天眼查基本信息

comData/Tyc/baseinfo0130_tyc.py

@@ -97,7 +97,7 @@ def baseinfo(com_soup):
     span_list = briefTag.find_all('span')
     for span in span_list:
         if len(span.attrs) == 0:
-            data['简介'] = span.text
+            data['简介'] = span.text.split('通过天眼查大数据分析')[0]
             break
     return data
 
@@ -106,107 +106,107 @@ def dic_handle(result_dic):
     try:
         company_name = result_dic['企业名称']
     except:
-        company_name = ''
+        company_name = None
 
     try:
         CreditCode = result_dic['统一社会信用代码']
     except:
-        CreditCode = ''
+        CreditCode = None
 
     try:
         OperName = result_dic['法定代表人']
     except:
-        OperName = ''
+        OperName = None
 
     try:
         PhoneNumber = result_dic['电话']
     except:
-        PhoneNumber = ''
+        PhoneNumber = None
 
     try:
         WebSite = result_dic['网址']
     except:
-        WebSite = ''
+        WebSite = None
 
     try:
         Email = result_dic['邮箱']
     except:
-        Email = ''
+        Email = None
 
     try:
         Desc = result_dic['简介']
     except:
-        Desc = ''
+        Desc = None
 
     try:
         Status = result_dic['经营状态']
     except:
-        Status = ''
+        Status = None
 
     try:
         StartDate = result_dic['成立日期']
     except:
-        StartDate = ''
+        StartDate = None
 
     try:
         RecCap = result_dic['实缴资本']
     except:
-        RecCap = ''
+        RecCap = None
 
     try:
         RegistCapi = result_dic['注册资本']
     except:
-        RegistCapi = ''
+        RegistCapi = None
 
     try:
         CheckDate = result_dic['核准日期']
     except:
-        CheckDate = ''
+        CheckDate = None
 
     try:
         OrgNo = result_dic['组织机构代码']
     except:
-        OrgNo = ''
+        OrgNo = None
 
     try:
         No = result_dic['工商注册号']
     except:
-        No = ''
+        No = None
 
     try:
         taxpayerNo = result_dic['纳税人识别号']
     except:
-        taxpayerNo = ''
+        taxpayerNo = None
 
     try:
         EconKind = result_dic['企业类型']
     except:
-        EconKind = ''
+        EconKind = None
 
     try:
         TermStart = result_dic['营业期限'].split('至')[0]
     except:
-        TermStart = ''
+        TermStart = None
 
     try:
         TeamEnd = result_dic['营业期限'].split('至')[1]
     except:
-        TeamEnd = ''
+        TeamEnd = None
 
     try:
         TaxpayerType = result_dic['纳税人资质']
     except:
-        TaxpayerType = ''
+        TaxpayerType = None
 
-    try:
-        SubIndustry = result_dic['国标行业']
-    except:
-        SubIndustry = ''
+    # try:
+    #     SubIndustry = result_dic['国标行业']
+    # except:
+    #     SubIndustry = ''
 
     try:
         region = result_dic['所属地区']
     except:
-        region = ''
+        region = None
     try:
         pattern = r'^(.*?省|.*?自治区)?(.*?市|.*?自治州)?(.*?区|.*?县|.*?自治县|.*?市辖区)?(.*?区|.*?县|.*?自治县|.*?市辖区)?$'
         matches = re.match(pattern, region)
@@ -220,53 +220,53 @@ def dic_handle(result_dic):
                     break
 
     except:
-        Province = ''
-        City = ''
-        County = ''
+        Province = None
+        City = None
+        County = None
 
     try:
         BelongOrg = result_dic['登记机关']
     except:
-        BelongOrg = ''
+        BelongOrg = None
 
     try:
         Info = result_dic['人员规模']
     except:
-        Info = ''
+        Info = None
 
     try:
         can_bao = result_dic['参保人数']
     except:
-        can_bao = ''
+        can_bao = None
 
     try:
         OriginalName = result_dic['曾用名']
     except:
-        OriginalName = ''
+        OriginalName = None
 
