def get_info(self, data):
tr_list = data.xpath(".//tr")
info = {}
for i, singledata in enumerate(tr_list):
temp = {}
td_list = singledata.xpath("./td")
# 针对表头和为信息的情况进行特殊对待
if len(td_list) == 0 or len(td_list) == 1:
continue
temp["name"] = deal_html_code.remove_symbol(
td_list[1].xpath("string(.)"))
reg_amount = deal_html_code.remove_symbol(
td_list[2].xpath("string(.)"))
temp["reg_amount"] = deal_html_code.match_float(reg_amount)
reg_date = deal_html_code.remove_symbol(
td_list[3].xpath("string(.)"))
temp["reg_date"] = deal_html_code.change_chinese_date(reg_date)
temp["reg_way"] = deal_html_code.remove_symbol(
td_list[4].xpath("string(.)"))
ac_amount = deal_html_code.remove_symbol(
td_list[5].xpath("string(.)"))
temp["ac_amount"] = deal_html_code.match_float(ac_amount)
ac_date = deal_html_code.remove_symbol(
td_list[6].xpath("string(.)"))
temp["ac_date"] = deal_html_code.change_chinese_date(ac_date)
temp["ac_way"] = deal_html_code.remove_symbol(
td_list[7].xpath("string(.)"))
info[i] = temp
return info
def get_info(self, data):
info = {}
for key, value in dict.iteritems():
info[value] = deal_html_code.get_match_info(key, data)
# 调用juege_if_public判断是否公示，判断是否公示是根据所取到的该部分信息中是否含有元
# 进行判断的，可能总结不够全面，以后见到不含元的再更改函数
info["if_asset"] = self.judge_if_public(info["asset"])
# 判断完是否公示后调用match_float取里面的数字
info["asset"] = deal_html_code.match_float(info["asset"])
info["if_benifit"] = self.judge_if_public(info["benifit"])
info["benifit"] = deal_html_code.match_float(info["benifit"])
info["if_main"] = self.judge_if_public(info["main_income"])
info["main_income"] = deal_html_code.match_float(info["main_income"])
info["if_net"] = self.judge_if_public(info["net_income"])
info["net_income"] = deal_html_code.match_float(info["net_income"])
info["if_tax"] = self.judge_if_public(info["tax"])
info["tax"] = deal_html_code.match_float(info["tax"])
info["if_loan"] = self.judge_if_public(info["loan"])
info["loan"] = deal_html_code.match_float(info["loan"])
info["if_subsidy"] = self.judge_if_public(info["subsidy"])
info["subsidy"] = deal_html_code.match_float(info["subsidy"])
info["if_income"] = deal_html_code.match_float(info["income"])
info["income"] = self.judge_if_public(info["income"])
info["if_profit"] = deal_html_code.match_float(info["profit"])
info["profit"] = self.judge_if_public(info["profit"])
info["if_debt"] = deal_html_code.match_float(info["debt"])
info["debt"] = self.judge_if_public(info["debt"])
return info
def deal_single_info(self, data, info):
for i, singledata in enumerate(data):
org = singledata["ORG"]
id = singledata["ID"]
seqid = singledata["SEQ_ID"]
code = singledata["GUARANTY_REG_NO"]
dates = singledata["START_DATE"]
dates = deal_html_code.change_chinese_date(dates)
dept = singledata["CREATE_ORG"]
amount = singledata["ASSURE_CAPI"]
amount = deal_html_code.match_float(amount)
status = singledata["STATUS"]
cates = singledata["ASSURE_KIND"]
start_date = singledata["ASSURE_START_DATE"]
start_date = deal_html_code.change_date_style(start_date)
end_date = singledata["ASSURE_END_DATE"]
end_date = deal_html_code.change_date_style(end_date)
period = "自" + start_date + "至" + end_date
ranges = singledata["ASSURE_SCOPE"]
remark = singledata["REMARK"]
