Skip to content

The official Ruby API for Scale AI, the data platform for AI.

License

Notifications You must be signed in to change notification settings

scaleapi/scaleapi-ruby

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

72 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Scale

Scale Logo

This is the official Scale RubyGem (scaleapi).

Scale is an API for Human Intelligence. Businesses like Alphabet (Google), Uber, Proctor & Gamble, Houzz, and many more use us to power tasks such as:

  • Draw bounding boxes and label parts of images (to train ML algorithms for self-driving cars)
  • Transcribe documents, images, and webpages
  • Scrape websites
  • Triage support tickets
  • Categorize and compare images, documents, and webpages

Scale is actively hiring software engineers - apply here.

Installation

Add this line to your application's Gemfile:

gem 'scaleapi'

And then execute:

$ bundle

Or install it yourself as:

$ gem install scaleapi

Usage

First, initialize the Scale client:

require 'scale'

scale = Scale.new(api_key: 'SCALE_API_KEY')

Note that you can optionally provide a callback_auth_key and callback_url when initializing the Scale client. You can also set default_request_params which is a Hash that will be included in every request sent to Scale (either as a query string param or part of the request body).

If you're having trouble finding your API Key or Callback Auth Key, then go to the Scale Dashboard. If you set a default callback_url in your account settings, you won't need to pass it in everytime.

Creating Tasks

This gem supports two ways of creating tasks. You can call create_#{tasktype}_task on the scale object or you can call scale.tasks.create and pass in the corresponding type. Upon success, it will return the appropriate object for that task type. Upon failure, it will raise an application-level error.

For every type of task, you can pass in the following options when creating:

  • callback_url: a URL to send the webhook to upon completion. This is required when there is no default callback URL set when either initializing the Scale object or in your account settings.
  • urgency: a string indicating how long the task should take, options are immediate, day, or week. The default is day.
  • metadata: a Hash that contains anything you want in it. Use it for storing data relevant to that task, such as an internal ID for your application to associate the task with. Note that the keys of the Hash will be returned as String rather than Symbol.

Categorization Tasks

To create a categorization task, run the following:

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

scale.create_categorization_task({
  callback_url: 'http://www.example.com/callback',
  instruction: 'Is this company public or private?',
  attachment_type: 'website',
  attachment: 'https://www.google.com',
  categories: ['public', 'private']
})

Upon success, this will return a Scale::Api::Tasks::Categorization object. It will raise one of the errors if it's not successful.

Alternatively, you can also create a task this way

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

scale.tasks.create({
  type: 'categorization',
  callback_url: 'http://www.example.com/callback',
  instruction: 'Is this company public or private?',
  attachment_type: 'website',
  attachment: 'https://www.google.com',
  categories: ['public', 'private']
})

This will also return a Scale::Api::Tasks::Categorization object.

Read more about creating categorization tasks

Comparison Tasks

To create a comparison task, run the following:

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

scale.create_comparison_task({
  callback_url: 'http://www.example.com/callback',
  instruction: 'Do the objects in these images have the same pattern?',
  attachments: [
    'http://i.ebayimg.com/00/$T2eC16dHJGwFFZKjy5ZjBRfNyMC4Ig~~_32.JPG',
    'http://images.wisegeek.com/checkered-tablecloth.jpg'
  ],
  attachment_type: 'image',
  choices: ['yes', 'no']
})

Upon success, this will return a Scale::Api::Tasks::Comparison object. If it fails, it will raise one of the errors.

Alternatively, you can also create a task this way

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

scale.tasks.create({
  type: 'comparison',
  callback_url: 'http://www.example.com/callback',
  instruction: 'Do the objects in these images have the same pattern?',
  attachments: [
    'http://i.ebayimg.com/00/$T2eC16dHJGwFFZKjy5ZjBRfNyMC4Ig~~_32.JPG',
    'http://images.wisegeek.com/checkered-tablecloth.jpg'
  ],
  attachment_type: 'image',
  choices: ['yes', 'no']
})

This will also return a Scale::Api::Tasks::Comparison object.

Read more about creating comparison tasks

Datacollection Tasks

To create a datacollection task, run the following:

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

scale.create_datacollection_task({
  callback_url: 'http://www.example.com/callback',
  instruction: 'Find the URL for the hiring page for the company with attached website.',
  attachment: 'https://scale.com/',
  attachment_type: 'website',
  fields: {
    hiring_page: 'Hiring Page URL'
  }
})

Upon success, this will return a Scale::Api::Tasks::Datacollection object. If it fails, it will raise one of the errors.

