def goal_detail(url):
content = http_to_char.get_content_by_url(url)
root = etree.HTML(content)
aa = root.xpath('//table[@class="mtable"]//tr')
list=[]
for data in aa:
if len(data.xpath('./td/img[@src="/images/row/1.gif"]'))+len(data.xpath('./td/img[@src="/images/row/2.gif"]'))+len(data.xpath('./td/img[@src="/images/row/3.gif"]')) != 0:
list.append(data)
num_zhu = 0
num_ke = 0
str_goal = ""
for data in list:
if data.xpath('./td/text()')[0] == '\xa0':
num_ke += 1
str_goal = str_goal + str(data.xpath('./td/text()')[2][:-1]) + "分 " + str(num_zhu) + "-" + str(num_ke) + ','
else:
if len(list[0].xpath('./td/text()')) == 3:
num_zhu += 1
str_goal = str_goal + str(data.xpath('./td/text()')[0][:-1]) + "分 " + str(num_zhu) + "-" + str(num_ke) + ','
else:
num_zhu += 1
str_goal = str_goal + str(data.xpath('./td/text()')[1][:-1]) + "分 " + str(num_zhu) + "-" + str(num_ke) + ','
try:
if list[0].xpath('./td/text()')[0] == '\xa0':
return '客队',list[0].xpath('./td/text()')[2][:-1],str_goal
else:
if len(list[0].xpath('./td/text()')) == 3:
return '主队',list[0].xpath('./td/text()')[0][:-1],str_goal
else:
return '主队',list[0].xpath('./td/text()')[1][:-1],str_goal
except IndexError:
return "",-1,""
def get_game_end(url):
content = http_to_char.get_content_by_url(url)
root = etree.HTML(content)
table_game = root.xpath('//tr[@yy]')
result = []
for data_game in table_game:
if data_game.xpath('./td/span[@class="red"]/text()')[0] == '完':
num = data_game.xpath('./@id')[0][1:]
duiwu = data_game.xpath('./@gy')[0]
game = str(duiwu).split(',')
game_name = game[0]
changci = data_game.xpath('./td[3]/text()')[0]
zhu_name = game[1]
ke_name = game[2]
num_zhu = data_game.xpath('./td/div[@class="pk"]/a/text()')[0]
num_ke = data_game.xpath('./td/div[@class="pk"]/a/text()')[2]
str_quan = str(num_zhu)+'-'+str(num_ke)
str_ban = data_game.xpath('./td[9]/text()')[0]
num_goal = int(num_zhu)+int(num_ke)
first_goal_team = None
firt_goal_time = None
goal_decs = None
if int(num_zhu)+int(num_ke) != 0:
first_goal_team,firt_goal_time,goal_decs = goal_detail("http://live.500.com/detail.php?fid="+num+"&r=1")
result.append((num,game_name,changci,zhu_name,ke_name,str_quan,str_ban,num_goal,first_goal_team,firt_goal_time,goal_decs))
return result
def get_game_list(url, page, count):
content = http_to_char.get_content_by_url(url)
root = etree.HTML(content)
table_game = root.xpath('//tr[@yy]')
for data_game in table_game:
if data_game.xpath('./td/span[@class="red"]/text()')[0] == '完':
game_id = str(data_game.xpath('./@id')[0][1:])
num_zhu = int(data_game.xpath('./td/div[@class="pk"]/a/text()')[0])
pankou = str(data_game.xpath('./td/div[@class="pk"]/a/text()')[1])
num_ke = int(data_game.xpath('./td/div[@class="pk"]/a/text()')[2])
# 欧盘输赢情况
out = tongji.tongji_yapan(num_zhu, num_ke, pankou)
result_list = oupei.get_oupei(game_id)
result_list.append(out)
if ((count != 0) & (count % 50 == 0)):
page = page + 1
with open("oupei" + str(page) + ".csv", 'a',
encoding="utf-8") as f:
f.writelines(','.join(str(i) for i in result_list))
f.write('\n')
count = count + 1
return page, count
def get_oupei(game_id):
url = 'http://odds.500.com/fenxi/ouzhi-' + game_id + '.shtml'
content = http_to_char.get_content_by_url(url)
root = etree.HTML(content)
result_list = []
for bet_company in list:
one = root.xpath('//*[@title="' + bet_company + '"]/..')
