A Ruby client for the official GitHub API.
Supports all the API methods. It's built in a modular way. You can either instantiate the whole API wrapper Github.new or use parts of it i.e. Github::Client::Repos.new if working solely with repositories is your main concern. Intuitive query methods allow you easily call API endpoints.
- Intuitive GitHub API interface navigation.
- It's comprehensive. You can request all GitHub API resources.
- Modular design allows for working with parts of API.
- Fully customizable including advanced middleware stack construction.
- Supports OAuth2 authorization.
- Flexible argument parsing. You can write expressive and natural queries.
- Requests pagination with convenient DSL and automatic options.
- Easy error handling split for client and server type errors.
- Supports multithreaded environment.
- Custom media type specification through the 'media' parameter.
- Request results caching
- Fully tested with unit and feature tests hitting the live api.
Install the gem by running
gem install github_api
or put it in your Gemfile and run bundle install
gem "github_api"
To start using the gem, you can either perform requests directly on Github
namespace:
Github.repos.list user: 'piotrmurach'
or create a new client instance like so
github = Github.new
and then call api methods, for instance, to list a given user repositories do
github.repos.list user: 'piotrmurach'
The github_api closely mirrors the GitHub API hierarchy. For example, if you want to create a new file in a repository, look up the GitHub API spec. In there you will find contents sub category underneath the repository category. This would translate to the request:
github = Github.new
github.repos.contents.create 'piotrmurach', 'finite_machine', 'hello.rb',
path: 'hello.rb',
content: "puts 'hello ruby'"
The whole library reflects the same api navigation. Therefore, if you need to list releases for a repository do:
github.repos.releases.list 'piotrmurach', 'finite_machine'
or to list a user's followers:
github.users.followers.list 'piotrmurach'
The code base has been extensively documented with examples of how to use each method. Please refer to the documentation under the Github::Client
class name.
Alternatively, you can find out which methods are supported by an api by calling actions
on a class or instance. For example, in order to find out available endpoints for Github::Client::Repos::Contents
api call actions
method:
Github::Client::Repos::Contents.actions
=> [:archive, :create, :delete, :find, :get, :readme, :update]
The code base is modular. This means that you can work specifically with a given part of GitHub API. If you want to only work with activity starring API do the following:
starring = Github::Client::Activity::Starring.new oauth_token: token
starring.star 'piotrmurach', 'github'
Please refer to the documentation and look under Github::Client
to see all available classes.
The github_api library allows for flexible argument parsing.
Arguments can be passed directly inside the method called. The required
arguments are passed in first, followed by optional parameters supplied as hash options:
issues = Github::Client::Issues.new
issues.milestones.list 'piotrmurach', 'github', state: 'open'
In the previous example, the order of arguments is important. However, each method also allows you to specify required
arguments using hash symbols and thus remove the need for ordering. Therefore, the same example could be rewritten like so:
issues = Github::Client::Issues.new
issues.milestones.list user: 'piotrmurach', repo: 'github', state: 'open'
Furthermore, required
arguments can be passed during instance creation:
issues = Github::Client::Issues.new user: 'piotrmurach', repo: 'github'
issues.milestones.list state: 'open'
Similarly, the required
arguments for the request can be passed inside the current scope such as:
issues = Github::Client::Issues.new
issues.milestones(user: 'piotrmurach', repo: 'github').list state: 'open'
But why limit ourselves? You can mix and match arguments, for example:
issues = Github::Client::Issues.new user: 'piotrmurach'
issues.milestones(repo: 'github').list
issues.milestones(repo: 'tty').list
You can also use a bit of syntactic sugar whereby "username/repository" can be passed as well:
issues = Github::Client::Issues.new
issues.milestones('piotrmurach/github').list
issues.milestones.list 'piotrmurach/github'
Finally, use the with
scope to clearly denote your requests
issues = Github::Client::Issues.new
issues.milestones.with(user: 'piotrmurach', repo: 'github').list
Please consult the method documentation or GitHub specification to see which arguments are required and what are the option parameters.
The response is of type Github::ResponseWrapper
and allows traversing all the json response attributes like method calls. In addition, if the response returns more than one resource, these will be automatically yielded to the provided block one by one.
