def name(self,data):
info = {}
if len(data)!=0:
for i,single in enumerate(data):
if "regNo" in single.keys():
code = single["regNo"]
code = deal_html_code.remove_symbol(code)
else:
code = ''
if "uniScid" in single.keys():
ccode = single["uniScid"]
else:
ccode = ''
ccode = deal_html_code.remove_symbol(ccode)
if "brName" in single.keys():
name = single["brName"]
name = deal_html_code.remove_symbol(name)
else:
name = ''
if "regOrganName" in single.keys():
gov_dept = single["regOrganName"]
gov_dept = deal_html_code.remove_symbol(gov_dept)
else:
gov_dept = ''
info[i] = [name, code, gov_dept,ccode]
return info
def name(self, data):
information = {}
for i, singledata in enumerate(data):
types = '黑名单'
if "bulletinListed" in singledata.keys():
in_reason = singledata["bulletinListed"]
in_reason = deal_html_code.remove_symbol(in_reason)
else:
in_reason = ''
if "abnTime" in singledata.keys():
in_date = singledata["abnTime"]
in_date = change_chinese_date(in_date)
else:
in_date = '0000-00-00'
if "bulletinRemoved" in singledata.keys():
out_reason = singledata["bulletinRemoved"]
out_reason = deal_html_code.remove_symbol(out_reason)
else:
out_reason = ''
if "remTime" in singledata.keys():
out_date = singledata["remTime"]
out_date = change_chinese_date(out_date)
else:
out_date = '0000-00-00'
if "remOrganInterpreted" in singledata.keys():
gov_dept = singledata["remOrganInterpreted"]
else:
gov_dept = ''
information[i] = [
types, in_reason, in_date, out_reason, out_date, gov_dept
]
return information
def name(self,data):
information = {}
for i in xrange(len(data)):
singledata = data[i]
types = '经营异常'
if "speCauseInterpreted" in singledata.keys():
in_reason = singledata["speCauseInterpreted"]
else:
in_reason = ''
if 'abnTime' in singledata.keys():
in_date = singledata["abnTime"]
in_date = deal_html_code.change_chinese_date(in_date)
else:
in_date = '0000-00-00'
if "remExcpResInterpreted" in singledata.keys():
out_reason = singledata["remExcpResInterpreted"]
out_reason = deal_html_code.remove_symbol(out_reason)
else:
out_reason = ''
if 'remDate' in singledata.keys():
out_date = singledata["remDate"]
out_date = deal_html_code.change_chinese_date(out_date)
else:
out_date = '0000-00-00'
if "decOrgInterpreted" in singledata.keys():
gov_dept = singledata["decOrgInterpreted"]
else:
gov_dept = ''
information[i] = [types, in_reason, in_date, out_reason, out_date, gov_dept]
return information
def name(self,data):
information = {}
for i,singledata in enumerate(data):
# nodeNum = singledata["nodeNum"]
ia_zch = singledata["regNum"]
if "intCls" in singledata.keys():
ia_flh = singledata["intCls"]
else:
ia_flh = ''
if "regAnncIssue" in singledata.keys():
ia_zcgg = singledata["regAnncIssue"]
else:
ia_zcgg = ''
if "goodsCnName" in singledata.keys():
ia_servicelist = deal_html_code.remove_symbol(singledata["goodsCnName"])
else:
ia_servicelist = ''
if "propertyBgnDate" in singledata.keys():
begin = singledata["propertyBgnDate"]
begin = deal_html_code.change_chinese_date(begin)
else:
begin = ''
if "propertyEndDate" in singledata.keys():
end = singledata["propertyEndDate"]
end = deal_html_code.change_chinese_date(end)
else:
end = ''
if "uniScid" in singledata.keys():
regNo = singledata["uniScid"]
else:
regNo = singledata["regNo"]
province = judge_province(regNo)
if begin== '' and end =='':
ia_zyqqx = ''
else:
ia_zyqqx = begin + '至' + end
if "regAnncDate" in singledata.keys():
ia_zcdate = singledata["regAnncDate"]
ia_zcdate = deal_html_code.change_chinese_date(ia_zcdate)
else:
ia_zcdate = ''
if "tmImage" in singledata.keys():
tmImage = singledata["tmImage"]
else:
tmImage = ''
information[i] = [ia_zch, ia_flh, ia_zcgg, ia_servicelist, ia_zyqqx, ia_zcdate,province,tmImage]
return information
def get_single_goods(self,data):
information = {}
if data != '':
for i,singledata in enumerate(data):
name = singledata["guaName"]
if "own" in singledata.keys():
ownership = singledata["own"]
else:
ownership = ''
situation = singledata["guaDesc"]
situation = deal_html_code.remove_symbol(situation)
if "remark" in singledata.keys():
remark = singledata["remark"]
else:
remark = ''
information[i] = [name, ownership, situation, remark]
return information
def deal_info(self, item):
info = {}
href = item.xpath("./@href")[0]
url = url_first + href
company = item.xpath('./h1[@class="f20"]')[0].xpath('string(.)')
