def parse(self,response):
# print(response.text)
item = QiyemuluItem()
try:
dataJson= json.loads(response.text)
alldata = dataJson.get("data","")
if alldata:
'''股东信息'''
result = alldata.get("result", "")
if result:
infoList = []
for info in result:
tags = []
infoDict = {}
infoDict['id'] = info.get("id", '')
infoDict['name'] = info.get("name", '')
if info.get("tagList", ''):
for tag in info.get("tagList", ''):
tags.append(tag.get("name", "0"))
infoDict['tags'] = ','.join(tags)
if info.get("capital", ''):
for cap in info.get("capital", ''):
infoDict['amomon'] = cap.get("amomon", "")
infoDict['percent'] = cap.get("percent", "")
infoList.append(infoDict)
item['companyId'] = response.meta['id']
item["holderInfo"] = str(infoList)
yield item
except Exception as e:
print(e.args)
def content_parse(self, response):
item = QiyemuluItem()
infoDict = response.meta['infoDict']
content = response.css("div.container ::text").extract()
infoDict['container'] = ''.join(content).replace(" ", "")
item['companyId'] = response.meta['id']
item['allInfo'] = str(infoDict)
yield item
def parse(self,response):
# print(response.text)
item = QiyemuluItem()
try:
dataJson= json.loads(response.text)
alldata = dataJson.get("data","")
if alldata:
result = alldata.get("resultList", "")
if result:
infoList = []
for info in result:
infoDict = {}
plaintiffs = info.get("plaintiff","")
defendants = info.get("defendant","")
if plaintiffs:
plaintiff_list = []
for pla in plaintiffs:
plaintiff_list.append(pla.get('name',''))
infoDict['plaintiff'] = ','.join(plaintiff_list)
if defendants:
defendant_list = []
for defend in defendants:
defendant_list.append(defend.get('name',''))
infoDict['defendant'] = ','.join(defendant_list)
infoDict['caseReason'] = info.get("caseReason", '')
infoDict['caseNo'] = info.get("caseNo", '')
infoDict['startDate'] = info.get("startDate", '')
infoDict['litigant'] = info.get("litigant", '')
infoDict['contractors'] = info.get("contractors", '')
infoDict['judge'] = info.get("judge", '')
infoDict['court'] = info.get("court", '')
infoDict['courtroom'] = info.get("courtroom", '')
infoList.append(infoDict)
item['companyId'] = response.meta['id']
item["allInfo"] = str(infoList)
yield item
except Exception as e:
print(e.args)
def parse(self, response):
# print(response.text)
item = QiyemuluItem()
try:
dataJson = json.loads(response.text)
alldata = dataJson.get("data", "")
if alldata:
'''高管信息'''
result = alldata.get("result", "")
if result:
infoList = []
for info in result:
infoDict = {}
infoDict['id'] = info.get("id", '')
infoDict['name'] = info.get("name", '')
infoDict['typeJoin'] = ','.join(
info.get("typeJoin", ""))
infoList.append(infoDict)
item['companyId'] = response.meta['id']
item["allInfo"] = str(infoList)
yield item
except Exception as e:
print(e.args)
def parse(self,response):
# print(response.text)
item = QiyemuluItem()
try:
dataJson= json.loads(response.text)
alldata = dataJson.get("data","")
if alldata:
result = alldata.get("result", "")
if result:
infoList = []
for info in result:
infoDict = {}
infoDict['changeItem'] = info.get("changeItem", '')
infoDict['changeTime'] = info.get("changeTime", '')
infoDict['createTime'] = info.get("createTime", '')
infoDict['contentBefore'] = remove_tags(info.get("contentBefore", ''))
infoDict['contentAfter'] = remove_tags(info.get("contentAfter", ''))
infoList.append(infoDict)
item['companyId'] = response.meta['id']
item["allInfo"] = str(infoList)
yield item
# print(infoList)
except Exception as e:
print(e.args)
def parse(self, response):
item = QiyemuluItem()
if response.status == 200:
json_data = json.loads(response.text)
state = json_data.get("state", '')
try:
if state == "ok":
data = json_data.get('data', '')
if data:
for info in data[:-1]:
item['cname'] = response.meta['name']
item['companyId'] = info.get('id', "")
item['companyName'] = remove_tags(
