def deal_single_info(self, i, single, cookies, item):
# 文献标识
nrdAn = single.xpath(".//input[@name = 'nrdAnHidden']")[0].xpath(
'./@value')[0]
# 文献唯一标识
cid = single.xpath(".//input[@name='idHidden']")[0].xpath(
'./@value')[0]
sid = cid
nrdPn = single.xpath(".//input[@name ='nrdPnHidden']")[0].xpath(
'./@value')[0]
str = u'代理机构'
agency = self.deal_info(str, single)
str = u'代理人'
agent = self.deal_info(str, single)
str = u'申请号'
code = self.deal_info(str, single)
code = code.split('CN')[-1]
str = u'申请日'
app_date = self.deal_info(str, single)
app_date = deal_html_code.change_date(app_date)
str = u'申请（专利权）人'
applicant = self.deal_info(str, single)
address = single.xpath(".//input[@name ='appAddrHidden']")[0].xpath(
'./@value')[0]
str = u'发明人'
inventor = self.deal_info(str, single)
str = u'IPC分类号'
main_cate = self.deal_info(str, single)
str = u'IPC分类号'
sub_cate = self.deal_info(str, single)
str = u'公开（公告）号'
pub_code = self.deal_info(str, single)
str = u'公开（公告）日'
pub_date = self.deal_info(str, single)
pub_date = deal_html_code.change_date(pub_date)
str = u'优先权日'
priority_date = self.deal_info(str, single)
str = u'优先权号'
priority_code = self.deal_info(str, single)
priority = priority_date + ' ' + priority_code
name = single.xpath(".//input[@name ='titleHidden']")[0].xpath(
'./@value')[0]
name = deal_html_code.remove_symbol(name)
remark = self.get_remark(nrdPn, sid, cid, cookies)
source = 'pss-system'
law_search_info = self.get_law_info(nrdAn, nrdPn, cookies)
string = u'同族'
finger = single.xpath(".//a[contains(.,'%s')]" %
string)[0].xpath("string(.)")
finger = finger.split("：")[-1]
if int(finger) == 0:
same_info = {}
else:
same_info = self.get_cognation_info(nrdPn, cookies)
item[i] = [
name, code, app_date, applicant, address, inventor, main_cate,
sub_cate, pub_code, pub_date, priority, remark, agent, agency,
source, law_search_info, same_info
]
def get_law_info(self, nrdAn, nrdPn, cookies):
law_info = {}
try:
start_flag = 0
string = lawparams % (nrdPn, nrdAn, start_flag)
user_agent = random.choice(config.USER_AGENTS)
headers["User-Agent"] = user_agent
url = config.lawurl
result = requests.post(url,
string,
headers=headers,
cookies=cookies).content
lawdata = json.loads(result)["lawStateList"]
for i, single in enumerate(lawdata):
dates = single["prsDate"]
dates = deal_html_code.change_date(dates)
code = single["nrdAn"]
code = code.split("CN")[-1]
vcode = single["nrdPn"]
# print vcode
if vcode != '' and vcode != None:
vcode = vcode.split("CN")[-1]
else:
vcode = ''
content = single["lawStateExtendMeaning"]
law_info[i] = [dates, code, vcode, content]
except Exception, e:
logging.error("law error:%s" % e)
print e