Files
4grayandClaude Fable 5 3326f73388 fix(stalker): send cmd in the reference MAG wire format
A real MAG sends cmd unencoded and the portal decodes its query exactly
once, so a cmd that already contains percent sequences (%3A tokens,
pre-encoded path segments) must pass through untouched. The previous
encodeURIComponent transport (2c032cd3c, 0.22) double-encoded such cmds
(%3A -> %253A): strict portals and reseller panels that compare cmd
literally, and stock create_link handlers matching the decoded value,
saw a different string than a real STB sends.

The new shared encodeStalkerCmdValue() reproduces the reference wire
bytes: % passes through verbatim, characters the WHATWG URL serializer
keeps raw in a query stay raw (so the bytes survive the axios/new URL
transport unchanged), and everything else is percent-encoded. That
preserves the 0.22 injection protection - &, # (and ; for PHP setups
with a ; argument separator) inside cmd cannot append or truncate query
parameters; they decode back to the original byte server-side.

Both transports now share the format: the Electron query builder is
extracted to buildStalkerRequestUrl() and the web-backend /stalker
proxy appends cmd to the portal URL itself instead of letting axios
turn slashes into %2F (the opposite divergence).

Also unifies the two divergent response-side cmd normalizers: the
cross-portal collection resolver now uses the Stalker store's
normalizeStalkerPlaybackCommand/resolveStalkerPlaybackUrl, so playing
from Favorites/global collections resolves relative (/media/...) and
query-only (?token=...) create_link replies against the portal base
instead of handing the player a bare relative path.

The mock portal's create_link response gains mock-only cmd_received/
query_keys_received diagnostics; a new Electron e2e pins the contract
end-to-end (single decode, injection blocked). Unit corpus tests cover
the encoder, the Electron builder, the web-backend proxy, and the
resolver.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-08-01 22:33:47 +02:00
..

Stalker Mock Server

A local mock implementation of the Stalker/Ministra portal API for development and end-to-end testing of IPTVnator.

Overview

The mock server speaks the same portal.php HTTP protocol as a real Stalker portal, generating deterministic fake data using @faker-js/faker seeded from the connecting MAC address. This means:

  • The same MAC address always returns the same data (consistent across page refreshes and test runs).
  • Different MAC addresses produce different datasets — use predefined scenario MACs for specific test conditions.
  • Data is generated once per MAC on first request and cached in memory for the server's lifetime. Restart to regenerate.

Quick Start

# Start the mock server (port 3210)
nx serve stalker-mock-server

# Or with file watching (auto-restarts on source changes)
nx run stalker-mock-server:serve-with-watch

# Or run both the mock server + Angular dev server in parallel
nx run-many --targets=serve --projects=stalker-mock-server,web

Then in IPTVnator, add a new Stalker portal:

  • Portal URL: http://localhost:3210/portal.php (tolerant panel-style endpoint) or http://localhost:3210/stalker_portal/server/load.php (canonical Ministra endpoint — see Two endpoints below)
  • MAC Address: one of the predefined scenarios below (or any MAC for auto-generated data)

Two endpoints: tolerant vs strict

The same actions are served at two paths with deliberately different strictness, because the app treats them differently: a URL containing /stalker_portal is imported as a full portal (handshake + token + watchdog), anything else as a simple portal (no authentication at all).

Path Behaviour
/portal.php Tolerant. Ignores the Bearer token and the MAC format, like most reseller panels in the wild.
/stalker_portal/server/load.php Strict. Enforces the token and the MAC format exactly like the real middleware.
/server/load.php Strict. The second full-portal URL shape the app recognizes; enforced identically.

Known app inconsistency: StalkerSessionService.isFullStalkerPortal classifies /server/load.php as a full portal, but the import dialog's isFullStalkerPortalUrl checks only for /stalker_portal, so importing a bare …/server/load.php URL persists isFullStalkerPortal: false and the app skips the handshake. The mock is deliberately faithful to a real portal here (that path enforces auth), which makes it the right fixture to drive the upcoming fix that unifies those two predicates. Until then, import full portals through a /stalker_portal/... URL.

The strict endpoint reproduces the parts of Stalker 4.9.35 that a client can actually get wrong:

  • Every action except handshake, get_profile, get_localization and do_auth requires Authorization: Bearer <token>.
  • A token only counts once get_profile has adopted it — a handshake alone is not a session. Adoption is deliberately stricter than the stock server: only tokens the mock actually issued (or the already-bound one) are accepted, so a client with a broken token pipeline fails loudly.
  • Auth failures come back as HTTP 200 with a plain-text body (Authorization failed., Unauthorized request.), never a 401/403. Clients that only check status codes will silently render nothing.
  • The handshake is idempotent: presenting the MAC's current token returns that same token instead of rotating it.
  • device_id/device_id2 are pinned to the MAC on first non-empty value; any later change — including sending them empty again — is a permanent device conflict with the "Your STB is damaged." block message.
  • signature, metrics and prehash are accepted and ignored, exactly as the stock server does.
  • The MAC must match the Infomir OUI format (00:1A:79:XX:XX:XX) or get_profile answers with a bare { status: 1 }.

