<?php
/**
 * Builds the search/replace pairs for moving a site to a new address and folder.
 *
 * For each source URL (home, and siteurl if different) it adds the common spellings found
 * in WordPress data: http and https, protocol-relative (//host/path), JSON-escaped
 * (https:\/\/host\/path), URL-encoded (https%3A%2F%2Fhost%2Fpath), and optionally the
 * www / non-www twin. Server paths (ABSPATH, wp-content, uploads) are mapped too.
 *
 * @package AccessNow\Replica
 */

namespace AccessNow\Replica\Replace;

defined( 'ABSPATH' ) || exit;

/**
 * Replacement pairs.
 */
final class Replacement_Plan {

	/**
	 * Builds the pairs.
	 *
	 * @param array $src  Source site facts (home, siteurl, abspath, content_dir, uploads_dir).
	 * @param array $dest Destination facts (same keys).
	 * @param array $opts {www: bool, paths: bool, extra: [[search, replace], ...]}.
	 * @return array{pairs: array<string, string>, needles: string[], rows: array[]}
	 */
	public static function build( array $src, array $dest, array $opts = array() ) {
		$opts  = array_merge(
			array(
				'www'   => true,
				'paths' => true,
				'extra' => array(),
			),
			$opts
		);
		$pairs = array();
		$rows  = array(); // For the dry-run table: [search, replace, kind].

		$urls = array();
		foreach ( array( 'siteurl', 'home' ) as $key ) {
			$from = untrailingslashit( (string) ( $src[ $key ] ?? '' ) );
			$to   = untrailingslashit( (string) ( $dest[ $key ] ?? '' ) );
			if ( '' !== $from && '' !== $to && $from !== $to ) {
				$urls[ $from ] = $to;
			}
		}
		foreach ( $urls as $from => $to ) {
			$variants = self::url_variants( $from, ! empty( $opts['www'] ) );
			foreach ( $variants as $variant ) {
				self::add_url_pair( $pairs, $variant, $to );
			}
			$rows[] = array( $from, $to, 'url' );
		}

		if ( ! empty( $opts['paths'] ) ) {
			foreach ( array( 'uploads_dir', 'content_dir', 'abspath' ) as $key ) {
				$from = untrailingslashit( (string) ( $src[ $key ] ?? '' ) );
				$to   = untrailingslashit( (string) ( $dest[ $key ] ?? '' ) );
				if ( strlen( $from ) > 4 && '' !== $to && $from !== $to && ! isset( $pairs[ $from ] ) ) {
					$pairs[ $from ]                          = $to;
					$pairs[ str_replace( '/', '\\/', $from ) ] = str_replace( '/', '\\/', $to );
					$rows[]                                  = array( $from, $to, 'path' );
				}
			}
		}

		foreach ( (array) $opts['extra'] as $extra ) {
			if ( isset( $extra[0], $extra[1] ) && '' !== (string) $extra[0] && $extra[0] !== $extra[1] ) {
				$pairs[ (string) $extra[0] ] = (string) $extra[1];
				$rows[]                      = array( (string) $extra[0], (string) $extra[1], 'custom' );
			}
		}

		return array(
			'pairs'   => $pairs,
			'needles' => self::needles( array_keys( $pairs ) ),
			'rows'    => $rows,
		);
	}

	/**
	 * Source URL spellings: scheme and www twins.
	 *
	 * @param string $url URL without trailing slash.
	 * @param bool   $www Include the www / non-www twin.
	 * @return string[]
	 */
	private static function url_variants( $url, $www ) {
		$parts = wp_parse_url( $url );
		if ( empty( $parts['host'] ) ) {
			return array( $url );
		}
		$rest  = ( isset( $parts['port'] ) ? ':' . $parts['port'] : '' ) . ( isset( $parts['path'] ) ? rtrim( $parts['path'], '/' ) : '' );
		$hosts = array( $parts['host'] );
		if ( $www ) {
			$hosts[] = 0 === strpos( $parts['host'], 'www.' ) ? substr( $parts['host'], 4 ) : 'www.' . $parts['host'];
		}
		$out = array();
		foreach ( $hosts as $host ) {
			foreach ( array( 'https://', 'http://' ) as $scheme ) {
				$out[] = $scheme . $host . $rest;
			}
		}
		return $out;
	}

	/**
	 * Adds a URL pair plus its protocol-relative, JSON-escaped and URL-encoded forms.
	 *
	 * @param array  $pairs Pairs (by reference).
	 * @param string $from  Source URL.
	 * @param string $to    Destination URL.
	 */
	private static function add_url_pair( array &$pairs, $from, $to ) {
		$pairs[ $from ] = $to;

		$from_rel = preg_replace( '#^https?:#', '', $from );
		$to_rel   = preg_replace( '#^https?:#', '', $to );
		$pairs[ $from_rel ] = $to_rel;

		$pairs[ str_replace( '/', '\\/', $from ) ] = str_replace( '/', '\\/', $to );
		$pairs[ str_replace( '/', '\\/', $from_rel ) ] = str_replace( '/', '\\/', $to_rel );

		$enc_from = rawurlencode( $from );
		$enc_to   = rawurlencode( $to );
		$pairs[ $enc_from ] = $enc_to;
		$pairs[ str_replace( array( '%3A', '%2F' ), array( '%3a', '%2f' ), $enc_from ) ] = str_replace( array( '%3A', '%2F' ), array( '%3a', '%2f' ), $enc_to );
	}

	/**
	 * Pre-check needles: for each search string, its longest run without / \ % or :.
	 * Every spelling of a URL (escaped, encoded) still contains that run (usually the host).
	 *
	 * @param string[] $searches Search strings.
	 * @return string[]
	 */
	private static function needles( array $searches ) {
		$out = array();
		foreach ( $searches as $search ) {
			$best = '';
			foreach ( preg_split( '#[/\\\\%:]+#', $search ) as $run ) {
				if ( strlen( $run ) > strlen( $best ) ) {
					$best = $run;
				}
			}
			// Encoded forms (%2F...) leave fragments like "2Flocalhost"; the plain host is also added.
			if ( '' !== $best ) {
				$out[] = preg_replace( '/^(2F|2f|3A|3a)+/', '', $best );
			}
		}
		return array_values( array_unique( array_filter( $out, 'strlen' ) ) );
	}
}