<?php
if ( ! defined( 'ABSPATH' ) ) {
	exit;
}

// phpcs:disable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching -- Reads the plugin's own knowledge tables once, when starter questions are created.

/**
 * Creates starter questions from the website it is installed on (no AI).
 *
 * Sources, all from public content already read by Replyvane_Knowledge:
 *  - questions and answers already written on the site (FAQ sections,
 *    accordions, toggles, FAQ blocks, headings ending in "?")
 *  - common page types recognised by their address or title: home, about,
 *    contact, prices/plans, services, booking, refunds, shipping, privacy, terms
 *  - contact details and opening hours found in the text
 *  - WooCommerce products that visitors can see
 *
 * Generated questions are marked "auto". Running it again refreshes them,
 * but never touches questions you added or edited yourself.
 */
class Replyvane_Generator {

	const MAX_HARVESTED = 60;
	const MAX_PRODUCTS  = 30;

	private static $topics = array(
		'about'           => 'About us',
		'getting-started' => 'Getting started',
		'services'        => 'Services',
		'pricing'         => 'Prices and plans',
		'products'        => 'Products',
		'policies'        => 'Policies',
		'contact'         => 'Contact',
	);

	public static function hooks() {
		add_action( 'replyvane_kb_scan_complete', array( __CLASS__, 'maybe_auto' ) );
	}

	/** After the first scan (or when asked), create questions automatically. */
	public static function maybe_auto() {
		if ( get_option( 'replyvane_autogen_pending' ) ) {
			delete_option( 'replyvane_autogen_pending' );
			$report = self::generate();
			update_option( 'replyvane_autogen_report', $report, false );
		}
	}

	/* ------------------------------------------------------------------ */
	/* Reading helpers                                                     */
	/* ------------------------------------------------------------------ */

	private static function pages() {
		global $wpdb;
		$table = Replyvane_Knowledge::table();
		$rows  = $wpdb->get_results( $wpdb->prepare( "SELECT post_id, url, MIN(title) AS title FROM %i GROUP BY post_id, url ORDER BY MIN(id)", $table ), ARRAY_A );
		$pages = array();
		foreach ( (array) $rows as $r ) {
			// The page's own path ("about-us"), which works with any permalink setting.
			$uri             = get_page_uri( (int) $r['post_id'] );
			$path            = trim( strtolower( $uri ? $uri : (string) wp_parse_url( $r['url'], PHP_URL_PATH ) ), '/' );
			$r['path']       = $path;
			$r['type']       = get_post_type( $r['post_id'] );
			$r['match_text'] = strtolower( $path . ' ' . $r['title'] );
			$pages[]         = $r;
		}
		return $pages;
	}

	private static function page_text( $post_id ) {
		global $wpdb;
		$table = Replyvane_Knowledge::table();
		$rows  = $wpdb->get_col( $wpdb->prepare( "SELECT content FROM %i WHERE post_id = %d ORDER BY id", $table, $post_id ) );
		return implode( "\n", (array) $rows );
	}

	/** Lines of a page without section markers or link addresses. */
	private static function lines( $text ) {
		$out = array();
		foreach ( explode( "\n", (string) $text ) as $line ) {
			$line = trim( preg_replace( '/\s*\[https?:[^\]]+\]/', '', $line ) );
			if ( '' === $line || 0 === strpos( $line, '## ' ) || '(continued)' === $line ) {
				continue;
			}
			$out[] = $line;
		}
		return array_values( array_unique( $out ) );
	}

	private static function headings( $post_id ) {
		global $wpdb;
		$table = Replyvane_Knowledge::table();
		$out   = array();
		foreach ( (array) $wpdb->get_col( $wpdb->prepare( "SELECT content FROM %i WHERE post_id = %d ORDER BY id", $table, $post_id ) ) as $content ) {
			foreach ( explode( "\n", $content ) as $line ) {
				if ( 0 === strpos( $line, '## ' ) ) {
					$parts = explode( ' › ', substr( $line, 3 ) );
					$h     = trim( end( $parts ) );
					if ( '' !== $h && mb_strlen( $h ) <= 70 && '?' !== mb_substr( $h, -1 ) ) {
						$out[ $h ] = true;
					}
				}
			}
		}
		return array_keys( $out );
	}

	/** The first real sentences of a page (skips short labels and buttons). */
	private static function intro( $text, $max = 600 ) {
		$picked = array();
		$length = 0;
		foreach ( self::lines( $text ) as $line ) {
			if ( mb_strlen( $line ) < 60 || 0 === strpos( $line, '- ' ) ) {
				continue;
			}
			$picked[] = $line;
			$length  += mb_strlen( $line );
			if ( $length >= $max || count( $picked ) >= 3 ) {
				break;
			}
		}
		$text = implode( "\n\n", $picked );
		return mb_strlen( $text ) > $max + 200 ? rtrim( mb_substr( $text, 0, $max ) ) . '…' : $text;
	}

