#!/usr/bin/env python3 """Generate the plain-text X Account Monitor report from merged scan results.""" import json import os from datetime import datetime, timezone # Config CHUNK1 = "/home/kiro/trend-engine/trends/x/account-monitor/scan_results_2026-06-16_CHUNK1.json" CHUNK2 = "/home/kiro/trend-engine/trends/x/account-monitor/scan_results_2026-06-16_CHUNK2.json" OUTPUT_DIR = "/home/kiro/trend-engine/trends/x/account-monitor" DATE_STR = "2026-06-16" # IRRELEVANT HANDLES (filter from Top 10, show [FILTERED] in breakdown) IRRELEVANT = { "CINEMA505", # Brazilian cinema account (Portuguese), non-AI "TheAIColony", # General AI news aggregator "GoogleLabs", # Large corporate, general Google products "GoogleAI", # Large corporate, Gemini launches "Ai_Vaidehi", # RT aggregator, 0 original posts "AI_with_jasmin", # RT aggregator, remote job lists "CharaspowerAI", # RT-heavy general AI "AIBuzzNews", # General AI news, not filmmaking "AlexStudio44", # No qualifying tweets, unclear niche "FaithJaiye2187", # 0 qualifying tweets consistently "KammelaBala", # 0 qualifying tweets consistently "MyStageAI", # 0 qualifying tweets consistently "ChrisLaubAI", # 100% RTs of non-filmmaking (finance/news) "AIGuideHQ", # 0 qualifying tweets consistently } # These accounts are actually relevant AI filmmaking despite being auto-discovered # (kept for comparison — only filter the IRRELEVANT set above) BIO_INFO = { "runwayml": "Runway AI — professional AI filmmaking (~260K)", "CuriousRefuge": "Curious Refuge — home for AI filmmakers", "nofilmschool": "No Film School — filmmaking news (~241.7K)", "venturetwins": "Justine Moore — a16z partner AI video (~149.5K)", "Kling_ai": "Kling AI — AI video generation (~130K)", "paultrillo": "Paul Trillo — LA director/artist (~15K)", "c_valenzuelab": "Cristóbal Valenzuela — Runway CEO (~42.7K)", "CorridorDigital": "Corridor Digital — VFX/filmmaking (15M+ YouTube)", "karenxcheng": "Karen X. Cheng — Director, brand AI video (~58K)", "higgsfield": "Higgsfield AI — multi-model platform (~63K)", "pika_labs": "Pika — AI video for social creators (~143K)", "EHuanglu": "el.cine / Elsa Huang — Film director, AI consultant (~128K)", "azed_ai": "Amira Zairi — Adobe & Leonardo AI ambassador (~37K)", "ChrisLaubAI": "RT-only account — FILTERED, zero AI filmmaking content", "AIFrontliner": "AI news aggregator — Seedance, Claude, AI filmmaking mentions", "AIWarper": "AI creator — Seedance 2.0 prompts, ComfyUI workflows", "AITECHio": "AI/crypto hybrid — ACN token, AI infrastructure", "DonutAI": "AI tool ecosystem — small but on-topic", "JeffSynthesized": "Jeff Synthesized — NeoCinema AI Studio, Runway/Kling CPP (~8.5K)", "ai_filmmaker": "Dmitry Alekseev — AI filmmaker, Curious Refuge featured", "visiblemaker": "Christian Fleischer — AI cinema director, The Great Catspy (50M+ views)", "Magiermogul": "Particle Panic — AI filmmaker, $1M AI Film Award CPP", "WuxiaRocks": "WuxIA Rocks — AI filmmaking, character consistency (~6.5K)", "JiamingSuen": "Jiaming Sun — AI/VFX creator, CorridorDigital collaborator (~900)", "moonvalley": "Moonvalley — clean-trained AI film model, Paul Trillo partnership", "higgsfield_ai": "Higgsfield AI — ops account, Kling integrations, prompt drops", "Chandan_Perla": "Chandan Perla — India AI Film Festival organizer", "AITOPIAai": "AI TOPIA — Grok Imagine Video integrations, tool content", "AI_Video_Dumps": "AI Video Dumps — AI filmmaking clips and demos", "cladlabs": "Clad Labs — AI/tech (small account)", "4aibsc": "4AI BSC — AI content (auto-discovered)", } def format_snippet(text, max_len=65): """Truncate tweet text for display.""" # Remove newlines text = text.replace("\n", " ").replace("\r", " ").strip() if len(text) <= max_len: return text return text[:max_len-3] + "..." def main(): # Merge chunks all_accounts = [] for chunk_path in [CHUNK1, CHUNK2]: with open(chunk_path) as f: data = json.load(f) all_accounts.extend(data["accounts"]) # Build