ParadeDB Ecto

A experimental downstream fork that extends Ecto with support for ParadeDB.

Installation

This repo should work as a mostly drop-in replacement for Ecto in any existing project that uses it and Postgresql:

def deps do
  [
    {:ecto, git: "https://github.com/moosieus/ecto", branch: "master"}
    {:ecto_sql, git: "https://github.com/moosieus/ecto_sql", branch: "master"}
  ]
end

Usage

Once installed, you'll need to create a search index within a migration as shown below. Tentatively, all search indexes must follow the #{schema}_search_idx naming convention.

defmodule Project.Repo.Migrations.MigrationName do
  use Ecto.Migration

  def up do
    execute("""
    CALL paradedb.create_bm25(
      index_name => 'mock_items_search_idx',
      table_name => 'mock_items',
      key_field => 'id',
      text_fields => paradedb.field('description'),
      numeric_fields => paradedb.field('rating')
    );
    """)
  end

  def down do
    execute("""
    CALL paradedb.drop_bm25(
      index_name => 'mock_items_search_idx',
      schema_name => 'public'
    )
    """)
  end
end

Afterwards, search queries can be performed via the search: and or_search: macros:

from(
  i in MockItems,
  search: parse(i, "description:shoes")
)

Hexdocs for this repo can be generated with MIX_ENV=docs mix docs.

Miscellaneous

Goals of this project:

• Figure out what a composable API for ParadeDB would look like in Ecto.
• Create an implementation that works, and accomodates those who especially want to use ParadeDB in Elixir projects.
• Identify ways to minimize the maintenance, packaging, and adoption burdens for a wider audience.
• Potentially serve as ParadeDB's ordained "ORM" for Elixir.

Why a downstream fork of Ecto?

Ecto provides extensibility in the form of its fragments API. Additionally Postgrex.Extension is available for adding additional PostgreSQL types to be encoded and decoded.

ParadeDB reuses all of Postgres' existing data types, so there is no need to define further types.

While fragments are useful for interpolating singular expressions and function calls, ParadeDB's queries more often than not require functionality well beyond the scope of what fragment offers.