python写一个双色球彩票计算器

#!/usr/bin/python
# coding: UTF-8

'''
打开网址
  读取内容
内容解析
  根据源码得到需爬取内容
  1、开奖日期：2018年8月26日
  2、红球
                                <li class="ball_red">03</li>
                                <li class="ball_red">07</li>
                                <li class="ball_red">08</li>
                                <li class="ball_red">14</li>
                                <li class="ball_red">25</li>
                                <li class="ball_red">32</li>
  3、篮球
                                <li class="ball_blue">06</li>
打开数据库连接
  爬取内容写入数据库
    共3个字段
    1、开奖日期
    2、红球，红球使用分号“;”分隔，方便调用和导出
    3、篮球
create table tow_color_ball(open_date varchar(10),red_n varchar(20),blue_n varchar(2))
'''

import urllib
import urllib2
import re
import numpy as np
import operator
import MySQLdb

# 连接mysql
def conn_db():
  db = 'pythondb'
  host = 'localhost'
  iuser = 'xxx'
  passwd = 'xxxxxx'
  conn = MySQLdb.connect(db = db, host = host, user = iuser, passwd = passwd)
  cursor = conn.cursor()
  return cursor

# 处理网页获取页面源码
def get_html_values(url):
  url_open = urllib.urlopen(url)
  url_read = url_open.read()
  return url_read

# 处理源码，获取日期、红球、篮球
def manage_html(html_values):
  red_no_re = re.compile('(?<=\<li class\=\"ball_red\"\>)[0-9]+(?=\<\/li\>)')
  blue_no_re = re.compile('(?<=\<li class\=\"ball_blue\"\>)[0-9]+(?=\<\/li\>)')
  date_re = re.compile('(?<=开奖日期：)[0-9]+年[0-9]+月[0-9]+日')
  red_no_list = re.findall(red_no_re,html_values)
  red_numbers = ';'.join(red_no_list)
  blue_number = re.search(blue_no_re,html_values)
  blue_number = blue_number.group()
  date_value = re.search(date_re,html_values)
  date_value = date_value.group()
  return date_value, red_numbers, blue_number

# 可恶的日期，竟然是YYYY年MM月DD日，需要改成YYYY-MM-DD
def manage_date(date_value):
  date_value = date_value.replace('年','-').replace('月','-').replace('日','')
  return date_value

# 处理页面编号，每次编号-1，也就是说end_page要小于url中的页码
def get_page(url,end_page):
  url_num = re.search('(?<=\/)[0-9]+(?=\.)',url)
  url_num = url_num.group()
  if int(end_page) > int(url_num):
    return 'end'
  url_num_1 = int(url_num) - 1
  url = url.replace(url_num,str(url_num_1))
  return url

# 查看库中是否已存在开奖日期，防止重复写入
def check_open_date(open_date):
  conn = conn_db()
  check_sql = 'select 1 from tow_color_ball where open_date = %r' %open_date
  conn.execute(check_sql)
  excur = conn.fetchall()
  conn.close()
  #如过未查到excur == ()
  return excur

# 写入数据库
def write_db(date_value, red_numbers, blue_number):
  conn = conn_db()
  in_sql = "insert into tow_color_ball(open_date,red_n,blue_n) values(%r,%r,%r)" %(date_value, red_numbers, blue_number)
  conn.execute(in_sql)
  conn.execute('commit')
  conn.close()

# 彩票主程序，用来爬取彩票号码
def ball_main(url,end_page):
  while True:
    html_values = get_html_values(url)
    date_value, red_numbers, blue_number = manage_html(html_values)
    date_value = manage_date(date_value)
    data_check = check_open_date(date_value)
    if data_check == ():
      write_db(date_value, red_numbers, blue_number)
    url = get_page(url,end_page)
    if url == 'end':
      print 'url_page已到达end_page,获取完成'
      return 0

