<?php
/**
 * Same-origin proxy for layer legend images.
 *
 * WHY THIS EXISTS
 * The application is served cross-origin isolated — index.php and the container
 * both send `Cross-Origin-Embedder-Policy: require-corp`, which SQLite in a
 * worker needs. Under that policy a cross-origin image is blocked unless the
 * provider sends `Cross-Origin-Resource-Policy`, and most OGC servers do not.
 * A CORS request satisfies COEP on its own, so the client tries the provider
 * directly first (see _registerLegend in MapView.js); this proxy is the
 * fallback for providers where that still fails. Fetching the image here makes
 * it same-origin, which sidesteps the policy entirely.
 *
 * SECURITY
 * This script runs inside the Docker network, where it can reach ApiEngine,
 * MainBackend and the database — none of which are reachable from outside. An
 * open `?url=` proxy would therefore be a server-side request forgery hole
 * straight into them. Hosts are allow-listed, and the resolved address is
 * checked against private and reserved ranges as a second line of defence, so
 * a DNS entry that later points inward cannot be used either. Redirects are not
 * followed, because a 302 to an internal host would defeat both checks.
 *
 * Legend URLs come from the application's own layer definitions in main.js,
 * never from user input, so the allow-list costs nothing today. Adding a host
 * is a deliberate decision: it grants that host the right to have its content
 * served from our origin.
 */

declare(strict_types=1);

/** Hosts whose legend images may be proxied. */
const ALLOWED_HOSTS = [
    'ows.digitalearth.africa',          // datacube-ows — sends ACAO but no CORP
    'geoserver.digitalearth.africa',    // GeoServer — sends neither, so needs this proxy
];

const MAX_BYTES = 2097152;   // 2 MB — legends are a few KB; this is a sanity cap
const TIMEOUT_S = 10;

/**
 * Refuse without saying why, and without reflecting the requested URL back.
 *
 * @param int $code
 * @return void
 */
function deny($code = 404)
{
    http_response_code($code);
    header('Content-Type: text/plain; charset=utf-8');
    header('Cache-Control: no-store');
    echo "Legend not available\n";
    exit;
}

if (($_SERVER['REQUEST_METHOD'] ?? 'GET') !== 'GET') {
    deny(405);
}

$raw = isset($_GET['url']) ? (string) $_GET['url'] : '';
if ($raw === '' || strlen($raw) > 2048) {
    deny(400);
}

$parts = parse_url($raw);
if ($parts === false || empty($parts['scheme']) || empty($parts['host'])) {
    deny(400);
}

$scheme = strtolower($parts['scheme']);
if ($scheme !== 'http' && $scheme !== 'https') {
    deny(400);
}

$host = strtolower($parts['host']);
if (!in_array($host, ALLOWED_HOSTS, true)) {
    deny(403);
}

// Second line of defence: even an allow-listed name must not resolve inward.
$ip = gethostbyname($host);
if ($ip === $host) {
    deny(502);                      // name did not resolve
}
if (!filter_var($ip, FILTER_VALIDATE_IP, FILTER_FLAG_NO_PRIV_RANGE | FILTER_FLAG_NO_RES_RANGE)) {
    deny(403);                      // resolves to a private or reserved address
}

$ch = curl_init($raw);
curl_setopt_array($ch, [
    CURLOPT_RETURNTRANSFER => true,
    CURLOPT_FOLLOWLOCATION => false,     // a redirect could point inward
    CURLOPT_CONNECTTIMEOUT => 5,
    CURLOPT_TIMEOUT        => TIMEOUT_S,
    CURLOPT_PROTOCOLS      => CURLPROTO_HTTP | CURLPROTO_HTTPS,
    CURLOPT_USERAGENT      => 'LUPMIS2-legend-proxy/1.0',
    CURLOPT_ACCEPT_ENCODING => '',
    // Stop reading once the cap is passed, rather than buffering a huge body.
    CURLOPT_BUFFERSIZE     => 16384,
    CURLOPT_NOPROGRESS     => false,
    CURLOPT_PROGRESSFUNCTION => function ($res, $dlTotal, $dlNow) {
        return ($dlTotal > MAX_BYTES || $dlNow > MAX_BYTES) ? 1 : 0;
    },
]);

$body   = curl_exec($ch);
$status = (int) curl_getinfo($ch, CURLINFO_RESPONSE_CODE);
$type   = (string) curl_getinfo($ch, CURLINFO_CONTENT_TYPE);
curl_close($ch);

if ($body === false || $status !== 200) {
    deny(502);
}
if (strlen($body) > MAX_BYTES) {
    deny(502);
}
// Only ever hand back an image, whatever the upstream claims to be.
if (stripos($type, 'image/') !== 0) {
    deny(502);
}

$clean = explode(';', $type)[0];
header('Content-Type: ' . trim($clean));
header('Content-Length: ' . strlen($body));
header('Cache-Control: public, max-age=86400');
header('X-Content-Type-Options: nosniff');
header('Cross-Origin-Resource-Policy: same-origin');
echo $body;