company = remove_symbol(company)
status = item.xpath('./div[@class="wrap-corpStatus"]')[0].xpath(
'string(.)')
status = remove_symbol(status)
code = item.xpath('.//div[@class="div-map2"]')[0].xpath('string(.)')
code = remove_symbol(code)
# print code
if len(code.split(':')) == 2:
code = code.split(':')[1]
else:
code = code.split('：')[1]
daibiao = item.find('.//div[@class="div-user2"]').xpath('string(.)')
daibiao = remove_symbol(daibiao)
dates = item.find('.//div[@class="div-info-circle2"]').xpath(
'string(.)')
dates = remove_symbol(dates)
dates = dates.split('：')[1]
dates = change_chinese_date(dates)
history_name = item.xpath('.//div[@class="div-info-circle3"]')
if len(history_name) != 0:
history = item.xpath(
'.//div[@class="div-info-circle3"]/span[@class="g3"]'
)[0].xpath('string(.)')
else:
history = None
history = remove_symbol(history)
if history != None:
list = re.split('；', str(history))
templist = []
for k, temp in enumerate(list):
if temp != u'':
templist.append(temp)
history = ';'.join(templist)
info[code] = [url, company, status, code, daibiao, dates, history]
return info
def get_all_info(self, url):
information = {}
result = requests.get(url).content
data = json.loads(result)
if "regNo" in data.keys():
code = data["regNo"]
else:
code = ''
if "uniScid" in data.keys():
ccode = data["uniScid"]
else:
ccode = ''
uuid = data["annlId"]
name = data["entName"]
if len(data["entAnnlEtpsSet"]) != 0:
entAnnlEtps = data["entAnnlEtpsSet"][0]
if 'tel' in entAnnlEtps.keys():
tel = entAnnlEtps["tel"]
else:
tel = ''
if 'email' in entAnnlEtps.keys():
email = entAnnlEtps["email"]
else:
email = ''
if 'postalCode' in entAnnlEtps.keys():
postcode = entAnnlEtps["postalCode"]
else:
postcode = ''
if 'ifInvest' in entAnnlEtps.keys():
if_invest = int(entAnnlEtps["ifInvest"])
else:
if_invest = 0
if 'ifStocktrans' in entAnnlEtps.keys():
if_sharetrans = int(entAnnlEtps["ifStocktrans"])
else:
if_sharetrans = 0
else:
tel = ''
email = ''
postcode = ''
if_invest = 0
if_sharetrans = 0
if "hasFwarnnt" in data.keys():
if_fwarnnt = data["hasFwarnnt"]
else:
if_fwarnnt = 0
if "addr" in data.keys():
address = data["addr"]
else:
address = ''
status = data["opStateInterpreted"]
if len(data["entAnnlEmployeeSet"]) != 0:
employees = data["entAnnlEmployeeSet"][0]
employee = employees["empNum"]
if_empnum = int(employees["empNumSign"])
else:
employee = ''
if_empnum = 0
if "womEmpNum" in data.keys():
womennum = data["womEmpNum"]
if_womennum = int(data["womEmpNumSign"])
else:
womennum = ''
if_womennum = 0
if "holdingsMsg" in data.keys():
holding = data["holdingsMsg"]
holding = config.holding[holding]
if_holding = int(data["holdingsMsgSign"])
else:
holding = ''
if_holding = 0
if "mainBusiact" in data.keys():
mainbus = data["mainBusiact"]
mainbus = deal_html_code.remove_symbol(mainbus)
else:
mainbus = ''
if "priPid" in data.keys():
pripid = data["priPid"]
else:
pripid = ''
refuuid = data["refUuid"]
if "hasWebsite" in data.keys():
if_website = int(data["hasWebsite"])
else:
if_website = 0
if "ifNet" in data.keys():
if_net = int(data["ifNet"])
else:
if_net = 0
if "reportMode" in data.keys():
report_mode = data["reportMode"]
else:
report_mode = ''
if "reportType" in data.keys():
types = config.company_type[data["reportType"]]
else:
types = ''
if len(data["entAnnlPeSet"]) != 0:
entAnnlPeSet = data["entAnnlPeSet"][0]
if "psnName" in entAnnlPeSet.keys():
runner = entAnnlPeSet["psnName"]
else:
runner = ''
if "fundAm" in entAnnlPeSet.keys():
amount = entAnnlPeSet["fundAm"]
if "tel" in entAnnlPeSet.keys():
tel = entAnnlPeSet["tel"]
else:
tel = ''
else:
runner = ''
amount = ''
fill_date = data["anCheDate"]
fill_date = deal_html_code.change_chinese_date(fill_date)
information[0] = [
