<?php
/**
 * Segments: named filters such as "Brand", "Commercial intent" or "Blog".
 *
 * @package TurtleReport
 */

namespace TurtleReport\Policy;

use TurtleReport\Support\Text;

defined( 'ABSPATH' ) || exit;

/**
 * What a segment is and how it matches.
 *
 * A segment matches either search queries or pages, using "contains any of
 * these words" or a regular expression (RE2 syntax, which Google uses).
 * Every segment can also be used in reverse ("Not brand"). Turning a segment
 * into a Google API filter is Reports\SegmentFilter's job.
 */
final class Segments {

	const TYPE_QUERY = 'query';
	const TYPE_PAGE  = 'page';

	const MODE_CONTAINS = 'contains';
	const MODE_REGEX    = 'regex';

	const MAX_SEGMENTS = 12;
	const MAX_VALUE    = 500;

	/**
	 * Cleans stored or posted segments.
	 *
	 * @param array $raw Segments.
	 * @return array[] Items with id, name, type, mode and value.
	 */
	public static function normalize( array $raw ) {
		$out = array();
		$ids = array();
		foreach ( $raw as $seg ) {
			if ( ! is_array( $seg ) ) {
				continue;
			}
			$name  = isset( $seg['name'] ) && is_scalar( $seg['name'] ) ? trim( sanitize_text_field( (string) $seg['name'] ) ) : '';
			$value = isset( $seg['value'] ) && is_scalar( $seg['value'] ) ? trim( sanitize_text_field( (string) $seg['value'] ) ) : '';
			if ( '' === $name || '' === $value ) {
				continue;
			}
			$id = isset( $seg['id'] ) && is_string( $seg['id'] ) && preg_match( '/^g[a-z0-9]{4,12}$/', $seg['id'] ) ? $seg['id'] : '';
			while ( '' === $id || isset( $ids[ $id ] ) ) {
				$id = 'g' . strtolower( wp_generate_password( 6, false ) );
			}
			$ids[ $id ] = true;
			$item       = array(
				'id'    => $id,
				'name'  => Text::substr( $name, 0, 40 ),
				'type'  => isset( $seg['type'] ) && self::TYPE_PAGE === $seg['type'] ? self::TYPE_PAGE : self::TYPE_QUERY,
				'mode'  => isset( $seg['mode'] ) && self::MODE_REGEX === $seg['mode'] ? self::MODE_REGEX : self::MODE_CONTAINS,
				'value' => Text::substr( $value, 0, self::MAX_VALUE ),
			);
			if ( self::MODE_REGEX === $item['mode'] && ! self::valid_regex( $item['value'] ) ) {
				continue;
			}
			$out[] = $item;
			if ( count( $out ) >= self::MAX_SEGMENTS ) {
				break;
			}
		}
		return $out;
	}

	/**
	 * Whether a regular expression is safe to send to Google.
	 *
	 * Google uses RE2, which has no look-arounds or back-references, so
	 * those are rejected up front rather than failing at Google.
	 *
	 * @param string $pattern Pattern.
	 * @return bool
	 */
	public static function valid_regex( $pattern ) {
		if ( '' === $pattern || preg_match( '/\(\?[=!<]|\\\\[1-9]/', $pattern ) ) {
			return false;
		}
		// phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged -- Testing user input; a warning means invalid.
		return false !== @preg_match( '~' . str_replace( '~', '\~', $pattern ) . '~u', '' );
	}

	/**
	 * RE2 pattern for a segment, without flags.
	 *
	 * @param array $segment Segment.
	 * @return string
	 */
	public static function pattern( array $segment ) {
		if ( self::MODE_REGEX === $segment['mode'] ) {
			return $segment['value'];
		}
		$terms = array_filter( array_map( 'trim', preg_split( '/[,\n]+/', $segment['value'] ) ) );
		$terms = array_map(
			static function ( $term ) {
				return preg_replace( '/[.\\\\+*?()|\[\]{}^$]/', '\\\\$0', $term );
			},
			$terms
		);
		return '(' . implode( '|', $terms ) . ')';
	}

	/**
	 * Ready-made segments offered in the site editor.
	 *
	 * @return array[]
	 */
	public static function presets() {
		$presets = array(
			array(
				'label' => __( 'Brand searches', 'turtlereport' ),
				'name'  => __( 'Brand', 'turtlereport' ),
				'type'  => self::TYPE_QUERY,
				'mode'  => self::MODE_CONTAINS,
				'value' => '',
				'hint'  => __( 'Enter your brand and product names, separated by commas.', 'turtlereport' ),
			),
			array(
				'label' => __( 'Commercial intent', 'turtlereport' ),
				'name'  => __( 'Commercial', 'turtlereport' ),
				'type'  => self::TYPE_QUERY,
				'mode'  => self::MODE_CONTAINS,
				'value' => 'buy, price, pricing, cost, cheap, quote, hire, near me, best, top, review, vs, compare, alternative, service, company, agency, software, tool, discount, deal, order, book, demo, trial',
				'hint'  => '',
			),
			array(
				'label' => __( 'Informational', 'turtlereport' ),
				'name'  => __( 'Informational', 'turtlereport' ),
				'type'  => self::TYPE_QUERY,
				'mode'  => self::MODE_CONTAINS,
				'value' => 'how, what, why, when, where, who, guide, tutorial, tips, ideas, examples, meaning, definition, learn, template',
				'hint'  => '',
			),
			array(
				'label' => __( 'Blog pages', 'turtlereport' ),
				'name'  => __( 'Blog', 'turtlereport' ),
				'type'  => self::TYPE_PAGE,
				'mode'  => self::MODE_CONTAINS,
				'value' => '/blog/',
				'hint'  => __( 'Change /blog/ if your articles live elsewhere, e.g. /resources/ or /news/.', 'turtlereport' ),
			),
			array(
				'label' => __( 'Product or service pages', 'turtlereport' ),
				'name'  => __( 'Product & service pages', 'turtlereport' ),
				'type'  => self::TYPE_PAGE,
				'mode'  => self::MODE_CONTAINS,
				'value' => '/services/, /product/, /products/, /pricing, /features',
				'hint'  => '',
			),
		);

		/**
		 * Filters the segment presets offered in the site editor.
		 *
		 * @param array[] $presets Presets with label, name, type, mode, value and hint.
		 */
		return (array) apply_filters( 'turtlereport_segment_presets', $presets );
	}
}