<?php
/**
 * Message text helpers: cleaning, length, billing units and placeholders.
 *
 * @package VibrateSMS
 */

namespace VibrateSMS;

defined( 'ABSPATH' ) || exit;

/**
 * Text rules shared by every send path.
 *
 * Vibrate bills one unit per 256 characters of the trimmed message, in any
 * language. "Characters" are JavaScript string length (UTF-16 code units),
 * because that is what the Vibrate API measures, so an emoji counts as 2.
 */
final class Text {

	const UNIT_LENGTH = 256;

	/** Longest message the plugin will send (about 8 units). */
	const MAX_LENGTH = 2048;

	/** Longest value a customer controlled name placeholder may expand to. */
	const MAX_NAME_LENGTH = 40;

	/**
	 * Whitespace that JavaScript's String.prototype.trim() removes, which is
	 * how the API trims a message before counting it.
	 */
	const TRIM_CLASS = '[\s\x{00A0}\x{1680}\x{2000}-\x{200A}\x{2028}\x{2029}\x{202F}\x{205F}\x{3000}\x{FEFF}]';

	/**
	 * Normalise text before it is stored or sent: valid UTF-8, "\n" line
	 * breaks, no control characters.
	 *
	 * Deliberately does NOT strip tags or percent sequences the way
	 * sanitize_textarea_field() does: an SMS is plain text, "<3" and
	 * percent-encoded (Hebrew) URLs are legitimate content, and every place
	 * that prints a message escapes it.
	 *
	 * @param mixed $text Raw text.
	 */
	public static function clean( $text ): string {
		if ( ! is_scalar( $text ) ) {
			return '';
		}
		$text = wp_check_invalid_utf8( (string) $text, true );
		$text = str_replace( array( "\r\n", "\r" ), "\n", $text );
		// C0 controls except tab and newline, DEL, and C1 controls.
		$text = preg_replace( '/[\x00-\x08\x0B\x0C\x0E-\x1F\x7F\x{0080}-\x{009F}]/u', '', $text );
		return null === $text ? '' : $text;
	}

	/**
	 * Remove HTML from text that is saved for later SMS (sender, templates):
	 * script and style elements with their content, comments, and anything
	 * shaped like a tag. Unlike strip_tags(), a "<" that does not open a tag
	 * ("<3", "a < b") stays, because it is legitimate in an SMS.
	 *
	 * @param string $text Text.
	 */
	public static function without_markup( string $text ): string {
		$patterns = array(
			'#<(script|style)\b[^>]*>.*?</\1\s*>#is',
			'#<!--.*?-->#s',
			'#</?[a-z][a-z0-9-]*(?:\s[^<>]*)?/?>#i',
		);
		$clean = preg_replace( $patterns, '', $text );
		return null === $clean ? '' : $clean;
	}

	/**
	 * Trim the way the API does.
	 *
	 * @param string $text Text.
	 */
	public static function trim( string $text ): string {
		$trimmed = preg_replace( '/^' . self::TRIM_CLASS . '+|' . self::TRIM_CLASS . '+$/u', '', $text );
		return null === $trimmed ? trim( $text ) : $trimmed;
	}

	/**
	 * Length in UTF-16 code units, matching JavaScript's string length.
	 *
	 * @param string $text Text.
	 */
	public static function length( string $text ): int {
		$chars = preg_match_all( '/./us', $text );
		if ( false === $chars ) {
			return strlen( $text );
		}
		$astral = preg_match_all( '/[\x{10000}-\x{10FFFF}]/u', $text );
		return (int) $chars + (int) $astral;
	}

	/**
	 * Billing units for one recipient.
	 *
	 * @param string $text Message.
	 */
	public static function units( string $text ): int {
		return max( 1, (int) ceil( self::length( self::trim( $text ) ) / self::UNIT_LENGTH ) );
	}

	/**
	 * Fill {placeholders}. strtr() makes a single pass, so a value that itself
	 * contains "{something}" is never expanded again.
	 *
	 * @param string               $template Template text.
	 * @param array<string, mixed> $vars     Placeholder values, keyed without braces.
	 */
	public static function render( string $template, array $vars ): string {
		$pairs = array();
		foreach ( $vars as $key => $value ) {
			$pairs[ '{' . $key . '}' ] = self::plain( $value );
		}
		return self::trim( strtr( $template, $pairs ) );
	}

	/**
	 * Turn any value (including formatted HTML such as wc_price()) into a
	 * single line of plain text.
	 *
	 * @param mixed $value Value.
	 */
	public static function plain( $value ): string {
		if ( is_array( $value ) || is_object( $value ) || null === $value ) {
			return '';
		}
		if ( is_bool( $value ) ) {
			return $value ? '1' : '0';
		}
		$text = html_entity_decode( wp_strip_all_tags( (string) $value ), ENT_QUOTES | ENT_HTML5, 'UTF-8' );
		$text = str_replace( "\xC2\xA0", ' ', $text );
		$text = preg_replace( '/\s+/u', ' ', $text );
		return trim( (string) $text );
	}

	/**
	 * A customer typed name, made safe to put in an SMS sent from the shop's
	 * own sender.
	 *
	 * Anyone can place an order with someone else's phone number, so a name
	 * like "Your account is locked, log in at evil.example" would otherwise
	 * reach that person as a message from the shop. Only letters (any script),
	 * spaces, apostrophes and hyphens survive, which removes links, digits and
	 * punctuation, and the result is capped in length.
	 *
	 * @param mixed $value Raw name.
	 */
	public static function safe_name( $value ): string {
		$name = self::plain( $value );
		$name = preg_replace( "/[^\\p{L}\\p{M}\\s'\\x{2019}\\x{05F3}\\-]/u", '', $name );
		$name = trim( (string) preg_replace( '/\s+/u', ' ', (string) $name ) );
		if ( self::length( $name ) > self::MAX_NAME_LENGTH ) {
			$name = trim( (string) preg_replace( '/^(.{' . self::MAX_NAME_LENGTH . '}).*$/us', '$1', $name ) );
		}
		return $name;
	}

	/**
	 * Shorten text for a table cell.
	 *
	 * @param string $text  Text.
	 * @param int    $limit Characters to keep.
	 */
	public static function excerpt( string $text, int $limit = 80 ): string {
		$text = self::plain( $text );
		if ( preg_match_all( '/./us', $text ) <= $limit ) {
			return $text;
		}
		return trim( (string) preg_replace( '/^(.{' . $limit . '}).*$/us', '$1', $text ) ) . '...';
	}
}