拉取年报附件

comData/annualReport1023/拉取年报附件.py

+"""
+"""
+从redis中读出企业信用代码，根据信用代码查询企业信息，拼接链接下载。
+文件存储格式： 企业名称：
+                年报.pdf
+"""
+import os
+from pathlib import Path
+
+from base.BaseCore import BaseCore
+from datetime import datetime
+baseCore = BaseCore()
+import requests
+import urllib3
+urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)
+log = baseCore.getLogger()
+# 11数据库
+cnx_ = baseCore.cnx_
+cursor_ = baseCore.cursor_
+r = baseCore.r
+
+def search_att(social_code):
+    sql = """select name, full_path from clb_sys_attachment where item_id=%s and type_id=1"""
+    cursor_.execute(sql, social_code)
+    results = cursor_.fetchall()
+    return results
+
+
+
+if __name__ == "__main__":
+    while True:
+        item = baseCore.redicPullData('NianBao:socialcode')
+
+        # item = '9111000010001002XD|中国华能'
+        social_code = item.split('|')[0]
+        com_name = item.split('|')[1]
+        results = search_att(social_code)
+
+        if results:
+            pass
+        else:
+            log.info(f'{item}---没有年报附件')
+        for result in results:
+            file_name = result[0]
+            full_path = result[1]
+            # 创建文件夹
+            file_dir = f'D:\\kkwork\\企业数据\\年报附件\\{com_name}'
+            Path(file_dir).mkdir(parents=True, exist_ok=True)
+            file_name = file_name.replace(':', '：')
+            pdf_path = os.path.join(file_dir, file_name)
+
+            # 下载
+            req = requests.get(full_path)
+            if req.status_code == 200:
+                with open(pdf_path, 'wb') as f:
+                    f.write(req.content)
+                log.info(f'{file_name}下载完成')
+            else:
+                log.info(f'{item}下载失败')
+            # break
+        # break
+
+    pass