导入企业年报

comData/annualReport1014/BaseCore.py

@@ -666,7 +666,7 @@ class BaseCore:
         self.cnx_.commit()
 
     # 插入到att表 返回附件id
-    def tableUpdate(self, retData, com_name, year, pdf_name, num, pub_time):
+    def tableUpdate(self, retData, com_name, year, pdf_name, num, pub_time,origin):
         item_id = retData['item_id']
         type_id = retData['type_id']
         group_name = retData['group_name']
@@ -688,12 +688,12 @@ class BaseCore:
             return id
         else:
 
-            Upsql = '''insert into clb_sys_attachment(year,name,type_id,item_id,group_name,path,full_path,category,file_size,order_by,status,create_by,create_time,page_size,object_key,bucket_name,publish_time) values(%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s)'''
+            Upsql = '''insert into clb_sys_attachment(year,name,type_id,item_id,group_name,path,full_path,category,file_size,order_by,status,create_by,create_time,page_size,object_key,bucket_name,publish_time,source) values(%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s)'''
 
             values = (
                 year, pdf_name, type_id, item_id, group_name, path, full_path, category, file_size, order_by,
                 status, create_by,
-                create_time, page_size,full_path.split('https://zzsn.obs.cn-north-1.myhuaweicloud.com/')[1],'zzsn',pub_time)
+                create_time, page_size,full_path.split('https://zzsn.obs.cn-north-1.myhuaweicloud.com/')[1],'zzsn',pub_time,origin)
 
             self.cursor_.execute(Upsql, values)  # 插入
             self.cnx_.commit()  # 提交

comData/annualReport1014/report.py

@@ -29,7 +29,7 @@ type_id = 1
 create_by = 'XueLingKun'
 taskType = '企业年报'
 #付俊雪的需要改为巨潮资讯网1_福布斯2000_PDF_60_付
-file_path = 'D:\\年报\\欧盟记分牌2500_年报补充_87_20231020'
+file_path = 'D:\\年报\\福布斯2000强_年报补充_20231018'
 log.info(f'=============当前pid为{baseCore.getPID()}==============')
 
 def sendKafka(dic_news):
@@ -146,9 +146,9 @@ if __name__=='__main__':
         social_code = data[1]
         ename = data[2]
         cname = data[3]
-        file_name = ename + ':' + file_year + '年年度报告' + '.pdf'
+        file_name = cname + ':' + file_year + '年年度报告' + '.pdf'
         content = ''
-        origin = ename + '官网'
+        origin = cname + '官网'
         #解析文件页数和内容
         log.info(f"-----------正在处理{file_name}--------------")
         with open(pdf_path, 'rb') as file:
@@ -178,7 +178,7 @@ if __name__=='__main__':
             retData_f = uptoOBS(retData, pathType, taskType, start_time,file_name,pdf_path)
             if retData_f['state']:
                 #retData, com_name, year, pdf_name, num, pub_time
-                att_id= baseCore.tableUpdate(retData_f, cname,file_year,file_name, num,file_year+'-12-31')
+                att_id= baseCore.tableUpdate(retData_f, cname,file_year,file_name, num,file_year+'-12-31',origin)
                 if att_id:
                     dic_news = {
                         'attachmentIds': att_id,