@profullstack/x402-gateway 0.2.0 → 0.2.2

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@profullstack/x402-gateway",
-  "version": "0.2.0",
+  "version": "0.2.2",
   "type": "module",
   "description": "Sell crawl access to AI training crawlers by the day over x402, settled by CoinPay. One middleware: 402 with an offer, a sales page with CLI instructions, signed passes, and a robots.txt that keeps search crawlers welcome.",
   "keywords": [

package/src/edge.js

@@ -59,22 +59,42 @@ export function inCidrs(ip, compiled) {
 /**
  * The caller's address, as the edge reported it.
  *
- * `x-forwarded-for` is a list the client can seed; the LAST hop appended by
- * our own edge is trustworthy and the first is not, but every platform in
- * front of these sites (Railway, a bare droplet behind nginx) puts the real
- * client first and nothing else, so first is what is used. `x-real-ip` is the
- * nginx spelling of the same thing.
+ * `x-forwarded-for` is a list the client can seed, and a denylist read from
+ * its first entry is a denylist any client can step around by sending one.
+ * The entry our own edge appends is the LAST, on every platform in front of
+ * these sites (Railway's proxy, nginx with `$proxy_add_x_forwarded_for`), so
+ * last is what is used. `x-real-ip` is nginx's spelling of the same hop and
+ * is preferred when present, because nginx sets it from the socket and
+ * nothing a client sends survives into it.
+ *
+ * A CDN in front of the edge would make the last hop the CDN's; put its
+ * ranges nowhere near `denyCidrs` and this still fails safe: nothing is
+ * refused, nothing is charged, by this check.
  */
 export function clientIp(request) {
+  const real = request.headers.get('x-real-ip')?.trim();
+  if (real) return real;
   const xff = request.headers.get('x-forwarded-for');
-  if (xff) return xff.split(',')[0].trim();
-  return request.headers.get('x-real-ip')?.trim() ?? '';
+  if (!xff) return '';
+  const hops = xff.split(',').map((h) => h.trim()).filter(Boolean);
+  return hops[hops.length - 1] ?? '';
 }
 
 /* -------------------------------------------------------------- spoofing -- */
 
 const CLAIMS_CHROMIUM = /\bChrome\/\d+/;
 
+/**
+ * A crawler that says so. Googlebot's evergreen string is
+ * "Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible;
+ * Googlebot/2.1; +http://www.google.com/bot.html) Chrome/W.X.Y.Z Safari/537.36"
+ * -- it claims Chrome, it sends no Sec-Fetch-Mode, and it is the last thing
+ * on earth to charge. Bingbot is built the same way. Anything that declares
+ * itself is judged by the lists, never by this check: the whole point of the
+ * check is the client that declares nothing.
+ */
+const DECLARES_ITSELF = /compatible;|\bbot\b|bot\/|crawler|spider|slurp/i;
+
 /**
  * A request that claims a Chromium user agent but carries none of the
  * fetch-metadata headers Chromium cannot omit.
@@ -88,5 +108,6 @@ const CLAIMS_CHROMIUM = /\bChrome\/\d+/;
 export function isSpoofedBrowser(request) {
   const ua = request.headers.get('user-agent') ?? '';
   if (!CLAIMS_CHROMIUM.test(ua)) return false;
+  if (DECLARES_ITSELF.test(ua)) return false;
   return !request.headers.has('sec-fetch-mode');
 }