feat: 支持从文件中读取官网数据库

mycode/main.py

@@ -111,43 +111,46 @@ def get_cbma_info_from_db_and_ana(year: str = '2023'):
     df = pd.read_sql_query(query_gzhs, conn)
     conn.close()
 
-    # 尝试连接官网库进行查询
+    file_path = os.path.join(BASE_DIR, f'cbma_site/{year}.xlsx')
-    import psycopg2
+    if os.path.exists(file_path):
-    conn_web = None
+        df_web = pd.read_excel(file_path, sheet_name='Result 1')
-    df_web = None
+    else:
-    try:
+        import psycopg2
-        conn_web = psycopg2.connect(
+        conn_web = None
-            "dbname={} user={} password={} host={} port={}".format('edn_cms', 'auditor', 'Lde78B3_cbma', '10.65.253.10', '54321'))
+        df_web = None
-        query_web = f"""
+        try:
-        SELECT
+            conn_web = psycopg2.connect(
-            a_outer.id,
+                "dbname={} user={} password={} host={} port={}".format('edn_cms', 'auditor', 'Lde78B3_cbma', '10.65.253.10', '54321'))
-            TO_CHAR(a_outer.ctime, 'YYYY') AS pub_year,
+            query_web = f"""
-            TO_CHAR(a_outer.ctime, 'MM') AS pub_month,
+            SELECT
-            TO_CHAR(a_outer.ctime, 'DD') AS pub_day,
+                a_outer.id,
-            a_outer.title,
+                TO_CHAR(a_outer.ctime, 'YYYY') AS pub_year,
-            a_outer.source,
+                TO_CHAR(a_outer.ctime, 'MM') AS pub_month,
-            a_outer.hits,
+                TO_CHAR(a_outer.ctime, 'DD') AS pub_day,
-            t.title as bankuai,
+                a_outer.title,
-            a_outer.src
+                a_outer.source,
-        FROM
+                a_outer.hits,
-            "a_article" a_outer
+                t.title as bankuai,
-        left join (
+                a_outer.src
-            select id, title, father, path
+            FROM
-            from a_article
+                "a_article" a_outer
-            where father in (20110528, 19080024)
+            left join (
-        ) t on a_outer.father = t.id
+                select id, title, father, path
-        WHERE
+                from a_article
-            a_outer.TYPE = 3
+                where father in (20110528, 19080024)
-            and a_outer.deleted is NULL
+            ) t on a_outer.father = t.id
-            and EXTRACT ( YEAR FROM a_outer.ctime ) = {year}
+            WHERE
-            and bankuai is not NULL
+                a_outer.TYPE = 3
-        ORDER BY
+                and a_outer.deleted is NULL
-            a_outer.ctime;
+                and EXTRACT ( YEAR FROM a_outer.ctime ) = {year}
-        """
+                and bankuai is not NULL
-        df_web = pd.read_sql_query(query_web, conn_web)
+            ORDER BY
-        conn_web.close()
+                a_outer.ctime;
-    except Exception as e:
+            """
-        print(e)
+            df_web = pd.read_sql_query(query_web, conn_web)
+            conn_web.close()
+        except Exception as e:
+            print(e)
 
     df['source'] = ''
     # 追加总院数据来源