用Python从sina下载复权因子-经管之家官网！

提供复权信息的网站不多
用Python从sina下载复权因子，稍加修改就可以下载其他复权信息
import os
import urllib2
import re
#找到特定str所在位置
def find_str(str_list, str):
idx = -1
row = 1
for l in str_list:
idx = l.find(str)
if idx != -1:
break
row += 1
return row
#第二类解析
def deco_url2(url_list, str):
s='\xc0\xfa\xca\xb7\xbd\xbb\xd2\xd7begin-->'
e='\xc0\xfa\xca\xb7\xbd\xbb\xd2\xd7end-->'
s_idx = find_str(url_list, s)
e_idx = find_str(url_list, e)
factor=[]
if e_idx - s_idx == 966-899:
return factor
else:
x='\xbc\xbe\xb6\xc8\xb8\xb4\xc8\xa8\xc0\xfa\xca\xb7\xbd\xbb\xd2\xd7'
x_idx=find_str(url_list,x)
time_row=x_idx+16
factor_row=x_idx+23
delta = 11
while factor_row <= e_idx-7:
t = url_list[time_row-1][6:16]
#print t, url_list[factor_row-1]
f_re = re.search('\d+(\.\d*)?',url_list[factor_row-1])
temp = t+','+f_re.group()+','+str
factor.append(temp.strip())
time_row += delta
factor_row += delta
return factor
#第一类解析
def deco_url1(url_list, str):
s='\xc0\xfa\xca\xb7\xbd\xbb\xd2\xd7begin-->'
e='\xc0\xfa\xca\xb7\xbd\xbb\xd2\xd7end-->'
s_idx = find_str(url_list, s)
e_idx = find_str(url_list, e)
factor=[]
if e_idx - s_idx == 966-899:
return factor
else:
x='\xbc\xbe\xb6\xc8\xb8\xb4\xc8\xa8\xc0\xfa\xca\xb7\xbd\xbb\xd2\xd7'
x_idx=find_str(url_list,x)
time_row=x_idx+18
factor_row=x_idx+26
delta = 14
while factor_row <= e_idx-7:
if delta == 14:
t = url_list[time_row-1][5:15]
if delta == 11:
t = url_list[time_row-1][6:16]
#print t, url_list[factor_row-1]
f_re = re.search('\d+(\.\d*)?',url_list[factor_row-1])
if f_re == None:
time_row -= 2
factor_row -= 3
delta = 11
ti = url_list[time_row-1].find('2')
t = url_list[time_row-1][ti:(10+ti)]
#print url_list[factor_row-1]
f_re = re.search('\d+(\.\d*)?',url_list[factor_row-1])
temp = t+','+f_re.group()+','+str
factor.append(temp.strip())
time_row += delta
factor_row += delta
return factor
#得到股票代码
fopen = open('python/symbol.txt','r')
symbol=fopen.readlines()
fopen.close()
#得到年份
fopen = open('python/year.txt','r')
year=fopen.readlines()
fopen.close()
#一年四个季度
jidu=['4','3','2','1']
l1="http://vip.stock.finance.sina.com.cn/corp/go.php/vMS_FuQuanMarketHistory/stockid/"
l2=".phtml?year="
l3="&jidu="
for sym in symbol[0:1571]:
for y in year[0:24]:
for jd in jidu:
ls=len(sym)
ly=len(y)
link=l1+sym[0:(ls-1)]+l2+y[0:(ly-1)]+l3+jd
#print step, link
url_test = urllib2.urlopen(link)
url_list=url_test.readlines()
time = y[0:(ly-1)]+jd
print time, sym,
if time >= '20062':
s = deco_url1(url_list,sym)
else:
s = deco_url2(url_list,sym)
f='python/factor/'+sym[0:(ls-1)]+'.txt'
fobj=open(f,'a')
for ss in s:
fobj.write(ss+'\n')
fobj.close()