momentry_core/scripts/verify_charade_pipeline.py

#!/opt/homebrew/bin/python3.11
"""
Verify Charade pipeline completion.
Usage: python3 scripts/verify_charade_pipeline.py <file_uuid>
"""

import sys
import psycopg2
import subprocess
import json

DATABASE_URL = "postgres://accusys@localhost:5432/momentry"
OUTPUT_DIR = "/Users/accusys/momentry/output_dev"


def check_file_outputs(file_uuid):
    """Check all expected output files exist"""
    expected_files = [
        "cut.json",
        "yolo.json",
        "face.json",
        "face_traced.json",
        "pose.json",
        "asrx.json",
        "visual_chunk.json",
        "scene.json",
        "scene_meta.json",
        "story_llm.json",
        "story_story.json",
        "tmdb.json",
    ]
    
    results = []
    for ext in expected_files:
        path = f"{OUTPUT_DIR}/{file_uuid}.{ext}"
        try:
            size = subprocess.check_output(["stat", "-f%z", path]).decode().strip()
            results.append({"file": ext, "exists": True, "size": int(size)})
        except:
            results.append({"file": ext, "exists": False, "size": 0})
    
    return results


def check_db_records(file_uuid, schema="dev"):
    """Check database records"""
    conn = psycopg2.connect(DATABASE_URL)
    cur = conn.cursor()
    
    checks = [
        ("videos", f"SELECT status FROM {schema}.videos WHERE file_uuid = '{file_uuid}'"),
        ("monitor_jobs", f"SELECT status, completed_processors FROM {schema}.monitor_jobs WHERE uuid = '{file_uuid}'"),
        ("pre_chunks", f"SELECT COUNT(*) FROM {schema}.pre_chunks WHERE file_uuid = '{file_uuid}'"),
        ("face_detections_total", f"SELECT COUNT(*) FROM {schema}.face_detections WHERE file_uuid = '{file_uuid}'"),
        ("face_detections_embedding", f"SELECT COUNT(embedding) FROM {schema}.face_detections WHERE file_uuid = '{file_uuid}'"),
        ("face_detections_trace", f"SELECT COUNT(trace_id) FROM {schema}.face_detections WHERE file_uuid = '{file_uuid}'"),
        ("face_detections_identity", f"SELECT COUNT(identity_id) FROM {schema}.face_detections WHERE file_uuid = '{file_uuid}'"),
        ("chunks_total", f"SELECT COUNT(*) FROM {schema}.chunk WHERE file_uuid = '{file_uuid}'"),
        ("chunks_embedding", f"SELECT COUNT(embedding) FROM {schema}.chunk WHERE file_uuid = '{file_uuid}'"),
    ]
    
    results = []
    for name, query in checks:
        try:
            cur.execute(query)
            result = cur.fetchone()
            results.append({"check": name, "value": result[0] if result else None})
        except Exception as e:
            results.append({"check": name, "error": str(e)})
    
    cur.close()
    conn.close()
    return results


def check_identity_bindings(file_uuid, schema="dev"):
    """Check identity bindings"""
    conn = psycopg2.connect(DATABASE_URL)
    cur = conn.cursor()
    
    checks = [
        ("audrey_faces", f"""
            SELECT COUNT(*) FROM {schema}.face_detections fd 
            JOIN {schema}.identities i ON fd.identity_id = i.id 
            WHERE fd.file_uuid = '{file_uuid}' AND i.name = 'Audrey Hepburn'
        """),
        ("cary_faces", f"""
            SELECT COUNT(*) FROM {schema}.face_detections fd 
            JOIN {schema}.identities i ON fd.identity_id = i.id 
            WHERE fd.file_uuid = '{file_uuid}' AND i.name = 'Cary Grant'
        """),
        ("top_identities", f"""
            SELECT i.name, COUNT(*) as count 
            FROM {schema}.face_detections fd 
            JOIN {schema}.identities i ON fd.identity_id = i.id 
            WHERE fd.file_uuid = '{file_uuid}' AND fd.identity_id IS NOT NULL 
            GROUP BY i.name 
            ORDER BY count DESC 
            LIMIT 10
        """),
    ]
    
    results = []
    for name, query in checks:
        try:
            cur.execute(query)
            if name == "top_identities":
                rows = cur.fetchall()
                results.append({"check": name, "value": rows})
            else:
                result = cur.fetchone()
                results.append({"check": name, "value": result[0] if result else 0})
        except Exception as e:
            results.append({"check": name, "error": str(e)})
    
    cur.close()
    conn.close()
    return results


def print_report(file_uuid, file_outputs, db_records, identity_bindings):
    """Print verification report"""
    print(f"\n{'='*60}")
    print(f"Charade Pipeline Verification Report")
    print(f"File UUID: {file_uuid}")
    print(f"{'='*60}\n")
    
    print("## File Outputs")
    for f in file_outputs:
        status = "✅" if f["exists"] and f["size"] > 0 else "❌"
        size_kb = f["size"] / 1024
        print(f"  {status} {f['file']}: {size_kb:.1f} KB")
    
    print("\n## Database Records")
    for r in db_records:
        value = r.get("value", r.get("error", "N/A"))
        if isinstance(value, (list, tuple)):
            value = ", ".join(str(v) for v in value)
        print(f"  {r['check']}: {value}")
    
    print("\n## Identity Bindings")
    for r in identity_bindings:
        value = r.get("value", r.get("error", "N/A"))
        if isinstance(value, list):
            print(f"  {r['check']}:")
            for row in value:
                print(f"    - {row[0]}: {row[1]} faces")
        else:
            print(f"  {r['check']}: {value}")
    
    print(f"\n{'='*60}\n")


def main():
    if len(sys.argv) < 2:
        file_uuid = "c3c635e3641da80dde10cc555ffcdda5"
    else:
        file_uuid = sys.argv[1]
    
    print("Verifying pipeline...")
    
    file_outputs = check_file_outputs(file_uuid)
    db_records = check_db_records(file_uuid)
    identity_bindings = check_identity_bindings(file_uuid)
    
    print_report(file_uuid, file_outputs, db_records, identity_bindings)


if __name__ == "__main__":
    main()