python爬取新浪股票数据—绘图【原创分享】

  1 """
  2     新浪财经，爬取历史股票数据
  3 """
  4 
  5 # -*- coding:utf-8 -*-
  6 
  7 import numpy as np
  8 import urllib.request, lxml.html
  9 from urllib.request import urlopen
 10 from bs4 import BeautifulSoup
 11 import re, time
 12 import matplotlib.pyplot as plt
 13 from datetime import datetime
 14 # 绘图显示中文设置
 15 plt.rcParams['font.sans-serif'] = ['SimHei']
 16 plt.rcParams['axes.unicode_minus'] = False
 17 
 18 
 19 # 公共模块，请求头信息
 20 def public(link):
 21     r = urllib.request.Request(link)
 22 
 23     ug = 'Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/49.0.2623.221 Safari/537.36 SE 2.X MetaSr 1.0'
 24 
 25     r.add_header('User-Agent', ug)
 26 
 27     cookie = "SUB=_2AkMsqZjif8NxqwJRmfkRxG7nZYpzyg_EieKa9Wk5JRMyHRl-yD83qkJatRB6Bym2DDqPE870e3uMsySIjHjrMbMNxNqk; " \
 28              "SUBP=0033WrSXqPxfM72-Ws9jqgMF55529P9D9WFXmxLGpAG5k05lCJw6qgYe; " \
 29              "SINAGLOBAL=172.16.92.24_1542789082.401113; " \
 30              "Apache=172.16.92.24_1542789082.401115; UOR=www.baidu.com,blog.sina.com.cn,; " \
 31              "ULV=1542789814434:1:1:1:172.16.92.24_1542789082.401115:; U_TRS1=000000d1.1f4d3546.5bf53673.955fa32e; " \
 32              "U_TRS2=000000d1.1f593546.5bf53673.736853cc; FINANCE2=661413ac85cadaab72ec7e3d842d6a3a; _s_upa=1"
 33 
 34     r.add_header("Cookie", cookie)
 35 
 36     html = urllib.request.urlopen(r, timeout=500).read()
 37 
 38     bsObj = BeautifulSoup(html, "lxml")  # 将html对象转化为BeautifulSoup对象
 39 
 40     return bsObj
 41 
 42 
 43 # 获取股票价格
 44 def shares_price(code, year, quarter):
 45     link = "http://money.finance.sina.com.cn/corp/go.php/vMS_MarketHistory/stockid/%s.phtml?year=%d&jidu=%d" % (code, year, quarter)
 46 
 47     bsObj = public(link)
 48     # print(bsObj)
 49 
 50     a = 0
 51     # date_list为日期列表，open_list为开盘价列表，high_list为最高价列表，close_list为收盘价列表，low_list为最低价列表
 52     price_list, date_list, open_list, high_list, close_list, low_list = [], [], [], [], [], []
 53     # 获取股票信息
 54     jpg_title = re.findall("(.*?\))", bsObj.title.text)
 55 
 56     prices_bs = bsObj.find_all(name='div', attrs={"align": 'center'})
 57     # 获取并处理价格信息
 58     for price_bs in prices_bs:
 59         # 去除空格
 60         price_bs_1 = price_bs.text.replace("\n\r\n\t\t\t", "")
 61         price_bs_2 = price_bs_1.replace("\t\t\t\n", "")
 62 
 63         # 6个字符串为一个列表
 64         if a != 6:
 65             price_list.append(price_bs_2)
 66             a = a + 1
 67         else:
 68             date_list.append(price_list[0])
 69             open_list.append(price_list[1])
 70             high_list.append(price_list[2])
 71             close_list.append(price_list[3])
 72             low_list.append(price_list[4])
 73             a = 0
 74             price_list = []
 75     # 删除列表头
 76     for b in (date_list, open_list, high_list, close_list, low_list):
 77         b.pop(0)
 78 
 79     # 全部倒序排列（由日期远到近，从左到右排列）
 80     for c in (date_list, open_list, high_list, close_list, low_list):
 81         c.reverse()
 82 
 83     return date_list, open_list, high_list, close_list, low_list, jpg_title
 84 
 85 
 86 # 输入股票代码，年份，季度
 87 code = "002925"
 88 year = "2018"
 89 quarter = 4
 90 # 以下为手动输入模式，因调试方便默认上面固定模式。
 91 # code = input("code：")  # 002925
 92 # year = input("year：")    # 2018
 93 # quarter = int(input("quarter："))
 94 
 95 # 列表字符串转为数值date
 96 x = [datetime.strptime(d, '%Y-%m-%d').date() for d in shares_price(code, int(year), quarter)[0]]
 97 # 将爬取的数据（字符串）转化为浮点型
 98 open_list = [float(i) for i in shares_price(code, int(year), quarter)[1]]
 99 high_list = [float(i) for i in shares_price(code, int(year), quarter)[2]]
100 close_list = [float(i) for i in shares_price(code, int(year), quarter)[3]]
101 low_list = [float(i) for i in shares_price(code, int(year), quarter)[4]]
102 
103 # 线条设置
104 plt.plot(x, open_list, label='open', linewidth=1, color='red', marker='o', markerfacecolor='blue', markersize=2)
105 plt.plot(x, high_list, label='high', linewidth=1, color='green', marker='o', markerfacecolor='blue', markersize=2)
106 plt.plot(x, close_list, label='close', linewidth=1, color='blue', marker='o', markerfacecolor='blue', markersize=2)
107 plt.plot(x, low_list, label='low', linewidth=1, color='black', marker='o', markerfacecolor='blue', markersize=2)
108 
109 # 取数列最大数值与最小值做图表的边界值。
110 plt.ylim(min(low_list)-1, max(high_list)+1)
111 plt.gcf().autofmt_xdate()  # 自动旋转日期标记
112 
113 # 打印表头
114 plt.xlabel('time')
115 plt.ylabel('price')
116 # shares_price(code, int(year), quarter)[5][0]为title中的股票名称与代码
117 plt.title('gp_1_{0}.jpg'.format(shares_price(code, int(year), quarter)[5][0]))
118 plt.legend()
119 plt.show()