def updateShowDetail(self,s_id):
#用来仅仅更新一部剧的所有季和集的方法
db = Database(self.log,self.config)
urlTarget = self.config.url+db.getOneLinkBySid(s_id)
cookie = cookielib.CookieJar()
opener = urllib2.build_opener(urllib2.HTTPCookieProcessor(cookie))
req = urllib2.Request(
url = urlTarget
)
htmlData = ""
htmlData = opener.open(req).read()
if htmlData:
bsContent = BeautifulSoup(htmlData)
pinfo = bsContent.find('p',attrs={'class':'sumtext'}).get_text() #要取到的剧的介绍
pinfo = pinfo.replace("'", "\\'")
DivLarge = bsContent.find('aside',attrs={'class':'quikinfo'})
DivSmall = DivLarge.findAll('li')
#处理每周日期
update_time = DivSmall[0].a.get_text()
#处理每集长度
length = DivSmall[1].get_text()
length = length[17:]
#查找地区、电视台
area = DivSmall[3].get_text()
area = area[10:]
channel = DivSmall[2].get_text()
channel = channel[10:]
status = DivSmall[5].get_text()
status = status[14:]
#print status
DetailOfShow = {
's_id' : s_id,
's_description' : pinfo,
'update_time' : update_time,
'length' : length,
'area' : area,
'channel' : channel,
'status' : status
}
print DetailOfShow
def workWithOneShowsEp(self,s_id):
#用来仅仅更新一部剧的所有季和集的方法
db = Database(self.log,self.config)
urlTarget = self.config.url+db.getOneLinkBySid(s_id)
cookie = cookielib.CookieJar()
opener = urllib2.build_opener(urllib2.HTTPCookieProcessor(cookie))
req = urllib2.Request(
url = urlTarget
)
htmlData = ""
htmlData = opener.open(req).read()
if htmlData:
bsContent = BeautifulSoup(htmlData)
bsLists = bsContent.findAll('li',attrs = {'class':'parent'})
print ('the len of biLists is '+str(len(bsLists)))
for oneSeason in bsLists:
#print oneSeason
se_id = oneSeason.strong.get_text()
se_id = re.search('Season\s\d{1,2}',se_id).group()
se_id = se_id[7:]
epList = oneSeason.findAll('li',attrs = {'class':'ep info RAWR'}) + oneSeason.findAll('li',attrs = {'class':'ep info '})
for oneEpisode in epList:
#print oneEpisode
#集数
e_num = oneEpisode.find('span',attrs = {'class':'pnumber'}).get_text()
if e_num[0] == '0' and len(e_num) > 1:
e_num = e_num[1:]
#集名
e_name = oneEpisode.find('a',attrs = {'itemprop':'url'}).get_text()
#e_name = MySQLdb.escape_string(e_name)
#播放时间
time_temp = oneEpisode.find('span',attrs = {'class':'datepub'})
e_time = time_temp['content']
time_temp = time_temp.get_text()
time = time_temp[-7:]
hour = re.search('\d{1,2}:',time).group()
hour = hour[:-1]
hour = string.atoi(hour)
minute = re.search(':\d{2}[a|p]m',time).group()
#注意调整am和pm的时间差，另外需要注意的是这里的时间都是标准UTC时间，天朝使用需要+8
if (minute[-2] == 'p') and (hour != 12):
hour += 12
minute = minute[1:-2]
if len(str(hour)) < 2:
hour = '0' + str(hour)
e_time += ' ' + str(hour) + ':' + minute + ':00'
status_temp = oneEpisode.find('span',attrs = {'class':'paired'})
if status_temp:
e_status = u'已播放'
else:
e_status = u'即将播出'
episodeInfoToBeAired = {
's_id' : s_id,
'se_id' : se_id,
'e_num' : e_num,
'e_name' : e_name,
'e_status' : e_status,
'e_description' : '',
'e_time' : e_time
}
print episodeInfoToBeAired
flag = True
if flag == False:
break
