dconger/algolia-elixir

Elixir implementation of Algolia search API

Algolia

This is the elixir implementation of Algolia search API, it is purely functional

Add to your dependencies

  defp deps do
    [{:algolia, "~> 0.6.4"}]
  end

(Pre-Elixir-1.4) Add :algolia to your applications

  def application do
    [applications: [:algolia]]
  end

Configuration

Using environment variables:

ALGOLIA_APPLICATION_ID=YOUR_APPLICATION_ID
ALGOLIA_API_KEY=YOUR_API_KEY

Using config:

config :algolia,
  application_id: YOUR_APPLICATION_ID,
  api_key: YOUR_API_KEY

NOTE: You must use ADMIN API_KEY instead of SEARCH API_KEY to enable write access

The Client

You don't need to initiate an index with this client unlike other OO Algolia clients. However, Most of the client search/write functions all use the syntax

operation(index, args....)

So you can easy emulate the index.function() syntax using piping

"my_index" |> operation(args)

Return values

All functions are serialized into maps before returning these responses

• {:ok, response}
• {:error, error_code, response}
• {:error, "Cannot connect to Algolia"}: The client implements retry strategy on all Algolia hosts with increasing timeout, It should only return this error when it has tried all 4 hosts. More Details here.

Examples

Searching

Searching an index

    "my_index" |> search("some query")

With Options

    "my_index" |> search("some query", %{attributesToRetrieve: "firstname", hitsPerPage: 20})

See all available search options here

Multiple queries at once

    multi([%{index_name => "my_index1", query: "search query"},
            %{index_name => "my_index2", query: "another query", hitsPerPage: 3,},
            %{index_name => "my_index3", query: "3rd query", tagFilters: "promotion"}])

You can specify a strategy to optimize your multiple queries

• :none: Execute the sequence of queries until the end.
• stop_if_enough_matches: Execute the sequence of queries until the number of hits is reached by the sum of hits.

    multi([query1, query2], strategy: :stop_if_enough_matches)

Saving

All save_* operations will overrides the object at the objectID

Save a single object to index without specifying objectID, must have objectID inside object, or use the id_attribute option (see below)

    "my_index" |> save_object(%{objectID: "1"})

Save a single object with a given objectID

    "my_index" |> save_object(%{title: "hello"}, "12345")

Save multiple objects to an index

    "my_index" |> save_objects([%{objectID: "1"}, %{objectID: "2"}])

Updating

Partially updates a single object

    "my_index" |> partial_update_object(%{title: "hello"}, "12345")

Update multiple objects, must have objectID in each object, or use the id_attribute option (see below)

    "my_index" |> partial_update_objects([%{objectID: "1"}, %{objectID: "2"}])

Partial update by default creates a new object if an object does not exist at the objectID, you can turn this off by passing false to the :upsert? option

    "my_index" |> partial_update_object(%{title: "hello"}, "12345", upsert?: false)
    "my_index" |> partial_update_objects([%{id: "1"}, %{id: "2"}], id_attribute: :id, upsert?: false)

Bonus for this Elixir client only: id_attribute option

All write functions such as save_object and partial_update_object comes with an id_attribute option that lets the you specifying an objectID from an existing field in the object, so you do not have to generate it yourself

    "my_index" |> save_object(%{id: "2"}, id_attribute: :id)

It also works for batch operations, such as save_objects and partial_update_objects

    "my_index" |> save_objects([%{id: "1"}, %{id: "2"}], id_attribute: :id)

However, this cannot be used together with an ID specifying argument together

    "my_index" |> save_object(%{id: "1234"}, "1234", id_attribute: :id)
    > Error

Wait for task

All write operations can be waited on by simply piping the response into wait/1

    "my_index" |> save_object(%{id: "123"}) |> wait

Since the client polls the server to check for publishing status, You can specify a time between each tick of the poll, the default is 100 ms

    "my_index" |> save_object(%{id: "123"}) |> wait(2_000)

You can also use the underlying wait_task function explicitly

    {:ok, %{"taskID" => task_id, "indexName" => index}}
      = "my_index" |> save_object(%{id: "123"}

    wait(index, task_id)

or with option

    wait(index, task_id, 1_000)

Index related operations

Listing all indexes

   list_indexes()

move_index/2

Moves an index to a new one

   move_index(source_index, destination_index)

copy_index/2

Copies an index to a new one

    copy_index(source_index, destination_index)

Clear an index

    clear_index(index)

Settings

get_settings/1

    get_settings(index)

Example response

{:ok,
  %{"minWordSizefor1Typo" => 4,
    "minWordSizefor2Typos" => 8,
    "hitsPerPage" => 20,
    "attributesToIndex" => nil,
    "attributesToRetrieve" => nil,
    "attributesToSnippet" => nil,
    "attributesToHighlight" => nil,
    "ranking" => [
        "typo",
        "geo",
        "words",
        "proximity",
        "attribute",
        "exact",
        "custom"
    ],
    "customRanking" => nil,
    "separatorsToIndex" => "",
    "queryType" => "prefixAll"}
}

set_settings/2

    set_settings(index, %{"hitsPerPage" => 20})

     > %{"updatedAt" => "2013-08-21T13:20:18.960Z",
        "taskID" => 10210332.
        "indexName" => "my_index"}

TODOS:

• get_object
• save_object
• save_objects
• update_object
• partial_update_object
• partial_update_objects
• delete_object
• delete_objects
• list_indexes
• clear_index
• wait_task
• wait (convenience function for piping response into wait_task)
• set_settings
• get_settings
• list_user_keys
• get_user_key
• add_user_key
• update_user_key
• delete_user_key