<?php
/**
 * Collects the site's public content (published pages, posts and WooCommerce products) for the AI to learn.
 * Drafts, private posts, orders and user data are never read.
 *
 * @package Clientager
 */

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

/**
 * Builds and sends the /learn payload.
 */
class Clientager_Learn {

	const MAX_PAGES    = 20;
	const MAX_CONTENT  = 8000;
	const MAX_PRODUCTS = 100;
	const MAX_BYTES    = 1500000;
	const EVENT        = 'clientager_auto_learn';

	/**
	 * Auto-update (paid plans): a minute after a page, post or product is saved, trashed or deleted, the site is
	 * learned again, once for a burst of edits.
	 */
	public static function init() {
		add_action( self::EVENT, array( __CLASS__, 'auto_run' ) );
		add_action( 'save_post', array( __CLASS__, 'changed' ), 20, 2 );
		add_action( 'before_delete_post', array( __CLASS__, 'changed' ), 20, 2 );
		add_action( 'wp_trash_post', array( __CLASS__, 'changed' ), 20 );
		add_action( 'woocommerce_update_product', array( __CLASS__, 'changed' ), 20 );
		add_action( 'woocommerce_product_set_stock_status', array( __CLASS__, 'changed' ), 20 );
	}

	/**
	 * Something on the site changed: schedule one learn a minute from now (if the plan updates automatically).
	 *
	 * @param int          $post_id Post id.
	 * @param WP_Post|null $post    Post.
	 */
	public static function changed( $post_id, $post = null ) {
		$post = $post instanceof WP_Post ? $post : get_post( $post_id );
		if ( ! $post || wp_is_post_revision( $post ) || wp_is_post_autosave( $post ) || ! in_array( $post->post_type, array( 'page', 'post', 'product', 'product_variation' ), true ) ) {
			return;
		}
		if ( ! Clientager_Api::token() || wp_next_scheduled( self::EVENT ) ) {
			return;
		}
		if ( empty( self::known_status()['auto_update'] ) ) {
			return;
		}
		wp_schedule_single_event( time() + MINUTE_IN_SECONDS, self::EVENT );
	}

	/**
	 * The last account status this site saw, without asking Clientager (saving a post never waits on the network).
	 *
	 * @return array
	 */
	private static function known_status() {
		$hit = get_transient( Clientager_Api::STATUS_CACHE );
		if ( is_array( $hit ) ) {
			return $hit;
		}
		$last = get_option( Clientager_Api::OPT_LAST_STATUS );
		return is_array( $last ) && isset( $last['status'] ) && is_array( $last['status'] ) ? $last['status'] : array();
	}

	/**
	 * The scheduled learn.
	 */
	public static function auto_run() {
		$s = Clientager_Api::status( true );
		if ( ! empty( $s['auto_update'] ) ) {
			self::run();
		}
	}

	/**
	 * Pages and products this plan learns (from the account status; the smallest plan's when unknown).
	 *
	 * @return array{pages:int,products:int}
	 */
	private static function limits() {
		$s = Clientager_Api::status( true );
		$l = isset( $s['learn_limits'] ) && is_array( $s['learn_limits'] ) ? $s['learn_limits'] : array();
		return array(
			'pages'    => isset( $l['pages'] ) ? max( 1, min( 500, (int) $l['pages'] ) ) : self::MAX_PAGES,
			'products' => isset( $l['products'] ) ? max( 1, min( 5000, (int) $l['products'] ) ) : self::MAX_PRODUCTS,
		);
	}

	/**
	 * Sends the site's content to Clientager.
	 *
	 * @return array API result.
	 */
	public static function run() {
		$payload = self::payload();
		$res     = Clientager_Api::request( 'POST', '/learn', $payload, array( 'timeout' => 60 ) );
		if ( $res['ok'] ) {
			Clientager_Api::clear_status();
		}
		return $res;
	}

	/**
	 * The full payload.
	 *
	 * @return array
	 */
	public static function payload() {
		return array(
			'site'     => self::site(),
			'pages'    => self::pages(),
			'products' => self::products(),
		);
	}

	/**
	 * Basic facts about the site.
	 *
	 * @return array
	 */
	private static function site() {
		return array(
			'name'        => wp_strip_all_tags( get_bloginfo( 'name' ) ),
			'tagline'     => wp_strip_all_tags( get_bloginfo( 'description' ) ),
			'url'         => home_url( '/' ),
			'timezone'    => wp_timezone_string(),
			'currency'    => function_exists( 'get_woocommerce_currency' ) ? get_woocommerce_currency() : '',
			'language'    => get_locale(),
			'admin_email' => get_option( 'admin_email' ),
		);
	}

	/**
	 * Published pages first, then the latest posts, up to the plan's number, as plain text.
	 *
	 * @return array
	 */
	private static function pages() {
		$out   = array();
		$max   = self::limits()['pages'];
		$chars = max( 1500, min( self::MAX_CONTENT, (int) floor( self::MAX_BYTES / $max ) ) );
		$pages = get_posts(
			array(
				'post_type'        => 'page',
				'post_status'      => 'publish',
				'has_password'     => false,
				'numberposts'      => $max,
				'orderby'          => 'menu_order title',
				'order'            => 'ASC',
				'suppress_filters' => false,
			)
		);
		$left  = $max - count( $pages );
		$posts = $left > 0 ? get_posts(
			array(
				'post_type'        => 'post',
				'post_status'      => 'publish',
				'has_password'     => false,
				'numberposts'      => $left,
				'orderby'          => 'date',
				'order'            => 'DESC',
				'suppress_filters' => false,
			)
		) : array();
		foreach ( array_merge( $pages, $posts ) as $p ) {
			$text = self::plain( $p->post_content );
			if ( strlen( $text ) < 40 ) {
				continue;
			}
			$out[] = array(
				'title'   => wp_strip_all_tags( get_the_title( $p ) ),
				'url'     => get_permalink( $p ),
				'content' => self::cut( $text, $chars ),
			);
		}
		return $out;
	}