For example, when request is issued to list all the branches on a given repository, each branch will be yielded one by one:
repos = Github::Client::Repos.new
repos.branches user: 'piotrmurach', repo: 'github' do |branch|
puts branch.name
end
The ResponseWrapper
allows you to call json attributes directly as method calls. there is no magic here, all calls are delegated to the response body. Therefore, you can directly inspect request body by calling body
method on the ResponseWrapper
like so:
response = repos.branches user: 'piotrmurach', repo: 'github'
response.body # => Array of branches
Each response comes packaged with methods allowing for inspection of HTTP start line and headers. For example, to check for rate limits and status codes do:
response = Github::Client::Repos.branches 'piotrmurach', 'github'
response.headers.ratelimit_limit # "5000"
response.headers.ratelimit_remaining # "4999"
response.headers.status # "200"
response.headers.content_type # "application/json; charset=utf-8"
response.headers.etag # "\"2c5dfc54b3fe498779ef3a9ada9a0af9\""
response.headers.cache_control # "public, max-age=60, s-maxage=60"
If you want to verify if the response was success, namely, that the 200
code was returned call the success?
like so:
response = Github::Client::Repos.branches 'piotrmurach', 'github'
response.success? # => true
It is possible to specify additional header information which will be added to the final request.
For example, to set etag
and X-Poll_Interval
headers, use the :headers
hash key inside the :options
hash like in the following:
events = Github::Client::Activity::Events.new
events.public headers: {
'X-Poll-Interval': 60,
'ETag': "a18c3bded88eb5dbb5c849a489412bf3"
}
In order to set custom media types for a request use the accept header. By using the :accept
key you can determine media type like in the example:
issues = Github::Client::Issues.new
issues.get 'piotrmurach', 'github', 108, accept: 'application/vnd.github.raw'
The github_api provides ability to specify global configuration options. These options will be available to all api calls.
The configuration options can be set by using the configure
helper
Github.configure do |c|
c.basic_auth = "login:password"
c.adapter = :typheous
c.user = 'piotrmurach'
c.repo = 'finite_machine'
end
Alternatively, you can configure the settings by passing a block to an instance like:
Github.new do |c|
c.endpoint = 'https://github.company.com/api/v3'
c.site = 'https://github.company.com'
c.upload_endpoint = 'https://github.company.com/api/uploads'
end
or simply by passing hash of options to an instance like so
github = Github.new basic_auth: 'login:password',
adapter: :typheous,
user: 'piotrmurach',
repo: 'finite_machine'
The following is the full list of available configuration options:
adapter # Http client used for performing requests. Default :net_http
auto_pagination # Automatically traverse requests page links. Default false
basic_auth # Basic authentication in form login:password.
client_id # Oauth client id.
client_secret # Oauth client secret.
connection_options # Hash of connection options.
endpoint # Enterprise API endpoint. Default: 'https://api.github.com'
oauth_token # Oauth authorization token.
org # Global organization used in requests if none provided
per_page # Number of items per page. Max of 100. Default 30.
repo # Global repository used in requests in none provided
site # enterprise API web endpoint
ssl # SSL settings in hash form.
user # Global user used for requests if none provided
user_agent # Custom user agent name. Default 'Github API Ruby Gem'
The github_api will use the default middleware stack which is exposed by calling stack
on a client instance. However, this stack can be freely modified with methods such as insert
, insert_after
, delete
and swap
. For instance, to add your CustomMiddleware
do:
Github.configure do |c|
c.stack.insert_after Github::Response::Helpers, CustomMiddleware
end
Furthermore, you can build your entire custom stack and specify other connection options such as adapter
by doing:
Github.new do |c|
c.adapter :excon
c.stack do |builder|
builder.use Github::Response::Helpers
builder.use Github::Response::Jsonize
end
end
By default requests over SSL are set to OpenSSL::SSL::VERIFY_PEER. However, you can turn off peer verification by
github = Github.new ssl: { verify: false }
If your client fails to find CA certs, you can pass other SSL options to specify exactly how the information is sourced
ssl: {
client_cert: "/usr/local/www.example.com/client_cert.pem"
client_key: "/user/local/www.example.com/client_key.pem"
ca_file: "example.com.cert"
ca_path: "/etc/ssl/"
}
For instance, download CA root certificates from Mozilla cacert and point ca_file at your certificate bundle location. This will allow the client to verify the github.com ssl certificate as authentic.