name, uuid, tel, address, email, postcode, status, employee,
if_empnum, womennum, if_womennum, holding, if_holding, mainbus,
code, ccode, pripid, refuuid, if_invest, if_sharetrans, if_fwarnnt,
if_website, if_net, report_mode, types, runner, amount, province,
fill_date
]
runinfo = data["entAnnlRunSet"]
assureinfo = data["entAnnlFwarnntSet"]
investinfo = data["entAnnlOutinvSet"]
labinfo = data["entAnnlSocSet"]
permitinfo = data["entAnnlPermitSet"]
partyinfo = data["entAnnlPartySet"]
schangeinfo = data["entAnnlStocktransSet"]
shareinfo = data["entAnnlInvestorSet"]
webinfo = data["entAnnlWebsiteSet"]
return information, runinfo, assureinfo, investinfo, labinfo, permitinfo, partyinfo, schangeinfo, shareinfo, webinfo
def update_basic(information, connect, cursor, gs_basic_id):
if '企业名称' in information.keys():
name = information[u"企业名称"]
elif '名称' in information.keys():
name = information[u"名称"]
if '统一社会信用代码' in information.keys():
code = information[u"统一社会信用代码"]
ccode = information[u"统一社会信用代码"]
ccode = str(ccode).replace('\t', '').replace('\n', '').replace(' ', '')
elif '注册号' in information.keys():
code = information[u"注册号"]
ccode = ''
elif '统一社会信用代码/注册号' in information.keys():
code = information[u"统一社会信用代码/注册号"]
pattern = re.compile('^91.*|92.*|93.*')
ccode = re.findall(pattern, code)
if len(ccode) == 0:
ccode = ''
elif len(ccode) == 1:
ccode = ccode[0]
if '登记状态' in information.keys():
status = information[u"登记状态"]
if '类型' in information.keys():
types = information[u'类型']
if '组成形式' in information.keys():
jj_type = information[u'组成形式']
else:
jj_type = None
if '法定代表人' in information.keys():
legal_person = information[u"法定代表人"]
responser = None
investor = None
runner = None
elif '经营者' in information.keys():
runner = information[u"经营者"]
legal_person = None
responser = None
investor = None
elif '负责人' in information.keys():
responser = information[u'负责人']
runner = None
legal_person = None
investor = None
elif '投资人' in information.keys():
investor = information[u"投资人"]
responser = None
runner = None
legal_person = None
elif '执行事务合伙人' in information.keys():
legal_person = information[u"执行事务合伙人"]
investor = None
runner = None
responser = None
if '成立日期' in information.keys():
sign_date = information[u"成立日期"]
sign_date = change_chinese_date(sign_date)
elif '注册日期' in information.keys():
sign_date = information[u"注册日期"]
sign_date = change_chinese_date(sign_date)
else:
sign_date = None
if '注册资本' in information.keys():
reg_amount = information[u"注册资本"]
elif '成员出资额' in information.keys():
reg_amount = information[u"成员出资额"]
else:
reg_amount = None
if '核准日期' in information.keys():
appr_date = information[u"核准日期"]
appr_date = change_chinese_date(appr_date)
else:
appr_date = None
if '营业期限自' in information.keys():
start_date = information[u"营业期限自"]
start_date = change_chinese_date(start_date)
elif '合伙期限自' in information.keys():
start_date = information[u"合伙期限至"]
start_date = change_chinese_date(start_date)
else:
start_date = None
if '营业期限至' in information.keys():
end_date = information[u"营业期限至"]
end_date = change_chinese_date(end_date)
elif '合伙期限至' in information.keys():
end_date = information[u"合伙期限至"]
end_date = change_chinese_date(end_date)
else:
end_date = None
if '登记机关' in information.keys():
reg_zone = information[u"登记机关"]
if '住所' in information.keys():
reg_address = information[u"住所"]
elif '经营场所' in information.keys():
reg_address = information[u"经营场所"]
elif '主要经营场所' in information.keys():
reg_address = information[u"主要经营场所"]
elif '营业场所' in information.keys():
reg_address = information[u"营业场所"]
if '业务范围' in information.keys():
scope = information[u"业务范围"]
elif '经营范围' in information.keys():
scope = information[u"经营范围"]
