import pandas as pd


old_df = pd.read_csv("public_new_increase_in_cps_research_base_export_2025-06-24_172044.csv",low_memory=False)
new_df = pd.read_csv("new_increase_in_cps_research_base_export_2025-06-25_123717.csv",low_memory=False)


KEY_COLUMNS = ['increase_in_cps_research_base_id']  

old_df = old_df.set_index(KEY_COLUMNS).sort_index()
new_df = new_df.set_index(KEY_COLUMNS).sort_index()


new_only = new_df[~new_df.index.isin(old_df.index)]     
common = new_df[new_df.index.isin(old_df.index)]        


modified = common[common.ne(old_df).any(axis=1)]

# 💾 Save results
new_only.to_csv("new_rows.csv")
modified.to_csv("modified_rows.csv")

print("✅ Comparison complete.")
print(f"🆕 New rows: {len(new_only)}")
print(f"✏️ Modified rows: {len(modified)}")