best-tweet-per-account for Top 10 account_best = {} # handle -> best entry for acc in all_accounts: handle = acc["handle"] tweets = acc.get("tweets", []) for t in tweets: tweet_id = t.get("id", "") url = f"https://x.com/{handle}/status/{tweet_id}" if tweet_id else "" entry = { "handle": handle, "tweet": t, "text": t.get("text", ""), "score": t.get("score", 0), "url": url, "date": t.get("date", ""), "views": t.get("views", 0), "likes": t.get("likes", 0), "retweets": t.get("retweets", 0), "replies": t.get("replies", 0), } if handle not in account_best or entry["score"] > account_best[handle]["score"]: account_best[handle] = entry # Build flat entries list for full breakdown entries = [] for acc in all_accounts: handle = acc["handle"] tweets = acc.get("tweets", []) for t in tweets: tweet_id = t.get("id", "") url = f"https://x.com/{handle}/status/{tweet_id}" if tweet_id else "" entries.append({ "handle": handle, "tweet": t, "text": t.get("text", ""), "score": t.get("score", 0), "url": url, "date": t.get("date", ""), "views": t.get("views", 0), "likes": t.get("likes", 0), "retweets": t.get("retweets", 0), "replies": t.get("replies", 0), }) # Top 10: best per account, sorted by score top10_candidates = [e for e in account_best.values() if e["handle"] not in IRRELEVANT] top10_candidates.sort(key=lambda x: x["score"], reverse=True) # Build lines lines = [] lines.append(f"Title: X Account Monitor -- {DATE_STR}") lines.append("") lines.append("Top 10 by Engagement Score") lines.append("") lines.append("#, Account, Best Tweet, Views, Date") rank = 0 for entry in top10_candidates[:10]: rank += 1 handle = entry["handle"] snippet = format_snippet(entry["text"]) line = f"{rank}, @{handle}, {snippet} ({entry['url']}), {entry['views']}, {entry['date']}" lines.append(line) lines.append("") lines.append("---") lines.append("") lines.append("Full Account Breakdown") lines.append("") # Group by account acc_map = {} for entry in entries: h = entry["handle"] if h not in acc_map: acc_map[h] = [] acc_map[h].append(entry) total_scans = len(all_accounts) accounts_with_content = sum(1 for acc in all_accounts if acc.get("qualifying_count", 0) > 0) total_qualifying = sum(acc.get("qualifying_count", 0) for acc in all_accounts) # Sort accounts by their best tweet score account_order = sorted(acc_map.keys(), key=lambda h: max(e["score"] for e in acc_map[h]), reverse=True) for handle in account_order: entries = acc_map[handle] best = entries[0] bio = BIO_INFO.get(handle, "") is_irrelevant = handle in IRRELEVANT tag = " [FILTERED]" if is_irrelevant else "" qual_count = len(entries) lines.append(f"@{handle}{tag} — {qual_count} qualifying tweets") if bio: lines.append(f" {bio}") for i, e in enumerate(entries[:5]): snippet = format_snippet(e["text"], max_len=80) lines.append(f" [{e['date']}] L{e['likes']} RT{e['retweets']} R{e['replies']} V{e['views']} | {snippet}") lines.append(f" {e['url']}") lines.append("") lines.append("---") lines.append(f"Summary: {total_scans} scans total, {accounts_with_content} accounts with qualifying tweets, {total_qualifying} qualifying tweets found.") # Count filtered filtered_qual = sum( acc.get("qualifying_count", 0) for acc in all_accounts if acc["handle"] in IRRELEVANT ) lines.append(f" {len(IRRELEVANT)} accounts filtered from Top 10 as irrelevant ({filtered_qual} qualifying tweets excluded).") # Errors error_count = 0 for chunk_path in [CHUNK1, CHUNK2]: with open(chunk_path) as f: data = json.load(f) error_count += len(data.get("errors", [])) if error_count: lines.append(f" {error_count} API errors (non-blocking).") report = "\n".join(lines) # Save to disk output_path = os.path.join(OUTPUT_DIR, f"{DATE_STR}_account-monitor.txt") with open(output_path, "w") as f: f.write(report) print(f"Report saved to {output_path}") print(f" Total entries: {len(entries)}") print(f" Top 10 shown: {len(top10_candidates[:10])}") print() print(report) if __name__ == "__main__": main()