Skip to main content

此版本的 GitHub Enterprise 已停止服务 2022-06-03. 即使针对重大安全问题,也不会发布补丁。 要获得更好的性能、改进的安全性和新功能,请升级到 GitHub Enterprise 的最新版本。 如需升级方面的帮助,请联系 GitHub Enterprise 支持

Getting started with the REST API

Learn the foundations for using the REST API, starting with authentication and some endpoint examples.

Let's walk through core API concepts as we tackle some everyday use cases.

注意:以下指南使用 REST API 进行 GitHub.com。

  • 使用 http(s)://[hostname]/api/v3 访问 GitHub Enterprise Server 的 API。

  • 指南指定了可能在 您的 GitHub Enterprise Server 实例 上不存在的用户名和仓库。 您可能需要使用不同的名称才能看到类似的输出。

Overview

Most applications will use an existing wrapper library in the language of your choice, but it's important to familiarize yourself with the underlying API HTTP methods first.

There's no easier way to kick the tires than through cURL.

Hello World

Let's start by testing our setup. Open up a command prompt and enter the following command:

$ curl https://api.github.com/zen

> Keep it logically awesome.

The response will be a random selection from our design philosophies.

Next, let's GET Chris Wanstrath's GitHub profile:

# GET /users/defunkt
$ curl https://api.github.com/users/defunkt

> {
>   "login": "defunkt",
>   "id": 2,
>   "node_id": "MDQ6VXNlcjI=",
>   "avatar_url": "https://avatars.githubusercontent.com/u/2?v=4",
>   "gravatar_id": "",
>   "url": "https://api.github.com/users/defunkt",
>   "html_url": "https://github.com/defunkt",
>   ...
> }

Mmmmm, tastes like JSON. Let's add the -i flag to include headers:

$ curl -i https://api.github.com/users/defunkt

> HTTP/2 200
> server: GitHub.com
> date: Thu, 08 Jul 2021 07:04:08 GMT
> content-type: application/json; charset=utf-8
> cache-control: public, max-age=60, s-maxage=60
> vary: Accept, Accept-Encoding, Accept, X-Requested-With
> etag: W/"61e964bf6efa3bc3f9e8549e56d4db6e0911d8fa20fcd8ab9d88f13d513f26f0"
> last-modified: Fri, 01 Nov 2019 21:56:00 GMT
> x-github-media-type: github.v3; format=json
> access-control-expose-headers: ETag, Link, Location, Retry-After, X-GitHub-OTP, X-RateLimit-Limit, X-RateLimit-Remaining, X-RateLimit-Used, X-RateLimit-Resource, X-RateLimit-Reset, X-OAuth-Scopes, X-Accepted-OAuth-Scopes, X-Poll-Interval, X-GitHub-Media-Type, Deprecation, Sunset
> access-control-allow-origin: *
> strict-transport-security: max-age=31536000; includeSubdomains; preload
> x-frame-options: deny
> x-content-type-options: nosniff
> x-xss-protection: 0
> referrer-policy: origin-when-cross-origin, strict-origin-when-cross-origin
> content-security-policy: default-src 'none'
> x-ratelimit-limit: 60
> x-ratelimit-remaining: 53
> x-ratelimit-reset: 1625731053
> x-ratelimit-resource: core
> x-ratelimit-used: 7
> accept-ranges: bytes
> content-length: 1305
> x-github-request-id: 9F60:7019:ACC5CD5:B03C931:60E6A368
>
> {
>  "login": "defunkt",
>  "id": 2,
>  "node_id": "MDQ6VXNlcjI=",
>  "avatar_url": "https://avatars.githubusercontent.com/u/2?v=4",
>  "gravatar_id": "",
>  "url": "https://api.github.com/users/defunkt",
>  "html_url": "https://github.com/defunkt",
>
>   ...
> }

There are a few interesting bits in the response headers. As expected, the Content-Type is application/json.

Any headers beginning with X- are custom headers, and are not included in the HTTP spec. For example:

  • X-GitHub-Media-Type has a value of github.v3. This lets us know the media type for the response. Media types have helped us version our output in API v3. We'll talk more about that later.
  • Take note of the X-RateLimit-Limit and X-RateLimit-Remaining headers. This pair of headers indicate how many requests a client can make in a rolling time period (typically an hour) and how many of those requests the client has already spent.

Authentication

Unauthenticated clients can make 60 requests per hour. To get more requests per hour, we'll need to authenticate. In fact, doing anything interesting with the GitHub Enterprise Server API requires authentication.

Using personal access tokens

The easiest and best way to authenticate with the GitHub Enterprise Server API is by using Basic Authentication via OAuth tokens. OAuth tokens include personal access tokens.

Use a -u flag to set your username:

$ curl -i -u your_username http(s)://[hostname]/api/v3/users/octocat

When prompted, you can enter your OAuth token, but we recommend you set up a variable for it:

You can use -u "your_username:$token" and set up a variable for token to avoid leaving your token in shell history, which should be avoided.

