Ruby Sequel switch databases on the fly

_problem.md

1. I wanted to keep global user data in one database and org data in separate database
2. I want to use Sequel models in the same way as I was using a single databases
3. I need simple interface to define that model data resides in Organization and not Global DB

Only possible solution I found is by patching Sequel native methods. If anyone knows better way to do it, like injecting DB connection just before dataset executes a query, please comment and explain.

I tried to use "sharding" and "arbitrary_servers", but that is not what I want. I have 1000nds of orgs, and 1000nds of DBs. I need connections to be fuly dynamic and I want to define request context once when request starts in Thread.current.

code.rb

module Sequel::Plugins::OrgData
  module ClassMethods
    def __modelize data
      new.tap do |model|
        model.instance_exec { @values = data }
      end
    end

    def __conn
      Thread.current[:org_db]
    end

    def find id
      __modelize __conn[to_s.tableize.to_sym].where(id:id).first
    end
  end

  module DatasetMethods
    def count
      custom = sql.sub 'SELECT * ', 'select count(id) as cnt '
      model.__conn.fetch(custom).first[:cnt]
    end

    def each &block
      model.__conn.fetch(sql).each do |row|
        block.call self.model.__modelize(row)
      end
    end
  end
end

###

# User and Invoice data is in org database shards
User.plugin :org_data
Invoice.plugin :org_data

###

org = Org[123]     # comes from default DB, "postgres://some_server:5432/app_data"
inv = org.invoices # comes from ORG DB,     "postgres://some_server:5432/app_org_data_123"

###

def get_it
  puts [
    User.find(1).name,
    User.where(org_id: 123).count,
    User.first(3).map(&:name).join(', ')
  ].join(' --- ')
end

Thread.current[:org_db] = Sequel.connect('postgres://localhost:5432/app_org_data_1')
get_it

Thread.current[:org_db] = Sequel.connect('postgres://localhost:5432/app_org_data_2')
get_it