<?php
/**
 * SMS length and number of parts, the same rules as web/lib/sms.ts.
 *
 * @package Smsportal
 */

namespace Smsportal;

defined( 'ABSPATH' ) || exit;

/**
 * GSM-7: 160 characters in one part, 153 per part when split; extension characters take two places
 * and never straddle two parts. Anything outside GSM-7 → UCS-2: 70 / 67 UTF-16 code units.
 */
class SmsLength {

	const GSM7_CHARS    = "@£\$¥èéùìòÇ\nØø\rÅåΔ_ΦΓΛΩΠΨΣΘΞÆæßÉ !\"#¤%&'()*+,-./0123456789:;<=>?¡ABCDEFGHIJKLMNOPQRSTUVWXYZÄÖÑÜ§¿abcdefghijklmnopqrstuvwxyzäöñüà";
	const GSM7_EXTENDED = '^{}\\[~]|€';

	const GSM7_SINGLE = 160;
	const GSM7_MULTI  = 153;
	const UCS2_SINGLE = 70;
	const UCS2_MULTI  = 67;

	/**
	 * Length, parts and encoding of a message.
	 *
	 * @param string $text Message (UTF-8).
	 * @return array{length:int,segments:int,per_segment:int,encoding:string}
	 */
	public static function info( string $text ): array {
		$chars = self::chars( $text );
		$units = self::gsm7_units( $chars );
		$gsm   = null !== $units;

		$single = $gsm ? self::GSM7_SINGLE : self::UCS2_SINGLE;
		$multi  = $gsm ? self::GSM7_MULTI : self::UCS2_MULTI;
		$length = $gsm ? array_sum( $units ) : self::utf16_length( $chars );

		if ( 0 === $length ) {
			$segments = 0;
		} elseif ( $length <= $single ) {
			$segments = 1;
		} else {
			$segments = $gsm ? self::gsm7_parts( $units, $multi ) : (int) ceil( $length / $multi );
		}

		return array(
			'length'      => $length,
			'segments'    => $segments,
			'per_segment' => $length > $single ? $multi : $single,
			'encoding'    => $gsm ? 'GSM-7' : 'UCS-2',
		);
	}

	/**
	 * Split into characters (code points); invalid UTF-8 is dropped.
	 *
	 * @param string $text Text.
	 * @return string[]
	 */
	private static function chars( string $text ): array {
		$chars = preg_split( '//u', wp_check_invalid_utf8( $text, true ), -1, PREG_SPLIT_NO_EMPTY );
		return is_array( $chars ) ? $chars : array();
	}

	/**
	 * Places each character takes in GSM-7, or null when one is not in the alphabet.
	 *
	 * @param string[] $chars Characters.
	 * @return int[]|null
	 */
	private static function gsm7_units( array $chars ): ?array {
		$units = array();
		foreach ( $chars as $char ) {
			if ( false !== strpos( self::GSM7_EXTENDED, $char ) ) {
				$units[] = 2;
			} elseif ( false !== strpos( self::GSM7_CHARS, $char ) ) {
				$units[] = 1;
			} else {
				return null;
			}
		}
		return $units;
	}

	/**
	 * UTF-16 code units: characters outside the BMP (4 bytes in UTF-8, e.g. emoji) count two.
	 *
	 * @param string[] $chars Characters.
	 * @return int
	 */
	private static function utf16_length( array $chars ): int {
		$length = 0;
		foreach ( $chars as $char ) {
			$length += 4 === strlen( $char ) ? 2 : 1;
		}
		return $length;
	}

	/**
	 * Parts of a split GSM-7 message; an escaped character that does not fit opens the next part.
	 *
	 * @param int[] $units    Places per character.
	 * @param int   $per_part Places per part.
	 * @return int
	 */
	private static function gsm7_parts( array $units, int $per_part ): int {
		$parts = 1;
		$used  = 0;
		foreach ( $units as $size ) {
			if ( $used + $size > $per_part ) {
				++$parts;
				$used = $size;
			} else {
				$used += $size;
			}
		}
		return $parts;
	}
}