天眼查脚本维护

comData/Tyc/getTycId.py

@@ -47,11 +47,11 @@ cnx_ = baseCore.cnx
 cursor_ = baseCore.cursor
 taskType = '天眼查企业id/天眼查'
 #根据信用代码获取天眼查id 企业名字等信息
-def getTycIdByXYDM(xydm):
+def getTycIdByXYDM(com_name):
     retData={'state':False,'tycData':None,'reput':True}
     url=f"https://capi.tianyancha.com/cloud-tempest/search/suggest/v3?_={baseCore.getNowTime(3)}"
     ip = baseCore.get_proxy()
-    paramJsonData = {'keyword':xydm}
+    paramJsonData = {'keyword':com_name}
     try:
         # headers['User-Agent'] = baseCore.getRandomUserAgent()
         # headers['X-AUTH-TOKEN'] = baseCore.GetTYCToken()
@@ -62,21 +62,21 @@ def getTycIdByXYDM(xydm):
         if retJsonData['data'] and retJsonData['state']== 'ok':
             pass
         else:
-            log.error(f"---{xydm}-未查询到该企业---")
+            log.error(f"---{com_name}-未查询到该企业---")
             retData['reput'] = False
             return retData
         matchType=retJsonData['data'][0]['matchType']
-        if matchType=='信用代码匹配':
+        if matchType =='公司名称匹配':
             retData['state'] = True
             retData['tycData'] = retJsonData['data'][0]
             response.close()
             return retData
         else:
-            log.error(f"{xydm}------{retJsonData}")
+            log.error(f"{com_name}------{retJsonData}")
             response.close()
             return retData
     except Exception as e:
-        log.error(f"---{xydm}--{e}---")
+        log.error(f"---{com_name}--{e}---")
         return retData
 
 

comData/Tyc/newsbucai.py

@@ -52,7 +52,7 @@ headers = {
 cnx_ = baseCore.cnx
 cursor_ = baseCore.cursor
 
-taskType = '企业动态/天眼查/补采专精特新'
+taskType = '企业动态/天眼查/'
 
 def reqDetailmsg(url,headers):
 
