racket-chatgpt-bootstrap/private/b64parts.php

<?php
/*
 * b64parts.php
 *
 * Zet een gegeven bestand, bijvoorbeeld een zip, om naar base64-parts.
 *
 * Regels:
 *   - Inputbestand mag binary zijn.
 *   - Output-parts zijn plain base64 tekst.
 *   - Geen chunk_split; dus geen extra newline-overhead.
 *   - Max part size geldt voor de base64-tekst, niet voor de binary input.
 *
 * Voorbeeld:
 *
 *   require_once __DIR__ . '/b64parts.php';
 *
 *   $parts = new Base64Parts(__DIR__ . '/data');
 *   $manifest = $parts->splitFile(
 *       __DIR__ . '/data/html-parsing.zip',
 *       'html-parsing'
 *   );
 *
 *   print_r($manifest);
 */

class Base64PartsException extends Exception
{
}

class Base64Parts
{
    private $dataDir;
    private $maxBase64PartBytes;

    public function __construct($dataDir, $maxBase64PartBytes = null)
    {
        $this->dataDir = rtrim((string)$dataDir, '/');

        /*
         * Max grootte van de base64-text per part.
         * Houd dit op 2 MiB tenzij je fetcher kleiner nodig heeft.
         */
        $this->maxBase64PartBytes = $maxBase64PartBytes === null
            ? 2 * 1024 * 1024
            : (int)$maxBase64PartBytes;

        if ($this->maxBase64PartBytes < 1024) {
            throw new Base64PartsException('maxBase64PartBytes is te klein.');
        }
    }

    public function splitFile($sourceFile, $name, $clearOldParts = true)
    {
        $sourceFile = (string)$sourceFile;
        $name = $this->safeName($name);

        if (!is_file($sourceFile) || !is_readable($sourceFile)) {
            throw new Base64PartsException('Bronbestand ontbreekt of is niet leesbaar: ' . $sourceFile);
        }

        $this->ensureDataDir();

        if ($clearOldParts) {
            $this->removeParts($name);
        }

        /*
         * Base64 maakt van 3 binary bytes precies 4 tekstbytes.
         * Door de binary chunk op een veelvoud van 3 te houden, krijgen
         * alle niet-laatste parts een nette base64-lengte zonder padding.
         */
        $maxBinaryChunkBytes = intdiv($this->maxBase64PartBytes, 4) * 3;

        if ($maxBinaryChunkBytes < 3) {
            throw new Base64PartsException('Berekende binary chunk size is ongeldig.');
        }

        $in = fopen($sourceFile, 'rb');

        if ($in === false) {
            throw new Base64PartsException('Kan bronbestand niet openen: ' . $sourceFile);
        }

        $parts = array();
        $nr = 1;
        $totalBinaryBytes = 0;
        $totalBase64Bytes = 0;

        while (!feof($in)) {
            $bin = fread($in, $maxBinaryChunkBytes);

            if ($bin === false) {
                fclose($in);
                throw new Base64PartsException('Fout bij lezen van: ' . $sourceFile);
            }

            if ($bin === '') {
                break;
            }

            $b64 = base64_encode($bin);

            if (strlen($b64) > $this->maxBase64PartBytes) {
                fclose($in);
                throw new Base64PartsException('Interne fout: base64 part is groter dan maximum.');
            }

            $partNumber = sprintf('%06d', $nr);
            $partFile = $this->partFile($name, $partNumber);

            if (file_put_contents($partFile, $b64, LOCK_EX) === false) {
                fclose($in);
                throw new Base64PartsException('Kan part niet schrijven: ' . $partFile);
            }

            $binaryBytes = strlen($bin);
            $base64Bytes = strlen($b64);

            $parts[] = array(
                'number'       => $partNumber,
                'file'         => $partFile,
                'basename'     => basename($partFile),
                'binary_bytes' => $binaryBytes,
                'base64_bytes' => $base64Bytes,
            );

