用pyodbc输入merge语句时出现无限循环

Question

我目前正在使用 AWS Lambda 实现一个 API，将数据插入 SQL Server 数据库。 我正在使用Python 3.11。 代码执行成功并返回200状态码， 但是当我在 SSMS 中运行 SELECT 查询时， 它会导致无限执行。 我需要帮助来解决这个问题。

这是我的代码：

conn_str = (
        f'Driver={{ODBC Driver 17 for SQL Server}};'
        f'Database=TMP;'
        f'Encrypt=no;'
    )
    data = [{
        "sfid" : "aws1123124",
        "lastmodifieddate" : "a551231",
        "comments" : "a11231",
        "keyword" : "a41231231",
        "productcode" : "s55234234",
        "serialNumber" : "d1123123"
    }]
    conn = pyodbc.connect(conn_str)
    cursor = conn.cursor()
    cursor.execute("BEGIN TRANSACTION")
    try :
        for record in data:
                print("in record")
                print(record['sfid'])
                merge_sql = """
                MERGE INTO [dbo].[LocalDB] AS Target
                USING (SELECT ? AS sfid, ? AS lastmodifieddate, ? AS comments, ? AS keyword, ? AS productcode, ? AS serialNumber) AS Source
                ON Target.sfid = Source.sfid
                WHEN MATCHED THEN
                    UPDATE SET 
                        lastmodifieddate = Source.lastmodifieddate, 
                        comments = Source.comments, 
                        keyword = Source.keyword, 
                        productcode = Source.productcode, 
                        serialNumber = Source.serialNumber
                WHEN NOT MATCHED THEN
                    INSERT (sfid, lastmodifieddate, comments, keyword, productcode, serialNumber)
                    VALUES (Source.sfid, Source.lastmodifieddate, Source.comments, Source.keyword, Source.productcode, Source.serialNumber);
                """
                cursor.execute(merge_sql, 
                               record['sfid'], 
                               record['lastmodifieddate'], 
                               record['comments'], 
                               record['keyword'],
                               record['productcode'],
                               record['serialNumber'])
        conn.commit()
    except Exception as e:
        cursor.execute("ROLLBACK")
        print(f"An error occurred: {e}")
        raise e
    finally:
        if cursor is not None:
            try:
                cursor.close()
            except Exception as ex:
                print(f"Error closing cursor: {ex}")
        
        if conn is not None:
            try:
                conn.close()
            except Exception as ex:
                print(f"Error closing connection: {ex}")

    # TODO implement
    return {
        'statusCode': 200,
        'body': json.dumps('Hello from Lambda!')
    }

Answer 1

我只是更改了您的代码以删除

cursor.execute("BEGIN TRANSACTION")

和

cursor.execute("ROLLBACK")

，然后在

cursor.commit()

中添加

try

，没有任何问题。我还按照链接的 文章列表 中的建议，向

HOLDLOCK

添加了

MERGE

提示（不过，最好还是使用 upsert）。这是有意压缩的，但这实际上变成了：

conn = pyodbc.connect(conn_str)
cursor = conn.cursor()
try :
    for record in data:
        print("in record")
        print(record['sfid'])
        merge_sql = """
        MERGE INTO [dbo].[LocalDB] WITH (HOLDLOCK) AS Target 
        ...
        """
        cursor.execute(merge_sql, 
                       record['sfid'], 
                       record['lastmodifieddate'], 
                       record['comments'], 
                       record['keyword'],
                       record['productcode'],
                       record['serialNumber'])
        cursor.commit()
except Exception as e:
    print(f"An error occurred: {e}")
    raise e
finally:
...