def firefox_ver(self, value):
# 如果格式不符合，保持原来的值，不做任何修改
if not helper.match_expect_type(value, 'dict'):
return
if 'min' in value and helper.match_expect_type(value['min'], 'int'):
self._firefox_ver['min'] = value['min']
# 使用range进行list生成时，会会忽略最大值，所以需要+1
# [range(74, 75)] => [74]
if 'max' in value and helper.match_expect_type(value['max'], 'int'):
self._firefox_ver['max'] = value['max'] + 1
def chrome_max_release_year(self, value):
# 是否为整数
if not helper.match_expect_type(value, 'int'):
return
# 是否大于0
if value < 0:
return
# 是否小于当前年-2008
if CHROME_MAX_RELEASE_YEAR < value:
return
self._chrome_max_release_year = value
def site_max_page_no(self, value):
if not gbh_helper.match_expect_type(value, 'int'):
raise ValueError('site_max_page_no的值必须是整数')
if value < 1 or value > 10:
raise ValueError('site_max_page_no的值必须在1到9之间')
# 实际使用列表表达式生成url，因此site_max_page_no要+1，符合感受
self._site_max_page_no = value + 1
self._site = self._generate_site(
enumset_site=self._raw_site,
enumset_protocol=self._protocol,
int_site_max_page_no=self._site_max_page_no)
def time_interval_in_seconds(self, old_date_time, new_date_time):
'''
计算old_date_time和new_date_time之间时间间隔，单位秒
:param old_date_time:
:param new_date_time:
:return: int
'''
if not helper.match_expect_type(old_date_time, 'datetime.datetime'):
if helper.match_expect_type(old_date_time, 'str'):
old_date_time = datetime.datetime.strptime(
old_date_time, '%Y-%m-%d %H:%M:%S')
else:
raise ValueError('old_date_time的格式不正确')
if not helper.match_expect_type(new_date_time, 'datetime.datetime'):
if helper.match_expect_type(new_date_time, 'str'):
new_date_time = datetime.datetime.strptime(
new_date_time, '%Y-%m-%d %H:%M:%S')
else:
raise ValueError('new_date_time的格式不正确')
# datetime.datetime.now()+datetime.timedelta(days=1)
return int((new_date_time - old_date_time).total_seconds())
def getExtractDataFunction(webenum):
'''
根据enum SupportedWeb的名字，返回对应的prefilter函数，如果没有出错，返回None
:param webenum:
:return: function
'''
if not gbh_helper.match_expect_type(webenum, 'SupportedWeb'):
return
if webenum.name == gfp_self_enum.SupportedWeb.Xici.name:
return gen_proxy_from_page.extra_data_from_page_xicidaili
if webenum.name == gfp_self_enum.SupportedWeb.Kuai.name:
return gen_proxy_from_page.extra_data_from_page_kuaidaili
if webenum.name == gfp_self_enum.SupportedWeb.Proxylist.name:
return gen_proxy_from_page.extra_data_from_page_proxylist
if webenum.name == gfp_self_enum.SupportedWeb.Hidemy.name:
return gen_proxy_from_page.extra_data_from_page_hidemy
return
def getPrefilterFunction(webenum):
'''
根据enum SupportedWeb的名字，返回对应的prefilter函数，如果没有出错，返回None
:param webenum:
:return: function
'''
if not gbh_helper.match_expect_type(webenum, 'SupportedWeb'):
return
if webenum.name == gfp_self_enum.SupportedWeb.Xici.name:
return prefilter.pre_filter_xicidaili
if webenum.name == gfp_self_enum.SupportedWeb.Kuai.name:
return prefilter.pre_filter_kuaidaili
if webenum.name == gfp_self_enum.SupportedWeb.Proxylist.name:
return prefilter.pre_filter_proxy_list
if webenum.name == gfp_self_enum.SupportedWeb.Hidemy.name:
return prefilter.pre_filter_hidemy
return
def valid_time_in_db(self, value):
if not gbh_helper.match_expect_type(value, 'int'):
raise ValueError('valid_time_in_db的值必须是整数')
if value < 300 or value > 86400 * 5:
raise ValueError('valid_time_in_db的值必须在300到86400×5之间')
self._valid_time_in_db = value