nx_cam_chaturbate.nx

nx_cam_chaturbate.nx -- /compare/mediaingest R9 (lr_watch) SITE ADAPTER: Chaturbate. WHY THIS EXISTS, MEASURED 2026-09-01 -- not recalled, and it corrects the record. The retired Python detector whose output sits in ~/.ctbrec/logs tried 8 regex patterns, 4 JS-object patterns and one API endpoint against the ROOM PAGE, then reported: "Room Status: PRIVATE (found 'private show') / Streams Found: 0 / DETECTION FAILED" "3. JavaScript rendering required ... Consider using Selenium/WebDriver" Re-reading the HTML that same run SAVED (page_dj_woman_20251029_110425.html, 184,898 B) refutes every part of that conclusion: - the body carries next="/?next=/dj_woman/" plus roomlist / endless_page markers, i.e. the request was BOUNCED TO THE LANDING PAGE and the room page was never retrieved at all; - the page <title> is the generic Chaturbate homepage title, not a room title; - the literal "private show" it reports matching is homepage boilerplate (3 hits, none a room status field), so the PRIVATE verdict was read off the wrong document. ALL THIRTEEN "pattern misses" therefore have ONE cause, and it is not JavaScript. A check that ran on the wrong subject produced a confident, complete-looking false negative, and its remedy ("render JavaScript") would have bought a browser engine to fix a fetch that never landed. INDEPENDENT SECOND WITNESS, found in our own tree: nx_http_client.nx's build_request_cookie comment already says "chaturbate's /?next= age-gate hands back csrftoken/sbr/AG_Key that must be replayed to reach the room -- without a jar we land on the gate page forever." THE REAL MECHANISM needs no room page and no browser. Two endpoints: 1. STATUS IN BULK -- GET /affiliates/api/onlinerooms/?format=json returns EVERY online room in ONE request. This is the rate-limit answer: N watched performers cost 1 call, not N. Paying one ajax per performer across a large fleet is precisely what flags an exit IP into a site-wide 403, so bulk is not an optimisation here, it is the correctness path. 2. URL RESOLVE -- POST /get_edge_hls_url_ajax/ (application/x-www-form-urlencoded, room_slug=<name>&bandwidth=high, X-Requested-With: XMLHttpRequest) -> JSON carrying room_status, url and cmaf_edge. *** THE EDGE TOKEN IN THAT url IS SINGLE-USE / CONNECTION-BOUND. *** ANY fetch of the tokenised playlist -- even a HEAD, even a "quick validity check" -- CONSUMES it, and the next consumer gets HTTP 403. So this organ RESOLVES AND RETURNS; it never probes, never validates and never pre-fetches the URL it hands back. That is not style: our own nx_hls_get fetches the MASTER playlist and THEN fetches the variant media playlist, which spends the token on the first fetch and 403s on the second. cb_capture_contract() states the handoff rule in the return value rather than in a comment, because a rule that lives only in a comment is invisible to every caller -- the same lesson nx_https_client.nx's NOT_IMPLEMENTED verdict records. EGRESS ETHICS (standing estate bound, /compare/mediaingest R12): sovereign-owned egress only.