python爬取连续一字板股票及当时日期数据【原创分享】

import tushare as ts
import pandas as pd
import time


# 筛选一字板的策略
def gp_rules(code):
    # 获取某只股票的历史数据
    data = ts.get_hist_data(code, start='2018-01-01', end='2019-07-26')
    # 排除该时间段无数据的股票
    if data is None:
        return

    # 筛选表头【开盘价，收盘价, 涨幅】
    data_2 = data.iloc[:, [0, 2, 6]]
    # 筛选出涨幅大于9%且开盘价等于收盘价的股票，即一字涨停的股票（有待优化~）
    data_3 = data_2[(data_2['p_change'] > 9) & (data_2['open'] == data_2['close'])]
    # 去除不存在一字板的股票（即空数据)
    if data_3.index is None:
        return
    line = len(data_3)
    f, s = 1, 0
    # 遍历检查是否连续的一字板
    for n in range(0, line):

        # 检查是否连续的一字板
        if int(data_3['open'][n] * (data_3['p_change'][n]/100 + 1)) == int(data_3['open'][n-1]):
            f = f + 1
            # 若4连或4连以上一字板则打印代码及当时日期
            if f >= 5 and s == 0:
                print(code, data_3.index[n-4])
                s = 1
        else:
            f = 1


# 筛选股票代码
def get_code():
　　 # 自己获取
    TOKEN = 'xxxx'
    pro = ts.pro_api(token=TOKEN)
    # 筛选代码，上市日期
    get_codedata = pro.stock_basic(exchange='', list_status='L', fields='symbol, list_date')
    # 将list_date改为日期格式并转为索引
    get_codedata['list_date'] = pd.to_datetime(get_codedata['list_date'])
    get_codedata = get_codedata.set_index('list_date')
    # 筛选2018-01-01前的数据，取code，转为list
    symbol_list = get_codedata[:'2018-01-01']['symbol'].tolist()
    return symbol_list


# 测试运行时间
start = time.clock()


code_list = get_code()
print("正在爬取符合策略的股票，请耐心等待......")

for code in list(code_list):
    time.sleep(1.4)
    gp_rules(code)


print("爬取筛选完成！")
end = time.clock()

print('运行时间：%s Seconds' % (end-start))