Alternatively, you can also create a task this way

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

scale.tasks.create({
  type: 'datacollection',
  callback_url: 'http://www.example.com/callback',
  instruction: 'Find the URL for the hiring page for the company with attached website.',
  attachment: 'https://scale.com/',
  attachment_type: 'website',
  fields: {
    hiring_page: 'Hiring Page URL'
  }
})

This will also return a Scale::Api::Tasks::Datacollection object.

Read more about creating datacollection tasks

Image Recognition Tasks

To create an image recognition task, run the following:

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

scale.create_annotation_task({
  callback_url: 'http://www.example.com/callback',
  instruction: 'Draw a box around each **baby cow** and **big cow**',
  attachment_type: 'image',
  attachment: 'http://i.imgur.com/v4cBreD.jpg',
  objects_to_annotate: ['baby cow', 'big cow'],
  with_labels: true,
  examples: [
    {
      correct: true,
      image: 'http://i.imgur.com/lj6e98s.jpg',
      explanation: 'The boxes are tight and accurate'
    },
    {
      correct: false,
      image: 'http://i.imgur.com/HIrvIDq.jpg',
      explanation: 'The boxes are neither accurate nor complete'
    }
  ]
})

Upon success, this will return a Scale::Api::Tasks::ImageRecognition object. If it fails, it will raise one of the errors.

Note: create_annotation_task is also aliased to create_image_recognition_task, to help avoid confusion.

Alternatively, you can also create a task this way

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

scale.tasks.create({
  type: 'annotation',
  callback_url: 'http://www.example.com/callback',
  instruction: 'Draw a box around each **baby cow** and **big cow**',
  attachment_type: 'image',
  attachment: 'http://i.imgur.com/v4cBreD.jpg',
  objects_to_annotate: ['baby cow', 'big cow'],
  with_labels: true,
  examples: [
    {
      correct: true,
      image: 'http://i.imgur.com/lj6e98s.jpg',
      explanation: 'The boxes are tight and accurate'
    },
    {
      correct: false,
      image: 'http://i.imgur.com/HIrvIDq.jpg',
      explanation: 'The boxes are neither accurate nor complete'
    }
  ]
})

This will also return a Scale::Api::Tasks::ImageRecognition object.

Read more about creating image recognition tasks

Transcription Tasks

To create a transcription task, run the following:

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

scale.create_transcription_task({
  callback_url: 'http://www.example.com/callback',
  instruction: 'Transcribe the given fields.',
  attachment_type: 'website',
  attachment: 'http://news.ycombinator.com/',
  fields: {
    title: 'Title of Webpage',
    top_result: 'Title of the top result'
  }
})

Upon success, this will return a Scale::Api::Tasks::Transcription object. If it fails, it will raise one of the errors.

Alternatively, you can also create a task this way

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

scale.tasks.create({
  type: 'transcription',
  callback_url: 'http://www.example.com/callback',
  instruction: 'Transcribe the given fields.',
  attachment_type: 'website',
  attachment: 'http://news.ycombinator.com/',
  fields: {
    title: 'Title of Webpage',
    top_result: 'Title of the top result'
  }
})

This will also return a Scale::Api::Tasks::Transcription object.

Read more about creating transcription tasks

Audio Transcription Tasks

To create an audio transcription task, run the following:

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

scale.create_audiotranscription_task({
  callback_url: 'http://www.example.com/callback',
  attachment_type: 'audio',
  attachment: 'https://storage.googleapis.com/deepmind-media/pixie/knowing-what-to-say/second-list/speaker-3.wav',
  verbatim: false
})

Upon success, this will return a Scale::Api::Tasks::AudioTranscription object. If it fails, it will raise one of the errors.

Alternatively, you can also create a task this way

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

scale.tasks.create({
  type: 'audiotranscription',
  callback_url: 'http://www.example.com/callback',
  attachment_type: 'audio',
  attachment: 'https://storage.googleapis.com/deepmind-media/pixie/knowing-what-to-say/second-list/speaker-3.wav',
  verbatim: false
})

This will also return a Scale::Api::Tasks::AudioTranscription object.

Read more about creating audio transcription tasks

Listing Tasks

To get a list of tasks, run the following command:

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

scale.tasks.list

This will return a Scale::Api::TaskList object.

Scale::Api::TaskList implements Enumerable, meaning you can do fun stuff like this:

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

scale.tasks.list.map(&:id)

This will return an array containing the last 100 tasks' task_id.

You can also access it like a normal array:

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

scale.tasks.list[0]

This will return the appropriate Task object (or nil if empty).

You can filter this list by:

  • start_time (which expects a Time object)
  • end_time (which expects a Time object)
  • type (which expects one of the tasks types)
  • status (which expects a string which is either completed, pending, or canceled)

For example:

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

scale.tasks.list(end_time: Time.parse('January 20th, 2017'), status: 'completed')

This will return a Scale::Api::TaskList object up to 100 tasks that were completed by January 20th, 2017.

