put_call_ratio.py

import pandas as pd
from io import StringIO
import requests
import time
import random

def get_put_call_ratio_df(q_start_date: str, q_end_date:str):
    url = 'https://www.taifex.com.tw/cht/3/pcRatioDown'
    payload = {
        'queryStartDate': q_start_date,
        'queryEndDate': q_end_date,
    }
    res = requests.post(url, data=payload)

    data_file = StringIO(res.text)
    df = pd.read_csv(data_file, index_col=False)
    return df

start_date = '2003-11-20'
end_date = '2023-11-15'
date_range = pd.date_range(start=start_date, end=end_date, freq='30D')

final_pc_ratio_df = pd.DataFrame()
for i in range(1, len(date_range)):
    q_start_date = date_range[i-1].strftime('%Y/%m/%d')
    q_end_date = date_range[i].strftime('%Y/%m/%d')

    pc_ratio_df = get_put_call_ratio_df(q_start_date, q_end_date)
    final_pc_ratio_df = pd.concat([final_pc_ratio_df, pc_ratio_df], axis=0)
    time.sleep(random.uniform(1, 2))
    final_pc_ratio_df.to_csv('put_call_ratio.csv')
    print(f'{q_end_date} done')

final_pc_ratio_df = final_pc_ratio_df.drop_duplicates()
final_pc_ratio_df.columns = ['Date', 'Put Volume', 'Call Volume', 'Put/Call Volume Ratio', 'Put OI', 'Call OI', 'Put/Call OI Ratio']
final_pc_ratio_df.index = pd.to_datetime(final_pc_ratio_df.index)
final_pc_ratio_df.sort_index().to_csv('put_call_ratio.csv')