filter_csv.py

import pandas as pd

# Read the CSV file
df = pd.read_csv('gt1_8kElo_all.zip')

# Filter the DataFrame based on the conditions
filtered_df = df[(df['Result'] == '1-0') & 
                 (df['WhiteElo'] > 1900) & 
                 (df['WhiteElo'] < 2300) & 
                 (df['BlackElo'] < 2600)]

# Select only the 'transcript' column
transcript_df = filtered_df[['transcript']]

# Save the filtered 'transcript' column to a new CSV file
transcript_df.to_csv('NEW_lichess_filtered.csv', index=False)