<?php
// SPDX-License-Identifier: GPL-3.0-or-later
declare(strict_types=1);

namespace Aiwend\Provider;

/**
 * Upper bound on the input size of a request, used to reserve budget before the provider reports real usage.
 *
 * The bound is the UTF-8 length of the JSON form of everything that will be sent as input. A token always covers at least one byte, so
 * the number of tokens can never exceed the number of bytes: this is a true upper bound without a tokenizer. (Phase 91a, measured:
 * escaping non-ASCII text as \uXXXX, which PHP does by default, inflated Persian, Russian and Japanese text by 2 to 3 times and emoji
 * by 3 times without making the bound any safer, because the escapes are not sent as such: the provider receives and bills the characters.)
 * Slashes are left unescaped for the same reason. Invalid UTF-8 still throws, exactly as before.
 */
final class InputEstimate
{
    public static function bytes(ModelRequest $request): int
    {
        return strlen(json_encode([$request->instruction, $request->messages, $request->context], JSON_THROW_ON_ERROR | JSON_UNESCAPED_UNICODE | JSON_UNESCAPED_SLASHES));
    }
}