鉴于以下声明:
p = db.query(Profile).options(joinedload('*')).filter_by(id=p.id).limit(1).one()
我将得到一个子查询+一个连接,而不是一个“纯”连接:
SELECT [...]
FROM (SELECT profile.id AS profile_id, ...
FROM profile
WHERE profile.id = %(id_1)s
LIMIT %(param_1)s) AS anon_1 LEFT OUTER JOIN city AS city_1 ON city_1.id = anon_1.profile_city LEFT OUTER JOIN country AS country_1 ON country_1.id = city_1.country LEFT OUTER JOIN state AS state_1 ON country_1.id = state_1.country LEFT OUTER JOIN state AS state_2 ON state_2.id = city_1.state LEFT OUTER JOIN country AS country_2 ON country_2.id = state_2.country LEFT OUTER JOIN state AS state_3 ON state_3.id = city_1.state LEFT OUTER JOIN country AS country_3 ON country_3.id = state_3.country LEFT OUTER JOIN starred AS starred_1 ON anon_1.profile_id = starred_1.star LEFT OUTER JOIN profiletext AS profiletext_1 ON anon_1.profile_id = profiletext_1.profile LEFT OUTER JOIN starred AS starred_2 ON anon_1.profile_id = starred_2.idprofile LEFT OUTER JOIN photo AS photo_1 ON anon_1.profile_id = photo_1.profile LEFT OUTER JOIN gps AS gps_1 ON anon_1.profile_id = gps_1.idprofile
但我真正需要的是:
SELECT ...
FROM profile LEFT OUTER JOIN city AS city_1 ON city_1.id = profile.city LEFT OUTER JOIN country AS country_1 ON country_1.id = city_1.country LEFT OUTER JOIN state AS state_1 ON country_1.id = state_1.country LEFT OUTER JOIN state AS state_2 ON state_2.id = city_1.state
LEFT OUTER JOIN country AS country_2 ON country_2.id = state_2.country LEFT OUTER JOIN state AS state_3 ON state_3.id = city_1.state LEFT OUTER JOIN country AS country_3 ON country_3.id = state_3.country LEFT OUTER JOIN starred AS starred_1 ON profile.id = starred_1.star LEFT OUTER JOIN profiletext AS profiletext_1 ON profile.id = profiletext_1.profile LEFT OUTER JOIN starred AS starred_2 ON profile.id = starred_2.idprofile LEFT OUTER JOIN photo AS photo_1 ON profile.id = photo_1.profile LEFT OUTER JOIN gps AS gps_1
ON profile.id = gps_1.idprofile
WHERE profile.id = 4
limit 1;
即。没有子查询。
数据库:postgresql 9.2
答案 0 :(得分:2)
使用联合预先加载时,如果查询包含修饰符 影响连接外部返回的行,例如使用时 DISTINCT,LIMIT,OFFSET或同等学历,完成的陈述是 首先包装在子查询中,并且连接专门用于 join eager loading应用于子查询。 SQLAlchemy加入了 急切的装载更进一步,然后再往前十英里 绝对确保它不会影响查询的最终结果, 只有集合和相关对象的加载方式,无论如何 查询的格式是什么。
我意识到这是一个老问题,但是生成的SQL不起作用的原因是什么?