Question

使用SQLAlchemy我想从子查询sq_objects重现可用的aggregate function jsonb_object_agg：

from sqlalchemy import select, func
s = select([
    sq_objects.c.object_id,
    func.jsonb_object_agg(
        sq_objects.c.keys, sq_objects.c.values).over(
        partition_by=sq_objects.c.object_id).label("attributes"),
]).\
    distinct(sq_objects.c.object_id)

但是，execute返回：

（psycopg2.ProgrammingError）无法适应类型＆＃39;方法＆＃39;

[ SQL: "SELECT DISTINCT ON (sq_objects.object_id)
     sq_objects.object_id,
     jsonb_object_agg(
         %(jsonb_object_agg_1)s,
         %(jsonb_object_agg_2)s
     ) OVER (PARTITION BY sq_objects.object_id) AS attributes
 FROM (SELECT ... ) AS sq_objects"
] [
parameters: {'jsonb_object_agg_1': <bound method Properties.keys of <sqlalchemy.sql.base.ImmutableColumnCollection object at 0x7f0ffb7aa828>>,
             'jsonb_object_agg_2': <bound method Properties.values of <sqlalchemy.sql.base.ImmutableColumnCollection object at 0x7f0ffb7aa828>>}]

这是我要重现的SQL代码：

SELECT DISTINCT ON (sq_objects.object_id)
    sq_objects.object_id,
    jsonb_object_agg(
        sq_objects.keys,
        sq_objects.values
    ) OVER (PARTITION BY sq_objects.object_id) AS attributes
FROM (SELECT ... ) AS sq_objects

Answer 1

不是最优雅的实现，但是因为我现在是子查询列的标签;原始text输入解决了问题：

func.jsonb_object_agg(
    text("sq_objects.keys"),
    text("sq_objects.values")
).over(
    partition_by=sq_objects.c.object_id).label("attributes"),

Answer 2

问题在于属性keys和values是不可变列集合sq_objects.c的方法。解决该问题的另一种方法是使用项目访问符号来获取实际的列：

func.jsonb_object_agg(
    sq_objects.c["keys"],
    sq_objects.c["values"]
).over(
    partition_by=sq_objects.c.object_id
).label("attributes")

JSONB sqlalchemy聚合函数

2 个答案: