Enhanced ChatGPT Clone: Features Agents, MCP, Skills, DeepSeek, Anthropic, AWS, OpenAI, Responses API, Azure, Groq, o1, GPT-5, Mistral, OpenRouter, Vertex AI, Gemini, Artifacts, AI model switching, message search, Code Interpreter, langchain, DALL-E-3, OpenAPI Actions, Functions, Secure Multi-User Auth, Presets, open-source for self-hosting. Active
TypeScript aianthropicartifactsawsazurechatgptchatgpt-cloneclaudeclonedeepseekgeminigooglegpt-5librechatmcpo1openairesponses-apivisionwebui
import { setTokenHeader } from 'librechat-data-provider';
import { getSessionPrincipal } from '../session';
/** Shaped like the real access token, whose payload carries an `id` claim. */
const jwtWith = (claims: Record<string, unknown>) =>
`header.${btoa(JSON.stringify(claims)).replace(/=+$/, '')}.signature`;
/** A LibreChat access token, whose payload carries `id`. */
const tokenFor = (userId: string, issuedAt = 1) => jwtWith({ id: userId, iat: issuedAt });
/** An OpenID `id_token`, which names its subject as `sub` under an issuer. */
const openIdTokenFor = (subject: string, issuer = 'https://idp.example.com') =>
jwtWith({ sub: subject, iss: issuer });
describe('getSessionPrincipal', () => {
afterEach(() => {
setTokenHeader(undefined);
});
it('reads the account from the installed credentials', () => {
setTokenHeader(tokenFor('user-a'));
expect(getSessionPrincipal()).toBe('user:user-a');
});
/* `setUserContext` installs the header in one synchronous call while React
* state catches up later, so this has to change with the header, not after. */
it('changes in the same turn as the header', () => {
setTokenHeader(tokenFor('user-a'));
setTokenHeader(tokenFor('user-b'));
expect(getSessionPrincipal()).toBe('user:user-b');
});
it('reports no account once signed out', () => {
setTokenHeader(tokenFor('user-a'));
setTokenHeader(undefined);
expect(getSessionPrincipal()).toBeUndefined();
});
/* A silent refresh rotates the token for the same account. Comparing tokens
* would read that as a different session and throw away queued work the user
* is still entitled to, so the claim is what matters. */
it('sees a refreshed token as the same account', () => {
setTokenHeader(tokenFor('user-a', 1));
const before = getSessionPrincipal();
setTokenHeader(tokenFor('user-a', 2));
expect(getSessionPrincipal()).toBe(before);
expect(getSessionPrincipal()).toBe('user:user-a');
});
/* With OpenID token reuse the bearer is the provider's own id_token, which
* has no LibreChat `id`. Reading those as "nobody" would make two different
* accounts compare equal, which is exactly how a queued write reaches the
* wrong one. */
it('identifies an OpenID session by its subject and issuer', () => {
setTokenHeader(openIdTokenFor('subject-a'));
expect(getSessionPrincipal()).toBe('oidc:https://idp.example.com:subject-a');
});
it('keeps two OpenID accounts distinct', () => {
setTokenHeader(openIdTokenFor('subject-a'));
const first = getSessionPrincipal();
setTokenHeader(openIdTokenFor('subject-b'));
expect(getSessionPrincipal()).not.toBe(first);
expect(getSessionPrincipal()).toBeDefined();
});
it('keeps the same subject from different issuers distinct', () => {
setTokenHeader(openIdTokenFor('shared', 'https://one.example.com'));
const first = getSessionPrincipal();
setTokenHeader(openIdTokenFor('shared', 'https://two.example.com'));
expect(getSessionPrincipal()).not.toBe(first);
});
/* An opaque token names no one, so the credential itself has to separate the
* sessions rather than both reading as nobody. */
it('keeps two opaque sessions distinct', () => {
setTokenHeader('opaque-token-a');
const first = getSessionPrincipal();
setTokenHeader('opaque-token-b');
expect(first).toBeDefined();
expect(getSessionPrincipal()).not.toBe(first);
});
it('falls back to the credential when the payload carries no subject', () => {
setTokenHeader(jwtWith({ email: '[email protected]' }));
expect(getSessionPrincipal()).toBeDefined();
expect(getSessionPrincipal()).not.toBe('user:undefined');
});
it('decodes a payload whose length needs base64 padding restored', () => {
/* Ids vary in length, so the dropped `=` padding has to be put back. */
for (const id of ['a', 'ab', 'abc', 'abcd', '6a89f66dd0b2baf4d4b02917']) {
setTokenHeader(tokenFor(id));
expect(getSessionPrincipal()).toBe(`user:${id}`);
}
});
});