	/**
	 * Lines that mention prices, labelled with the plan or product name they
	 * sit under ("Starter: $89 / month – Up to 8 classes a month.").
	 */
	private static function price_lines( $text, $page_title = '' ) {
		$raw     = explode( "\n", (string) $text );
		$out     = array();
		$heading = '';
		$prev    = '';
		$price   = '/[$€£₹]\s?\d|\d\s?(usd|eur|gbp|inr)\b|per (month|year|session|hour)|\/\s?(mo|month|yr|year)\b/i';
		$count   = count( $raw );
		for ( $i = 0; $i < $count; $i++ ) {
			$line = trim( preg_replace( '/\s*\[https?:[^\]]+\]/', '', $raw[ $i ] ) );
			if ( '' === $line || '(continued)' === $line ) {
				continue;
			}
			if ( 0 === strpos( $line, '## ' ) ) {
				$parts   = explode( ' › ', substr( $line, 3 ) );
				$heading = trim( end( $parts ) );
				$prev    = '';
				continue;
			}
			if ( ! preg_match( $price, $line ) ) {
				$prev = $line;
				continue;
			}
			$item = ltrim( $line, '- ' );
			if ( mb_strlen( $item ) < 30 ) {
				// A bare price: label it with its heading, or the short line above it.
				$label = '';
				if ( $heading && 0 !== strcasecmp( $heading, $page_title ) && ! preg_match( '/\d/', $heading ) ) {
					$label = $heading;
				} elseif ( $prev && mb_strlen( $prev ) < 40 && ! preg_match( '/\d|[.!?]$/', $prev ) ) {
					$label = $prev;
				}
				if ( $label ) {
					$item = rtrim( $label, ':' ) . ': ' . $item;
				}
				// Add the short description that follows, if there is one.
				$next = isset( $raw[ $i + 1 ] ) ? trim( $raw[ $i + 1 ] ) : '';
				if ( $next && 0 !== strpos( $next, '## ' ) && mb_strlen( $next ) < 90 && ! preg_match( $price, $next ) ) {
					$item .= ' – ' . $next;
					$i++;
				}
			}
			$item = mb_substr( $item, 0, 180 );
			if ( ! in_array( $item, $out, true ) ) {
				$out[] = $item;
			}
			$prev = $line;
			if ( count( $out ) >= 12 ) {
				break;
			}
		}
		return $out;
	}

	private static function contact_details( $text ) {
		preg_match_all( '/[A-Z0-9._%+\-]+@[A-Z0-9.\-]+\.[A-Z]{2,}/i', $text, $emails );
		$phones = array();
		if ( preg_match_all( '/(?<![\w$.])\+?\(?\d[\d\s().\-]{7,}\d(?![\w])/', $text, $m ) ) {
			foreach ( $m[0] as $candidate ) {
				$digits = preg_replace( '/\D/', '', $candidate );
				if ( strlen( $digits ) >= 9 && strlen( $digits ) <= 15 && ! preg_match( '/^(19|20)\d{2}[\-\/.]/', trim( $candidate ) ) ) {
					$phones[] = trim( $candidate );
				}
			}
		}
		$hours = array();
		foreach ( self::lines( $text ) as $line ) {
			if ( mb_strlen( $line ) < 120
				&& preg_match( '/\b(mon|tue|wed|thu|fri|sat|sun)[a-z]*\b|\bweekdays?\b|\bdaily\b/i', $line )
				&& preg_match( '/\d{1,2}(:\d{2})?\s*(am|pm)|\b\d{1,2}:\d{2}\b|\bclosed\b/i', $line ) ) {
				$hours[] = ltrim( $line, '- ' );
			}
		}
		return array(
			'emails' => array_slice( array_values( array_unique( array_map( 'strtolower', $emails[0] ) ) ), 0, 3 ),
			'phones' => array_slice( array_values( array_unique( $phones ) ), 0, 3 ),
			'hours'  => array_slice( array_values( array_unique( $hours ) ), 0, 8 ),
		);
	}

