带有分页的Rails复杂过滤器

Question

我正在做一个项目，我必须在其中展示组。

涉及以下模型

组

class Group < ApplicationRecord
  has_many :user_groups, dependent: :destroy
  has_many :users, through: :user_groups

  belongs_to :creator, :class_name => "User", :foreign_key => "creator_id"

  belongs_to :group_status

  has_many :matches, dependent: :destroy
  has_many :liked, :through => :matches, :source => :macther

  has_many :inverse_matches , :class_name => "Match" , :foreign_key => "matcher_id", dependent: :destroy
  has_many :liked_by, :through => :inverse_matches, :source => :group

  enum status: %w[active inactive]

  reverse_geocoded_by :latitude, :longitude
end

用户

class User < ApplicationRecord
  
  has_one :setting, dependent: :destroy

  has_many :friendships, dependent: :destroy #, :class_name => "Friendship", :foreign_key => 'user_id'
  has_many :invited, :through => :friendships, :source => :friend

  has_many :inverse_friendships , :class_name => "Friendship" , :foreign_key => "friend_id", dependent: :destroy
  has_many :invited_by, :through => :inverse_friendships, :source => :user

  has_many :user_groups, dependent: :destroy
  has_many :groups, through: :user_groups

  has_many :group_creations, :foreign_key => :creator_id, :class_name => 'Group'
  has_many :group_statuses

  
  has_many :matches, dependent: :destroy
  
  enum gender: %w[male female other]
end

用户组

class UserGroup < ApplicationRecord
    belongs_to :user
    belongs_to :group

    enum status: %w[pending joined reported]
    attr_accessor :invite_id
end

匹配

class Match < ApplicationRecord
  belongs_to :group
  belongs_to :matcher, :class_name => "Group"

  belongs_to :user
  enum status: %w[like match unlike block blocked report reported]
end

首先根据用户的年龄、性别和组员人数等偏好显示组，然后显示其余用户。

我所做的是首先排除所有不符合用户偏好的组并显示剩余的组，而不是显示那些不符合用户偏好的组。

这就是算法的样子

1- 查找与群组成员计数不匹配的群组

group_member_count_check = Group.joins(:users).joined_group.group('groups.id')
                                        .having("count(groups.id) != #{setting.group_member_count}").pluck(:id)

2- 查找不符合性别偏好的组

if setting.male_only
  gender_check = Group.joins(:users).joined_group.where('users.gender !=?', 0).pluck(:id)
elsif setting.female_only
  gender_check = Group.joins(:users).joined_group.where('users.gender !=?', 1).pluck(:id)
end

3- 比删除当前用户包含被阻止用户的组

blocked_users = current_user.friendships.where(friendships: {status: 2}).pluck(:friend_id) + 
                current_user.inverse_friendships.where(friendships: {status: 2}).pluck(:user_id)
blocked_group_ids = Group.joins(:users).joined_group.where(users: {id: blocked_users}).pluck(:id)

4- 删除已经喜欢/不喜欢的组

already_liked_or_matched_id = (group.matches.where(status: [0,1,3,4,5,6]).pluck(:matcher_id)  
      + current_user.matches.where(status: [2]).pluck(:matcher_id)).uniq

现在首先返回不包括上述提取组的组

groups = Group.includes(:group_status, :users)
                      .where(status: 'active', spot_light_enabled: false)
                      .where.not(id: group_member_count_check +  gender_check +blocked_group_ids + already_liked_or_matched_id  )
                      .near([latitude, longitude], setting.max_distace, order: 'random()').page(params[:page]).per(params[:per_page])

如果没有首选组，则显示其他组

if groups.length.zero?
          groups = Group.includes(:group_status, :users).where(status: 'active')
                              .where(id: gender_check + group_member_count_check)
                              .where.not(id: blocked_group_ids + already_liked_or_matched_id )
                              .near([latitude, longitude], 10000, order: 'random()').page(params[:page]).per(params[:per_page])
        end

回复是这样的

{ 
   groups: groups, 
   pagination: pagination_dict(Group.where.not(id: blocked_group_ids ).page(params[:page]).per(params[:per_page]))
}