@@ -81,14 +81,14 @@ def beinWork(tyc_code, social_code,start_time):
         for m in range(0,3):
             ip = baseCore.get_proxy()
             headers['User-Agent'] = baseCore.getRandomUserAgent()
-            response = requests.get(url=url, headers=headers, proxies=ip, verify=False)
+            response = requests.get(url=url, headers=headers, verify=False)
             time.sleep(random.randint(3, 5))
             break
         if (response.status_code == 200):
             pass
     except Exception as e:
         #todo:重新放入redis中
-        baseCore.rePutIntoR('NoticeEnterprise:gnqy_socialCode',social_code)
+        baseCore.rePutIntoR('NewsResend:newsInfo',social_code)
         log.error(f"{tyc_code}-----获取总数接口失败")
         error = '获取总数接口失败'
         state = 0
@@ -125,7 +125,7 @@ def beinWork(tyc_code, social_code,start_time):
     up_okCount = 0
     up_errorCount = 0
     up_repetCount = 0
-    for num in range(1, totalPage + 1):
+    for num in range(1, 10):
         time.sleep(3)
         log.info(f"获取分页数据--{tyc_code}----分页{num}----开始")
         start_page = time.time()
@@ -134,7 +134,7 @@ def beinWork(tyc_code, social_code,start_time):
             try:
                 ip = baseCore.get_proxy()
                 headers['User-Agent'] = baseCore.getRandomUserAgent()
-                response_page = requests.get(url=url_page, headers=headers, proxies=ip, verify=False)
+                response_page = requests.get(url=url_page, headers=headers, verify=False)
                 # time.sleep(3)
                 break
             except:
@@ -168,24 +168,15 @@ def beinWork(tyc_code, social_code,start_time):
             source = info_page['website']
             link = info_page['uri']
             try:
-                sel_sql = '''select social_credit_code from brpa_source_article_news where source_address = %s and social_credit_code=%s and type='2' '''
-                cursor_.execute(sel_sql, (link, social_code))
-            except Exception as e:
-                print(e)
-            selects = cursor_.fetchone()
-            if selects:
-                log.info(f'{tyc_code}-----{social_code}----{link}:已经存在')
-
-                # todo:如果该条数据存在则说明该条数据之后的都已经采集完成，就可以跳出函数，执行下一个企业
-
+                time_struct = time.localtime(int(info_page['rtm'] / 1000))  # 首先把时间戳转换为结构化时间
+                time_format = time.strftime("%Y-%m-%d %H:%M:%S", time_struct)  # 把结构化时间转换为格式化时间
+                if time_format > '2023-12-10 00:00':
+                    pass
+                else:
                     retData['up_okCount'] = up_okCount
                     retData['up_errorCount'] = up_errorCount
                     retData['up_repetCount'] = up_repetCount
-                # return retData
-                continue
-            try:
-                time_struct = time.localtime(int(info_page['rtm'] / 1000))  # 首先把时间戳转换为结构化时间
-                time_format = time.strftime("%Y-%m-%d %H:%M:%S", time_struct)  # 把结构化时间转换为格式化时间
+                    return retData
             except:
                 time_format = baseCore.getNowTime(1)
             try:
@@ -303,8 +294,8 @@ def beinWork(tyc_code, social_code,start_time):
 def doJob():
     while True:
         # 根据从Redis中拿到的社会信用代码,在数据库中获取对应基本信息
-        social_code = baseCore.redicPullData('NewsEnterprise:gnqybc_socialCode')
-        # social_code = '912301001275921118'
+        social_code = baseCore.redicPullData('NewsResend:newsInfo')
+        # social_code = '91320000733334390E'
         # 判断 如果Redis中已经没有数据，则等待
         if social_code == None:
             time.sleep(20)
@@ -316,14 +307,15 @@ def doJob():
                 pass
             else:
                 #数据重新塞入redis
-                baseCore.rePutIntoR('NewsEnterprise:gnqybc_socialCode',social_code)
+                baseCore.rePutIntoR('NewsResend:newsInfo',social_code)
                 continue
             id = data[0]
+            com_name= data[1]
             xydm = data[2]
             tycid = data[11]
             if tycid == None or tycid == '':
                 try:
-                    retData = getTycIdByXYDM(xydm)
+                    retData = getTycIdByXYDM(com_name)
                     if retData['tycData'] and retData['reput']:
                         tycid = retData['tycData']['id']
                         # todo:写入数据库
@@ -335,7 +327,7 @@ def doJob():
                         takeTime = baseCore.getTimeCost(start, time.time())
                         baseCore.recordLog(social_code, taskType, state, takeTime, '', '获取天眼查id失败')
                         log.info(f'======={social_code}====重新放入redis====')
-                        baseCore.rePutIntoR('NewsEnterprise:gnqybc_socialCode', social_code)
+                        baseCore.rePutIntoR('NewsResend:newsInfo', social_code)
                         continue
                     elif not retData['reput'] and not retData['tycData']:
                         continue
@@ -343,7 +335,7 @@ def doJob():
                     state = 0
                     takeTime = baseCore.getTimeCost(start, time.time())
                     baseCore.recordLog(social_code, taskType, state, takeTime, '', '获取天眼查id失败')
-                    baseCore.rePutIntoR('NewsEnterprise:gnqybc_socialCode', social_code)
+                    baseCore.rePutIntoR('NewsResend:newsInfo', social_code)
                     continue
             count = data[17]
             log.info(f"{id}---{xydm}----{tycid}----开始处理")
@@ -363,7 +355,7 @@ def doJob():
         except Exception as e:
             log.info(f'==={social_code}=====获取企业信息失败====')
             #重新塞入redis
-            baseCore.rePutIntoR('NewsEnterprise:gnqybc_socialCode',social_code)
+            baseCore.rePutIntoR('NewsResend:newsInfo',social_code)
             state = 0
             takeTime = baseCore.getTimeCost(start, time.time())
             baseCore.recordLog(social_code, taskType, state, takeTime, '', f'获取企业信息失败--{e}')