cancel_cause = singledata["WRITEOFF_REASON"]
RN = singledata["RN"]
types = 'mort_person'
person_href = person_url + params.format(org, id, seqid)
person_info = self.get_detail_info(person_href, types)
types = 'mort_goods'
goods_href = goods_url + params.format(org, id, seqid)
# print goods_href
goods_info = self.get_detail_info(goods_href, types)
info[RN] = [
code, dates, dept, amount, status, cates, period, ranges,
remark, cancel_cause, person_info, goods_info
]
def get_info(self, data):
info = {}
for key, value in config.report_lab_dict.iteritems():
info[value] = deal_html_code.get_match_info(key, data)
# 这两种情况的采集没有太大意义就不再入库
if info["birth_num"] == '' or info["birth_num"] == '人':
info = {}
else:
# 判定欠费金额，实际缴费金额，缴费基数 是否公示
# 判定标准选取生育的各个对应信息进行标准，
# 即认为如果生育、医疗、养老、失业中有一个欠费，实缴，基数是不公示的
# 则其他的也是不公示的
if info["birth_owe"] > 0:
if_owe = 0
else:
if_owe = 1
info["if_owe"] = if_owe
if info["birth_base"] == 0:
if_basenum = 0
else:
if_basenum = 1
info["if_basenum"] = if_basenum
if info["birth"] == 0:
if_periodamount = 0
else:
if_periodamount = 1
info["if_periodamount"] = if_periodamount
for key, value in info.iteritems():
print key, value
if "if" in key:
continue
info[key] = deal_html_code.match_float(value)
return info
def deal_single_info(self, data, info): for i, singledata in enumerate(data): name = singledata["STOCK_NAME"] reg_amount = singledata["SHOULD_CAPI"] reg_amount = deal_html_code.match_float(reg_amount) reg_date = singledata["SHOULD_CAPI_DATE"] reg_date = deal_html_code.change_chinese_date(reg_date) reg_way = singledata["SHOULD_CAPI_TYPE"] ac_amount = singledata["REAL_CAPI"] ac_amount = deal_html_code.match_float(ac_amount) ac_date = singledata["REAL_CAPI_DATE"] ac_date = deal_html_code.change_chinese_date(ac_date) ac_way = singledata["REAL_CAPI_TYPE"] uuid = singledata["ID"] RN = singledata["RN"] info[RN] = [name, uuid,reg_amount, reg_date, reg_way, ac_amount, ac_date, ac_way]
def get_info(self):
result, status_code = Send_Request(self.url,
self.headers).send_request()
info = {}
runinfo = {}
if status_code == 200:
flag = 1
data = json.loads(result.content)
code = data["REG_NO"]
if code.startswith("9"):
ccode = code
else:
ccode = ''
name = data["CORP_NAME"]
tel = data["TEL"]
address = data["ADDR"]
email = data["E_MAIL"]
postcode = data["ZIP"]
status = data["PRODUCE_STATUS"]
employee = data["PRAC_PERSON_NUM"]
if employee == None:
if_empnum = 0
elif u"不公示" in employee:
if_empnum = 0
else:
if_empnum = 1
womennum = data["WOM_EMP_NUM"]
if womennum == None:
if_womennum = 0
elif u"不公示" in womennum:
if_womennum = 0
else:
if_womennum = 1
holding = data["HOLDINGS_MSG"]
if holding == None:
if_holding = 0
elif u"不公示" in holding:
if_holding = 0
else:
if_holding = 1
mainbus = data["MAIN_BUSIACT"]
runner = data["OPER_MAN_NAME"]
amount = data["REG_CAPI"]
fill_date = data["REPORT_DATE"]
info[0] = [
name, tel, address, email, postcode, status, employee,
if_empnum, womennum, if_womennum, holding, if_holding, mainbus,
code, ccode, runner, amount, fill_date
]
asset = data["NET_AMOUNT"]
if_asset = self.judge_if_public(asset)
asset = deal_html_code.match_float(asset)
benifit = data["TOTAL_EQUITY"]
if_benifit = self.judge_if_public(benifit)
