<?php
/**
 * String helpers without regular expressions (PHP 7.4: no str_starts_with).
 *
 * @package Timaub\AfterHackChecklist
 */

namespace Timaub\AfterHackChecklist\Domain;

final class Text {

	public static function startsWith( string $haystack, string $needle ): bool {
		return '' === $needle || 0 === strncmp( $haystack, $needle, strlen( $needle ) );
	}

	public static function endsWith( string $haystack, string $needle ): bool {
		$len = strlen( $needle );
		return 0 === $len || ( strlen( $haystack ) >= $len && substr( $haystack, -$len ) === $needle );
	}

	/** Cuts to at most $max bytes without splitting a UTF-8 character when mbstring is present. */
	public static function cut( string $s, int $max ): string {
		if ( strlen( $s ) <= $max ) {
			return $s;
		}
		if ( function_exists( 'mb_strcut' ) ) {
			return mb_strcut( $s, 0, $max, 'UTF-8' );
		}
		return substr( $s, 0, $max );
	}

	/**
	 * First marker found (case-insensitive substring), with its byte offset.
	 *
	 * @param string   $haystack Bounded text.
	 * @param string[] $markers  Markers.
	 * @return array{0:string,1:int}|null
	 */
	public static function firstMarker( string $haystack, array $markers ) {
		$best = null;
		foreach ( $markers as $m ) {
			$pos = stripos( $haystack, $m );
			if ( false !== $pos && ( null === $best || $pos < $best[1] ) ) {
				$best = array( $m, $pos );
			}
		}
		return $best;
	}

	/**
	 * Up to Assumptions::EXCERPTS_MAX excerpts around distinct occurrences of the
	 * markers, the first at $first, joined by a newline, at most
	 * Assumptions::EXCERPTS_TOTAL_MAX_CHARS bytes (D-09). An occurrence inside the
	 * previous excerpt does not open a new one.
	 *
	 * @param string   $haystack Bounded text.
	 * @param string[] $markers  Markers.
	 * @param int      $first    Offset of the first occurrence.
	 */
	public static function excerptsAround( string $haystack, array $markers, int $first ): string {
		$parts = array();
		$pos   = $first;
		$left  = Assumptions::EXCERPTS_MAX;
		while ( null !== $pos && $left > 0 ) {
			--$left;
			$parts[] = self::excerptAround( $haystack, $pos );
			$end     = max( 0, $pos - 60 ) + Assumptions::EXCERPT_MAX_CHARS;
			$pos     = null;
			foreach ( $markers as $m ) {
				$p = $end < strlen( $haystack ) ? stripos( $haystack, $m, $end ) : false;
				if ( false !== $p && ( null === $pos || $p < $pos ) ) {
					$pos = $p;
				}
			}
		}
		return self::cut( implode( "\n", $parts ), Assumptions::EXCERPTS_TOTAL_MAX_CHARS );
	}

	/** Excerpt of at most Assumptions::EXCERPT_MAX_CHARS bytes around $pos, newlines flattened. */
	public static function excerptAround( string $haystack, int $pos ): string {
		$start = max( 0, $pos - 60 );
		$part  = substr( $haystack, $start, Assumptions::EXCERPT_MAX_CHARS + 8 );
		$part  = str_replace( array( "\r\n", "\r", "\n", "\t" ), ' ', $part );
		return self::cut( $part, Assumptions::EXCERPT_MAX_CHARS );
	}

	/** Lowercased file name of a '/' path. */
	public static function basename( string $path ): string {
		$pos = strrpos( $path, '/' );
		return false === $pos ? $path : substr( $path, $pos + 1 );
	}

	/**
	 * Extensions of a file name, final last: "x.php.jpg" => array('php','jpg').
	 * The part before the first dot is never an extension (".htaccess" => array('htaccess')).
	 *
	 * @return string[] lowercased
	 */
	public static function extensions( string $path ): array {
		$parts = explode( '.', strtolower( self::basename( $path ) ) );
		array_shift( $parts );
		return array_values( array_filter( $parts, 'strlen' ) );
	}

	/** Final extension, lowercased, '' when none. */
	public static function finalExtension( string $path ): string {
		$ext = self::extensions( $path );
		return $ext ? (string) end( $ext ) : '';
	}

	/** "Fichier PHP" of the SPEC: final extension in EXEC. */
	public static function isPhpFile( string $path ): bool {
		return in_array( self::finalExtension( $path ), Assumptions::EXEC, true );
	}

	/** Host of an absolute URL (http, https or protocol-relative), lowercased; '' when not absolute. */
	public static function hostOf( string $url ): string {
		$url = trim( $url, " \t\"'" );
		if ( self::startsWith( $url, '//' ) ) {
			$url = 'http:' . $url;
		}
		$lower = strtolower( $url );
		if ( ! self::startsWith( $lower, 'http://' ) && ! self::startsWith( $lower, 'https://' ) ) {
			return '';
		}
		$host = parse_url( $url, PHP_URL_HOST ); // phpcs:ignore WordPress.WP.AlternativeFunctions.parse_url_parse_url -- pure domain without WordPress functions (PurityTest); wp_parse_url() only works around PHP < 5.4.7, the plugin requires 7.4.
		return is_string( $host ) ? strtolower( $host ) : '';
	}

	/**
	 * Same host as the site, a "www." variant counting as the same (SPEC
	 * contrôle 6: redirection vers `www.` du même hôte reste muette).
	 *
	 * @param string   $host      Host to test.
	 * @param string[] $siteHosts Hosts of home and siteurl.
	 */
	public static function isSiteHost( string $host, array $siteHosts ): bool {
		$bare = self::stripWww( strtolower( $host ) );
		foreach ( $siteHosts as $site ) {
			if ( '' !== $site && self::stripWww( strtolower( $site ) ) === $bare ) {
				return true;
			}
		}
		return false;
	}

	private static function stripWww( string $host ): string {
		return self::startsWith( $host, 'www.' ) ? substr( $host, 4 ) : $host;
	}

	/**
	 * Lines of a bounded text, each cut to Assumptions::LINE_MAX_CHARS.
	 *
	 * @return string[]
	 */
	public static function lines( string $text ): array {
		$lines = explode( "\n", str_replace( array( "\r\n", "\r" ), "\n", $text ) );
		foreach ( $lines as $i => $line ) {
			$lines[ $i ] = substr( $line, 0, Assumptions::LINE_MAX_CHARS );
		}
		return $lines;
	}
}