if len(one) == 0:
get_data_simple(result_list)
else:
get_data(one[0], result_list)
return result_list
def get_game(url):
content = http_to_char.get_content_by_url(url)
root = etree.HTML(content)
table_game = root.xpath('//tr[@yy]')
for data_game in table_game:
if data_game.xpath('./td/span[@class="red"]/text()')[0] == 0:
num = data_game.xpath('./@id')[0][1:]
duiwu = data_game.xpath('./@gy')[0]
game = str(duiwu).split(',')
game_name = game[0]
zhu_name = game[1]
ke_name = game[2]
num_zhu = data_game.xpath('./td/div[@class="pk"]/a/text()')[0]
num_ke = data_game.xpath('./td/div[@class="pk"]/a/text()')[2]
str_ban = data_game.xpath('./td[9]/text()')
num_goal = int(num_zhu) + int(num_ke)
def get_yazhi(url):
content = http_to_char.get_content_by_url(url)
root = etree.HTML(content)
yazhi = []
yazhi_str = ""
born_3 = root.xpath('//*[@id="3"]/td[5]/table/tbody/tr/td[1]/text()')[0]
born_1 = root.xpath('//*[@id="3"]/td[5]/table/tbody/tr/td[2]/text()')[0]
born_0 = root.xpath('//*[@id="3"]/td[5]/table/tbody/tr/td[3]/text()')[0]
yazhi.append([born_3, born_1, born_0])
dead_3 = root.xpath('//*[@id="3"]/td[3]/table/tbody/tr/td[1]/text()')[0]
dead_1 = root.xpath('//*[@id="3"]/td[3]/table/tbody/tr/td[2]/text()')[0]
dead_0 = root.xpath('//*[@id="3"]/td[3]/table/tbody/tr/td[3]/text()')[0]
yazhi.append([dead_3, dead_1, dead_0])
# for data in yazhi:
# yazhi_str += data[0]+","+data[1]+","+data[2]+"|"
# return yazhi_str
return yazhi
def get_ouzhi(url):
content = http_to_char.get_content_by_url(url)
root = etree.HTML(content)
ouzhi_str = ""
ouzhi = []
born_3 = root.xpath('//*[@id="3"]/td[3]/table/tbody/tr[1]/td[1]/text()')[0]
born_1 = root.xpath('//*[@id="3"]/td[3]/table/tbody/tr[1]/td[2]/text()')[0]
born_0 = root.xpath('//*[@id="3"]/td[3]/table/tbody/tr[1]/td[3]/text()')[0]
ouzhi.append([born_3, born_1, born_0])
dead_3 = root.xpath('//*[@id="3"]/td[3]/table/tbody/tr[2]/td[1]/text()')[0]
dead_1 = root.xpath('//*[@id="3"]/td[3]/table/tbody/tr[2]/td[2]/text()')[0]
dead_0 = root.xpath('//*[@id="3"]/td[3]/table/tbody/tr[2]/td[3]/text()')[0]
ouzhi.append([dead_3, dead_1, dead_0])
ceri_born_3 = root.xpath(
'//*[@id="3"]/td[6]/table/tbody/tr[1]/td[1]/text()')[0]
ceri_born_1 = root.xpath(
'//*[@id="3"]/td[6]/table/tbody/tr[1]/td[2]/text()')[0]
ceri_born_0 = root.xpath(
'//*[@id="3"]/td[6]/table/tbody/tr[1]/td[3]/text()')[0]
ouzhi.append([ceri_born_3, ceri_born_1, ceri_born_0])
ceri_dead_3 = root.xpath(
'//*[@id="3"]/td[6]/table/tbody/tr[2]/td[1]/text()')[0]
ceri_dead_1 = root.xpath(
'//*[@id="3"]/td[6]/table/tbody/tr[2]/td[2]/text()')[0]
ceri_dead_0 = root.xpath(
'//*[@id="3"]/td[6]/table/tbody/tr[2]/td[3]/text()')[0]
ouzhi.append([ceri_dead_3, ceri_dead_1, ceri_dead_0])
# for data in ouzhi:
# ouzhi_str += data[0]+","+data[1]+","+data[2]+"|"
#
# return ouzhi_str
return ouzhi