Alibaba Tablestore adapter for Ecto

Module Version Hex Docs Total Download License Last Updated

Ecto 3.x adapter for Alibaba Tablestore, this is built on top of ex_aliyun_ots to implement Ecto.Adapter and Ecto.Adapter.Schema behaviours.

Supported features:

  • Compatible Ecto.Repo API.
  • Support schema's timestamps() macro, make inserted_at and updated_at as an integer UTC timestamps.
  • Support :map | {:map, _} | :array | {:array, _} field type, use Jason to encode the field value into :string when save, use Jason to simply decode the field value into :map | :array when read, the :keys option of Jason's decode always use :string.
  • Support the partition key is autoincrementing, use the sequence feature provided by ex_aliyun_ots.
  • Automatically converts the returned original row results into corresponding schema(s).
  • Automatically generate the provided attribute-column field(s) of schema entity into the filter expression option of GetRow (see EctoTablestore.Repo.one/2) and BatchGet (see EctoTablestore.Repo.batch_get/1) use entity_full_match: true, by default this option is false.
  • Automatically generate the provided attribute-column field(s) of schema entity into the condition expression option of BatchWrite.
  • Automatically map changeset's attribute-column field(s) into UpdateRow operation when call EctoTablestore.Repo.update/2:
    • Use atomic increment via {:increment, integer()} in changeset, and return the increased value in the corresponding field(s) by default;
    • Set any attribute(s) of schema changeset as nil will :delete_all that attribute-column field(s);
    • Set existed attribute(s) in schema changeset will :put to save.

Implement Tablestore row related functions in EctoTablestore.Repo module, please see document for details:

  • PutRow
  • GetRow
  • UpdateRow
  • DeleteRow
  • GetRange
  • StreamRange
  • BatchGetRow
  • BatchWriteRow
  • Search

Migration

Provide a simple migration to create table, please see EctoTablestore.Migration for details.

Usage

1, Configure My instance(s) information of Alibaba Tablestore product.

use Mix.Config

# config for `ex_aliyun_ots`

config :ex_aliyun_ots, MyInstance,
  name: "MY_INSTANCE_NAME",
  endpoint: "MY_INSTANCE_ENDPOINT",
  access_key_id: "MY_OTS_ACCESS_KEY",
  access_key_secret: "MY_OTS_ACCESS_KEY_SECRET"

config :ex_aliyun_ots,
  instances: [MyInstance]

# config for `ecto_tablestore`

config :my_otp_app, EctoTablestore.MyRepo,
  instance: MyInstance

2, Create the EctoTablestore.MyRepo module mentioned earlier in the configuration, use EctoTablestore.Repo and set required otp_app option with your OTP application's name.

defmodule EctoTablestore.MyRepo do
  use EctoTablestore.Repo,
    otp_app: :my_otp_app
end

3, Each repository in Ecto defines a start_link/0 function that needs to be invoked before using the repository. In general, this function is not called directly, but used as part of your application supervision tree.

Add EctoTablestore.MyRepo into your application start callback that defines and start your supervisor. You just need to edit start/2 function to start the repo as a supervisor on your application's supervisor:

def start(_type, _args) do
  children = [
    {EctoTablestore.MyRepo, []}
  ]

  opts = [strategy: :one_for_one, name: MyApp.Supervisor]
  Supervisor.start_link(children, opts)
end

4, After finish the above preparation, we can use MyRepo to operate data store.

Integrate Hashids

Base on unique integer of atomic-increment sequence, provides a way to simply integrate Hashids to generate your hash ids when insert row(s), for Repo.insert/2 or Repo.batch_write/1.

Use in schema

defmodule Module do
  use EctoTablestore.Schema

  tablestore_schema "table_name" do
    field(:id, :hashids, primary_key: true, autogenerate: true,
      hashids: [salt: "123", min_len: 2, alphabet: "..."])
    field(:content, :string)
  end

end

Options:

  • The primary_key as true is required;
  • The autogenerate as true is required;
  • The salt, min_len, and alphabet of hashids option is used for configuration options from Hashids.new/1.

Use in migration

Use :hashids as a type in add operation to define the partition key, the principle behind this will use ecto_tablestore_default_seq as a global default table to maintain the sequences of all tables, if ecto_tablestore_default_seq is not existed, there will create this, if it is existed, please ignore the "OTSObjectAlreadyExist" error of requested table already exists.

defmodule EctoTablestore.TestRepo.Migrations.TestHashids do
  use EctoTablestore.Migration

  def change do
    create table("table_name") do
      add :id, :hashids, partition_key: true, auto_increment: true
      add :oid, :integer
    end
  end

end

Options:

  • The partition_key as true is required;
  • The auto_increment as true is required.

References

Alibaba Tablestore product official references:

License

This project is licensed under the MIT license. Copyright (c) 2018- Xin Zou.