<?php
/**
 * Erkennung: Tor-Exit, Cloud-/Hosting-Provider, Rechenzentrum vs. Privatanschluss.
 * Ausschließlich frei verfügbare Signale (Tor-Exit-Liste, ASN-Heuristik).
 *
 * @package SafeMyIdent\Api
 */

declare(strict_types=1);

require_once __DIR__ . '/net.php';

/**
 * Tor-Exit-Node-Liste (täglich gecacht) als Set.
 *
 * @return array<string,bool>
 */
function smi_tor_exit_set(): array
{
    $hit = smi_cache_get('tor|exitlist');
    if (is_array($hit)) {
        return $hit;
    }
    $body = smi_provider_get('https://check.torproject.org/torbulkexitlist', 8, ['Accept' => 'text/plain']);
    $set = [];
    if ($body !== '') {
        foreach (preg_split('/\R/', $body) as $line) {
            $line = trim($line);
            if ($line !== '' && smi_ip_version($line) > 0) {
                $set[$line] = true;
            }
        }
    }
    // Auch leere Antwort cachen (kürzer), um Wiederholungen zu vermeiden.
    smi_cache_set('tor|exitlist', $set, $set ? 86400 : 900);
    return $set;
}

/**
 * Ist die IP ein bekannter Tor-Exit-Node?
 *
 * @param string $ip IP.
 * @return bool
 */
function smi_is_tor_exit(string $ip): bool
{
    return isset(smi_tor_exit_set()[$ip]);
}

/**
 * Cloud-/Hosting-Provider anhand AS-Name/ASN erkennen.
 *
 * @param string $asName AS-Name/Holder.
 * @param string $asn    ASN-Zahl.
 * @return array{is_hosting:bool,provider:string,type:string}
 *         type: cloud | hosting | cdn | isp | unknown
 */
function smi_detect_network(string $asName, string $asn): array
{
    $n = strtolower($asName);
    $asn = preg_replace('/\D/', '', $asn);

    // Bekannte Cloud/CDN nach ASN (Auszug) und Namensmuster.
    $cloudAsn = [
        '16509' => ['Amazon AWS', 'cloud'], '14618' => ['Amazon AWS', 'cloud'],
        '15169' => ['Google Cloud', 'cloud'], '396982' => ['Google Cloud', 'cloud'],
        '8075'  => ['Microsoft Azure', 'cloud'],
        '13335' => ['Cloudflare', 'cdn'], '14061' => ['DigitalOcean', 'cloud'],
        '16276' => ['OVH', 'hosting'], '24940' => ['Hetzner', 'hosting'],
        '20473' => ['Vultr / Choopa', 'hosting'], '63949' => ['Akamai / Linode', 'cloud'],
        '14618' => ['Amazon AWS', 'cloud'], '20940' => ['Akamai', 'cdn'],
        '45102' => ['Alibaba Cloud', 'cloud'], '132203' => ['Tencent Cloud', 'cloud'],
        '14421' => ['Oracle Cloud', 'cloud'], '31898' => ['Oracle Cloud', 'cloud'],
        '51167' => ['Contabo', 'hosting'], '60781' => ['LeaseWeb', 'hosting'],
        '9009'  => ['M247', 'hosting'], '35916' => ['MULTA (M247)', 'hosting'],
    ];
    if ($asn !== '' && isset($cloudAsn[$asn])) {
        return ['is_hosting' => true, 'provider' => $cloudAsn[$asn][0], 'type' => $cloudAsn[$asn][1]];
    }

    $map = [
        'amazon'      => ['Amazon AWS', 'cloud'], 'aws'      => ['Amazon AWS', 'cloud'],
        'google'      => ['Google Cloud', 'cloud'], 'microsoft' => ['Microsoft Azure', 'cloud'], 'azure' => ['Microsoft Azure', 'cloud'],
        'cloudflare'  => ['Cloudflare', 'cdn'], 'akamai'   => ['Akamai', 'cdn'], 'fastly' => ['Fastly', 'cdn'],
        'hetzner'     => ['Hetzner', 'hosting'], 'ovh'      => ['OVH', 'hosting'], 'digitalocean' => ['DigitalOcean', 'cloud'],
        'linode'      => ['Linode', 'cloud'], 'vultr'    => ['Vultr', 'hosting'], 'choopa' => ['Vultr / Choopa', 'hosting'],
        'contabo'     => ['Contabo', 'hosting'], 'leaseweb' => ['LeaseWeb', 'hosting'], 'm247' => ['M247', 'hosting'],
        'alibaba'     => ['Alibaba Cloud', 'cloud'], 'tencent' => ['Tencent Cloud', 'cloud'], 'oracle' => ['Oracle Cloud', 'cloud'],
        'scaleway'    => ['Scaleway', 'hosting'], 'gcore'   => ['G-Core', 'cdn'], 'ionos' => ['IONOS', 'hosting'],
        'netcup'      => ['netcup', 'hosting'], 'servers'  => ['Hosting', 'hosting'],
    ];
    foreach ($map as $needle => $info) {
        if (str_contains($n, $needle)) {
            return ['is_hosting' => true, 'provider' => $info[0], 'type' => $info[1]];
        }
    }

    // Heuristik: Hosting-typische Wörter.
    foreach (['hosting', 'datacenter', 'data center', 'dedicated', 'server', 'vps', 'colo', 'cloud'] as $kw) {
        if (str_contains($n, $kw)) {
            return ['is_hosting' => true, 'provider' => $asName, 'type' => 'hosting'];
        }
    }
    // Typische Consumer-ISP-Wörter.
    foreach (['telekom', 'vodafone', 'telefonica', 'o2', 'kabel', 'comcast', 'at&t', 'verizon', 'orange', 'telecom', 'broadband', 'mobile', 'wireless', 'dsl', 'fiber'] as $kw) {
        if (str_contains($n, $kw)) {
            return ['is_hosting' => false, 'provider' => $asName, 'type' => 'isp'];
        }
    }
    return ['is_hosting' => false, 'provider' => $asName, 'type' => 'unknown'];
}