为什么 SQL Alchemy ORM 的多对多 JOIN 对我来说效率低下？

Question

我运行正在我的一个类中跟踪查询，它采用 ORM 类以便它可以与几个类似的表一起工作。

(
            self.db.query(self.orm_contact_class)
            .options(
                load_only(
                    self.orm_contact_class.id,
                    self.orm_contact_class.name,
                    self.orm_contact_class.email_attempts_dict,
                ),
                joinedload(
                    self.orm_contact_class.__dict__[self.access_from_contact_to_company]
                ).load_only(self.orm_company_class.domain)
            )
            .where(
                self.orm_contact_class.email == None,
                self.orm_contact_class.name != None,
                self.orm_company_class.domain != None,
                catch_all_conditions
            )
        )

这导致了这个可怕的查询：

SELECT test.crunchbase_people.id AS test_crunchbase_people_id, 
       test.crunchbase_people.name AS test_crunchbase_people_name, 
       test.crunchbase_people.email_attempts_dict AS test_crunchbase_people_email_attempts_dict, 
       crunchbase_companies_1.id AS crunchbase_companies_1_id, 
       crunchbase_companies_1.domain AS crunchbase_companies_1_domain 
FROM test.crunchbase_companies, 
     test.crunchbase_people 
         LEFT OUTER JOIN (
             test.crunchbase_people_crunchbase_companies AS crunchbase_people_crunchbase_companies_1 
                 JOIN test.crunchbase_companies AS crunchbase_companies_1 
                 ON crunchbase_companies_1.id = crunchbase_people_crunchbase_companies_1.crunchbase_companies_id) 
             ON test.crunchbase_people.id = crunchbase_people_crunchbase_companies_1.crunchbase_people_id 
WHERE test.crunchbase_people.email IS NULL AND test.crunchbase_people.name IS NOT NULL AND test.crunchbase_companies.domain IS NOT NULL AND (test.crunchbase_companies.is_domain_catch_all = false OR test.crunchbase_companies.is_domain_catch_all IS NULL)

它最终永远不会完成，如果我在 Postgres 控制台中运行它只是 returns 完全相同的行的重复，一遍又一遍！

所以它永远不会映射对象，因为查询会永远持续下去。在没有 ORM 的情况下，有一种简单的方法可以做到这一点，查询运行s 在 0.5 秒内完成（并且没有像上面那样的重复项），但是我的对象没有映射，这会导致我重构很多代码。

有人知道这样的查询可能有什么问题吗？

Answer 1

正如评论员对原始 post 的建议，我缺少一个显式连接，因为 joinedload 不会替换它，而只是告诉应该预加载哪些字段。

(
            self.db.query(self.orm_contact_class)
            .join(self.orm_contact_class.__dict__[self.access_from_contact_to_company]) # added this line
            .options(
                load_only(
                    self.orm_contact_class.id,
                    self.orm_contact_class.name,
                    self.orm_contact_class.email_attempts_dict,
                ),
                joinedload(
                    self.orm_contact_class.__dict__[self.access_from_contact_to_company]
                ).load_only(self.orm_company_class.domain)
            )
            .where(
                self.orm_contact_class.email == None,
                self.orm_contact_class.name != None,
                self.orm_company_class.domain != None,
                catch_all_conditions
            )
        )

为什么 SQL Alchemy ORM 的多对多 JOIN 对我来说效率低下？

Why is SQL Alchemy ORM's Many to Many JOIN this inefficient for me?

python

sql

sqlalchemy