample dataset throughout the app; keep this view consistent.
		$settings = AICitare_Data::get_settings();
		if ( ! empty( $settings['demo_mode'] ) ) {
			return self::demo_stats( $days );
		}

		global $wpdb;
		$since = gmdate( 'Y-m-d H:i:s', time() - $days * DAY_IN_SECONDS );
		$hits  = $wpdb->prefix . 'aicitare_bot_hits';
		$refs  = $wpdb->prefix . 'aicitare_ai_refs';

		$by_bot = $wpdb->get_results( $wpdb->prepare( "SELECT bot_name, COUNT(*) c, MAX(hit_at) last_at FROM {$hits} WHERE hit_at >= %s GROUP BY bot_name ORDER BY c DESC", $since ), ARRAY_A ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared, PluginCheck.Security.DirectDB.UnescapedDBParameter -- table name is $wpdb->prefix + a hardcoded literal; the value is bound via prepare().
		$top_pages = $wpdb->get_results( $wpdb->prepare( "SELECT url, COUNT(*) c FROM {$hits} WHERE hit_at >= %s GROUP BY url ORDER BY c DESC LIMIT 8", $since ), ARRAY_A ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared, PluginCheck.Security.DirectDB.UnescapedDBParameter -- table name is $wpdb->prefix + a hardcoded literal; the value is bound via prepare().
		$timeline  = $wpdb->get_results( $wpdb->prepare( "SELECT DATE(hit_at) d, COUNT(*) c FROM {$hits} WHERE hit_at >= %s GROUP BY DATE(hit_at) ORDER BY d", $since ), ARRAY_A ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared, PluginCheck.Security.DirectDB.UnescapedDBParameter -- table name is $wpdb->prefix + a hardcoded literal; the value is bound via prepare().
		$ref_rows  = $wpdb->get_results( $wpdb->prepare( "SELECT source, COUNT(*) c, MAX(ref_at) last_at FROM {$refs} WHERE ref_at >= %s GROUP BY source ORDER BY c DESC", $since ), ARRAY_A ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared, PluginCheck.Security.DirectDB.UnescapedDBParameter -- table name is $wpdb->prefix + a hardcoded literal; the value is bound via prepare().
		$ref_recent = $wpdb->get_results( $wpdb->prepare( "SELECT source, url, ref_at FROM {$refs} WHERE ref_at >= %s ORDER BY ref_at DESC LIMIT 10", $since ), ARRAY_A ); // phpcs:ignore WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching, WordPress.DB.PreparedSQL.InterpolatedNotPrepared, PluginCheck.Security.DirectDB.UnescapedDBParameter -- table name is $wpdb->prefix + a hardcoded literal; the value is bound via prepare().

		$bots = AICitare_Data::get_bots();
		$meta = array();
		foreach ( AICitare_Data::bot_meta() as $b ) {
			$meta[ $b['id'] ] = $b;
		}

		$crawlers = array();
		$total    = 0;
		foreach ( (array) $by_bot as $r ) {
			$total += (int) $r['c'];
			$crawlers[] = array(
				'bot'     => $r['bot_name'],
				'owner'   => isset( $meta[ $r['bot_name'] ]['owner'] ) ? $meta[ $r['bot_name'] ]['owner'] : '',
				'count'   => (int) $r['c'],
				'allowed' => ! empty( $bots[ $r['bot_name'] ] ),
				'last'    => human_time_diff( strtotime( $r['last_at'] ), time() ) . ' ago',
			);
		}

		$pages = array();
		foreach ( (array) $top_pages as $r ) {
			$pages[] = array( 'url' => $r['url'], 'count' => (int) $r['c'] );
		}

		$tl = array();
		foreach ( (array) $timeline as $r ) {
			$tl[] = array( 'date' => $r['d'], 'count' => (int) $r['c'] );
		}

