fix: make the SOCKS5 proxy work for BookMyShow too

BookMyShow blocks datacenter IPs on top of TLS fingerprinting, confirmed by CI
runners getting 403 with a working impersonation, so cloud hosts need the proxy
for BMS as well as District. impit resolves DNS locally for socks5:// and hands
an IPv6-only exit an unreachable IPv4 address, so the scheme is normalized to
socks5h://. Verified end to end: a proxied BMS check returns 200 and alerts.

Docs said the proxy did nothing for BookMyShow, which was wrong.

.env.example

@@ -60,7 +60,8 @@ CURRENCY=₹
 STATE_PATH=./state.json        # Docker image defaults this to /data/state.json
 PORT=4733                      # status endpoint (--loop mode only)
 
-# Routes District checks through a SOCKS5 proxy; notifications stay direct.
-# Needed on datacenter hosts, whose IPs District's WAF blocks. It does nothing for
-# BookMyShow, which blocks on TLS fingerprint rather than IP (impit handles that).
+# Routes both sites' checks through a SOCKS5 proxy; notifications stay direct.
+# Both District and BookMyShow block datacenter IPs, so a cloud host usually needs this.
+# socks5:// is rewritten to socks5h:// internally, so the proxy resolves DNS. That is
+# required for IPv6-only exits and harmless otherwise.
 # PROXY_URL=socks5://user:pass@proxy.example.net:1080

README.md

@@ -181,7 +181,7 @@ with comments in [`.env.example`](.env.example).
 | `EXPIRY_UTC` | *auto* | When to retire. Defaults to local midnight after `TARGET_DATE`. |
 | `STATE_PATH` | `./state.json` | Where dedupe state lives, `/data/state.json` in Docker |
 | `PORT` | `4733` | Status endpoint |
-| `PROXY_URL` | | SOCKS5 proxy for District checks, see below |
+| `PROXY_URL` | | SOCKS5 proxy for both sites' checks, see below |
 
 v1 used flat names (`MOVIE_CODE`, `CINEMA_ID` and friends) for District. Those still work, so an
 existing `.env` keeps running unchanged.
@@ -210,13 +210,21 @@ permanently, while a home connection was never touched once. Either run it from
 connection, or point `PROXY_URL` at a SOCKS5 proxy. Only the site checks go through the proxy.
 Notifications stay direct, so your alerts don't depend on it.
 
+One gotcha if your proxy is IPv6-only: a `socks5://` URL makes some clients resolve DNS locally
+and hand the proxy an IPv4 address it cannot reach. The monitor rewrites the scheme to `socks5h://`
+so the proxy does the resolving, which fixes that and changes nothing for other proxies.
+
 ### BookMyShow returns HTTP 403
 
-BookMyShow blocks on the TLS fingerprint of the client, not on your IP, which is why plain
-`curl` and Node's `fetch` both get 403 no matter which headers or cookies you send. A proxy
-won't help either. This is handled with [impit](https://github.com/apify/impit), which performs
-the request with a real Chrome fingerprint. If you see 403s here, the impersonation has probably
-stopped matching, so try updating the image or `npm update impit`.
+BookMyShow blocks on two things at once. First the TLS fingerprint of the client, which is why
+plain `curl` and Node's `fetch` get 403 no matter which headers or cookies you send. That part is
+handled by [impit](https://github.com/apify/impit), which performs the request with a real Chrome
+fingerprint. Second your IP, if it belongs to a datacenter: BookMyShow 403s cloud hosts even with
+a perfect fingerprint, which I confirmed from GitHub Actions runners.
+
+So on a cloud host you need both, and setting `PROXY_URL` covers the second half. If you are still
+getting 403s from a residential connection, the impersonation has probably stopped matching the
+current Chrome, so update the image or run `npm update impit`.
 
 ### ntfy returns HTTP 429
 

src/index.js

@@ -33,9 +33,11 @@ function writeState(cfg, state) {
 
 /**
  * Routes provider requests through a SOCKS5 proxy. Notifications stay direct, so the
- * alerting path never depends on the proxy. District needs this on datacenter hosts whose
- * IP the WAF blocks. It does nothing for BookMyShow, which blocks on TLS fingerprint
- * rather than IP; impit handles that instead.
+ * alerting path never depends on the proxy.
+ *
+ * Both sites block datacenter IPs, so a cloud host generally needs this for both. It is
+ * only half the story for BookMyShow, which also blocks on TLS fingerprint: impit deals
+ * with the fingerprint, the proxy deals with the IP, and BMS needs both.
  */
 async function attachProxy(cfg) {
   if (!cfg.proxyUrl) return;

src/providers/bookmyshow.js

@@ -18,9 +18,13 @@ let client = null;
 async function getClient(cfg) {
   if (client) return client;
   if (!Impit) ({ Impit } = await import("impit"));
+  // With socks5:// impit resolves DNS locally and asks the proxy for that address,
+  // which fails outright on an IPv6-only exit that cannot reach the IPv4 we picked.
+  // socks5h:// leaves resolution to the proxy, so normalize to it.
+  const proxyUrl = cfg.proxyUrl ? cfg.proxyUrl.replace(/^socks5:\/\//i, "socks5h://") : "";
   client = new Impit({
     browser: "chrome",
-    ...(cfg.proxyUrl ? { proxyUrl: cfg.proxyUrl } : {}),
+    ...(proxyUrl ? { proxyUrl } : {}),
   });
   return client;
 }