            $totalBinaryBytes += $binaryBytes;
            $totalBase64Bytes += $base64Bytes;
            $nr++;
        }

        fclose($in);

        if (count($parts) === 0) {
            throw new Base64PartsException('Bronbestand is leeg: ' . $sourceFile);
        }

        $manifest = array(
            'name'               => $name,
            'source_file'        => $sourceFile,
            'source_bytes'       => filesize($sourceFile),
            'source_sha256'      => hash_file('sha256', $sourceFile),
            'max_base64_bytes'   => $this->maxBase64PartBytes,
            'binary_chunk_bytes' => $maxBinaryChunkBytes,
            'part_count'         => count($parts),
            'total_binary_bytes' => $totalBinaryBytes,
            'total_base64_bytes' => $totalBase64Bytes,
            'parts'              => $parts,
            'created_at'         => gmdate('c'),
        );

        $manifestFile = $this->manifestFile($name);

        if (file_put_contents(
            $manifestFile,
            json_encode($manifest, JSON_PRETTY_PRINT | JSON_UNESCAPED_SLASHES) . "\n",
            LOCK_EX
        ) === false) {
            throw new Base64PartsException('Kan manifest niet schrijven: ' . $manifestFile);
        }

        $manifest['manifest_file'] = $manifestFile;

        return $manifest;
    }

    public function readPart($name, $partNumber)
    {
        $name = $this->safeName($name);

        if (!preg_match('/^[0-9]{6}$/', (string)$partNumber)) {
            throw new Base64PartsException('Ongeldig partnummer: ' . $partNumber);
        }

        $file = $this->partFile($name, $partNumber);

        if (!is_file($file) || !is_readable($file)) {
            throw new Base64PartsException('Part ontbreekt of is niet leesbaar: ' . $file);
        }

        $txt = file_get_contents($file);

        if ($txt === false) {
            throw new Base64PartsException('Kan part niet lezen: ' . $file);
        }

        return $txt;
    }

    public function loadManifest($name)
    {
        $name = $this->safeName($name);
        $file = $this->manifestFile($name);

        if (!is_file($file) || !is_readable($file)) {
            throw new Base64PartsException('Manifest ontbreekt: ' . $file);
        }

        $raw = file_get_contents($file);

        if ($raw === false) {
            throw new Base64PartsException('Kan manifest niet lezen: ' . $file);
        }

        $json = json_decode($raw, true);

        if (!is_array($json)) {
            throw new Base64PartsException('Manifest is geen geldige JSON: ' . $file);
        }

        return $json;
    }

    public function removeParts($name)
    {
        $name = $this->safeName($name);

        foreach (glob($this->dataDir . '/' . $name . '.part.*.b64') ?: array() as $file) {
            if (is_file($file)) {
                @unlink($file);
            }
        }

        $manifest = $this->manifestFile($name);

        if (is_file($manifest)) {
            @unlink($manifest);
        }
    }

    private function ensureDataDir()
    {
        if (!is_dir($this->dataDir)) {
            if (!mkdir($this->dataDir, 0755, true)) {
                throw new Base64PartsException('Kan data directory niet maken: ' . $this->dataDir);
            }
        }

        if (!is_writable($this->dataDir)) {
            throw new Base64PartsException('Data directory is niet schrijfbaar: ' . $this->dataDir);
        }
    }

    private function safeName($name)
    {
        $name = (string)$name;

        if (!preg_match('/^[A-Za-z0-9_.+-]+$/', $name)) {
            throw new Base64PartsException('Ongeldige naam: ' . $name);
        }

        return $name;
    }

    private function partFile($name, $partNumber)
    {
        return $this->dataDir . '/' . $name . '.part.' . $partNumber . '.b64';
    }

    private function manifestFile($name)
    {
        return $this->dataDir . '/' . $name . '.parts.json';
    }
}