<?php
/**
 * Settings, models and languages.
 *
 * @package Captionwright
 */

namespace Captionwright;

defined( 'ABSPATH' ) || exit;

const OPTION = 'captionwright_settings';

/**
 * Whisper models the browser can run. Sizes are the one-time download in MB.
 *
 * @return array<string, array<string, mixed>>
 */
function models() {
	return array(
		'tiny'  => array(
			'label'  => __( 'Tiny – fastest, rough', 'captionwright' ),
			'webgpu' => 113,
			'wasm'   => 38,
		),
		'base'  => array(
			'label'  => __( 'Base – fast, good for clear speech', 'captionwright' ),
			'webgpu' => 195,
			'wasm'   => 73,
		),
		'small' => array(
			'label'  => __( 'Small – better accuracy, slower', 'captionwright' ),
			'webgpu' => 390,
			'wasm'   => 237,
		),
		'turbo' => array(
			'label'  => __( 'Large v3 Turbo – best accuracy (needs a GPU)', 'captionwright' ),
			'webgpu' => 1533,
			'wasm'   => 0,
		),
	);
}

/**
 * Languages offered in the UI (Whisper supports more; these cover most sites).
 *
 * @return array<string, string> code => English name
 */
function languages() {
	return array(
		'en' => 'English',
		'sv' => 'Swedish',
		'no' => 'Norwegian',
		'da' => 'Danish',
		'fi' => 'Finnish',
		'de' => 'German',
		'nl' => 'Dutch',
		'fr' => 'French',
		'es' => 'Spanish',
		'pt' => 'Portuguese',
		'it' => 'Italian',
		'pl' => 'Polish',
		'cs' => 'Czech',
		'et' => 'Estonian',
		'lv' => 'Latvian',
		'lt' => 'Lithuanian',
		'uk' => 'Ukrainian',
		'ru' => 'Russian',
		'tr' => 'Turkish',
		'el' => 'Greek',
		'hu' => 'Hungarian',
		'ro' => 'Romanian',
		'ar' => 'Arabic',
		'he' => 'Hebrew',
		'hi' => 'Hindi',
		'ja' => 'Japanese',
		'ko' => 'Korean',
		'zh' => 'Chinese',
		'id' => 'Indonesian',
		'vi' => 'Vietnamese',
	);
}

/**
 * The site language as a Whisper code, falling back to English.
 *
 * @return string
 */
function site_language() {
	$code = strtolower( substr( get_locale(), 0, 2 ) );
	if ( 'nb' === $code || 'nn' === $code ) {
		$code = 'no';
	}
	return array_key_exists( $code, languages() ) ? $code : 'en';
}

/**
 * Saved settings merged with defaults.
 *
 * @return array<string, mixed>
 */
function settings() {
	$defaults = array(
		'model'           => 'base',
		'language'        => site_language(),
		'default_on'      => true,
		'auto_transcript' => false,
		'consent'         => false,
		'processor_only'  => false,
	);
	$saved    = get_option( OPTION, array() );
	return array_merge( $defaults, is_array( $saved ) ? $saved : array() );
}

/**
 * Clean posted settings.
 *
 * @param array<string, mixed> $input Raw input.
 * @return array<string, mixed>
 */
function sanitize_settings( $input ) {
	$input = is_array( $input ) ? $input : array();
	$model = isset( $input['model'] ) ? sanitize_key( $input['model'] ) : 'base';
	$lang  = isset( $input['language'] ) ? sanitize_key( $input['language'] ) : site_language();
	return array(
		'model'           => array_key_exists( $model, models() ) ? $model : 'base',
		'language'        => array_key_exists( $lang, languages() ) ? $lang : site_language(),
		'default_on'      => ! empty( $input['default_on'] ),
		'auto_transcript' => ! empty( $input['auto_transcript'] ),
		'consent'         => ! empty( $input['consent'] ),
		'processor_only'  => ! empty( $input['processor_only'] ),
	);
}