架构

  create_table "users", id: :serial, force: :cascade do |t|
    t.string "phone_number"
    t.string "user_name"
    t.string "authentication_token"
    t.string "first_name"
    t.string "last_name"
    t.text "bio"
    t.date "dob"
    t.integer "age"
    t.integer "gender"
    t.integer "group_id"
    t.decimal "latitude", precision: 15, scale: 6
    t.decimal "longitude", precision: 15, scale: 6
    t.string "city"
    t.string "state"
    t.boolean "creator", default: false
    t.boolean "is_premium", default: false
    t.datetime "created_at", null: false
    t.datetime "updated_at", null: false
    t.integer "instant_match_allow", default: 1
    t.integer "spot_light_allow", default: 1
    t.boolean "is_blocked", default: false
    t.boolean "is_new", default: true
    t.index ["authentication_token"], name: "index_users_on_authentication_token"
    t.index ["phone_number"], name: "index_users_on_phone_number", unique: true
    t.index ["user_name"], name: "index_users_on_user_name", unique: true
  end

  create_table "groups", force: :cascade do |t|
    t.integer "group_status_id"
    t.string "name"
    t.integer "status", default: 0
    t.decimal "latitude", precision: 15, scale: 6
    t.decimal "longitude", precision: 15, scale: 6
    t.string "city"
    t.string "state"
    t.datetime "created_at", null: false
    t.datetime "updated_at", null: false
    t.integer "creator_id"
    t.integer "instant_match_allow", default: 1
    t.integer "spot_light_allow", default: 1
    t.boolean "spot_light_enabled", default: false
    t.datetime "spot_light_time"
    t.integer "place_id"
    t.index ["group_status_id"], name: "index_groups_on_group_status_id"
  end

  create_table "user_groups", force: :cascade do |t|
    t.integer "user_id"
    t.integer "group_id"
    t.boolean "creator", default: false
    t.integer "status", default: 0
    t.datetime "created_at", null: false
    t.datetime "updated_at", null: false
  end

  create_table "matches", force: :cascade do |t|
    t.integer "group_id"
    t.integer "matcher_id"
    t.integer "status"
    t.integer "user_id"
    t.datetime "created_at", null: false
    t.datetime "updated_at", null: false
    t.integer "chat_id"
  end

现在的问题

1- 主要是查询太多（我必须进行更多检查但排除已显示的组等）

2- 分页没有按预期工作（由于从首选组切换到其他组而跳过一些页面）

管理此问题的首选方法是什么？

Answer 1

为了快速重构，让我们考虑命名范围和以下方法。

• 首先，按用户偏好查找组
• 通过排除首选组来查找其他组
• 连接两个组并应用分页

例如，Group 类可能具有以下范围。 请注意，我无法对其进行测试，您可能仍需要重构以使其正常工作。

class Group < ApplicationRecord
  delegate :setting, to: :user, prefix: true # for accessing setting with `user_setting` method

  scope :status_active, -> { where(status: 'active') }
  scope :spot_light_disalbed, -> { where(spot_light_enabled: false) }
  
  scope :filter_blocked_groups, ->(blocked_users_id) { where.not(users: { id: blocked_users_id }) }
  scope :filter_by_gender_setting, -> { where('users.gender == ?', user_setting.male_only ? 0 : 1) }
  scope :filter_by_group_member_count, -> { group('groups.id').having("count(groups.id) == #{user_setting.group_member_count}") }
  
  scope :paginate, ->(page, per_page) { page(page).per(per_page) }
  scope :near_groups, ->(distance) { near([latitude, longitude], distance, order: 'random()') }
end

# Find groups by user preference
preferred_groups = Group.includes(:group_status, :users).joins(:users).joined_group
         .status_active
         .spot_light_disalbed
         .filter_blocked_groups(blocked_users_id)
         .filter_by_gender_setting
         .filter_by_group_member_count
         .near_groups(setting.max_distace)

# Find other groups by excluding the preferred groups 
other_groups = Group.includes(:group_status, :users).joins(:users).joined_group
               .where.not(id: preferred_groups.ids)
               .near_groups(10_000)

# Concat the two groups and apply pagination
groups = (preferred_groups + other_groups).paginate(params[:page], params[:per_page])

我没有包含already_liked_or_matched_id 查询，因为我不清楚该查询。你也可以添加它。上述方法的好处是..

• 我们只对groups 表进行两次查询
• 可读性

希望这会有所帮助。我相信还有更多的优化空间。您还可以考虑通过将所有这些查询移至单独的类来实现 query pattern。