Caching is supported through the faraday-http-cache
gem.
Add the gem to your Gemfile:
gem 'faraday-http-cache'
You can now configure cache parameters as follows
Github.configure do |config|
config.stack = proc do |builder|
builder.use Faraday::HttpCache, store: Rails.cache
end
end
More details on the available options can be found in the gem's own documentation: https://github.com/plataformatec/faraday-http-cache#faraday-http-cache
To start making requests as authenticated user you can use your GitHub username and password like so
Github.new basic_auth: 'login:password'
Though this method is convenient you should strongly consider using OAuth
for improved security reasons.
To create an access token through the GitHub Authorizations API, you are required to pass your basic credentials and scopes you wish to have for the authentication token.
github = Github.new basic_auth: 'login:password'
github.auth.create scopes: ['repo'], note: 'admin script'
You can add more than one scope from the user
, public_repo
, repo
, gist
or leave the scopes parameter out, in which case, the default read-only access will be assumed (includes public user profile info, public repo info, and gists).
Furthermore, to create auth token for an application you need to pass :app
argument together with :client_id
and :client_secret
parameters.
github = Github.new basic_auth: 'login:password'
github.auth.app.create 'client-id', scopes: ['repo']
In order to revoke auth token(s) for an application you must use basic authentication with client_id
as login and client_secret
as password.
github = Github.new basic_auth: "client_id:client_secret"
github.auth.app.delete 'client-id'
Revoke a specific app token.
github.auth.app.delete 'client-id', 'access-token'
You can check OAuth scopes you have by:
github = Github.new oauth_token: 'token'
github.scopes.list # => ['repo']
or inidividually for a given user:
github = Github.new
github.scopes.list 'token'
To list the scopes that the particular GitHub API action checks for do:
repos = Github::Client::Repos.new
response = repos.list user: 'piotrmurach'
response.headers.accepted_oauth_scopes # => ['delete_repo', 'repo', 'public_repo']
To understand what each scope means refer to documentation
In order to authenticate your app through OAuth2 on GitHub you need to
-
Visit https://github.com/settings/applications/new and register your app. You will need to be logged in to initially register the application.
-
Authorize your credentials https://github.com/login/oauth/authorize
You can use convenience methods to help you achieve this using GithubAPI gem:
github = Github.new client_id: '...', client_secret: '...'
github.authorize_url redirect_uri: 'http://localhost', scope: 'repo'
# => "https://github.com/login/oauth/authorize?scope=repo&response_type=code&client_id='...'&redirect_uri=http://localhost"
After you get your authorization code, call to receive your access_token
token = github.get_token( authorization_code )
Once you have your access token, configure your github instance following instructions under Configuration.
Note: If you are working locally (i.e. your app URL and callback URL are localhost), do not specify a :redirect_uri
otherwise you will get a redirect_uri_mismatch
error.
In order to use Two-Factor authentication you need provide X-GitHub-OTP: required; :2fa-type
header.
You can add headers during initialization:
Github.new do |config|
config.basic_auth = "user:password"
config.connection_options = {headers: {"X-GitHub-OTP" => '2fa token'}}
end
or per request:
github = Github.new basic_auth: 'login:password'
github.oauth.create scopes: ["public_repo"],
headers: {"X-GitHub-OTP" => "2fa token"}
Any request that returns multiple items will be paginated to 30 items by default. You can specify custom page
and per_page
query parameters to alter default behavior. For instance:
repos = Github::Client::Repos.new
response = repos.list user: 'wycats', per_page: 10, page: 5
Then you can query the pagination information included in the link header by:
response.links.first # Shows the URL of the first page of results.
response.links.next # Shows the URL of the immediate next page of results.
response.links.prev # Shows the URL of the immediate previous page of results.
response.links.last # Shows the URL of the last page of results.
In order to iterate through the entire result set page by page, you can use convenience methods:
response.each_page do |page|
page.each do |repo|
puts repo.name
end
end
or use has_next_page?
and next_page
helper methods like in the following:
while response.has_next_page?
