Postgresql - 使用来自SELECT的sqlalchemy的INSERT插入到不存在的地方

Question

正如在此指出的那样,可以使用postgresql 9.1+进行以下操作

INSERT INTO example_table
    (id, name)
SELECT 1, 'John'
WHERE
    NOT EXISTS (
        SELECT id FROM example_table WHERE id = 1
    );

我一直在玩sqlalchemy的0.9版本,他们从SELECT方法引入了INSERT,理论上应该处理上述问题.

有可能,如果可能,怎么样？(因为我想利用在使用原始sql时不返回的result.inserted_primary_key)

我如何使用bindparams作为'from_select'部分,因为我可以使用它的唯一方法是在select中使用表列.

例如

insrt = example_table.insert().
    from_select(['id', 'name'],
    example_table.select().
    where(~exists(select([example_table.c.id],
    example_table.c.id == 1))))

result = session.execute(insrt)

if result.is_insert:
    print 'do something with result.inserted_primary_key'

Answer 1

from sqlalchemy import *

"""
INSERT INTO example_table
    (id, name)
SELECT 1, 'John'
WHERE
    NOT EXISTS (
        SELECT id FROM example_table WHERE id = 1
    );
"""

m = MetaData()

example_table = Table("example_table", m,
                        Column('id', Integer),
                        Column('name', String)
                    )

sel = select([literal("1"), literal("John")]).where(
           ~exists([example_table.c.id]).where(example_table.c.id == 1)
      )

ins = example_table.insert().from_select(["id", "name"], sel)
print(ins)

输出:

INSERT INTO example_table (id, name) SELECT :param_1 AS anon_1, :param_2 AS anon_2 
WHERE NOT (EXISTS (SELECT example_table.id 
FROM example_table 
WHERE example_table.id = :id_1))