A custom library for handling API requests.
Live demo — try a plain fetch, a CSRF-protected request, and the retrier against real responses.
npm install @internetarchive/fetch-handlerimport { FetchHandler } from '@internetarchive/fetch-handler';
@property({ type: Object }) data: any = null;
@property({ type: String }) error: string = '';
@property({ type: Boolean }) loading: boolean = false;
private fetchHandler: FetchHandler;
this.fetchHandler = new FetchHandler({
apiBaseUrl: 'https://archive.org',
});
async fetchData() {
this.loading = true;
this.error = '';
try {
const result =
await this.fetchHandler.fetchApiPathResponse('/metadata/goody');
this.data = result;
} catch (error) {
this.error = `Error fetching data: ${error}`;
} finally {
this.loading = false;
}
}See the demo directory for a more detailed example.
Pass queryParams instead of building the query string yourself. Values are stringified and URL-encoded, and undefined/null entries are dropped, so optional params can go straight in without a conditional:
// GET https://archive.org/services/content-flags/?identifier=goody
await fetchHandler.fetchApiPathResponse('/services/content-flags/', {
queryParams: { identifier: 'goody', mediatype: undefined },
});Params of the same name already on the URL are replaced. To repeat a key, pass a URLSearchParams:
const queryParams = new URLSearchParams();
queryParams.append('flag', 'spam');
queryParams.append('flag', 'violence');
// GET https://archive.org/services/content-flags/?flag=spam&flag=violence
await fetchHandler.fetchApiPathResponse('/services/content-flags/', {
queryParams,
});You can customize how you'd like FetchHandler to retry a request, both globally and per-request.
There are two included configurations, DefaultRetryConfiguration and NoRetryConfiguration, which can be accessed via FetchRetryConfig.default and FetchRetryConfig.noRetry. .default (the default) will retry twice with exponential backoff delay and .noRetry will not retry.
Example:
import { FetchHandler, FetchRetrier, FetchRetryConfig } from '@internetarchive/fetch-handler';
// setting default: retry twice with exponential backoff delay
const fetchRetrier = new FetchRetrier({
retryConfig: FetchRetryConfig.default
})
const fetchHandler = new FetchHandler({
fetchRetrier
})
// use default retry configuration
await fetchHander.fetch('https://foo.com/api')
// specify retry config per-request: will not retry this request
await fetchHandler.fetch('https://foo.com/api', {
retryConfig: FetchRetryConfig.noRetry
})Fetch retrying can be configured using a RetryConfiguring object. This interface has two methods:
interface RetryConfiguring {
shouldRetry(
response: Response | null,
retryNumber: number,
error?: unknown,
): boolean;
retryDelay(
retryNumber: number,
response?: Response | null,
): Milliseconds | null;
}
class MyCustomRetryConfig implements RetryConfiguring {
shouldRetry(
response: Response | null,
retryNumber: number
): boolean {
if (response === null) return false;
if (retryNumber > 1) return false;
return response.status !== 404;
}
retryDelay(retryNumber: number): Milliseconds | null {
return 1000;
}
}
const customRetryConfig = new MyCustomRetryConfig();
// use your custom retry config on a request or globally
await fetchHandler.fetch('https://foo.com/api', {
retryConfig: customRetryConfig
})FetchHandler can attach an X-CSRF-Token header to a request. Provide a getCsrfToken function when constructing the handler, then opt individual requests in with includeCsrfToken: true:
const fetchHandler = new FetchHandler({
apiBaseUrl: 'https://archive.org',
getCsrfToken: async () => {
const { token } = await fetch('/services/csrf-token').then(r => r.json());
return token;
},
});
// sends the X-CSRF-Token header — no need to include it in the body
await fetchHandler.fetchApiPathResponse('/favorites/add', {
method: 'POST',
body: JSON.stringify({ identifier: 'goody' }),
includeCsrfToken: true,
});This is opt-in and off by default: not every backend endpoint's CORS policy allow-lists that header yet, and getting it wrong fails loudly (a blocked CORS preflight) rather than just missing a security control, so each caller should only pass includeCsrfToken: true once its target endpoint is known to accept the header. A request without it (or with getCsrfToken unset entirely) behaves exactly as before.
If a caller already sets an X-CSRF-Token header explicitly, it's left as-is.
Add 127.0.0.1 local.archive.org to your /etc/hosts file
npm run startNOTE The first time you visit the site, the browser will say the site is insecure because it's using a self-signed certificate. Accept the certificate and visit the site and the browser will then accept the certificate in the future. It may always say the site is insecure, but for the purposes of development, it's secure enough.
To run a local development server that serves the basic demo located in demo/index.html
To run the suite of Web Test Runner tests, run
npm run testTo run the tests in watch mode (for <abbr title="test driven development">TDD</abbr>, for example), run
npm run test:watchTo scan the project for linting errors, run
npm run lintTo automatically fix many linting errors, run
npm run format