Predefined Scenario MAC Addresses

MAC Address Scenario Description
00:1A:79:00:00:01 default 8 categories per type, 40 items each — the balanced go-to for daily dev
00:1A:79:FF:FF:FF large 20 categories, 200 items each — stress-test pagination and virtual scroll
00:1A:79:00:00:02 series-heavy 15 series categories with 6 seasons × 10 episodes — test deep series navigation
00:1A:79:00:00:03 minimal 2 categories, 5 items — edge case testing (empty states, single items)
00:1A:79:00:00:04 is-series 60% of VOD items have is_series=1 — tests the Ministra lazy-season flow
00:1A:79:00:00:05 embedded-series 50% of VOD items have embedded series[] arrays — tests the embedded series flow
00:1A:79:00:00:06 legacy-pagination No get_all_channels support — tests the paginated get_ordered_list crawl fallback for the full ITV channel list
00:1A:79:00:00:07 marketing-demo 35 original poster movies with the newest 20 first — safe for screenshots and marketing
00:1A:79:00:00:08 login-required get_profile answers status: 2 until the client completes do_auth with non-empty credentials. The app cannot finish this flow yet (its do_auth path is dormant and sends empty credentials), so the scenario is exercised at the HTTP level only — it exists to receive the upcoming client-side do_auth work
<any other MAC> auto MAC bytes used as seed → deterministic unique dataset

Configuration

Environment Variable Default Description
PORT 3210 HTTP port the server listens on
NODE_ENV development Node environment

Utility Endpoints

Endpoint Method Description
/health GET Health check — returns { status: "ok" }
/reset POST Clear all in-memory data, favorites, sessions and watchdog counters (useful between test runs)
/invalidate-session?macAddress=<mac> POST Drop that MAC's tokens so the next portal call fails with Authorization failed. — lets tests assert the client re-handshakes and retries. Pinned device identity survives, as on a real portal

API Coverage

All endpoints are served at GET /portal.php?action=<action>&... matching the real Stalker protocol:

Action Description
handshake Issues the access token (idempotent) plus the 5.x random nonce and not_valid flag
get_profile Turns the handshake token into a session; enforces device-id pinning, and on the strict endpoint the MAC format
get_events Watchdog ping; records the call and returns an empty event set (never affects authorization, as on a real portal)
do_auth Boolean login step: {js:true} for non-empty credentials (recorded for the login-required scenario), {js:false} otherwise
get_categories Category list filtered by type (itv/vod/series)
get_genres Genre list (mirrors categories)
get_ordered_list Paginated content list; if movie_id is present → returns seasons
get_all_channels Complete ITV channel list in one response (type=itv only); excludes censored (adult) genres; disabled in the legacy-pagination scenario
create_link Returns a real public HLS stream URL for playback
favorites Add / remove / get favorites (in-memory, resets on restart)
get_short_epg Current-and-upcoming EPG window for a channel (ch_id, size)
get_epg_info Bulk EPG keyed by channel id for a requested period window

Cover Images

Generated scenarios use Picsum Photos for cover images and logos, so they need an internet connection to display artwork. The marketing-demo scenario instead uses the committed, screenshot-safe poster catalog shared with the Xtream mock. Stalker serves those PNGs itself from /assets/marketing/poster/<slug>.png, so screenshots remain deterministic and offline once the repository is checked out.

Stream URLs

create_link returns real public HLS test streams so video actually plays:

  • https://test-streams.mux.dev/x36xhzz/x36xhzz.m3u8
  • https://devstreaming-cdn.apple.com/videos/streaming/examples/bipbop_4x3/bipbop_4x3_variant.m3u8
  • https://playertest.longtailvideo.com/adaptive/oceans/oceans.m3u8
  • https://playertest.longtailvideo.com/adaptive/bbbfull/bbbfull.m3u8

The stream chosen for a given item is deterministic based on the item's cmd string.

Using with Playwright E2E Tests

The Playwright config in apps/web-e2e/playwright.config.ts starts the mock server automatically alongside the Angular dev server when running e2e tests. See apps/web-e2e/src/stalker.e2e.ts for example stalker tests.

# Run all e2e tests (starts mock server automatically)
nx e2e web-e2e

# Or run only stalker-specific e2e tests
nx e2e web-e2e --grep "@stalker"

The test suite uses 00:1A:79:00:00:01 (default scenario) for most tests, and calls POST /reset in beforeEach to ensure a clean state between tests.

EPG Behavior

The mock server generates a 7-day EPG schedule for every ITV channel using 2-hour slots starting at the current UTC day boundary.

  • get_short_epg returns the current program and upcoming items from that schedule, limited by size
  • get_epg_info returns bulk data in the shape { js: { data: Record<channelId, program[]> } }
  • get_epg_info filters the bulk response from the current UTC day start through now + period

Architecture

See docs/architecture/stalker-mock-server.md for full implementation details.

Project Structure

apps/stalker-mock-server/
├── src/
│   ├── main.ts                            # Express bootstrap
│   └── app/
│       ├── scenarios.ts                   # MAC → scenario config mapping
│       ├── data-generator.ts              # Seeded faker data generation
│       ├── data-store.ts                  # Lazy per-MAC in-memory cache
│       ├── routes/
│       │   ├── portal.route.ts            # /portal.php route
│       │   └── dispatch.ts                # Shared Stalker action dispatcher
│       └── handlers/
│           ├── handshake.handler.ts
│           ├── do-auth.handler.ts
│           ├── get-categories.handler.ts
│           ├── get-ordered-list.handler.ts
│           ├── get-seasons.handler.ts
│           ├── create-link.handler.ts
│           ├── favorites.handler.ts
│           ├── get-epg-info.handler.ts
│           ├── get-short-epg.handler.ts
│           └── get-genres.handler.ts
├── project.json
├── tsconfig.json
└── README.md