		$ref_total = 0;
		$sources   = array();
		foreach ( (array) $ref_rows as $r ) {
			$ref_total += (int) $r['c'];
			$sources[]  = array(
				'source' => $r['source'],
				'count'  => (int) $r['c'],
				'last'   => human_time_diff( strtotime( $r['last_at'] ), time() ) . ' ago',
			);
		}
		$recent = array();
		foreach ( (array) $ref_recent as $r ) {
			$recent[] = array(
				'source' => $r['source'],
				'page'   => $r['url'],
				'time'   => human_time_diff( strtotime( $r['ref_at'] ), time() ) . ' ago',
			);
		}

		return array(
			'days'         => $days,
			'demo'         => false,
			'totalCrawls'  => $total,
			'uniqueBots'   => count( $crawlers ),
			'crawlers'     => $crawlers,
			'topPages'     => $pages,
			'timeline'     => $tl,
			'referralTotal' => $ref_total,
			'sources'      => $sources,
			'recentRefs'   => $recent,
		);
	}

	/** Demo-derived stats so the view is populated when Demo mode is on. Scales with the window. */
	private static function demo_stats( $days ) {
		$bots = AICitare_Data::get_bots();
		$meta = array();
		foreach ( AICitare_Data::bot_meta() as $b ) {
			$meta[ $b['id'] ] = $b;
		}
		// Per-day averages; multiplied by the window so 7/30/90 differ visibly.
		$per_day = array(
			'GPTBot' => 4.3, 'PerplexityBot' => 3.2, 'ClaudeBot' => 2.5,
			'Google-Extended' => 2.0, 'OAI-SearchBot' => 1.5, 'Bytespider' => 0.7, 'CCBot' => 0.55,
		);
		$crawlers = array();
		$total    = 0;
		foreach ( $per_day as $bot => $rate ) {
			$c = max( 1, (int) round( $rate * $days ) );
			$total += $c;
			$crawlers[] = array(
				'bot'     => $bot,
				'owner'   => isset( $meta[ $bot ]['owner'] ) ? $meta[ $bot ]['owner'] : '',
				'count'   => $c,
				'allowed' => ! empty( $bots[ $bot ] ),
				'last'    => ( $rate > 3 ? '18 min ago' : ( $rate > 1.4 ? '2 hr ago' : '9 hr ago' ) ),
			);
		}
		$tl = array();
		for ( $i = $days - 1; $i >= 0; $i-- ) {
			$tl[] = array(
				'date'  => gmdate( 'Y-m-d', time() - $i * DAY_IN_SECONDS ),
				'count' => (int) round( 8 + 22 * ( 0.5 + 0.5 * sin( $i / 3.0 ) ) ),
			);
		}
		$page_rates = array(
			'/best-cooling-pillows' => 2.4, '/memory-foam-vs-latex' => 2.0, '/side-sleeper-guide' => 1.7,
			'/neck-pain-pillows' => 1.3, '/casper-original-review' => 1.1, '/pillow-loft-guide' => 0.9,
		);
		$pages = array();
		foreach ( $page_rates as $u => $rate ) {
			$pages[] = array( 'url' => $u, 'count' => max( 1, (int) round( $rate * $days ) ) );
		}
		$source_rates = array( 'ChatGPT' => 1.1, 'Perplexity' => 0.62, 'Gemini' => 0.37, 'Claude' => 0.23 );
		$sources   = array();
		$ref_total = 0;
		$lasts     = array( '22 min ago', '1 hr ago', '3 hr ago', '5 hr ago' );
		$li        = 0;
		foreach ( $source_rates as $src => $rate ) {
			$c = max( 1, (int) round( $rate * $days ) );
			$ref_total += $c;
			$sources[] = array( 'source' => $src, 'count' => $c, 'last' => $lasts[ $li++ ] );
		}
		$recent = array(
			array( 'source' => 'ChatGPT', 'page' => '/best-cooling-pillows', 'time' => '22 min ago' ),
			array( 'source' => 'Perplexity', 'page' => '/memory-foam-vs-latex', 'time' => '1 hr ago' ),
			array( 'source' => 'ChatGPT', 'page' => '/side-sleeper-guide', 'time' => '2 hr ago' ),
			array( 'source' => 'Gemini', 'page' => '/casper-original-review', 'time' => '3 hr ago' ),
		);
		return array(
			'days'          => $days,
			'demo'          => true,
			'totalCrawls'   => $total,
			'uniqueBots'    => count( $crawlers ),
			'crawlers'      => $crawlers,
			'topPages'      => $pages,
			'timeline'      => $tl,
			'referralTotal' => $ref_total,
			'sources'       => $sources,
			'recentRefs'    => $recent,
		);
	}
}