A long-lived project that still receives updates
Wrest is a fluent, easy-to-use, object oriented Ruby HTTP/REST client library with support for RFC2616 HTTP caching, multiple HTTP backends and async calls. It runs on CRuby and JRuby and is in production use at 10 * 🦄 scale.


~> 3.11
~> 2.0.0
~> 13.0
~> 6.4
~> 1.35.0


 Project Readme

Wrest 4.0.1

Gem Version Ruby JRuby

Wrest is a ruby REST/HTTP client library. It is currently in use at 10x🦄 scale across all Ruby/JRuby systems at Gojek.

$ gem install wrest
$ wrest # interactive shell
Ruby 3.1.2, 2022-04-12, arm64-darwin21
Loading Wrest 4.0.0
>> 'https://api2.binance.com/api/v3/ticker/24hr'.to_uri.get.deserialize

In your project Gemfile:

gem 'wrest', '~> 4.0.0'
  • Quick tool to wrangle APIs to powerful library to build complex production grade systems, Wrest does it all
  • Clean, object-oriented API with URLs as first class entities
  • Supports RFC 2616 based caching
  • Async http calls using Threads (truly useful only on JRuby due to GIL limitations on CRuby) or EventMachine
  • Is spec driven, strongly favours immutable objects and avoids class methods and setters making it better suited for use as a library, especially in multi-threaded environments
  • Provides convenient HTTP wrappers, redirect handling, serialisation, deserialisation and xpath based lookup


For Facebook, Twitter, Delicious, GitHub and other API examples, see http://github.com/kaiwren/wrest/tree/master/examples

Basic Http Calls


  • Basic API calls

    # Works with json and xml out of the box to give you a hash
    # See lib/wrest/components/translators to add other formats
  • Timeout support

    'https://api.github.com/repos/c42/wrest/issues'.to_uri(timeout: 5).get.body
  • Redirect support

    'http://google.com'.to_uri(follow_redirects: false).get
    'http://google.com'.to_uri(follow_redirects_limit: 1).get

    :follow_redirects_limit defaults to 5 if not specified.

  • Deserialise with XPath filtering

                                                  xpath: '//user/name/text()'
  • More complex request with parameters and a custom deserialiser

                  lat: 35,
                  lon: 139
  • Basic HTTP auth and URI extensions using Wrest::Uri#[]

    base_uri = 'https://api.del.icio.us/v1'.to_uri(username: 'kaiwren', password: 'fupupp1es')
    bookmarks = base_uri['/posts/get'].get.deserialise
  • Detailed debugging and logging (NOT FOR USE IN PRODUCTION! SEE API DOCS!)

    'https://api.github.com/repos/c42/wrest/issues'.to_uri(detailed_http_logging: $stdout).get.deserialize


  • Regular, vanilla Post with a body and headers

    'http://my.api.com'.to_uri.post('YAML encoded body', 'Content-Type' => 'text/x-yaml')
  • Form encoded post

             username: 'kaiwren', password: 'fupupp1es'
             url: 'http://blog.sidu.in/search/label/ruby',
             description: 'The Ruby related posts on my blog!',
             extended: "All posts tagged with 'ruby'",
             tags: 'ruby hacking'
  • Multipart posts

     image: UploadIO.new(File.open(file_path), "image/png", file_path),
     key: imgur_key

Note: To enable Multipart support, you'll have to explicitly require 'wrest/multipart', which depends on the multipart-post gem.


To delete a resource:

                                              username: 'kaiwren',
                                              password: 'fupupp1es'
                                              url: 'http://c2.com'


Wrest supports caching with the following pluggable back-ends:

  • Hash
  • Memcached
  • Redis


Use the following method to enable caching for all requests, and set Hash as the default cache store. Note: Hash should NEVER be used in a production environment. It is unbounded and will keep increasing in size.

    c42 = 'http://c42.in'.to_uri.get

To use Hash as a cache store in an explicit request (without setting hash as default), use the following API:

    r1 = "http://c42.in".to_uri.using_hash.get


A Memcached based caching back-end is available in Wrest. You can get instructions on how to install Memcached on your system here. The Dalli gem is used by Wrest to interface with Memcached. Install dalli using 'gem install dalli'.

Use the following method to enable caching for all requests, and set Memcached as the default back-end.


To use Memcached as a cache store in an explicit request (without setting memcached as default), use the following API:

    r2 = "http://c42.in".to_uri.using_memcached.get


Wrest also supports a Redis based caching back-end. Follow the guide here to install Redis in your system. It uses redis-rd to interface with Redis. Install redis-rb using gem install redis.

Use the following method to enable caching for all requests, and set Redis as the default back-end.


To use Redis as a cache store in an explicit request (without setting redis as default), use the following API:

    r3 = "http://c42.in".to_uri.using_redis.get

A detailed writeup regarding caching as defined by RFC 2616, and how Wrest implements caching is at Wrest Caching Doc

You can create your own back-ends for Wrest caching by implementing the interface implemented in https://github.com/c42/wrest/blob/master/lib/wrest/caching/redis.rb

To explicitly disable caching for specific requests:



Uri level callbacks

You can define a set of callbacks that are invoked based on the http codes of the responses to any requests on a given uri.

