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

namespace Aiwend\Knowledge;

/**
 * How many chunks one site may index semantically, by embedding dimension (owner decision D4). The retrieval scan is linear in
 * chunks x dimension, so the ceiling shrinks as vectors grow: 5,000 chunks at the reference dimension 1536, 2,500 at 3072, and for
 * any other supported dimension the proportional share, never below FLOOR and never above the reference ceiling. Only dimensions
 * that EmbeddingConfig supports are ever built; a non-positive or implausible dimension has no ceiling (0), which refuses the build.
 *
 * The values are internal policy, not a setting: tests and a future rollout can inject different ones through the constructor, but
 * nothing reads them from an option or a request. Content is never truncated to fit; an oversized site is refused before any
 * embedding call (typed by GenerationBuilder as failure code chunk_ceiling, with the limit and the count).
 */
final class IndexLimits
{
    public const REFERENCE_DIMENSION = 1536;
    public const REFERENCE_CHUNKS = 5000;
    public const FLOOR = 500;
    public const MAX_DIMENSION = 8192;

    /** @param array<int, int> $overrides dimension => ceiling, for tests and reviewed rollouts */
    public function __construct(private array $overrides = [])
    {
    }

    public function maxChunks(int $dimension): int
    {
        if ($dimension < 1 || $dimension > self::MAX_DIMENSION) {
            return 0;
        }
        if (isset($this->overrides[$dimension])) {
            return max(0, $this->overrides[$dimension]);
        }
        if ($dimension <= self::REFERENCE_DIMENSION) {
            return self::REFERENCE_CHUNKS;
        }
        return max(self::FLOOR, intdiv(self::REFERENCE_CHUNKS * self::REFERENCE_DIMENSION, $dimension));
    }

    /** @return array{ok:bool, chunks:int, limit:int} */
    public function check(int $dimension, int $chunks): array
    {
        $limit = $this->maxChunks($dimension);
        return ['ok' => $limit > 0 && $chunks <= $limit, 'chunks' => $chunks, 'limit' => $limit];
    }
}