A Typed GitLab SDK for Browsers, Node.js, and Deno.
Table of Contents
Features
- Complete - All features of Gitlab's exposed APIs are covered up to version 16.0. See here for the full list.
- Universal - Works in all modern browsers, Node.js, and Deno.
- Tested - All libraries have > 80% test coverage.
- Typed - All libraries have extensive TypeScript declarations.
Usage
Browsers
|
Load @gitbeaker/rest directly from esm.sh
<script type="module">
import { Gitlab } from 'https://esm.sh/@gitbeaker/rest';
</script>
|
---|
Deno
|
Load @gitbeaker/rest directly from esm.sh
import { Gitlab } from 'https://esm.sh/@gitbeaker/rest?dts';
|
---|
Node 18+
|
Install with npm install @gitbeaker/rest , or yarn add @gitbeaker/rest
import { Gitlab } from '@gitbeaker/rest';
|
---|
API Client
Instantiate the library using a basic token created in your Gitlab Profile
const api = new Gitlab({
token: 'personaltoken',
});
Available instantiating options:
Name | Optional | Default | Description |
---|
host | Yes | https://gitlab.com | Gitlab Instance Host URL |
token | No* | N/A | Personal Token. Required (one of the three tokens are required) |
oauthToken | No* | N/A | OAuth Token. Required (one of the three tokens are required) |
jobToken | No* | N/A | CI Job Token. Required (one of the three tokens are required) |
rejectUnauthorized | Yes | true | Http Certificate setting, Only applies to non-browser releases and HTTPS hosts urls |
sudo | Yes | false | Sudo query parameter |
camelize | Yes | false | Camelizes all response body keys |
requesterFn | No | @gitbeaker/rest & @gitbeaker/cli : fetch-based, The @gitbeaker/core package does not have a default and thus must be set explicitly | Request Library Wrapper |
queryTimeout | Yes | 300000 | Query Timeout in ms |
profileToken | Yes | N/A | Requests Profiles Token |
profileMode | Yes | execution | Requests Profiles Token |
rateLimits | No | DEFAULT_RATE_LIMITS | Global and endpoint specific adjustable rate limits |
*One of these options must be supplied.
Expanded Payloads
For simplicity, only the response body is returned from the API methods. However, seeing additional response fields, such as the status, headers, etc., may be helpful. For this purpose, an additional optional parameter, showExpanded
can be passed for most API methods.
For methods that return non-paginated results, the payload has this structure:
type ResponseBodyTypes =
| Record<string, unknown>
| Record<string, unknown>[]
| ReadableStream
| Blob
| string
| string[]
| number
| void
| null;
interface FormattedResponse<T extends ResponseBodyTypes = ResponseBodyTypes> {
body: T;
headers: Record<string, string>;
status: number;
}
For methods that return paginated results, the payload also includes paginated information outlined in the Pagination documentation
Available pagination options:
Name | Keyset | Offset | Type | Default | Description |
---|
pagination | X | X | 'offset' or 'keyset' | 'offset' | Defines which pagination type should be used |
perPage | X | X | Number | 20 | Amount of results per request |
orderBy | X | | String | | What field the results should be ordered by |
sort | X | | 'asc' or 'desc' | 'asc' | The direction of sort for the results |
maxPages | | X | Number | N/A | Maximum amount of requests that should be made |
page | | X | Number | N/A | Specific page to be retrieved |
showExpanded | | X | Boolean | false | Returns with the pagination information in addition to the data |
For any .all() function on a resource, it will return all the items from Gitlab. This can be troublesome if there are many items, as the request itself can take a while to be fulfilled. As such, a maxPages option can be passed to limit the scope of the all function.