	/** First page whose address or title contains one of the words (shortest address wins). */
	private static function find_page( array $pages, array $words ) {
		$found = null;
		foreach ( $pages as $p ) {
			if ( 'page' !== $p['type'] ) {
				continue;
			}
			foreach ( $words as $w ) {
				if ( preg_match( '/(^|[^a-z])' . preg_quote( $w, '/' ) . '/', $p['match_text'] ) ) {
					if ( ! $found || strlen( $p['path'] ) < strlen( $found['path'] ) ) {
						$found = $p;
					}
					break;
				}
			}
		}
		return $found;
	}

	private static function button( $page, $label = '' ) {
		return array(
			array(
				'label' => mb_substr( $label ? $label : $page['title'], 0, 60 ),
				'url'   => $page['url'],
			),
		);
	}

	/* ------------------------------------------------------------------ */
	/* Building the questions                                              */
	/* ------------------------------------------------------------------ */

	private static function build() {
		$site  = wp_specialchars_decode( get_bloginfo( 'name' ), ENT_QUOTES );
		$pages = self::pages();
		$items = array();
		$topic = array();
		$add   = function ( $id, $cat, $question, $answer, $args = array() ) use ( &$items ) {
			if ( '' === trim( (string) $answer ) ) {
				return;
			}
			$items[ $id ] = array_merge(
				array(
					'id'       => 'auto-' . $id,
					'category' => $cat,
					'question' => $question,
					'answer'   => mb_substr( trim( $answer ), 0, 1800 ),
					'keywords' => array(),
					'links'    => array(),
					'featured' => false,
					'order'    => count( $items ) + 1,
				),
				$args
			);
		};

		// Home page: what the business does.
		$front = (int) get_option( 'page_on_front' );
		$home  = null;
		foreach ( $pages as $p ) {
			if ( ( $front && (int) $p['post_id'] === $front ) || untrailingslashit( $p['url'] ) === untrailingslashit( home_url( '/' ) ) ) {
				$home = $p;
				break;
			}
		}
		if ( $home ) {
			$tagline = trim( wp_specialchars_decode( get_bloginfo( 'description' ), ENT_QUOTES ) );
			$intro   = self::intro( self::page_text( $home['post_id'] ) );
			$add( 'what-we-do', 'about', "What does {$site} do?", ( $tagline ? $tagline . "\n\n" : '' ) . $intro, array( 'featured' => true, 'keywords' => array( 'about', 'overview', 'introduction' ) ) );
		}

		// About page.
		$about = self::find_page( $pages, array( 'about', 'our-story', 'who-we-are', 'our story', 'who we are' ) );
		if ( $about ) {
			$add( 'about', 'about', "Tell me about {$site}", self::intro( self::page_text( $about['post_id'] ), 700 ), array( 'links' => self::button( $about ), 'keywords' => array( 'about', 'story', 'history', 'mission', 'founded', 'founder' ) ) );
		}

		// Services.
		$services = self::find_page( $pages, array( 'services', 'service', 'what-we-do', 'treatments', 'solutions' ) );
		if ( $services ) {
			$title  = $services['title'];
			$heads  = array_slice( array_values( array_filter( self::headings( $services['post_id'] ), function ( $h ) use ( $title ) { return 0 !== strcasecmp( $h, $title ); } ) ), 0, 10 );
			$answer = self::intro( self::page_text( $services['post_id'] ), 60 );
			if ( $heads ) {
				$answer .= ( $answer ? "\n\n" : '' ) . '- ' . implode( "\n- ", $heads );
			}
			$add( 'services', 'services', 'What services do you offer?', $answer, array( 'featured' => true, 'links' => self::button( $services ), 'keywords' => array( 'services', 'offer', 'provide', 'options', 'treatments' ) ) );
		}

		// Prices and plans.
		$pricing = self::find_page( $pages, array( 'pricing', 'prices', 'price', 'plans', 'plan', 'membership', 'packages', 'subscription', 'rates', 'fees' ) );
		$price_source = $pricing ? $pricing : $home;
		if ( $price_source ) {
			$prices = self::price_lines( self::page_text( $price_source['post_id'] ), $price_source['title'] );
			if ( $prices ) {
				$add(
					'prices',
					'pricing',
					$pricing ? 'What are your prices and plans?' : 'How much does it cost?',
					'- ' . implode( "\n- ", $prices ),
					array( 'featured' => true, 'links' => $pricing ? self::button( $pricing ) : array(), 'keywords' => array( 'price', 'prices', 'cost', 'how much', 'plans', 'fees', 'rates' ) )
				);
			}
		}

		// Booking.
		$booking = self::find_page( $pages, array( 'book', 'booking', 'appointment', 'schedule', 'reserve' ) );
		if ( $booking ) {
			$add( 'booking', 'getting-started', 'How do I book?', self::intro( self::page_text( $booking['post_id'] ), 450 ) ?: "You can book on our {$booking['title']} page.", array( 'featured' => true, 'links' => self::button( $booking, 'Book now' ), 'keywords' => array( 'book', 'booking', 'appointment', 'schedule', 'reserve', 'start' ) ) );
		}