By default, scale.tasks.list only returns up to 100 tasks, but you can pass in the limit yourself.

It also supports pagination, here's an example:

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

first_page = scale.tasks.list
second_page = first_page.next_page

Scale::Api::TaskList#next_page returns the next page in the list of tasks (as a new Scale::Api::TaskList). You can see if there are more pages by calling Scale::Api::TaskList#has_more? on the object.

scale.tasks.list is aliased to scale.tasks.where and scale.tasks.all.

For more information, read our documentation

Finding tasks by ID

To find a task by ID, run the following:

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

task_id = 'TASK_ID'
scale.tasks.find(task_id)

This will return the appropriate Scaler::Api::Tasks object based on the task type

Canceling tasks

There are two ways to cancel a task.

Cancel by task_id:

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

task_id = 'TASK_ID'
scale.tasks.cancel(task_id)

Cancel on the task object:

require 'scale'
scale = Scale.new(api_key: 'SCALE_API_KEY')

task_id = 'TASK_ID'
scale.tasks.find(task_id).cancel!

Both ways will return a new task object for the type, with the status set to canceled and calling canceled? on the task will return true.

Task Object

All tasks return a task object for their type. Currently, this gem supports the following task types:

  • categorization (Scale::Api::Tasks::Categorization)
  • comparison (Scale::Api::Tasks::Comparison)
  • datacollection (Scale::Api::Tasks::Datacollection)
  • annotation (Scale::Api::Tasks::ImageRecognition)
  • transcription (Scale::Api::Tasks::Transcription)
  • audiotranscription (Scale::Api::Tasks::AudioTranscription)

At the time of writing, this is every task type that Scale supports.

Convenience Methods

Every one of the task type objects has the following convenience (instance) methods:

  • day?: returns true when a task's urgency is set to day
  • week?: returns true when a task's urgency is set to week
  • immediate?: returns true when a task's urgency is set to immediate
  • pending?: returns true when a task's status is set to pending
  • completed?: returns true when a task's status is set to completed
  • canceled?: returns true when a task's status is set to canceled
  • callback_succeeded?: returns true when the response from the callback was successful

You can also access all the properties of the task object directly, some examples:

irb(main):009:0> task.instruction
=> "Find the URL for the hiring page for the company with attached website."
irb(main):013:0> task.metadata
=> {"bagel"=>true}
irb(main):016:0> task.completed_at
=> 2017-02-10 20:41:12 UTC

Callbacks

This gem allows you to create and parse callback data, so it can be easily used for web applications:

For example, for Ruby on Rails:

# app/controllers/scale_api_controller.rb

require 'scale'

class ScaleApiController < ApplicationController
  # POST /scale_api
  def create
    scale = Scale.new(api_key: 'SCALE_API_KEY', callback_auth_key: 'SCALE_CALLBACK_AUTH_KEY')

    callback = scale.build_callback params, callback_key: request.headers['scale-callback-auth']
    return render status: 403 unless callback.verified? # Render forbidden if verifying the callback fails

    callback.response # Response content hash (code and result)
    callback.task     # Scale::Api::Tasks object for task type

  end
end

Errors

This gem will raise exceptions on application-level errors. Here are the list of errors:

Scale::Api::BadRequest
Scale::Api::TooManyRequests
Scale::Api::NotFound
Scale::Api::Unauthorized
Scale::Api::InternalServerError
Scale::Api::ConnectionError
Scale::Api::APIKeyInvalid

Development

After checking out the repo, run bin/setup to install dependencies. You can also run bin/console for an interactive prompt that will allow you to experiment.

To install this gem onto your local machine, run bundle exec rake install. To release a new version, update the version number in scaleapi-ruby.gemspec, and then run bundle exec rake release, which will create a git tag for the version, push git commits and tags, and push the .gem file to rubygems.org.

To test locally, run bundle exec rspec. By default, client - server communication is mocked by vcr, which uses local casettes recorded from production server. To test against production server, set a test api key as SCALE_TEST_API_KEY and run bundle exec rake spec:production, which sets environment variable SCALE_TEST_SERVER to production. It will also update new client - server communication in the cassette. Please note that changes in the cassettes will break existing specs. This is because 1) the server may return random results for test api key, and 2) some of the cassettes have been modified specifically for the specs.

Contributing

Bug reports and pull requests are welcome on GitHub at https://github.com/scaleapi/scaleapi-ruby.

Thanks to wikiti for creating the first unofficial Scale API Ruby Client!

License

The gem is available as open source under the terms of the MIT License.

About

The official Ruby API for Scale AI, the data platform for AI.

Resources

License

Stars

Watchers

Forks

Packages

No packages published