# updated_time = time.strftime("%Y-%m-%d %H:%M:%S", time.localtime(time.time()))
row_count = 0
flag = 0
ccode = remove_symbol(ccode)
try:
row_count = cursor.execute(
update_string,
(gs_basic_id, name, ccode, status, types, jj_type, legal_person,
responser, investor, runner, sign_date, appr_date, reg_amount,
start_date, end_date, reg_zone, reg_address, scope, gs_basic_id))
logging.info('update basic :%s' % row_count)
connect.commit()
except Exception, e:
flag = 100000004
logging.error("basic error:" % e)
def name(self,data):
info = {}
if len(data)!=0:
for i,single in enumerate(data):
if "invName" in single.keys():
name = single["invName"]
else:
name = ''
if "invTypeInterpreted" in single.keys():
types = single["invTypeInterpreted"]
else:
types = ''
if single["blicTypeInterpreted"]=='':
license_code = ''
license_type = ''
elif single["cetfTypeInterpreted"] =='':
license_type = ''
license_code = ''
elif single["blicTypeInterpreted"]!='':
license_type = single["blicTypeInterpreted"]
license_code = single["bLicNo"]
license_code = deal_html_code.remove_symbol(license_code)
elif single["cetfTypeInterpreted"]:
license_type = single["cetfTypeInterpreted"]
license_code = single["cetfId"]
encrypted = single["encrypted"]
if "cetfType" in single.keys():
cetfType = single["cetfType"]
elif "bLicType" in single.keys():
cetfType = single["bLicType"]
else:
cetfType = ''
if "subconAm" in single.keys():
reg_amount = single["subconAm"]
else:
reg_amount = ''
if "acconAm" in single.keys():
true_amount = single["acconAm"]
else:
true_amount = ''
if "conDate" in single.keys():
ta_date = single["conDate"]
ta_date = deal_html_code.change_chinese_date(ta_date)
else:
ta_date = '0000-00-00'
if "conForm" in single.keys():
ta_ways = single["conForm"]
else:
ta_ways = ''
if ta_ways == '1':
ta_ways = '货币'
if "countryInterpreted" in single.keys():
country = single["countryInterpreted"]
else:
country = ''
if "dom" in single.keys():
address = single["dom"]
else:
address = ''
info[i] = [name,types,license_code,license_type,reg_amount,true_amount,ta_date,ta_ways,country,address,encrypted,cetfType]
return info
def name(self,data):
info = []
name = data["entName"]
if "uniScid" in data.keys():
ccode = data["uniScid"]
else:
ccode = ''
if "opStateInterpreted" in data.keys():
status = data["opStateInterpreted"]
else:
status = ''
if "entTypeInterpreted" in data.keys():
types = data["entTypeInterpreted"]
else:
types = ''
if "lerep" in data.keys():
legal_person = data["lerep"]
else:
legal_person = ''
if "estDate" in data.keys():
reg_date = data["estDate"]
reg_date = deal_html_code.change_chinese_date(reg_date)
else:
reg_date = None
if "issBlicDate" in data.keys():
appr_date = data["issBlicDate"]
appr_date = deal_html_code.change_chinese_date(appr_date)
else:
appr_date = None
if "regCapInterpreted" in data.keys():
reg_amount = data["regCapInterpreted"]
else:
reg_amount = ''
if "opFrom" in data.keys():
start_date = data["opFrom"]
start_date = deal_html_code.change_chinese_date(start_date)
else:
start_date = None
if "opTo" in data.keys():
end_date = data["opTo"]
end_date = deal_html_code.change_chinese_date(end_date)
else:
end_date = None
if "regOrganInterpreted" in data.keys():
reg_zone = data["regOrganInterpreted"]
else:
reg_zone = ''
if "dom" in data.keys():
reg_address = data["dom"]
else:
reg_address = ''
if "opScope" in data.keys():
scope = data["opScope"]
scope = deal_html_code.remove_symbol(scope)
else:
scope = ''
if "canDate" in data.keys():
del_date = data["canDate"]
del_date = deal_html_code.change_chinese_date(del_date)
else:
del_date = None
info = [name,ccode,status,types,legal_person,reg_date,appr_date,reg_amount,start_date,end_date,reg_zone,reg_address,scope,del_date]
return info