     try:
         EnglishName = result_dic['英文名称']
     except:
-        EnglishName = ''
+        EnglishName = None
 
     try:
         IxCode = result_dic['进出口企业代码']
     except:
-        IxCode = ''
+        IxCode = None
 
     try:
         Address = result_dic['地址']
     except:
-        Address = ''
+        Address = None
 
     try:
         Scope = result_dic['经营范围']
     except:
-        Scope = ''
+        Scope = None
 
     aa_dict = {
         'name': company_name,  # 企业名称
-        'shortName': '',  # 企业简称
+        'shortName': None,  # 企业简称
         'socialCreditCode': CreditCode,  # 统一社会信用代码
         'legalPerson': OperName,  # 法定代表人
         'officialPhone': PhoneNumber,  # 电话
@@ -285,7 +285,7 @@ def dic_handle(result_dic):
         'businessStartDate': TermStart,  # 营业期限自
         'businessEndDate': TeamEnd,  # 营业期限至
         'taxpayerQualification': TaxpayerType,  # 纳税人资质
-        'industry': SubIndustry,  # 所属行业
+        'industry': None,  # 所属行业
         'region': region,
         'province': Province,  # 所属省
         'city': City,  # 所属市
@@ -565,7 +565,7 @@ if __name__ == '__main__':
         start_time = time.time()
         # 获取企业信息
         # company_field = baseCore.redicPullData('BaseInfoEnterprise:gnqy_socialCode')
-        company_field = '91130000738711917Q||'
+        company_field = '91110000710925016E||'
         if company_field == 'end':
             # 本轮处理完毕，需要发送邮件，并且进入下一轮
             baseCore.sendEmail(file_name)
@@ -604,14 +604,14 @@ if __name__ == '__main__':
         # category = company_field.split('|')[19]
         # exchange = company_field.split('|')[20]
         # listType = company_field.split('|')[21]
-        ynDomestic = '1'
-        countryName = ''
-        securitiesCode = ''
-        securitiesShortName = ''
-        listingDate = ''
-        category = ''
-        exchange = ''
-        listType = ''
+        ynDomestic = None
+        countryName = None
+        securitiesCode = None
+        securitiesShortName = None
+        listingDate = None
+        category = None
+        exchange = None
+        listType = None
 
         count = redaytowork(com_name, social_code, securitiesCode, securitiesShortName, listingDate, category, exchange,
                             listType, ynDomestic, countryName, file_name)

comData/Tyc/classtool.py

@@ -56,7 +56,7 @@ class Token():
         query = {
             'fenghaoTime': {'$lt': 'updateTime'},  # 封号时间小于更新时间
         }
-        result = db_storage.find_one(query, sort=[('updateTime', -1)])
+        result = db_storage.find_one(query, sort=[('updateTime', 1)])
         cookies = result['cookies']
         id_token = result['_id']
         return cookies, id_token
@@ -77,7 +77,7 @@ class Token():
             filter = {'_id': ObjectId(id_token)}
 
             # 更新操作
-            update = {'$set': {'fenghaoTime': datetime.datetime.now()}}
+            update = {'$set': {'fenghaoTime': datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S')}}
 
             # 执行更新操作
             db_storage.update_one(filter, update)
@@ -86,7 +86,7 @@ class Token():
             filter = {'_id': ObjectId(id_token)}
 
             # 更新操作
-            update = {'$set': {'updateTtime': datetime.datetime.now()}}
+            update = {'$set': {'updateTime': datetime.datetime.now().strftime('%Y-%m-%d %H:%M:%S')}}
 
             # 执行更新操作
             db_storage.update_one(filter, update)

comData/Tyc/get_tyc_cookies.py

@@ -26,7 +26,7 @@ if __name__ == "__main__":
     name = input('所属用户:')
     driver = create_driver()
     driver.get(url)
-    time.sleep(100)
+    time.sleep(60)
 
     cookies = driver.get_cookies()
     # print(driver.get_cookies())