	/**
	 * Published WooCommerce products (when WooCommerce is active).
	 *
	 * @return array
	 */
	private static function products() {
		if ( ! function_exists( 'wc_get_products' ) ) {
			return array();
		}
		$out      = array();
		$currency = function_exists( 'get_woocommerce_currency' ) ? get_woocommerce_currency() : '';
		$items    = wc_get_products(
			array(
				'status'  => 'publish',
				'limit'   => self::limits()['products'],
				'orderby' => 'popularity',
			)
		);
		foreach ( $items as $product ) {
			$desc  = $product->get_short_description() ? $product->get_short_description() : $product->get_description();
			$price = $product->get_price();
			$out[] = array(
				'id'          => (string) $product->get_id(),
				'name'        => wp_strip_all_tags( $product->get_name() ),
				'description' => self::cut( self::plain( $desc ), 1000 ),
				'price'       => '' === $price ? null : (float) $price,
				'currency'    => $currency,
				'sku'         => (string) $product->get_sku(),
				'url'         => get_permalink( $product->get_id() ),
				'image'       => $product->get_image_id() ? (string) wp_get_attachment_image_url( $product->get_image_id(), 'woocommerce_thumbnail' ) : '',
				'simple'      => $product->is_type( 'simple' ) && $product->is_purchasable(),
				'category'    => self::category( $product ),
				'options'     => self::options( $product ),
				'tags'        => self::tags( $product ),
				'in_stock'    => $product->is_in_stock(),
			);
		}
		return $out;
	}

	/**
	 * Product tags as text ("cotton, summer, unisex"), more words the chat can match.
	 *
	 * @param WC_Product $product Product.
	 * @return string
	 */
	private static function tags( $product ) {
		$terms = get_the_terms( $product->get_id(), 'product_tag' );
		if ( ! is_array( $terms ) ) {
			return '';
		}
		$names = array();
		foreach ( $terms as $term ) {
			$names[] = wp_strip_all_tags( $term->name );
		}
		return self::cut( implode( ', ', $names ), 300 );
	}

	/**
	 * Product attributes as text ("Color: Gray, Maroon; Size: S, M"), so the chat can answer "any in grey?".
	 *
	 * @param WC_Product $product Product.
	 * @return string
	 */
	private static function options( $product ) {
		$parts = array();
		foreach ( $product->get_attributes() as $attr ) {
			if ( ! is_object( $attr ) || ! method_exists( $attr, 'get_name' ) ) {
				continue;
			}
			$values = $attr->is_taxonomy() ? wc_get_product_terms( $product->get_id(), $attr->get_name(), array( 'fields' => 'names' ) ) : $attr->get_options();
			if ( $values ) {
				$parts[] = wc_attribute_label( $attr->get_name() ) . ': ' . implode( ', ', array_map( 'wp_strip_all_tags', (array) $values ) );
			}
		}
		return self::cut( implode( '; ', $parts ), 400 );
	}

	/**
	 * The product's main category name (the first one that is not "Uncategorized"), for the chat's category overview.
	 *
	 * @param WC_Product $product Product.
	 * @return string
	 */
	private static function category( $product ) {
		$terms = get_the_terms( $product->get_id(), 'product_cat' );
		if ( ! is_array( $terms ) ) {
			return '';
		}
		foreach ( $terms as $term ) {
			if ( 'uncategorized' !== $term->slug ) {
				return self::cut( wp_strip_all_tags( $term->name ), 60 );
			}
		}
		return '';
	}

	/**
	 * Post content as plain text: no shortcodes, blocks markup, scripts, styles or tags.
	 *
	 * @param string $html Raw content.
	 * @return string
	 */
	private static function plain( $html ) {
		$html = strip_shortcodes( (string) $html );
		$html = preg_replace( '#<(script|style|noscript|iframe|svg)[^>]*>.*?</\1>#is', ' ', $html );
		$html = preg_replace( '#<!--.*?-->#s', ' ', $html );
		$html = preg_replace( '#</(p|div|li|h[1-6]|tr|br)\s*>|<br\s*/?>#i', "\n", $html );
		$text = wp_strip_all_tags( $html );
		$text = html_entity_decode( $text, ENT_QUOTES | ENT_HTML5, 'UTF-8' );
		$text = preg_replace( '/\[[^\]]{0,200}\]/', ' ', $text );
		$text = preg_replace( "/[ \t\x{00A0}]+/u", ' ', $text );
		$text = preg_replace( "/\s*\n\s*/", "\n", $text );
		return trim( (string) $text );
	}

	/**
	 * Cuts a string to a number of characters without breaking a multibyte character.
	 *
	 * @param string $text Text.
	 * @param int    $max  Characters.
	 * @return string
	 */
	private static function cut( $text, $max ) {
		return function_exists( 'mb_substr' ) ? mb_substr( $text, 0, $max ) : substr( $text, 0, $max );
	}
}