info.get('name', ""))
item['alias'] = info.get('alias', "")
item['legalPersonName'] = info.get(
'legalPersonName', "")
item['phoneList'] = info.get('phone', "")
item['emailList'] = info.get('emails', "")
item['websiteList'] = info.get('websites', "")
item['estiblishTime'] = info.get(
'estiblishTime', "")
item['regCapital'] = info.get('regCapital', "")
item['creditCode'] = info.get('creditCode', "")
item['taxNumber'] = info.get('creditCode', "")
item['orgNumber'] = info.get('creditCode',
"")[9:18]
item['regStatus'] = info.get('regStatus', "")
item['companyOrgType'] = info.get(
'companyOrgType', "")
item['industry'] = info.get('categoryStr', "")
item['regLocation'] = info.get('regLocation', "")
item['logo'] = info.get('logo', "")
item['businessScope'] = info.get(
'businessScope', "")
item['base'] = info.get('base', "")
item['district'] = info.get('district', "")
item['historyNames'] = info.get('historyNames', "")
#后期补充数据
item['intro'] = info.get('intro', "")
item['regNumber'] = info.get('regNumber', "")
item['actualCapital'] = info.get(
'actualCapital', "")
item['taxQualification'] = info.get(
'taxQualification', "")
item['englishName'] = info.get('property3', "")
item['staffNumRange'] = info.get(
'staffNumRange', "")
item['socialStaffNum'] = info.get(
'socialStaffNum', "")
item['operatingPeriod'] = info.get(
'operatingPeriod', "")
item['approvedTime'] = info.get('approvedTime', "")
item['regInstitute'] = info.get('regInstitute', "")
item['updatetime'] = info.get('updatetime', "")
yield item
# break
else:
# self.redis_pool.srem("names:key",response.mata['name'])
print(response.mata['name'], "没有数据")
except Exception as e:
print(e.args)
def parse(self,response):
# print(response.text)
try:
dataJson= json.loads(response.text)
alldata = dataJson.get("data","")
if alldata:
item = QiyemuluItem()
#公司Id
item['companyId'] = alldata.get("id","")#response.meta['id']
#公司名称
item['companyName'] = alldata.get("name","")#response.meta['name']
#别名
item['alias'] = alldata.get("alias","")
#曾用名
item['historyNames'] = alldata.get("historyNames","")
#法定代表人
item['legalPersonName'] = alldata.get("legalPersonName","")
#联系方式列表
item['phoneList'] = ""
#邮箱列表
item['emailList'] = ""
#网址
item['websiteList'] = str(alldata.get("websiteList",""))
#成立日期
item['estiblishTime'] = alldata.get("estiblishTime","")
#工商注册号
item['regNumber'] = alldata.get("regNumber","")
#注册资本
item['regCapital'] = alldata.get("regCapital","")
#实缴资本
item['actualCapital'] = alldata.get("actualCapital","")
#统一信用代码
item['creditCode'] = alldata.get("creditCode","")
#纳税人识别号
item['taxNumber'] = alldata.get("taxNumber","")
#组织机构代码
item['orgNumber'] = alldata.get("orgNumber","")
#纳税人资质
item['taxQualification'] = alldata.get("taxQualification","")
#英文名称
item['englishName'] = alldata.get("property3","")
#经营状态
item['regStatus'] = alldata.get("regStatus","")
#人员规模
item['staffNumRange'] = alldata.get("staffNumRange","")
#参保人数
item['socialStaffNum'] = alldata.get("socialStaffNum","")
#公司类型
item['companyOrgType'] = alldata.get("companyOrgType","")
#行业
item['industry'] = alldata.get("industry","")
#营业期限
fromTime = alldata.get("fromTime","")
toTime = alldata.get("toTime","")
item['operatingPeriod'] = "%s/%s"%(fromTime,toTime)
#注册地址
item['regLocation'] = alldata.get("regLocation","")
#核准日期
item['approvedTime'] = alldata.get("approvedTime","")
#登记机关
item['regInstitute'] = alldata.get("regInstitute","")
#公司logo地址
item['logo'] = alldata.get("logo","")
#股权结构图
# item['equityUrl'] = baseInfo.get("equityUrl","")
#最新更新时间
item['updatetime'] = alldata.get("updatetime","")
#经营范围
item['businessScope'] = alldata.get("businessScope","")
#简介
item['intro'] = alldata.get("baseInfo","")
#所属地区简称
item['base'] = alldata.get("base","")
#所述区域简称
item['district'] = ""
yield item
# print(item)
except Exception as e:
print(e.args)
def parse(self,response):
print(response.text)
item = QiyemuluItem()