<?php

namespace Aimogen;

use WP_Filesystem_Base;
use Exception;

class Merges {

    public function __construct(
        private string $path = __DIR__ . '/pretrained_vocab_files/merges.txt'
    ) {}

    public function bpeMerges(): array
    {
        global $wp_filesystem;

        if ( ! is_a( $wp_filesystem, WP_Filesystem_Base::class ) ) {
            require_once ABSPATH . 'wp-admin/includes/file.php';
            $creds = request_filesystem_credentials( site_url() );
            wp_filesystem( $creds );
        }

        if ( ! $wp_filesystem->exists( $this->path ) ) {
            return [];
        }

        // Read file content safely
        $contents = $wp_filesystem->get_contents( $this->path );
        if ( $contents === false ) {
            throw new Exception( 'Error reading merges file: ' . esc_html(sanitize_text_field($this->path)) );
        }

        // Split into lines
        $lines_raw = explode( "\n", $contents );
        $lines     = [];

        // Skip first line (as original code does)
        array_shift( $lines_raw );

        foreach ( $lines_raw as $buffer ) {
            if ( trim( $buffer ) === '' ) {
                continue;
            }

            $line = array_filter(
                preg_split( "/(\s+)/", $buffer ),
                fn( $e ) => strlen( trim( $e ) ) > 0
            );

            if ( ! empty( $line ) ) {
                $lines[] = $line;
            }
        }

        return $lines;
    }
}