""" Probe Surebet247 for BTTS by injecting custom RPC calls into the live browser WS session. Approach: Step 1 - navigate normally and capture event IDs (no init script). Step 2 - navigate to a blank page so the WS closes cleanly. Step 3 - inject a WS-intercepting init script on the CONTEXT, then navigate again. Step 4 - send custom profile probes via page.evaluate(), capture replies. Usage: python3 tools/discover_surebet247_btts_profiles.py """ import sys, os sys.path.insert(0, os.path.join(os.path.dirname(__file__), '..')) import time, collections import msgpack from scrapers.surebet247 import _Surebet247Worker, _decode_frame BASE_SITE = 'https://www.surebet247.com' CONTEXT = ['en', 'MOBILE_WEB', 'CL38B1', '', 'NGN'] PROFILES_TO_TRY = [ 'pro_all', 'pro_extended', 'pro_popular', 'pro_full', 'pro_combo', 'pro_dc', 'pro_default', 'pro_gg', 'pro_btts', 'pro_goals', 'pro_specials', 'pro_1x2_dc', ] KNOWN_TYPES = { 1: 'Home/Away', 2: '1X2', 4: 'Asian Handicap', 5: 'Over/Under', } def encode_rpc(req_id: int, method: str, args: list) -> bytes: raw = msgpack.packb([4, {}, req_id, method, args], use_bin_type=True) return b'\x00' + raw def probe_profiles(page) -> dict: # ── Step 1: navigate WITHOUT init script, capture event IDs ────────────── print(' Step 1: Loading football page normally...') frames_recv_1 = [] frames_sent_1 = [] last_recv = [time.time()] def handle_ws_1(ws): if 'direct-feed' not in ws.url: return ws.on('framereceived', lambda p: (frames_recv_1.append(p), last_recv.__setitem__(0, time.time())) if isinstance(p, bytes) else None) ws.on('framesent', lambda p: frames_sent_1.append(p) if isinstance(p, bytes) else None) page.on('websocket', handle_ws_1) page.goto(f'{BASE_SITE}/sport/football', wait_until='domcontentloaded', timeout=60_000) time.sleep(8) for _ in range(5): page.mouse.wheel(0, 700) time.sleep(0.6) deadline = time.time() + 20 while time.time() < deadline: time.sleep(1) if time.time() - last_recv[0] > 3.0 and frames_recv_1: break try: page.remove_listener('websocket', handle_ws_1) except Exception: pass print(f' Step 1 done: {len(frames_recv_1)} recv / {len(frames_sent_1)} sent') # Extract event IDs event_ids = [] for frame in frames_sent_1: if not isinstance(frame, bytes): continue d = _decode_frame(frame) if not d or not isinstance(d, (list, tuple)) or len(d) < 5: continue if str(d[3]) == 'GetMainMarketsByProfileAndEventIds': ids = d[4][1] if isinstance(d[4], (list, tuple)) and len(d[4]) > 1 else [] if isinstance(ids, (list, tuple)): event_ids.extend(str(i) for i in ids) event_ids = list(dict.fromkeys(event_ids))[:6] if not event_ids: method_by_id = {} for frame in frames_sent_1: if isinstance(frame, bytes): d = _decode_frame(frame) if d and isinstance(d, (list, tuple)) and len(d) >= 4 and d[0] == 4: method_by_id[str(d[2])] = str(d[3]) rich_ids = {rid for rid, m in method_by_id.items() if m in ('GetRichEventsByTournamentIdAndStage', 'GetRichEventsByIds')} for frame in frames_recv_1: if not isinstance(frame, bytes): continue d = _decode_frame(frame) if not d or not isinstance(d, (list, tuple)) or d[0] != 2 or len(d) < 4: continue if str(d[2]) not in rich_ids: continue reply = d[3] if not isinstance(reply, (list, tuple)) or len(reply) < 2 or not reply[0]: continue for item in (reply[1] or []): if isinstance(item, (list, tuple)) and len(item) >= 2: eid = item[1] if isinstance(eid, (int, str)): event_ids.append(str(eid)) event_ids = list(dict.fromkeys(event_ids))[:6] print(f' Event IDs: {event_ids}') if not event_ids: print(' No event IDs — aborting.') return {} # ── Step 2: inject WS hook via context.add_init_script, then navigate again print('\n Step 2: Adding WS hook via context, re-navigating...') try: ctx = page.context ctx.add_init_script(""" (function() { var _orig = WebSocket.prototype.send; WebSocket.prototype.send = function(data) { if (!window._dfw && this.url && this.url.indexOf('direct-feed') !== -1) { window._dfw = this; } return _orig.apply(this, arguments); }; })(); """) print(' Context init script added.') except Exception as e: print(f' Context init script error: {e}') frames_recv_2 = [] frames_sent_2 = [] last_recv2 = [time.time()] def handle_ws_2(ws): if 'direct-feed' not in ws.url: return ws.on('framereceived', lambda p: (frames_recv_2.append(p), last_recv2.