... process response ...
res.next_page
end
One can also navigate straight to the specific page by:
res.count_pages # Number of pages
res.page 5 # Requests given page if it exists, nil otherwise
res.first_page # Get first page
res.next_page # Get next page
res.prev_page # Get previous page
res.last_page # Get last page
You can retrieve all pages in one invocation by passing the auto_pagination
option like so:
github = Github.new auto_pagination: true
Depending at what stage you pass the auto_pagination
it will affect all or only a single request. For example, in order to auto paginate all Repository API methods do:
Github::Сlient::Repos.new auto_pagination: true
However, to only auto paginate results for a single request do:
Github::Client::Repos.new.list user: '...', auto_pagination: true
The generic error class Github::Error::GithubError
will handle both the client (Github::Error::ClientError
) and service (Github::Error::ServiceError
) side errors.
For instance in your code you can catch errors like
begin
# Do something with github_api gem
rescue Github::Error::GithubError => e
puts e.message
if e.is_a? Github::Error::ServiceError
# handle GitHub service errors such as 404
elsif e.is_a? Github::Error::ClientError
# handle client errors e.i. missing required parameter in request
end
end
Any time Github client has a problem sending request a Github::Error::ClientError
is raised that will provide a summary of the problem and possible solutions.
When the Github client receives a HTTP response from GitHub service that indicates error then Github::Error::ServiceError
is raised.
There are number of specific error types such as Github::Error::NotAcceptable
when 406
status code is returned.
When Github::Error::ServiceError
is raised you can call data
to access it payload in JSON format.
Anytime there are error messages provided with Github::Error::ServiceError
you can access them by calling error_messages
helper.
A Rails controller that allows a user to authorize their GitHub account and then performs a request.
class GithubController < ApplicationController
def authorize
address = github.authorize_url redirect_uri: 'http://...', scope: 'repo'
redirect_to address
end
def callback
authorization_code = params[:code]
access_token = github.get_token authorization_code
access_token.token # => returns token value
end
private
def github
@github ||= Github.new client_id: '...', client_secret: '...'
end
end
In order to be able to create/update/remove files you need to use Contents API like so:
contents = Github::Client::Repos::Contents.new oauth_token: '...'
Having instantiated the contents, to create a file do:
contents.create 'username', 'repo_name', 'full_path_to/file.ext',
path: 'full_path_to/file.ext',
message: 'Your commit message',
content: 'The contents of your file'
Content is all Base64 encoded to/from the API, and when you create a file it encodes it automatically for you.
To update a file, first you need to find the file so you can get the SHA you're updating off of:
file = contents.find path: 'full_path_to/file.ext'
Then update the file just like you do with creating:
contents.update 'username', 'repo_name', 'full_path_to/file.ext',
path: 'full_path_to/file.ext'
message: 'Your commit message',
content: 'The contents to be updated',
sha: file.sha
Finally to remove a file, find the file so you can get the SHA you're removing:
file = contents.find path: 'full_path_to/file.ext'
Then delete the file like so:
github.delete 'username', 'tome-of-knowledge', 'full_path_to/file.ext',
path: 'full_path_to/file.ext',
message: 'Your Commit Message',
sha: file.sha
The test suite is split into two groups, live
and mock
.
The live
tests are in the features
folder and exercise the GitHub API directly by making live requests and then caching responses with VCR in directory named features\cassettes
. For details on how to get set up, please navigate to the features
folder.
To run all feature tests do:
bundle exec rake features
The mock
tests are in the spec
folder and their primary concern is to test the gem internals without the hindrance of external calls.
To run all specs do:
bundle exec rake spec
Finally to run all tests do:
bundle exec rake
Questions or problems? Please post them on the issue tracker. You can contribute changes by forking the project and submitting a pull request. You can ensure the tests are passing by running bundle
and rake
.
Bug reports and pull requests are welcome on GitHub at https://github.com/piotrmurach/github. This project is intended to be a safe, welcoming space for collaboration, and contributors are expected to adhere to the Contributor Covenant code of conduct.
Copyright (c) 2011 Piotr Murach. See LICENSE.txt for further details.