def process(self, params):
""""""
#过滤掉不在范围内的网站或url
patterns = [
'^http[s]{0,1}://(news|fashion|women|mil|health|cul|travel|history|learning|book|star.news|sports|(music\.)?yule|baobao|chihe|it|business|mgame)\.sohu\.com/.*',
'^http[s]{0,1}://pic\.\w+\.sohu\.com/.*',
'^http[s]{0,1}://gongyi\.sohu\.com/.*',
'^http[s]{0,1}://pic\.book\.sohu\.com/.*',
'^http[s]{0,1}://tv\.sohu\.com/.*',
'^http[s]{0,1}://my\.tv\.sohu\.com/.*',
'^http[s]{0,1}://www\.sohu\.com.*',
'^http[s]{0,1}://p\.weather\.com\.cn.*'
]
flag = False
for pattern in patterns:
if self.r.search(pattern, params.originalurl):
flag = True
break
if not flag:
Logger.log(params.originalurl, constant.ERRORCODE_SITE_NOGET_SITE)
return
if self.r.search('https?://www\.sohu\.com/a/\d+_\d+',
params.originalurl):
self.process_new_a(params)
elif self.r.search(
'^http[s]{0,1}://(www|news)\.sohu\.com.*|^http[s]{0,1}://p\.weather\.com\.cn.*',
params.originalurl):
ChangyanComments(self).process(params)
else:
self.process_video(params)
def process(self, params):
if self.r.search('http://sports\.le\.com/.*', params.originalurl):
ChangyanComments(self).process(params)
else:
self.process_ptv(params)
if params.step == None:
self.getclick(params)
elif params.step == self.STEP_PALY:
self.setclick(params)
class GfanComments(SiteComments):
STEP_1 = None
##############################################################################################
# @functions：__init__
# @param： none
# @return：none
# @author：Hedian
# @date：2016/12/15
# @note：GfanComments类的构造器，初始化内部变量
##############################################################################################
def __init__(self):
SiteComments.__init__(self)
self.bbs = None
self.news = None
def createobject(self):
if self.bbs is None:
self.bbs = CommenComments(self)
if self.news is None:
self.news = ChangyanComments(self)
##############################################################################################
# @functions：process
# @params：共通模块传入的参数（对象url, 原始url, 当前step数，自定义参数）
# @return：
# @author：Hedian
# @date：2016/12/15
# @note：
##############################################################################################
def process(self, params):
Logger.getlogging().info(params.originalurl)
self.createobject()
field = self.r.parse('^http://(\w+)\.?', params.originalurl)[0]
Logger.getlogging().debug(field)
if field == 'bbs':
# 机锋网论坛
self.bbs.process(params)
else:
# 机锋网其他处理
self.news.process(params)
def process(self, params):
Logger.getlogging().info(params.url)
field = self.r.parse('^http://(\w+)\.18183\.com*',
params.originalurl)[0]
if field == 'bbs':
CommenComments.getinstance(self).process(params)
else:
ChangyanComments(self).getcomments(params, '', 3, 2)
#重新设置部分网页的putime
if field == 'chanye':
self.setpubtime(params)
def process(self, params):
# 初始化内部子类对象
self.createobject()
# 论坛评论取得
if self.r.match('http://bbs\.ptbus\.com/.*', params.originalurl):
self.ptbusBbs.process(params)
# 新闻评论取得
elif self.r.match('http://.+\.ptbus\.com/.*', params.originalurl):
# self.ptbusNews.process(params)
# 非bbs页面评论改为畅言模式，调用畅言模块
ChangyanComments(self).process(params)
def __init__(self):
WebSite.__init__(self)
self.pattern = r'^http://\w+\.donews\.com.*'
self.setcommentimpl(ChangyanComments())
def __init__(self):
WebSite.__init__(self)
self.pattern = r'^http://www\.52tian\.net/.*'
self.setcommentimpl(ChangyanComments())
self.sets2queryimpl(tian52S2Query())
def __init__(self):
WebSite.__init__(self)
self.name = 'kumi'
self.pattern = r'^http://.*\.kumi\.cn/.*'
self.setcommentimpl(ChangyanComments())
self.sets2queryimpl(kumiS2Query())
def __init__(self):
WebSite.__init__(self)
self.name = 'sootoo'
self.pattern = '^http://www\.sootoo\.com/.*'
self.setcommentimpl(ChangyanComments())
def __init__(self):
WebSite.__init__(self)
self.pattern = r'^http://\w+\.edu-gov\.cn.*'
self.setcommentimpl(ChangyanComments())
def createobject(self):
if self.bbs is None:
self.bbs = CommenComments(self)
if self.news is None:
self.news = ChangyanComments(self)
def createobject(self):
if self.changyan is None:
self.changyan = ChangyanComments(self)
return self.changyan
def __init__(self):
WebSite.__init__(self)
self.name = 'qudong'
self.pattern = r'^http[s]{0,1}://.*\.qudong\.com'
self.setcommentimpl(ChangyanComments())