  "http://google.com".to_uri(callback: {
              200      => lambda {|response| Wrest.logger.info "Ok." },
              400..499 => lambda {|response| Wrest.logger.error "Invalid. #{response.body}"},
              300..302 => lambda {|response| Wrest.logger.debug "Redirected. #{response.message}" }

Per request callbacks

You can also define callbacks that are invoked based on the http code of the response to a particular request.

  "http://google.com".to_uri.get do |callback|
    callback.on_ok do |response|
      Wrest.logger.info "Ok."

    callback.on(202) do |response|
      Wrest.logger.info "Accepted."

    callback.on(200..206) do |response|
      Wrest.logger.info "Successful."

Please note that Wrest is a synchronous library. All requests are blocking, and will not return till the request is completed and appropriate callbacks executed.

Asynchronous requests

Asynchronous requests are non-blocking. They do not return a response and the request is executed on a separate thread. The only way to access the response while using asynchronous request is through callbacks.

Asynchronous requests support pluggable backends. The default backend used for asynchronous requests is ruby threads, which is only reliable when using JRuby.

  "http://c42.in".to_uri.get_async do |callback|
    callback.on_ok do |response|
      Wrest.logger.info "Ok."

  # Wait until the background threads finish execution before letting the program end.

You can change the default to eventmachine or to threads.




You can also override the default on Uri objects.

  "http://c42.in".to_uri.using_em.get_async do |callback|
    callback.on_ok do |response|
      Wrest.logger.info "Ok."

You can decide which AsyncBackend to use at runtime through to to_uri's options hash.

  "http://c42.in".to_uri(asynchronous_backend: ThreadBackend.new(number_of_threads)).get_async do |callback|
    callback.on_ok do |response|
      Wrest.logger.info "Ok."

Other useful stuff

Hash container with ActiveResource-like semantics

Allows any class to hold an attributes hash, somewhat like ActiveResource. It also supports several extensions to this base fuctionality such as support for typecasting attribute values. See examples/twitter.rb and examples/wow_realm_status.rb for more samples.


 class Demon
   include Wrest::Components::Container

   always_has       :id
   typecast         :age          =>  as_integer,
                    :chi          =>  lambda{|chi| Chi.new(chi)}

   alias_accessors  :chi => :energy

 kai_wren = Demon.new('id' => '1', 'age' => '1500', 'chi' => '1024', 'teacher' => 'Viss')
 kai_wren.id       # => '1'
 kai_wren.age      # => 1500
 kai_wren.chi      # => #<Chi:0x113af8c @count="1024">
 kai_wren.energy   # => #<Chi:0x113af8c @count="1024">
 kai_wren.teacher  # => 'Viss'

Opt-out of core extensions

Uncomfortable with extending String to add to_uri? Simply do

 gem "wrest", :require => "wrest_no_ext"

in your Gemfile. You can now do Uri.new('http://localhost') to build Uris.


The Wrest logger can be set and accessed through Wrest.logger and is configured by default to log to STDOUT. If you're using Wrest in a Rails application, you can configure logging by adding a config/initializers/wrest.rb file with the following contents :

  Wrest.logger = Rails.logger

Every request and response is logged at level debug.

Here is an sample request log message:

<- (POST 515036017 732688777 2010) http://localhost:3000/events.json

The request log consists of request type (POST), request hash (515036017), connection hash (732688777), thread id (2010), URI (http://localhost:3000/events.json)

Here is a sample response log message:

-> (POST 515036017 732688777 2010) 200 OK (0 bytes 0.01s)

The response log consists of request type that generated the response (POST), hash of the request that generated the response (515036017), hash of the connection (732688777), thread id (2010), status (200 OK), response body length (0 bytes) and time taken (0.01)s.

The thread id, request hash and connection hashes are used to track requests and their corresponding responses when using asynchronous requests and/or http connection pooling.

Detailed http debug logging can be turned on like so (DO NOT USE IN PRODUCTION! SEE API DOCS.):

'https://api.github.com/repos/c42/wrest/issues'.to_uri(detailed_http_logging: $stdout).get.deserialize


Standard options are available and can be listed using rake -T. Use rake:rcov for coverage and rake:rdoc to generate documentation. The link to the continuous integration build is over at the C42 Engineering open source page.


Wrest RDocs can be found at http://wrest.rubyforge.org


Features that are planned, in progress or already implemented are documented in the CHANGELOG starting from version 0.0.8.


The source is available at git://github.com/kaiwren/wrest.git

To install the Wrest gem, do (sudo) gem install wrest.

Wrest is currently available as a gem for for Ruby and JRuby.


You can launch the interactive Wrest shell by running bin/wrest if you have the source or invoking wrest from your prompt if you've installed the gem.

  $ wrest
  >> y 'http://twitter.com/statuses/public_timeline.json'.to_uri(timeout: 5).get.deserialise


Start the Sinatra test server for functional test. The dependencies for the test app are managed separately by a Gemfile under spec/sample_app.

To start the sample application:

  cd spec/sample_app
  bundle install
  bundle exec rake  # runs sample app on port 3000

Start a memcached daemon/process on port 11211 and redis on 6379 (both default ports)

  brew install memcached
  brew install redis
  brew services start memcached
  brew services start redis

Run the tests in a different terminal:

  # Run the normal test suite.
  bundle exec rake

  # Runs the functional test suite.
  bundle exec rake rspec:functional


©️ Copyright 2009-2022 Sidu Ponnappa. All Rights Reserved.