如何在 Python 中对多行使用单个 INSERT INTO 语句？

Question

我目前正在开发一个 Discord Python 机器人，我会循环遍历 ForumTags 列表，并为每个对象生成一个

INSERT INTO

SQL 语句，以将数据插入 MySQL 数据库。

但是，我想通过将所有这些单独的

INSERT INTO

语句组合到单个查询中来优化我的代码，如下所示：

INSERT INTO guild_support_tags (guild_id, tag_id, tag_name, category) VALUES 
(123, 1, "test", "test_category"), (456, 2, "another tag", "test_category2)

这是我当前的代码：

start = time.time()
for tag in forum.available_tags:
    await write_query("INSERT INTO guild_support_tags (guild_id, tag_id, tag_name, category) VALUES "
                      "(:guild_id, :tag_id, :tag_name, :category)", 
                      {"guild_id": interaction.guild_id, "tag_id": tag.id, "tag_name": tag.name, 
                       "category": category})

    print(f"loop done after {time.time() - start}")


# from other file - created by myself to execute MySQL statements
# I would like a solution where this part is untouched. But if it can be improved, its okay.
async def write_query(query: str, params: dict) -> None:
    async with async_session() as session:
        async with session.begin():
            await session.execute(text(query), params)

也许了解一下是件好事：我目前在 MySQL 数据库上使用 SQLAlchemy 以及 aiomysql 和 Python3.12。

Answer 1

我发现我仍然可以使用 execute() 函数，并且通过将参数

params

类型从

dict

更改为

dict | list

，它有效。

SQLAlchemy 将检测到这一点并更改其行为以插入所有需要的行。 这种方法允许高效插入而无需循环：

# Prepare the data for the query
query_values: list[dict] = [{"guild_id": interaction.guild_id, "tag_id": tag.id, 
                             "tag_name": tag.name, "category": category} 
                            for tag in forum.available_tags]

# Modify the function to allow "list" and "dict" as type for "params"
async def write_query(query: str, params: dict | list) -> None:
    async with async_session() as session:
        async with session.begin():
            await session.execute(text(query), params)

# execute it
await write_query("INSERT INTO guild_support_tags (guild_id, tag_id, tag_name, category) 
                   VALUES (:guild_id, :tag_id, :tag_name, :category)", query_values)