$ curl -i -u your_username:$token http(s)://[hostname]/api/v3/users/octocat

When authenticating, you should see your rate limit bumped to 5,000 requests an hour, as indicated in the X-RateLimit-Limit header. In addition to providing more calls per hour, authentication enables you to read and write private information using the API.

You can easily create a personal access token using your Personal access tokens settings page:

Personal Token selection

Get your own user profile

When properly authenticated, you can take advantage of the permissions associated with your account on 您的 GitHub Enterprise Server 实例. For example, try getting your own user profile:

$ curl -i -u your_username:your_token http(s)://[hostname]/api/v3/user

> {
>   ...
>   "plan": {
>     "space": 2516582,
>    "collaborators": 10,
>    "private_repos": 20,
>    "name": "medium"
>  }
>   ...
> }

This time, in addition to the same set of public information we retrieved for @defunkt earlier, you should also see the non-public information for your user profile. For example, you'll see a plan object in the response which gives details about the GitHub Enterprise Server plan for the account.

Using OAuth tokens for apps

Apps that need to read or write private information using the API on behalf of another user should use OAuth.

OAuth uses tokens. Tokens provide two big features:

  • Revokable access: users can revoke authorization to third party apps at any time
  • Limited access: users can review the specific access that a token will provide before authorizing a third party app

Tokens should be created via a web flow. An application sends users to GitHub Enterprise Server to log in. GitHub Enterprise Server then presents a dialog indicating the name of the app, as well as the level of access the app has once it's authorized by the user. After a user authorizes access, GitHub Enterprise Server redirects the user back to the application:

GitHub's OAuth Prompt

Treat OAuth tokens like passwords! Don't share them with other users or store them in insecure places. The tokens in these examples are fake and the names have been changed to protect the innocent.

Now that we've got the hang of making authenticated calls, let's move along to the Repositories API.

Repositories

Almost any meaningful use of the GitHub Enterprise Server API will involve some level of Repository information. We can GET repository details in the same way we fetched user details earlier:

$ curl -i http(s)://[hostname]/api/v3/repos/twbs/bootstrap

In the same way, we can view repositories for the authenticated user:

$ curl -i -H "Authorization: token 5199831f4dd3b79e7c5b7e0ebe75d67aa66e79d4" \
    http(s)://[hostname]/api/v3/user/repos

Or, we can list repositories for another user:

$ curl -i http(s)://[hostname]/api/v3/users/octocat/repos

Or, we can list repositories for an organization:

$ curl -i http(s)://[hostname]/api/v3/orgs/octo-org/repos

The information returned from these calls will depend on which scopes our token has when we authenticate:

  • A token with public_repo scope returns a response that includes all public repositories we have access to see on 您的 GitHub Enterprise Server 实例.
  • A token with repo scope returns a response that includes all public, private, or internal repositories we have access to see on 您的 GitHub Enterprise Server 实例.

As the docs indicate, these methods take a type parameter that can filter the repositories returned based on what type of access the user has for the repository. In this way, we can fetch only directly-owned repositories, organization repositories, or repositories the user collaborates on via a team.

$ curl -i "http(s)://[hostname]/api/v3/users/octocat/repos?type=owner"

In this example, we grab only those repositories that octocat owns, not the ones on which she collaborates. Note the quoted URL above. Depending on your shell setup, cURL sometimes requires a quoted URL or else it ignores the query string.

Create a repository

Fetching information for existing repositories is a common use case, but the GitHub Enterprise Server API supports creating new repositories as well. To create a repository, we need to POST some JSON containing the details and configuration options.

$ curl -i -H "Authorization: token 5199831f4dd3b79e7c5b7e0ebe75d67aa66e79d4" \
    -d '{
        "name": "blog",
        "auto_init": true,
        "private": true,
        "gitignore_template": "nanoc"
      }' \
    http(s)://[hostname]/api/v3/user/repos

In this minimal example, we create a new private repository for our blog (to be served on GitHub Pages, perhaps). Though the blog will be public, we've made the repository private. In this single step, we'll also initialize it with a README and a nanoc-flavored .gitignore template.

The resulting repository will be found at https://github.com/<your_username>/blog. To create a repository under an organization for which you're an owner, just change the API method from /user/repos to /orgs/<org_name>/repos.

Next, let's fetch our newly created repository:

$ curl -i http(s)://[hostname]/api/v3/repos/pengwynn/blog

> HTTP/2 404

> {
>    "message": "Not Found"
> }

Oh noes! Where did it go? Since we created the repository as private, we need to authenticate in order to see it. If you're a grizzled HTTP user, you might expect a 403 instead. Since we don't want to leak information about private repositories, the GitHub Enterprise Server API returns a 404 in this case, as if to say "we can neither confirm nor deny the existence of this repository."

Issues

The UI for Issues on GitHub Enterprise Server aims to provide 'just enough' workflow while staying out of your way. With the GitHub Enterprise Server Issues API, you can pull data out or create issues from other tools to create a workflow that works for your team.