		// Policies.
		$policies = array(
			'refunds'  => array( array( 'refund', 'return', 'cancellation', 'cancel' ), 'What is your refund and cancellation policy?', array( 'refund', 'return', 'cancel', 'money back', 'cancellation' ), true ),
			'shipping' => array( array( 'shipping', 'delivery' ), 'How does shipping work?', array( 'shipping', 'delivery', 'ship', 'postage', 'arrive' ), false ),
			'privacy'  => array( array( 'privacy' ), 'How do you handle my personal information?', array( 'privacy', 'data', 'personal information', 'gdpr' ), false ),
			'terms'    => array( array( 'terms', 'conditions' ), 'Where can I read your terms and conditions?', array( 'terms', 'conditions', 'agreement' ), false ),
		);
		foreach ( $policies as $id => $cfg ) {
			$page = self::find_page( $pages, $cfg[0] );
			if ( $page ) {
				$len    = 'terms' === $id || 'privacy' === $id ? 300 : 650;
				$answer = self::intro( self::page_text( $page['post_id'] ), $len );
				$answer = ( $answer ? $answer . "\n\n" : '' ) . "You can read the full details on our [{$page['title']}]({$page['url']}) page.";
				$add( $id, 'policies', $cfg[1], $answer, array( 'featured' => $cfg[3], 'links' => self::button( $page ), 'keywords' => $cfg[2] ) );
			}
		}

		// Contact details and hours.
		$contact = self::find_page( $pages, array( 'contact', 'get-in-touch', 'reach-us', 'locations', 'location' ) );
		$text    = $contact ? self::page_text( $contact['post_id'] ) : ( $home ? self::page_text( $home['post_id'] ) : '' );
		$details = self::contact_details( $text );
		$lines   = array();
		foreach ( $details['emails'] as $e ) {
			$lines[] = "- Email: [{$e}](mailto:{$e})";
		}
		foreach ( $details['phones'] as $ph ) {
			$lines[] = '- Phone: [' . $ph . '](tel:' . preg_replace( '/[^\d+]/', '', $ph ) . ')';
		}
		if ( $contact ) {
			$lines[] = "- Or send us a message on our [{$contact['title']}]({$contact['url']}) page.";
		}
		if ( $lines ) {
			$add( 'contact', 'contact', 'How can I contact you?', "You can reach us here:\n" . implode( "\n", $lines ), array( 'featured' => true, 'links' => $contact ? self::button( $contact, 'Contact us' ) : array(), 'keywords' => array( 'contact', 'email', 'phone', 'call', 'reach', 'message', 'support' ) ) );
		}
		if ( $details['hours'] ) {
			$add( 'hours', 'contact', 'What are your opening hours?', '- ' . implode( "\n- ", $details['hours'] ), array( 'links' => $contact ? self::button( $contact ) : array(), 'keywords' => array( 'hours', 'open', 'opening', 'times', 'closed', 'when' ) ) );
		}

		// Shop products (WooCommerce).
		if ( function_exists( 'wc_get_product' ) ) {
			$products = array();
			foreach ( $pages as $p ) {
				if ( 'product' === $p['type'] && count( $products ) < self::MAX_PRODUCTS ) {
					$product = wc_get_product( $p['post_id'] );
					if ( $product && $product->is_visible() ) {
						$products[] = array( $p, $product );
					}
				}
			}
			$clean = function ( $html ) {
				return trim( preg_replace( '/\s+/u', ' ', html_entity_decode( wp_strip_all_tags( (string) $html ), ENT_QUOTES, 'UTF-8' ) ) );
			};
			if ( $products ) {
				$list = array();
				foreach ( array_slice( $products, 0, 12 ) as $pp ) {
					$price  = $clean( $pp[1]->get_price_html() );
					$list[] = '- [' . $clean( $pp[1]->get_name() ) . '](' . $pp[0]['url'] . ')' . ( $price ? ' – ' . $price : '' );
				}
				$shop = function_exists( 'wc_get_page_permalink' ) ? wc_get_page_permalink( 'shop' ) : '';
				$add( 'products', 'products', 'What products do you sell?', implode( "\n", $list ), array( 'featured' => ! isset( $items['prices'] ), 'links' => $shop ? array( array( 'label' => 'Visit the shop', 'url' => $shop ) ) : array(), 'keywords' => array( 'products', 'shop', 'buy', 'sell', 'store', 'purchase' ) ) );
				foreach ( $products as $pp ) {
					$name    = $clean( $pp[1]->get_name() );
					$price   = $clean( $pp[1]->get_price_html() );
					$summary = mb_substr( $clean( $pp[1]->get_short_description() ), 0, 400 );
					if ( ! $price && ! $summary ) {
						continue;
					}
					$answer = ( $price ? "**{$name}** costs {$price}." : "**{$name}**" ) . ( $summary ? "\n\n" . $summary : '' );
					$add( 'product-' . $pp[0]['post_id'], 'products', "Tell me about {$name}", $answer, array( 'links' => self::button( $pp[0], 'View product' ), 'keywords' => array_merge( array( 'price', 'cost', 'how much' ), explode( ' ', mb_strtolower( $name ) ) ) ) );
				}
			}
		}

