<?php
/**
 * Export step 4: adds every listed file to the archive, a block at a time, resuming across slices.
 *
 * Bad entries are skipped and logged instead of failing the job: names that are not valid UTF-8 or that the archive
 * refuses (unsafe or duplicate paths), files that vanished after the scan, and files that changed while being added.
 * A file that fails part-way is rolled back: the slice ends at once and saves a writer state whose length is where that
 * file began, without touching the archive file. The next slice's resume() truncates to that length. Truncation thus
 * only ever happens to a saved length, so a crash or a replay of any saved state can always resume.
 *
 * @package Mench_Migrate
 */

defined( 'ABSPATH' ) || exit;
// phpcs:disable WordPress.WP.AlternativeFunctions -- Streams backups, uploads and job files (often several GB) chunk by chunk inside the plugin's own storage folder and the site being restored; WP_Filesystem has no streaming or locking API.

class Mench_Migrate_Step_Write_Files extends Mench_Migrate_Step_Export_Base {

	/** Writer errors that mean "this name cannot go in the archive"; nothing was written for it. */
	private const BAD_NAME = [ 'unsafe_path', 'duplicate_path' ];

	/** Writer errors that mean "the file went away or changed"; part of it may have been written. */
	private const BAD_SOURCE = [ 'source_unreadable', 'source_changed' ];

	public function label(): string {
		return __( 'Adding files', 'mench-migrate' );
	}

	public function run( Mench_Migrate_Job $job, float $deadline ): Mench_Migrate_Step_Result {
		if ( [] === $job->cursor ) {
			self::remove_db_dir( $job ); // The database step's last work files; its result is saved by now.
		}
		$c    = $job->cursor + [ 'pos' => 0, 'rel' => '', 'add' => [], 'start' => 0, 'files' => 0, 'bytes' => 0 ];
		$list = self::list_path( $job );
		$fh   = is_file( $list ) ? fopen( $list, 'rb' ) : false;
		if ( false === $fh ) {
			throw new Mench_Migrate_Storage_Exception( 'Cannot read the file list.', 'write-files' );
		}
		$total  = max( 1, (int) fstat( $fh )['size'] );
		$base   = self::content_dir();
		$writer = self::writer( $job );
		$beat   = microtime( true );
		$done   = false;
		$state  = null; // Set when a partial entry is rolled back; replaces the writer's own state.
		$files  = (int) ( $job->data['scan']['files'] ?? 0 );
		$bytes  = (int) ( $job->data['scan']['bytes'] ?? 0 );
		$report = static function ( int $partial ) use ( $job, &$c, $files, $bytes, $total ): void {
			$done_bytes = (int) $c['bytes'] + $partial;
			Mench_Migrate_Progress::report( $job, [
				'counts'   => [ 'files_done' => (int) $c['files'], 'files_total' => max( $files, (int) $c['files'] ), 'bytes_done' => $done_bytes, 'bytes_total' => max( $bytes, $done_bytes ) ],
				'fraction' => $bytes > 0 ? $done_bytes / $bytes : (int) $c['pos'] / $total,
			] );
		};
		$report( (int) ( $c['add']['offset'] ?? 0 ) );
		try {
			fseek( $fh, (int) $c['pos'] );
			while ( true ) {
				if ( [] === $c['add'] ) {
					$line = fgets( $fh );
					if ( false === $line ) {
						$done = true;
						break;
					}
					$c['pos']   = (int) ftell( $fh );
					$parts      = explode( "\t", rtrim( $line, "\n" ) );
					$c['rel']   = $parts[0];
					$c['start'] = $writer->length();
					if ( 3 !== count( $parts ) || 1 !== preg_match( '//u', $c['rel'] ) ) {
						$this->skip( $job, $c, 'invalid_name' );
						if ( microtime( true ) >= $deadline ) {
							break;
						}
						continue;
					}
				}
				try {
					$res = $writer->add_file( $base . '/' . $c['rel'], $c['rel'], $c['add'], Mench_Migrate_Progress::slice( $deadline ) );
				} catch ( Mench_Migrate_Archive_Exception $e ) {
					$code = $e->getMessage();
					if ( in_array( $code, self::BAD_SOURCE, true ) ) {
						$state            = $writer->state();
						$state['length']  = (int) $c['start'];
						$state['current'] = null;
						$this->skip( $job, $c, $code );
						break; // End the slice so the rolled-back state is saved before anything else is written.
					}
					if ( ! in_array( $code, self::BAD_NAME, true ) ) {
						throw $e;
					}
					$this->skip( $job, $c, $code );
					if ( microtime( true ) >= $deadline ) {
						break;
					}
					continue;
				}
				if ( ! $res['done'] ) {
					$c['add'] = $res['cursor'];
					$report( (int) $c['add']['offset'] );
					Mench_Migrate_Job_Runner::heartbeat( $job );
					if ( microtime( true ) >= $deadline ) {
						break;
					}
					continue; // The same large file, in another short sub-slice.
				}
				$c['files']++;
				$c['bytes'] += (int) $res['cursor']['size'];
				$c['add']    = [];
				$c['rel']    = '';
				$report( 0 );
				if ( microtime( true ) - $beat > 5 ) {
					Mench_Migrate_Job_Runner::heartbeat( $job );
					$beat = microtime( true );
				}
				if ( microtime( true ) >= $deadline ) {
					break;
				}
			}
		} finally {
			fclose( $fh );
		}
		$job->data['writer'] = $state ?? $writer->state();
		unset( $writer );

		if ( ! $done ) {
			return Mench_Migrate_Step_Result::more( $c, $c['pos'] / $total );
		}
		$job->log( 'info', 'Added ' . $c['files'] . ' files (' . size_format( $c['bytes'] ) . ').' );
		return Mench_Migrate_Step_Result::done( [ 'files' => [ 'count' => $c['files'], 'bytes' => $c['bytes'] ] ] );
	}

	/** Counts and logs a skipped file, then clears the in-progress entry from the cursor. */
	private function skip( Mench_Migrate_Job $job, array &$c, string $why ): void {
		$rel   = (string) $c['rel'];
		$shown = function_exists( 'mb_scrub' )
			? (string) preg_replace( '/[\x00-\x1f\x7f]/', '?', mb_scrub( $rel, 'UTF-8' ) )
			: (string) preg_replace( '/[^\x20-\x7e]/', '?', $rel );
		$job->log( 'warning', 'Skipped ' . $shown . ' (' . $why . ').' );
		$job->data['skipped_files'] = (int) ( $job->data['skipped_files'] ?? 0 ) + 1;
		$c['add'] = [];
		$c['rel'] = '';
	}
}