# 二项分布算法
class binomial_class(object):
  def __init__(self,case_count,real_count,p):
    self.case_count = case_count
    self.real_count = real_count
    self.p = p

  def multiply_fun(self,xlist):
    n = 1
    for x in xlist:
      n *= x
    return n
 
  def fact_fun(self,n):
    if n == 0:
      return 1
    n += 1
    fact_list = [i for i in range(1,n)]
    fact_num = self.multiply_fun(fact_list)
    return fact_num

  def c_n_x(self):
    fact_n = self.fact_fun(self.case_count)
    fact_x = self.fact_fun(self.real_count)
    fact_n_x = self.fact_fun(self.case_count - self.real_count)
    c_n_x_num = float(fact_n) / (fact_x * fact_n_x)
    return c_n_x_num

  def binomial_fun(self):
    c_n_k_num = self.c_n_x()
    pi = (self.p ** self.real_count) * ((1 - self.p) ** (self.case_count - self.real_count))
    binomial_num = c_n_k_num * pi
    return binomial_num

# 从库里获取彩票信息
def get_ball_infomation(start_dt,end_dt):
  conn = conn_db()
  sql = "select red_n,blue_n from tow_color_ball where date_format(open_date,'%%Y-%%m-%%d') >= %r and date_format(open_date,'%%Y-%%m-%%d') <= %r" %(start_dt,end_dt)
  conn.execute(sql)
  excur = conn.fetchall()
  conn.close()
  case_array = np.array(excur)
  row_count = case_array.shape[0]
  col_count = case_array.shape[1]
  red_ball_array = case_array[:,0]
  blue_ball_array = case_array[:,1]
  return red_ball_array,blue_ball_array,row_count,col_count

# 统计每个号码球的出现次数，这个应该在数据库里做，先放这，以后改
def every_ball_count(ball_array):
  ball_list = []
  for ball_char in ball_array:
    ball_list += ball_char.split(';')
  ball_count = {}
  for ball_num in ball_list:
    if ball_num in ball_count:
      ball_count[ball_num] += 1
    else:
      ball_count[ball_num] = ball_count.get(ball_num,0) + 1
  return ball_count

# 数据分析主函数，样本量必须大于等于7，否则不进行处理
def analysis_main(start_dt,end_dt):
  red_ball_array,blue_ball_array,row_count,col_count = get_ball_infomation(start_dt,end_dt)
  if row_count < 7:
    print '样本量不足以支持分析'
    return 1
  red_count_dict = every_ball_count(red_ball_array)
  blue_count_dict = every_ball_count(blue_ball_array)
  for red_case in red_count_dict:
    red_rate = binomial_class((red_count_dict[red_case] + 1),(row_count + 1),0.5)
    red_count_dict[red_case] = red_rate.binomial_fun()
  for blue_case in blue_count_dict:
    blue_rate = binomial_class((blue_count_dict[blue_case] + 1),(row_count + 1),0.5)
    blue_count_dict[blue_case] = blue_rate.binomial_fun()
  sorted_red_count = sorted(red_count_dict.iteritems(),key=operator.itemgetter(1),reverse=True)
  sorted_blue_count = sorted(blue_count_dict.iteritems(),key=operator.itemgetter(1),reverse=True)
  print sorted_blue_count[0]
  print '选择红球是:'
  n = 1
  for key,value in sorted_red_count:
    if n == 7:
      break
    print '%s,%s' %(key,str(value))
    n += 1
  print '选择蓝球是'
  print '%s,%s' %(sorted_blue_count[0][0],str(sorted_blue_count[0][1]))

if __name__ == '__main__':
  n = ''
  while n != '1' or n != '2':
    input_n = raw_input('''
请选择需要进行的功能
1、爬取页面的球号
2、进行球号分析
输入quit退出
请选择: ''')
    if input_n == '1':
      url = raw_input('''
请输入需要爬取的地址(此为开始地址,因此建议选择页码较大的地址)
输入: ''')
      end_page = raw_input('''
输入结束页码
(注意:如果结束页码大于输入地址的页码,则不会爬取任何页面)
输入: ''')
      ball_main(url,end_page)
    elif input_n == '2':
      analysis_main('2018-08-15','2018-09-09')
    elif input_n == 'quit':
      exit(0)