racket-chatgpt-bootstrap/lib/catalog-http.php

<?php

function catalog_http_normalize_header_name($name)
{
    return strtolower(trim($name));
}

function catalog_http_ensure_dir($dir)
{
    if (!is_dir($dir)) {
        if (!mkdir($dir, 0755, true)) {
            throw new RuntimeException('Kan cache directory niet maken: ' . $dir);
        }
    }

    if (!is_writable($dir)) {
        throw new RuntimeException('Cache directory is niet schrijfbaar: ' . $dir);
    }
}

function catalog_http_read_meta($metaFile)
{
    if ($metaFile === null || !is_file($metaFile)) {
        return array();
    }

    $json = file_get_contents($metaFile);

    if ($json === false || $json === '') {
        return array();
    }

    $meta = json_decode($json, true);

    return is_array($meta) ? $meta : array();
}

function catalog_http_write_meta($metaFile, $meta)
{
    if ($metaFile === null) {
        return;
    }

    $json = json_encode($meta, JSON_PRETTY_PRINT | JSON_UNESCAPED_SLASHES);

    if ($json === false) {
        throw new RuntimeException('Kan catalogus-cache metadata niet coderen.');
    }

    if (file_put_contents($metaFile, $json . "\n", LOCK_EX) === false) {
        throw new RuntimeException('Kan catalogus-cache metadata niet schrijven: ' . $metaFile);
    }
}

function catalog_http_validator_headers($meta)
{
    $headers = array();

    if (!empty($meta['etag'])) {
        $headers[] = 'If-None-Match: ' . $meta['etag'];
    }

    if (!empty($meta['last_modified'])) {
        $headers[] = 'If-Modified-Since: ' . $meta['last_modified'];
    }

    return $headers;
}

function catalog_http_fetch($url, $requestHeaders = array(), $userAgent = 'rktsndbx-catalog/1.0', $timeout = 180)
{
    if (function_exists('curl_init')) {
        $ch = curl_init($url);
        $responseHeaders = array();

        curl_setopt_array($ch, array(
            CURLOPT_RETURNTRANSFER => true,
            CURLOPT_FOLLOWLOCATION => true,
            CURLOPT_CONNECTTIMEOUT => 20,
            CURLOPT_TIMEOUT        => $timeout,
            CURLOPT_USERAGENT      => $userAgent,
            CURLOPT_FAILONERROR    => false,
            CURLOPT_HTTPHEADER     => $requestHeaders,
            CURLOPT_HEADERFUNCTION => function ($ch, $line) use (&$responseHeaders) {
                $pos = strpos($line, ':');

                if ($pos !== false) {
                    $name = catalog_http_normalize_header_name(substr($line, 0, $pos));
                    $value = trim(substr($line, $pos + 1));
                    $responseHeaders[$name] = $value;
                }

                return strlen($line);
            },
        ));

        $body = curl_exec($ch);

        if ($body === false) {
            $err = curl_error($ch);
            curl_close($ch);
            throw new RuntimeException('Catalogus ophalen mislukt: ' . $err);
        }

        $status = curl_getinfo($ch, CURLINFO_HTTP_CODE);
        curl_close($ch);

        if ($status !== 304 && ($status < 200 || $status >= 300)) {
            throw new RuntimeException('Catalogus gaf HTTP status ' . $status . ' voor ' . $url);
        }

        return array(
            'status' => $status,
            'headers' => $responseHeaders,
            'body' => $body,
        );
    }

    $headerLines = array('User-Agent: ' . $userAgent);

    foreach ($requestHeaders as $header) {
        $headerLines[] = $header;
    }

    $ctx = stream_context_create(array(
        'http' => array(
            'method' => 'GET',
            'timeout' => $timeout,
            'header' => implode("\r\n", $headerLines) . "\r\n",
            'ignore_errors' => true,
        ),
    ));

    $body = @file_get_contents($url, false, $ctx);

    if ($body === false) {
        throw new RuntimeException('Catalogus ophalen mislukt: ' . $url);
    }

    $status = 0;
    $responseHeaders = array();
    $rawHeaders = isset($http_response_header) && is_array($http_response_header)
        ? $http_response_header
        : array();

    foreach ($rawHeaders as $line) {
        if (preg_match('/^HTTP\/\S+\s+(\d+)/', $line, $m)) {
            $status = (int)$m[1];
            continue;
        }

        $pos = strpos($line, ':');

        if ($pos !== false) {
            $name = catalog_http_normalize_header_name(substr($line, 0, $pos));
            $value = trim(substr($line, $pos + 1));
            $responseHeaders[$name] = $value;
        }
    }

    if ($status !== 304 && ($status < 200 || $status >= 300)) {
        throw new RuntimeException('Catalogus gaf HTTP status ' . $status . ' voor ' . $url);
    }

    return array(
        'status' => $status,
        'headers' => $responseHeaders,
        'body' => $body,
    );
}

function catalog_http_fetch_cached($url, $cacheFile, $metaFile, $ttl, $userAgent, $timeout = 180)
{
    catalog_http_ensure_dir(dirname($cacheFile));

    if (is_file($cacheFile)) {
        $mtime = filemtime($cacheFile);

        if ($mtime !== false && $mtime >= time() - $ttl) {
            $data = file_get_contents($cacheFile);

            if ($data !== false && $data !== '') {
                return $data;
            }
        }
    }

    $meta = catalog_http_read_meta($metaFile);
    $requestHeaders = is_file($cacheFile) ? catalog_http_validator_headers($meta) : array();
    $response = catalog_http_fetch($url, $requestHeaders, $userAgent, $timeout);

    if ($response['status'] === 304) {
        $data = file_get_contents($cacheFile);

        if ($data !== false && $data !== '') {
            touch($cacheFile);
            $meta['checked_at'] = time();
            catalog_http_write_meta($metaFile, $meta);
            return $data;
        }
    }

    $data = $response['body'];

    if (file_put_contents($cacheFile, $data, LOCK_EX) === false) {
        throw new RuntimeException('Kan catalogus-cache niet schrijven: ' . $cacheFile);
    }

    $headers = $response['headers'];
    $now = time();
    $meta = array(
        'url' => $url,
        'fetched_at' => $now,
        'checked_at' => $now,
    );

    if (!empty($headers['etag'])) {
        $meta['etag'] = $headers['etag'];
    }

    if (!empty($headers['last-modified'])) {
        $meta['last_modified'] = $headers['last-modified'];
    }

    catalog_http_write_meta($metaFile, $meta);

    return $data;
}