#!/usr/bin/env python3 """Deep analysis for 2026-11-05 x-account-monitor run. Buckets assigned by HANDLE GROUP first (never AI-filtered inside the loop), disjoint-sum verified. Prints raw + organic core rankings with like rates, 24h deltas vs 11-04, per-handle freshness, originals-only dormancy, offer/pricing grep over in-window originals, and new-handle checks. """ import json, os, re from datetime import datetime, timedelta, timezone RAW = "/home/kiro/trend-engine/trends/x/raw" D = "2026-11-05" P = "2026-11-04" IRRELEVANT = {"AIBuzzNews","AIGuideHQ","AITECHio","AIFrontliner","TheAIColony","AITOPIAai", "GoogleLabs","GoogleAI","DonutAI","cladlabs","FaithJaiye2187","KammelaBala","MyStageAI", "AlexStudio44","AI_with_jasmin","Ai_Vaidehi","ChrisLaubAI","CINEMA505","OpenAI","AnthropicAI", "AIScottAdams","AISuperDomain","AI_Jasonyu","AIByLeo","Soaima_Ai","different_ai","via_ai_", "Teamily_AI","AIsaOneHQ","AiWithTariq","numerai"} NON_CORE = {"interesting_aIl","hubermanlab","NVIDIAAI"} FILM = {"DiscussingFilm","nofilmschool"} CORE = ["runwayml","CuriousRefuge","nofilmschool","venturetwins","Kling_ai","paultrillo", "c_valenzuelab","karenxcheng","higgsfield","pika_labs","EHuanglu","azed_ai","JeffSynthesized", "ai_filmmaker","Magiermogul","WuxiaRocks","JiamingSuen","higgsfield_ai","Chandan_Perla", "CharaspowerAI","AIWarper","ItsMaryamAi","ElsaSofia__AI","AI__TSUBAKI","heyglif"] CORESET = set(CORE) STRONG = re.compile(r"\b(AI|A\.I\.|artificial intelligence|generative|text-to-video|image-to-video|" r"Seedance|Midjourney|Sora|Veo|Kling|Runway|Pika|Higgsfield|ComfyUI|diffusion|" r"machine learning|prompt|ChatGPT|GPT|LLM|agentic|model)\b", re.I) def load(date): return json.load(open(os.path.join(RAW, f"scan_results_{date}_FULL.json"))) def pdate(s): try: return datetime.strptime(s, "%a %b %d %H:%M:%S %z %Y") except Exception: return None def score(q): return q["likes"] + 2*q["rt"] + 3*q["replies"] + q["views"]/100 def likerate(q): return (q["likes"]/q["views"]*100) if q["views"] else 0.0 def tid(url): m = re.search(r"/status/(\d+)", url or "") return m.group(1) if m else None cur = load(D); prev = load(P) q = cur["qualifying"] now = datetime.now(timezone.utc) print(f"=== RUN {D} === handles={len(cur['handles'])} fetched={len(cur['raw'])} " f"failures={len(cur['failures'])} qualifying={len(q)}") print(f"prev {P}: qualifying={len(prev['qualifying'])} -> delta {len(q)-len(prev['qualifying']):+d}") # ---- watchlist diff ---- wl_txt = open("/home/kiro/trend-engine/trends/x/raw/rising-accounts.md").read() wl = sorted({h for h in re.findall(r"@(\w+)", wl_txt)}) print(f"watchlist md handles={len(wl)} scan handles={len(cur['handles'])}") # ---- disjoint buckets by handle group ---- buckets = {"core":[], "film":[], "irrelevant":[], "labs":[], "offtopic":[], "health":[]} for x in q: h = x["handle"] if h == "NVIDIAAI" or h in ("OpenAI","AnthropicAI"): buckets["labs"].append(x) elif h == "interesting_aIl": buckets["offtopic"].append(x) elif h == "hubermanlab": buckets["health"].append(x) elif h in FILM: buckets["film"].append(x) elif h in IRRELEVANT: buckets["irrelevant"].append(x) else: buckets["core"].append(x) tot = sum(len(v) for v in buckets.values()) print("\n=== BUCKETS ===") for k,v in buckets.items(): print(f" {k:11s} {len(v):4d}") print(f" {'ASSIGNED':11s} {tot:4d} vs qualifying {len(q)} -> " f"{'OK' if tot==len(q) else 'MISMATCH'}") film_ai = [x for x in buckets["film"] if STRONG.search(x["text"])] irr_ai = [x for x in buckets["irrelevant"] if STRONG.search(x["text"])] print(f" film