SQLAlchemy：使用子查询和内部联接从sql查询创建sqlalchemy查询

Question

我是后端开发和python的新手。 我正在从具有子查询和内部联接的SQL查询创建SQLAlchemy查询。

我正在使用python> 3.5，“ oracle + cx_oracle”驱动程序和SQLAlchemy == 1.3.6

我需要转换的SQL查询是：

SELECT DISTINCT d.*, t.max_date, t.USER_ID
FROM DATABASE_ONE d
INNER JOIN (
    SELECT USER_ID, MAX(DATE_CREATED) max_date FROM (
        SELECT * FROM DATABASE_ONE
        WHERE USER_ID IN ('1','2','3')
        AND CATEGORY = 'Users'
        AND IS_DELETED IS NULL
    )
    GROUP  BY USER_ID
) t on t.USER_ID = d.USER_ID AND t.max_date = d.DATE_CREATED

下面是我尝试将上述SQL查询转换为SQLAlchemy查询的代码。

    # --------------------- first query -------------------

    first_query = (
        Session.query(DatabaseOne)
        .filter(
            DatabaseOne.user_id.in_('1', '2', '3'),
            DatabaseOne.category == "Obligor Rating Scorecards",
            DatabaseOne.is_deleted == None,
        )
        .subquery("first_query")
    )

    # -------------------- second query --------------------

    second_query = (
        Session.query(
            DatabaseOne.user_id,
            func.max(DatabaseOne.date_created).label("max_date"),
        )
        .select_from(first_query)
        .group_by(DatabaseOne.user_id)
        .subquery("second_query")
    )

    # -------------------- final query --------------------

    final_query = (
        Session.query(
            DatabaseOne, second_query.c.max_date, second_query.c.user_id
        )
        .join(
            second_query,
            and_(
                second_query.c.user_id == DatabaseOne.user_id,
                DatabaseOne.date_created == second_query.c.max_date,
            ),
        )
        .distinct(DatabaseOne)
        .all()
    )

• 我能够在first_query上获得预期的结果。
• second_query无法给出预期的结果。
• 我不确定final_query是否正确。

Answer 1

根据您的SQL，我认为您正在尝试查询提供了user_id且具有最大日期的行。 所以首先也许我们可以稍微更新一下SQL：

SELECT DISTINCT d.*, t.max_date, t.USER_ID
FROM DATABASE_ONE d
INNER JOIN (
    SELECT USER_ID, MAX(DATE_CREATED) max_date 
    FROM DATABASE_ONE
    WHERE 
        USER_ID IN ('1','2','3')
        AND CATEGORY = 'Users'
        AND IS_DELETED IS NULL
    GROUP  BY USER_ID
) t on t.USER_ID = d.USER_ID AND t.max_date = d.DATE_CREATED

在这种情况下，更容易阅读并转换为SQLAlchemy语言。

---

然后，关于第二个SQLAlchemy查询的问题是，您尝试从子查询中进行选择，但是选择是从原始表中进行选择。 如果将第二个SQLAlchemy查询转换为sql查询，则如下所示：

SELECT DATABASE_ONE.USER_ID, MAX(DATABASE_ONE.DATE_CREATED) max_date FROM (
    SELECT * FROM DATABASE_ONE
    WHERE USER_ID IN ('1','2','3')
    AND CATEGORY = 'Users'
    AND IS_DELETED IS NULL
)
GROUP  BY DATABASE_ONE.USER_ID

编写SQLAlchemy查询的更好方法：

    subquery = Session.query(
        DatabaseOne.user_id,
        func.max(DatabaseOne.date_created).label("max_date"),
    )
    .filter(
        DatabaseOne.user_id.in_('1', '2', '3'),
        DatabaseOne.category == "Obligor Rating Scorecards",
        DatabaseOne.is_deleted == None,
    )
    .group_by(DatabaseOne.user_id)
    .subquery()

    Session.query(
        DatabaseOne, subquery.c.user_id, subquery.c.max_date # see the .c method to access column in subquery
    ).select_from(subquery).join(DatabaseOne).filter(DatabaseOne.user_id = subquery.c.user_id,DatabaseOne.date_created = subquery.c.max_date).all()