		// Questions and answers already written on the site.
		global $wpdb;
		$harvest = $wpdb->get_results( $wpdb->prepare( 'SELECT * FROM %i ORDER BY id LIMIT %d', Replyvane_Knowledge::harvest_table(), self::MAX_HARVESTED ), ARRAY_A );
		foreach ( (array) $harvest as $h ) {
			$cat = 'faq-' . sanitize_title( $h['page_title'] );
			if ( ! isset( $topic[ $cat ] ) ) {
				$topic[ $cat ] = mb_substr( $h['page_title'], 0, 60 );
			}
			$add(
				'qa-' . substr( md5( $h['question'] . $h['post_id'] ), 0, 10 ),
				$cat,
				$h['question'],
				$h['answer'],
				array( 'links' => array( array( 'label' => mb_substr( $h['page_title'], 0, 60 ), 'url' => $h['url'] ) ) )
			);
		}

		// Related questions: others in the same topic.
		$by_cat = array();
		foreach ( $items as $key => $item ) {
			$by_cat[ $item['category'] ][] = $item['id'];
		}
		foreach ( $items as $key => $item ) {
			$items[ $key ]['related'] = array_slice( array_values( array_diff( $by_cat[ $item['category'] ], array( $item['id'] ) ) ), 0, 3 );
		}

		// Keep at most six popular questions.
		$featured = 0;
		foreach ( $items as $key => $item ) {
			if ( $item['featured'] && ++$featured > 6 ) {
				$items[ $key ]['featured'] = false;
			}
		}
		return array( array_values( $items ), $topic );
	}

	/* ------------------------------------------------------------------ */
	/* Saving                                                              */
	/* ------------------------------------------------------------------ */

	/**
	 * Creates or refreshes generated questions. Questions you added or edited
	 * yourself are left exactly as they are.
	 */
	public static function generate() {
		list( $items, $extra_topics ) = self::build();
		$report = array(
			'added'   => 0,
			'updated' => 0,
			'kept'    => 0,
			'removed' => 0,
			'time'    => current_time( 'mysql' ),
		);

		// Topics first, keeping any the site owner already has.
		$cats  = Replyvane_Store::categories();
		$have  = wp_list_pluck( $cats, 'id' );
		$order = count( $cats );
		$used  = array_unique( wp_list_pluck( $items, 'category' ) );
		foreach ( array_merge( self::$topics, $extra_topics ) as $id => $title ) {
			if ( in_array( $id, $used, true ) && ! in_array( $id, $have, true ) ) {
				$cats[] = array(
					'id'    => $id,
					'title' => $title,
					'order' => ++$order,
				);
			}
		}
		Replyvane_Store::save_categories( $cats );

		$made = array();
		foreach ( $items as $raw ) {
			$row = Replyvane_Store::clean_item( $raw );
			if ( is_wp_error( $row ) ) {
				continue;
			}
			$row['source'] = 'auto';
			$made[]        = $row['slug'];
			$existing      = Replyvane_Store::get_by_slug( $row['slug'] );
			if ( $existing && 'auto' !== $existing['source'] ) {
				$report['kept']++;
				continue;
			}
			$result = Replyvane_Store::save( $row, $existing ? (int) $existing['id'] : 0, false );
			if ( ! is_wp_error( $result ) ) {
				$report[ $existing ? 'updated' : 'added' ]++;
			}
		}

		// Remove generated questions the site no longer supports.
		foreach ( Replyvane_Store::items() as $row ) {
			if ( 'auto' === $row['source'] && ! in_array( $row['slug'], $made, true ) ) {
				Replyvane_Store::delete( $row['id'] );
				$report['removed']++;
			}
		}
		Replyvane_Plugin::bump_version();
		return $report;
	}
}
// phpcs:enable WordPress.DB.DirectDatabaseQuery.DirectQuery, WordPress.DB.DirectDatabaseQuery.NoCaching