AI-relevant: {len(film_ai)} of {len(buckets['film'])}") print(f" irrelevant AI-relevant: {len(irr_ai)} of {len(buckets['irrelevant'])}") # ---- core rankings ---- core = sorted(buckets["core"], key=score, reverse=True) print(f"\n=== CORE TOP 20 (raw score) of {len(core)} ===") for i,x in enumerate(core[:20],1): print(f"{i:2d}|@{x['handle']:16s}|{x['likes']:6d}L|{x['rt']:5d}RT|{x['replies']:4d}R|" f"{x['views']:9d}V|{likerate(x):5.2f}%|{x['date']}|{x['url']}") cut10 = score(core[9]) if len(core)>=10 else 0 cut11 = score(core[10]) if len(core)>10 else 0 print(f"raw core top-10 cutoff: {cut10:.0f} / rank 11: {cut11:.0f}") org = [x for x in core if likerate(x) >= 0.15] org.sort(key=score, reverse=True) print(f"\n=== CORE TOP 12 ORGANIC (like rate >= 0.15%) of {len(org)} ===") for i,x in enumerate(org[:12],1): print(f"{i:2d}|@{x['handle']:16s}|{x['likes']:6d}L|{x['views']:9d}V|{likerate(x):5.2f}%|" f"{x['date']}|{x['url']}") paid = [x for x in core if likerate(x) < 0.15 and x["views"] >= 100000] print(f"paid-reach core posts (>=100k views, <0.15% like rate): {len(paid)}") for x in paid: print(f" @{x['handle']} {x['likes']}L/{x['views']}V {likerate(x):.3f}% {x['date']}") # ---- 24h deltas ---- prevm = {} for x in prev["qualifying"]: i = tid(x["url"]) if i: prevm[i] = x print("\n=== 24H DELTAS (core, present both days) ===") rows=[] for x in core: i=tid(x["url"]); p=prevm.get(i) if not p: continue rows.append((x["likes"]-p["likes"], x["views"]-p["views"], x)) rows.sort(key=lambda r: r[0], reverse=True) for dl,dv,x in rows[:14]: print(f" +{dl:5d}L +{dv:8d}V @{x['handle']:15s} now {x['likes']}L/{x['views']}V " f"{likerate(x):.2f}% {x['date']}") print(" --- negative / flat ---") for dl,dv,x in rows: if dl<=5: print(f" {dl:+6d}L {dv:+9d}V @{x['handle']:15s} now {x['likes']}L/{x['views']}V " f"{likerate(x):.3f}%") # ---- freshness ---- print("\n=== FRESHNESS (hours since newest fetched tweet) ===") fr=[] for h,tw in cur["raw"].items(): ds=[pdate(t.get("createdAt","")) for t in tw] ds=[d for d in ds if d] if ds: fr.append(((now-max(ds)).total_seconds()/3600, h, len(ds))) fr.sort() print(" freshest:", ", ".join(f"@{h} {a:.1f}h" for a,h,n in fr[:12])) print(" stalest: ", ", ".join(f"@{h} {a:.0f}h" for a,h,n in fr[-12:])) print(" key accounts:") for a,h,n in fr: if h in {"higgsfield","higgsfield_ai","runwayml","Kling_ai","pika_labs","EHuanglu","Flovaai", "azed_ai","CharaspowerAI","heyglif","c_valenzuelab","venturetwins","openart_ai", "AIWarper","ItsMaryamAi","ElsaSofia__AI","NVIDIAAI","CuriousRefuge"}: print(f" @{h:16s} n={n:2d} newest {a:6.1f}h ago") # ---- dormancy: originals only, in-window ---- cutoff = now.replace(hour=0,minute=0,second=0,microsecond=0) - timedelta(days=3) print("\n=== DORMANCY (originals only) ===") for h in ["CuriousRefuge","karenxcheng","higgsfield","higgsfield_ai","WuxiaRocks","AI__TSUBAKI", "EinsiaAI","azed_ai","ItsMaryamAi","openart_ai","CuriousRefuge"]: tw = cur["raw"].get(h) or [] orig=[(pdate(t.get("createdAt","")),t) for t in tw if not t.get("retweeted_tweet") and t.get("type")!="retweet" and not (t.get("text") or "").startswith("RT @")] orig=[(d,t) for d,t in orig if d] if not orig: print(f" @{h:16s} NO ORIGINALS IN FETCHED SET"); continue newest=max(d for d,_ in orig) inw=[t for d,t in orig if d>=cutoff] best=max((t.get('likeCount') or 0) for t in inw) if inw else 0 print(f" @{h:16s} newest {(now-newest).total_seconds()/3600:8.1f}h originals={len(orig):2d}" f" in-window={len(inw):2d} best_likes={best:5d}" f"{' <-- DORMANT' if newest