__setitem__(0, time.time())) if isinstance(p, bytes) else None) ws.on('framesent', lambda p: frames_sent_2.append(p) if isinstance(p, bytes) else None) page.on('websocket', handle_ws_2) page.goto(f'{BASE_SITE}/sport/football', wait_until='domcontentloaded', timeout=60_000) time.sleep(10) for _ in range(3): page.mouse.wheel(0, 700) time.sleep(0.5) deadline = time.time() + 15 while time.time() < deadline: time.sleep(1) if time.time() - last_recv2[0] > 3.0 and frames_recv_2: break print(f' Step 2 load: {len(frames_recv_2)} recv / {len(frames_sent_2)} sent') ws_state = page.evaluate("""() => { if (!window._dfw) return 'not captured'; return 'readyState=' + window._dfw.readyState; }""") print(f' WS state in JS: {ws_state}') if 'readyState=1' not in ws_state: print(' WS not open — cannot inject probes.') return {} # ── Step 3: send probe frames ───────────────────────────────────────────── print('\n Step 3: Sending profile probes...') PROBE_BASE = 9000 pending = {} for i, profile in enumerate(PROFILES_TO_TRY): rid = PROBE_BASE + i pending[rid] = profile frame = encode_rpc(rid, 'GetMainMarketsByProfileAndEventIds', [profile, event_ids, 4, 1, CONTEXT]) result = page.evaluate("""(d) => { if (!window._dfw || window._dfw.readyState !== 1) return 'ws_closed'; window._dfw.send(new Uint8Array(d).buffer); return 'ok'; }""", list(frame)) print(f' {profile!r:25s} → {result}') time.sleep(0.4) time.sleep(7) # ── Step 4: parse replies ───────────────────────────────────────────────── results = {} all_recv = frames_recv_1 + frames_recv_2 for frame in all_recv: if not isinstance(frame, bytes): continue d = _decode_frame(frame) if not d or not isinstance(d, (list, tuple)) or d[0] != 2 or len(d) < 4: continue rid = d[2] if rid not in pending: continue reply = d[3] if not isinstance(reply, (list, tuple)) or len(reply) < 2 or not reply[0]: results[rid] = {} continue items = reply[1] if not isinstance(items, (list, tuple)): results[rid] = {} continue market_types = collections.defaultdict(list) for item in items: if not isinstance(item, (list, tuple)) or len(item) < 3: continue item_key = item[1] if not isinstance(item_key, (list, tuple)) or len(item_key) < 3: continue mtype = item_key[2] period = item_key[3] if len(item_key) > 3 else None mval = item[2] if not isinstance(mval, (list, tuple)) or not mval: continue for group in (mval[0] or []): if not isinstance(group, (list, tuple)) or len(group) < 2: continue spec = group[0] ocs = [] for oc in (group[1] or []): if isinstance(oc, (list, tuple)) and len(oc) >= 2: oc_type = oc[0][0] if isinstance(oc[0], (list, tuple)) and oc[0] else '?' price = round(oc[1] / 100, 2) if isinstance(oc[1], (int, float)) else oc[1] ocs.append((oc_type, price)) if ocs: market_types[(mtype, period, str(spec))].append(ocs) results[rid] = dict(market_types) return {'pending': pending, 'results': results} if __name__ == '__main__': print('Probing Surebet247 profiles for BTTS...\n') worker = _Surebet247Worker() data = worker.call(probe_profiles, timeout=250) if not data: print('No data returned.') else: pending = data.get('pending', {}) results = data.get('results', {}) print('\n\n═══ RESULTS ═══\n') for rid, profile in sorted(pending.items()): mkt = results.get(rid) if mkt is None: print(f' {profile!r:25s} → no reply') elif not mkt: print(f' {profile!r:25s} → empty / error') else: type_ids = sorted(set(t for (t, p, s) in mkt.keys())) new_types = [t for t in type_ids if t not in KNOWN_TYPES] label = ', '.join(KNOWN_TYPES.get(t, f'UNKNOWN({t})') for t in type_ids) flag = ' ← NEW TYPES!' if new_types else '' print(f' {profile!r:25s} → [{label}]{flag}') for (mtype, period, spec), samples in sorted(mkt.items()): s = samples[0] oc_ids = [o[0] for o in s] prices = [o[1] for o in s] known = KNOWN_TYPES.get(mtype, f'UNKNOWN({mtype})') print(f' type={mtype:<5} period={period} spec={spec:<15} {known:<20} oc_ids={oc_ids} prices={prices}')