Just like github.com, the API provides a few methods to view issues for the authenticated user. To see all your issues, call GET /issues:

$ curl -i -H "Authorization: token 5199831f4dd3b79e7c5b7e0ebe75d67aa66e79d4" \
    http(s)://[hostname]/api/v3/issues

To get only the issues under one of your GitHub Enterprise Server organizations, call GET /orgs/<org>/issues:

$ curl -i -H "Authorization: token 5199831f4dd3b79e7c5b7e0ebe75d67aa66e79d4" \
    http(s)://[hostname]/api/v3/orgs/rails/issues

We can also get all the issues under a single repository:

$ curl -i http(s)://[hostname]/api/v3/repos/rails/rails/issues

Pagination

A project the size of Rails has thousands of issues. We'll need to paginate, making multiple API calls to get the data. Let's repeat that last call, this time taking note of the response headers:

$ curl -i http(s)://[hostname]/api/v3/repos/rails/rails/issues

> HTTP/2 200

> ...
> Link: <http(s)://[hostname]/api/v3/repositories/8514/issues?page=2>; rel="next", <http(s)://[hostname]/api/v3/repositories/8514/issues?page=30>; rel="last"
> ...

The Link header provides a way for a response to link to external resources, in this case additional pages of data. Since our call found more than thirty issues (the default page size), the API tells us where we can find the next page and the last page of results.

Creating an issue

Now that we've seen how to paginate lists of issues, let's create an issue from the API.

To create an issue, we need to be authenticated, so we'll pass an OAuth token in the header. Also, we'll pass the title, body, and labels in the JSON body to the /issues path underneath the repository in which we want to create the issue:

$ curl -i -H 'Authorization: token 5199831f4dd3b79e7c5b7e0ebe75d67aa66e79d4' \
$    -d '{ \
$         "title": "New logo", \
$         "body": "We should have one", \
$         "labels": ["design"] \
$       }' \
$    http(s)://[hostname]/api/v3/repos/pengwynn/api-sandbox/issues

> HTTP/2 201
> Location: http(s)://[hostname]/api/v3/repos/pengwynn/api-sandbox/issues/17
> X-RateLimit-Limit: 5000

> {
>   "pull_request": {
>     "patch_url": null,
>     "html_url": null,
>     "diff_url": null
>   },
>   "created_at": "2012-11-14T15:25:33Z",
>   "comments": 0,
>   "milestone": null,
>   "title": "New logo",
>   "body": "We should have one",
>   "user": {
>     "login": "pengwynn",
>     "gravatar_id": "7e19cd5486b5d6dc1ef90e671ba52ae0",
>     "avatar_url": "https://secure.gravatar.com/avatar/7e19cd5486b5d6dc1ef90e671ba52ae0?d=https://a248.e.akamai.net/assets.github.com%2Fimages%2Fgravatars%2Fgravatar-user-420.png",
>     "id": 865,
>     "url": "http(s)://[hostname]/api/v3/users/pengwynn"
>   },
>   "closed_at": null,
>   "updated_at": "2012-11-14T15:25:33Z",
>   "number": 17,
>   "closed_by": null,
>   "html_url": "https://github.com/pengwynn/api-sandbox/issues/17",
>   "labels": [
>     {
>       "color": "ededed",
>       "name": "design",
>       "url": "http(s)://[hostname]/api/v3/repos/pengwynn/api-sandbox/labels/design"
>     }
>   ],
>   "id": 8356941,
>   "assignee": null,
>   "state": "open",
>   "url": "http(s)://[hostname]/api/v3/repos/pengwynn/api-sandbox/issues/17"
> }

The response gives us a couple of pointers to the newly created issue, both in the Location response header and the url field of the JSON response.

Conditional requests

A big part of being a good API citizen is respecting rate limits by caching information that hasn't changed. The API supports conditional requests and helps you do the right thing. Consider the first call we made to get defunkt's profile:

$ curl -i http(s)://[hostname]/api/v3/users/defunkt

> HTTP/2 200
> etag: W/"61e964bf6efa3bc3f9e8549e56d4db6e0911d8fa20fcd8ab9d88f13d513f26f0"

In addition to the JSON body, take note of the HTTP status code of 200 and the ETag header. The ETag is a fingerprint of the response. If we pass that on subsequent calls, we can tell the API to give us the resource again, only if it has changed:

$ curl -i -H 'If-None-Match: "61e964bf6efa3bc3f9e8549e56d4db6e0911d8fa20fcd8ab9d88f13d513f26f0"' \
$    http(s)://[hostname]/api/v3/users/defunkt

> HTTP/2 304

The 304 status indicates that the resource hasn't changed since the last time we asked for it and the response will contain no body. As a bonus, 304 responses don't count against your rate limit.

Now you know the basics of the GitHub Enterprise Server API!

  • Basic & OAuth authentication
  • Fetching and creating repositories and issues
  • Conditional requests

Keep learning with the next API guide Basics of Authentication!