8.2 KiB
plugin-paginate-rest.js
Octokit plugin to paginate REST API endpoint responses
Usage
| Browsers |
Load |
|---|---|
| Node |
Install with |
const MyOctokit = Octokit.plugin(paginateRest);
const octokit = new MyOctokit({ auth: "secret123" });
// See https://developer.github.com/v3/issues/#list-issues-for-a-repository
const issues = await octokit.paginate("GET /repos/{owner}/{repo}/issues", {
owner: "octocat",
repo: "hello-world",
since: "2010-10-01",
per_page: 100,
});
If you want to utilize the pagination methods in another plugin, use composePaginateRest.
function myPlugin(octokit, options) {
return {
allStars({owner, repo}) => {
return composePaginateRest(
octokit,
"GET /repos/{owner}/{repo}/stargazers",
{owner, repo }
)
}
}
}
octokit.paginate()
The paginateRest plugin adds a new octokit.paginate() method which accepts the same parameters as octokit.request. Only "List ..." endpoints such as List issues for a repository are supporting pagination. Their response includes a Link header. For other endpoints, octokit.paginate() behaves the same as octokit.request().
The per_page parameter is usually defaulting to 30, and can be set to up to 100, which helps retrieving a big amount of data without hitting the rate limits too soon.
An optional mapFunction can be passed to map each page response to a new value, usually an array with only the data you need. This can help to reduce memory usage, as only the relevant data has to be kept in memory until the pagination is complete.
const issueTitles = await octokit.paginate(
"GET /repos/{owner}/{repo}/issues",
{
owner: "octocat",
repo: "hello-world",
since: "2010-10-01",
per_page: 100,
},
(response) => response.data.map((issue) => issue.title),
);
The mapFunction gets a 2nd argument done which can be called to end the pagination early.
const issues = await octokit.paginate(
"GET /repos/{owner}/{repo}/issues",
{
owner: "octocat",
repo: "hello-world",
since: "2010-10-01",
per_page: 100,
},
(response, done) => {
if (response.data.find((issue) => issue.title.includes("something"))) {
done();
}
return response.data;
},
);
Alternatively you can pass a request method as first argument. This is great when using in combination with @octokit/plugin-rest-endpoint-methods:
const issues = await octokit.paginate(octokit.rest.issues.listForRepo, {
owner: "octocat",
repo: "hello-world",
since: "2010-10-01",
per_page: 100,
});
octokit.paginate.iterator()
If your target runtime environments supports async iterators (such as most modern browsers and Node 10+), you can iterate through each response
const parameters = {
owner: "octocat",
repo: "hello-world",
since: "2010-10-01",
per_page: 100,
};
for await (const response of octokit.paginate.iterator(
"GET /repos/{owner}/{repo}/issues",
parameters,
)) {
// do whatever you want with each response, break out of the loop, etc.
const issues = response.data;
console.log("%d issues found", issues.length);
}
Alternatively you can pass a request method as first argument. This is great when using in combination with @octokit/plugin-rest-endpoint-methods:
const parameters = {
owner: "octocat",
repo: "hello-world",
since: "2010-10-01",
per_page: 100,
};
for await (const response of octokit.paginate.iterator(
octokit.rest.issues.listForRepo,
parameters,
)) {
// do whatever you want with each response, break out of the loop, etc.
const issues = response.data;
console.log("%d issues found", issues.length);
}
composePaginateRest and composePaginateRest.iterator
The compose* methods work just like their octokit.* counterparts described above, with the differenct that both methods require an octokit instance to be passed as first argument
How it works
octokit.paginate() wraps octokit.request(). As long as a rel="next" link value is present in the response's Link header, it sends another request for that URL, and so on.
Most of GitHub's paginating REST API endpoints return an array, but there are a few exceptions which return an object with a key that includes the items array. For example:
- Search repositories (key
items) - List check runs for a specific ref (key:
check_runs) - List check suites for a specific ref (key:
check_suites) - List repositories for an installation (key:
repositories) - List installations for a user (key
installations)
octokit.paginate() is working around these inconsistencies so you don't have to worry about it.
If a response is lacking the Link header, octokit.paginate() still resolves with an array, even if the response returns a single object.
Types
The plugin also exposes some types and runtime type guards for TypeScript projects.
| Types |
|
|---|---|
| Guards |
|
PaginateInterface
An interface that declares all the overloads of the .paginate method.
PaginatingEndpoints
An interface which describes all API endpoints supported by the plugin. Some overloads of .paginate() method and composePaginateRest() function depend on PaginatingEndpoints, using the keyof PaginatingEndpoints as a type for one of its arguments.
import { Octokit } from "@octokit/core";
import {
PaginatingEndpoints,
composePaginateRest,
} from "@octokit/plugin-paginate-rest";
type DataType<T> = "data" extends keyof T ? T["data"] : unknown;
async function myPaginatePlugin<E extends keyof PaginatingEndpoints>(
octokit: Octokit,
endpoint: E,
parameters?: PaginatingEndpoints[E]["parameters"],
): Promise<DataType<PaginatingEndpoints[E]["response"]>> {
return await composePaginateRest(octokit, endpoint, parameters);
}
isPaginatingEndpoint
A type guard, isPaginatingEndpoint(arg) returns true if arg is one of the keys in PaginatingEndpoints (is keyof PaginatingEndpoints).
import { Octokit } from "@octokit/core";
import {
isPaginatingEndpoint,
composePaginateRest,
} from "@octokit/plugin-paginate-rest";
async function myPlugin(octokit: Octokit, arg: unknown) {
if (isPaginatingEndpoint(arg)) {
return await composePaginateRest(octokit, arg);
}
// ...
}
Contributing
See CONTRIBUTING.md