scratch/check_data_volume.py

import os

SENTI_AI_ROOT = r"C:\Users\LENOVO\Desktop\senti_ai"

print("--- CURRENT DATA VOLUME (JSON COUNT) ---")
for folder in os.listdir(SENTI_AI_ROOT):
    if os.path.isdir(folder) and folder.startswith("senti"):
        cleaned_dir = os.path.join(SENTI_AI_ROOT, folder, "cleaned")
        if os.path.exists(cleaned_dir):
            count = len([f for f in os.listdir(cleaned_dir) if f.endswith('.json')])
            print(f"{folder.ljust(20)} : {count} records")