benifit = deal_html_code.match_float(benifit)
income = data["SERV_FARE_INCOME"]
if_income = self.judge_if_public(income)
income = deal_html_code.match_float(income)
profit = data["PROFIT_TOTAL"]
if_profit = self.judge_if_public(profit)
profit = deal_html_code.match_float(profit)
main_income = data["SALE_INCOME"]
if_main = self.judge_if_public(main_income)
main_income = deal_html_code.match_float(main_income)
net_income = data["PROFIT_RETA"]
if_net = self.judge_if_public(net_income)
net_income = deal_html_code.match_float(net_income)
tax = data["TAX_TOTAL"]
if_tax = self.judge_if_public(tax)
tax = deal_html_code.match_float(tax)
debt = data["DEBT_AMOUNT"]
if_debt = self.judge_if_public(debt)
debt = deal_html_code.match_float(debt)
loan = data["LOAN"]
if_loan = self.judge_if_public(loan)
loan = deal_html_code.match_float(loan)
subsidy = data["SUBSIDY"]
if_subsidy = self.judge_if_public(subsidy)
subsidy = deal_html_code.match_float(subsidy)
runinfo[0] = [
asset, if_asset, benifit, if_benifit, income, if_income,
profit, if_profit, main_income, if_main, net_income, if_net,
tax, if_tax, debt, if_debt, loan, if_loan, subsidy, if_subsidy
]
else:
flag = 100000004
return info, runinfo, flag
def get_info(self):
result, status_code = Send_Request(self.url,self.headers).send_request()
info = {}
if status_code == 200:
flag = 1
data = json.loads(result.content)
if len(data) >0:
uuid = data["ID"]
birth_owe = data["PAYMENT_SY"]
birth_num = data["MATERNITY_NUM"]
birth = data["SOCIALINS_SY"]
birth_base = data["WAGES_SY"]
if u"不公示" in birth_owe:
if_owe = 0
else:
if_owe = 1
if u"不公示" in birth_base:
if_basenum = 0
else:
if_basenum = 0
if u"不公示" in birth:
if_periodamount = 0
else:
if_periodamount = 1
birth_owe = deal_html_code.match_float(birth_owe)
birth_num = deal_html_code.match_float(birth_num)
birth = deal_html_code.match_float(birth)
birth_base = deal_html_code.match_float(birth_base)
old_num = deal_html_code.match_float(data["ENDOWMENT_NUM"])
old_owe = deal_html_code.match_float(data["PAYMENT_JBYL"])
old = deal_html_code.match_float(data["SOCIALINS_JBYL"])
old_base = deal_html_code.match_float(data["WAGES_JBYL"])
unemploy = deal_html_code.match_float(data["SOCIALINS_SYBX"])
unemploy_base = deal_html_code.match_float(data["WAGES_SYBX"])
unemploy_owe = deal_html_code.match_float(data["PAYMENT_SYBX"])
unemploy_num = deal_html_code.match_float(data["UNEMPLOYED_NUM"])
medical = deal_html_code.match_float(data["SOCIALINS_YLBX"])
medical_base = deal_html_code.match_float(data["WAGES_YLBX"])
medical_owe = deal_html_code.match_float(data["PAYMENT_YLBX"])
medical_num = deal_html_code.match_float(data["MEDICARE_NUM"])
injury = deal_html_code.match_float(data["SOCIALINS_GSBX"])
injury_owe = deal_html_code.match_float(data["PAYMENT_GSBX"])
injury_num = deal_html_code.match_float(data["EMPLOYMENT_INJURY_NUM"])
info[0] = [uuid, if_owe, if_basenum, if_periodamount, birth_owe, birth_num, birth, birth_base, old_num,
old_owe, old, old_base, \
unemploy, unemploy_base, unemploy_owe, unemploy_num, medical, medical_base, medical_owe,
medical_num, injury, injury_owe, injury_num]
else:
logging.info("无社保信息")
else:
flag = 100000004
logging.info("打开社保链接失败！")
return info,flag