test_check_missing_rows.py

import pandas as pd
from datetime import timedelta
import os

def check_missing_timestamps(csv_path, interval="15min", log_path="missing_timestamps.log"):
    # Load CSV
    df = pd.read_csv(csv_path, parse_dates=['timestamp'], index_col='timestamp')

    # Ensure sorted index
    df = df.sort_index()

    # Infer interval and generate expected timestamp range
    start = df.index.min()
    end = df.index.max()
    expected = pd.date_range(start=start, end=end, freq=interval)

    # Find missing timestamps
    missing = expected.difference(df.index)

    # Write missing timestamps to log
    if missing.empty:
        print("✅ No missing timestamps found.")
    else:
        with open(log_path, "w") as log:
            for ts in missing:
                log.write(f"{ts.isoformat()}\n")
        print(f"❌ Found {len(missing)} missing timestamps.")
        print(f"📝 Log written to {log_path}")

if __name__ == "__main__":
    # Example usage
    symbol = "ADAUSDT"
    interval = "15min"  # or '1h', '1d', etc.
    filename = f"datasets/raw/full_scope/15m/{symbol}_15m_historical_data.csv"

    check_missing_timestamps(csv_path=filename, interval=interval)