<?php
/**
 * Backup sets on disk: sets/<YYYYMMDD-HHMMSS>-<host>-<rand>/
 *
 *   manifest.json            complete sets only (manifest.partial.json while in progress)
 *   db/tables.json           schema per table
 *   db/schema.sql            CREATE statements (readable, runnable by hand)
 *   db/NNNN-<table>.MMMMM.sql.gz
 *   objects/ab/<sha256>[.gz] each file's bytes, by checksum; scripts gzip-compressed (Set_Files)
 *   index/files.jsonl        {r, p|p64, t, s, m, h, gz?}: every file's real path, size and time
 *   (format 1 sets, 1.0.2 and earlier: files/<root>/<path>, a mirrored tree, and "lp" in rows)
 *   index/chunks.jsonl       {table, file, rows, sql_bytes, gz_bytes, sha256}
 *
 * Complete sets are never modified, so unchanged files' objects can be hardlinked between sets.
 *
 * @package AccessNow\Replica
 */

namespace AccessNow\Replica\Sets;

use AccessNow\Replica\Logger;
use AccessNow\Replica\Storage;

defined( 'ABSPATH' ) || exit;

/**
 * Set storage.
 */
final class Set_Store {

	/**
	 * Creates an empty set folder and returns its ID.
	 *
	 * @param string $source_home Source home URL.
	 */
	public static function create( $source_home ) {
		$host = (string) wp_parse_url( $source_home, PHP_URL_HOST );
		$path = trim( (string) wp_parse_url( $source_home, PHP_URL_PATH ), '/' );
		$slug = sanitize_title( $host . ( '' !== $path ? '-' . $path : '' ) );
		$id   = gmdate( 'Ymd-His' ) . '-' . ( '' !== $slug ? substr( $slug, 0, 60 ) : 'site' ) . '-' . strtolower( wp_generate_password( 4, false, false ) );
		// Format 2 keeps files in objects/, made as they arrive (Set_Files); nothing makes files/.
		foreach ( array( 'db', 'index' ) as $sub ) {
			wp_mkdir_p( self::dir( $id ) . $sub );
		}
		@chmod( self::dir( $id ), 0700 ); // phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged, WordPress.WP.AlternativeFunctions.file_system_operations_chmod
		return $id;
	}

	/**
	 * Absolute set directory with trailing slash.
	 *
	 * @param string $id Set ID.
	 */
	public static function dir( $id ) {
		return Storage::path( 'sets/' . $id . '/' );
	}

	/**
	 * Whether a string is a valid set ID.
	 *
	 * @param string $id ID.
	 */
	public static function valid_id( $id ) {
		return is_string( $id ) && (bool) preg_match( '/^\d{8}-\d{6}-[a-z0-9-]{1,64}$/', $id );
	}

	/**
	 * Manifest (complete or partial) or null.
	 *
	 * @param string $id Set ID.
	 * @return array|null
	 */
	public static function manifest( $id ) {
		if ( ! self::valid_id( $id ) ) {
			return null;
		}
		$m = Storage::read_json( self::dir( $id ) . 'manifest.json' );
		if ( is_array( $m ) ) {
			return $m;
		}
		$m = Storage::read_json( self::dir( $id ) . 'manifest.partial.json' );
		return is_array( $m ) ? $m : null;
	}

	/**
	 * All sets, newest first.
	 *
	 * @return array[] Manifests with 'set_id'.
	 */
	public static function all() {
		if ( ! Storage::exists() ) {
			return array(); // No backup yet, and listing must not create the folder.
		}
		$out = array();
		foreach ( (array) @scandir( Storage::path( 'sets' ) ) as $name ) { // phpcs:ignore WordPress.PHP.NoSilencedErrors.Discouraged
			if ( self::valid_id( (string) $name ) ) {
				$m = self::manifest( $name );
				if ( $m ) {
					$m['set_id'] = $name;
					$out[]       = $m;
				}
			}
		}
		usort(
			$out,
			static function ( $a, $b ) {
				return strcmp( $b['set_id'], $a['set_id'] );
			}
		);
		return $out;
	}

	/**
	 * Newest complete set of a source, for incremental pulls.
	 *
	 * @param string $source_home Source home URL.
	 * @param string $exclude     Set ID to skip.
	 * @return string|null
	 */
	public static function latest_complete( $source_home, $exclude = '' ) {
		foreach ( self::all() as $m ) {
			if ( $m['set_id'] !== $exclude && 'complete' === ( $m['status'] ?? '' ) && untrailingslashit( (string) ( $m['source']['home'] ?? '' ) ) === untrailingslashit( $source_home ) ) {
				return $m['set_id'];
			}
		}
		return null;
	}

	/**
	 * Deletes a set.
	 *
	 * @param string $id Set ID.
	 */
	public static function delete( $id ) {
		if ( self::valid_id( $id ) && is_dir( self::dir( $id ) ) ) {
			Storage::rrmdir( untrailingslashit( self::dir( $id ) ) );
			Logger::activity( 'backup', 'Backup set deleted: ' . $id );
		}
	}

	/**
	 * Keeps the newest $keep complete sets of a source; never deletes the last complete one.
	 *
	 * @param string $source_home Source home.
	 * @param int    $keep        Sets to keep (0 = all).
	 */
	public static function apply_retention( $source_home, $keep ) {
		if ( $keep <= 0 ) {
			return;
		}
		$complete = array();
		foreach ( self::all() as $m ) {
			if ( 'complete' === ( $m['status'] ?? '' ) && untrailingslashit( (string) ( $m['source']['home'] ?? '' ) ) === untrailingslashit( $source_home ) ) {
				$complete[] = $m['set_id'];
			}
		}
		foreach ( array_slice( $complete, max( 1, $keep ) ) as $id ) {
			self::delete( $id );
		}
	}
}