import { Gitlab } from '@gitbeaker/rest';
const api = new Gitlab({
host: 'http://example.com',
token: 'personaltoken',
});
let projects = await api.Projects.all({ maxPages: 2 });
You can also use this in conjunction with the perPage argument which would override the default of 30 per page set by Gitlab:
import { Gitlab } from '@gitbeaker/rest';
const api = new Gitlab({
host: 'http://example.com',
token: 'personaltoken',
});
let projects = await api.Projects.all({ maxPages: 2, perPage: 40 });
Additionally, if you would like to get back the pagination information, to know how many total pages there are for example, pass the option showExpanded
. If there are multiple results the pagination property will be included as shown below:
...
const { data, paginationInfo } = await api.Projects.all({
perPage:40,
maxPages:2,
showExpanded: true
});
...
This will result in a response in this format:
data: [
...
],
paginationInfo: {
next: 4,
current: 2,
previous: 1,
perPage: 3,
}
Note: Supplying any pagination restrictions is call intensive. Some resources will require many requests which can put a significant load on the Gitlab Server. The general best practice would be setting the page request option to only return the first page if all results are not required.
Similarly, support for Keyset pagination can be toggled on by passing a pagination parameter as a query option
const { data } = await api.Projects.all({
pagination: 'keyset',
sort: 'asc',
orderBy: 'created_at',
});
Rate Limits
Rate limits are completely customizable, and are used to limit the request rate between consecutive API requests within the library. By default, all non-specified endpoints use a 3000 rps rate limit, while some endpoints have much smaller rates as dictated by the Gitlab Docs. See below for the default values:
const DEFAULT_RATE_LIMITS = Object.freeze({
'**': 3000,
'projects/import': 6,
'projects/*/export': 6,
'projects/*/download': 1,
'groups/import': 6,
'groups/*/export': 6,
'groups/*/download': 1,
'projects/*/issues/*/notes': {
method: 'post',
limit: 300,
},
'projects/*/snippets/*/notes': {
method: 'post',
limit: 300,
},
'projects/*/merge_requests/*/notes': {
method: 'post',
limit: 300,
},
'groups/*/epics/*/notes': {
method: 'post',
limit: 300,
},
'projects/*/repository/archive*': 5,
'projects/*/jobs': 600,
'projects/*/members': 60,
'groups/*/members': 60,
});
Rate limits can be overridden when instantiating a API wrapper. For ease of use, these limits are configured using glob patterns, and can be formatted in two ways.
- The glob for the endpoint with the corresponding rate per second
- The glob for the endpoint, with an object specifying the specific method for the endpoint and the corresponding rate limit
const api = new Gitlab({
token: 'token',
rateLimits: {
'**': 30,
'projects/import/*': 40,
'projects/*/issues/*/notes': {
method: 'post',
limit: 300,
},
},
});
Error Handling
Request errors are returned back within a plain Error instance, using the cause to hold the original response and a text description of the error pulled from the response's error or message fields if JSON, or its plain text value:
class GitbeakerError extends Error {
constructor(
message: string,
options?: {
cause: {
description: string;
request: Request;
response: Response;
};
},
) {
super(message, options);
this.name = 'GitbeakerError';
}
}
Note, the message is assigned to the Response's statusText
, and the Request and Response types are from the NodeJS API.
Examples
Once you have your library instantiated, you can utilize many of the API's functionality:
Using the await/async method
import { Gitlab } from '@gitbeaker/rest';
const api = new Gitlab({
host: 'http://example.com',
token: 'personaltoken',
});
let users = await api.Users.all();
api.Projects.all().then((projects) => {
console.log(projects);
});
A general rule about all the function parameters:
- If it's a required parameter, it is a named argument in the functions
- If it's an optional parameter, it is defined in a options object following the named arguments
ie.
import { Projects } from '@gitbeaker/rest';
const projectsAPI = new Projects({
host: 'http://example.com',
token: 'personaltoken',
});
projectsAPI.create({
});
Contributors
This started as a fork from node-gitlab-legacy but I ended up rewriting much of the code. Here are the original work's contributors.