证监会公告判断文件是否已经存在文件服务器

comData/noticeReport/证监会-公告.py

-import json
+import json
@@ -86,9 +86,9 @@ def uptoOBS(pdf_url,pdf_name,type_id,social_code):
 
         return retData
 
-def secrchATT(item_id, name, type_id):
-    sel_sql = '''select id from clb_sys_attachment where item_id = %s and name = %s and type_id=%s '''
-    cursor_.execute(sel_sql, (item_id, name, type_id))
+def secrchATT(item_id, name, type_id,order_by):
+    sel_sql = '''select id from clb_sys_attachment where item_id = %s and name = %s and type_id=%s and order_by=%s '''
+    cursor_.execute(sel_sql, (item_id, name, type_id,order_by))
     selects = cursor_.fetchone()
     return selects
 
@@ -106,26 +106,26 @@ def tableUpdate(retData, com_name, year, pdf_name, num):
     page_size = retData['page_size']
     create_time = retData['create_time']
     order_by = num
-    selects = secrchATT(item_id, pdf_name, type_id)
-
-    if selects:
-        log.info(f'pdf_name:{pdf_name}已存在')
-        id = ''
-        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) values(%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s)'''
+    # selects = secrchATT(item_id, pdf_name, type_id)
+    #
+    # if selects:
+    #     log.info(f'pdf_name:{pdf_name}已存在')
+    #     id = ''
+    #     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) values(%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)
+    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)
 
-        cursor_.execute(Upsql, values)  # 插入
-        cnx_.commit()  # 提交
-        log.info("更新完成:{}".format(Upsql))
-        selects = secrchATT(item_id, pdf_name, type_id)
-        id = selects[0]
-        return id
+    cursor_.execute(Upsql, values)  # 插入
+    cnx_.commit()  # 提交
+    log.info("更新完成:{}".format(Upsql))
+    selects = secrchATT(item_id, pdf_name, type_id,order_by)
+    id = selects[0]
+    return id
 
 def RequestUrl(url, payload, social_code,start_time):
     # ip = get_proxy()[random.randint(0, 3)]
@@ -276,6 +276,17 @@ def InsterInto(social_code, pdf_url,pub_time):
         return insert
 
 def GetContent(pdf_url, pdf_name, social_code, year, pub_time, start_time,com_name,num):
+    #判断文件是否已经存在obs服务器中
+    # file_path = 'XQWAnnualReport/2023-10/浙江国祥股份有限公司首次公开发行股票并在主板上市暂缓发行公告.doc'
+    now_time = time.strftime("%Y-%m")
+    file_path = 'ZJH/'+now_time+'/'+pdf_name+'.pdf'
+    response = obsClient.getObjectMetadata('zzsn', file_path)
+    if response.status >= 300:
+        log.info('=====文件不存在obs=====')
+        pass
+    else:
+        log.info(f'=====文件存在obs========{file_path}')
+        return False
     #上传至华为云服务器
     retData = uptoOBS(pdf_url,pdf_name,8,social_code)
     #附件插入att数据库