什么是正确的使用方法包括（Rails 4、ActiveRecord、PostgreSQL）

Question

我有以下型号：

月球 -> 行星 -> 恒星 -> 银河 -> 宇宙

月亮belongs_to行星，行星belongs_to星星等等：

---

class Moon < ActiveRecord::Base
  belongs_to :planet, inverse_of: :moons

  has_one :star, through: :planet
  has_one :galaxy, through: :star
  has_one :universe, through: :galaxy
end

---

class Planet < ActiveRecord::Base
    belongs_to :star, inverse_of: :planets
    has_many :moons, inverse_of: :planet, dependent: :destroy
end

---

问题

我试图让 Rails 将这些对象的层次结构加载到内存中。我正在努力实现两件事：

1. 使用一个高效的查询加载所有内容
2. 无需额外调用数据库即可使用 moon.galaxy 而不是 moon.planet.star.galaxy 等调用。

我尝试了两种方法。第一个（m1）在Moon 上调用includes，所有关系都变平了。这导致查询效率非常低，但我可以调用m1.galaxy。第二个 (m2) 使用关系层次结构调用 includes。这会产生高效的查询，但我可以调用m2.galaxy 而无需访问数据库。

这样做的正确方法是什么？

---

示例

m1 - 低效查询，m2 - 高效查询

irb(main):152:0* m1 = Moon.includes(:planet, :star, :galaxy).where(galaxies: {name: 'The Milky Way'}).first
  SQL (1.9ms)  SELECT  "moons"."id" AS t0_r0, "moons"."name" AS t0_r1, "moons"."planet_id" AS t0_r2, "moons"."created_at" AS t0_r3, "moons"."updated_at" AS t0_r4, "planets"."id" AS t1_r0, "planets"."name" AS t1_r1, "planets"."star_id" AS t1_r2, "planets"."created_at" AS t1_r3, "planets"."updated_at" AS t1_r4, "stars"."id" AS t2_r0, "stars"."name" AS t2_r1, "stars"."galaxy_id" AS t2_r2, "stars"."created_at" AS t2_r3, "stars"."updated_at" AS t2_r4, "galaxies"."id" AS t3_r0, "galaxies"."name" AS t3_r1, "galaxies"."universe_id" AS t3_r2, "galaxies"."created_at" AS t3_r3, "galaxies"."updated_at" AS t3_r4 FROM "moons" LEFT OUTER JOIN "planets" ON "planets"."id" = "moons"."planet_id" LEFT OUTER JOIN "planets" "planets_moons_join" ON "planets_moons_join"."id" = "moons"."planet_id" LEFT OUTER JOIN "stars" ON "stars"."id" = "planets_moons_join"."star_id" LEFT OUTER JOIN "planets" "planets_moons_join_2" ON "planets_moons_join_2"."id" = "moons"."planet_id" LEFT OUTER JOIN "stars" "stars_moons_join" ON "stars_moons_join"."id" = "planets_moons_join_2"."star_id" LEFT OUTER JOIN "galaxies" ON "galaxies"."id" = "stars_moons_join"."galaxy_id" WHERE "galaxies"."name" = $1  ORDER BY "moons"."id" ASC LIMIT 1  [["name", "The Milky Way"]]
=> #<Moon id: 1, name: "The Moon", planet_id: 1, created_at: "2015-10-28 10:02:04", updated_at: "2015-10-28 10:02:04">

irb(main):153:0> m2 = Moon.includes(planet: {star: :galaxy}).where(galaxies: {name: 'The Milky Way'}).first
  SQL (0.5ms)  SELECT  "moons"."id" AS t0_r0, "moons"."name" AS t0_r1, "moons"."planet_id" AS t0_r2, "moons"."created_at" AS t0_r3, "moons"."updated_at" AS t0_r4, "planets"."id" AS t1_r0, "planets"."name" AS t1_r1, "planets"."star_id" AS t1_r2, "planets"."created_at" AS t1_r3, "planets"."updated_at" AS t1_r4, "stars"."id" AS t2_r0, "stars"."name" AS t2_r1, "stars"."galaxy_id" AS t2_r2, "stars"."created_at" AS t2_r3, "stars"."updated_at" AS t2_r4, "galaxies"."id" AS t3_r0, "galaxies"."name" AS t3_r1, "galaxies"."universe_id" AS t3_r2, "galaxies"."created_at" AS t3_r3, "galaxies"."updated_at" AS t3_r4 FROM "moons" LEFT OUTER JOIN "planets" ON "planets"."id" = "moons"."planet_id" LEFT OUTER JOIN "stars" ON "stars"."id" = "planets"."star_id" LEFT OUTER JOIN "galaxies" ON "galaxies"."id" = "stars"."galaxy_id" WHERE "galaxies"."name" = $1  ORDER BY "moons"."id" ASC LIMIT 1  [["name", "The Milky Way"]]
=> #<Moon id: 1, name: "The Moon", planet_id: 1, created_at: "2015-10-28 10:02:04", updated_at: "2015-10-28 10:02:04">

m1 - 我可以在内存中调用moon.galaxy.name，m2 - moon.galaxy.name 调用DB

irb(main):154:0> m1.galaxy.name
=> "The Milky Way"

irb(main):155:0> m2.galaxy.name
  Galaxy Load (0.5ms)  SELECT  "galaxies".* FROM "galaxies" INNER JOIN "stars" ON "galaxies"."id" = "stars"."galaxy_id" INNER JOIN "planets" ON "stars"."id" = "planets"."star_id" WHERE "planets"."id" = $1 LIMIT 1  [["id", 1]]
=> "The Milky Way"

m1 - moon.planet.star.galaxy.name 调用 DB，m2 - 我可以在内存中调用 moon.planet.star.galaxy.name

irb(main):156:0> m1.planet.star.galaxy.name
  Star Load (3.3ms)  SELECT  "stars".* FROM "stars" WHERE "stars"."id" = $1 LIMIT 1  [["id", 1]]
  Galaxy Load (0.3ms)  SELECT  "galaxies".* FROM "galaxies" WHERE "galaxies"."id" = $1 LIMIT 1  [["id", 1]]
=> "The Milky Way"

irb(main):157:0> m2.planet.star.galaxy.name
=> "The Milky Way"

---

查询的差异

Answer 1

嗯，我认为您的第二种方式（m2）是正确的。 Bay 还可以考虑 eager_load，因为 includes 可能会触发多个 SQL 查询，而不是为每个关联（planet、start 等）触发一个查询。

我认为您的问题在于这些不正确的关联 star、galaxy 和 universe。

class Moon < ActiveRecord::Base
  belongs_to :planet, inverse_of: :moons

  has_one :star, through: :planet
  has_one :galaxy, through: :star
  has_one :universe, through: :galaxy
end

只需删除它们，然后重试。

请注意您不能以这种方式使用has_one (example in documentation)