Enum ApiEndpoint 

pub enum ApiEndpoint {
    GoogleAI {
        api_key: String,
    },
    GoogleAIToken {
        access_token: AccessToken,
    },
    VertexAI(VertexConfig),
}

API endpoint selector — Google AI (direct), Google AI with OAuth token, or Vertex AI.

Google AI with API Key (default)

Uses an API key passed as a query parameter. The WebSocket URL is wss://generativelanguage.googleapis.com/ws/...?key={api_key} and model URIs are models/{model}.

Google AI with Access Token

Uses an OAuth2 access token (e.g. from gcloud auth print-access-token) passed as a query parameter. Same endpoint as Google AI but with access_token instead of key. This is the recommended approach when using gcloud credentials without an API key.

Vertex AI

Uses a regional endpoint with OAuth2 bearer-token authentication. The WebSocket URL is wss://{location}-aiplatform.googleapis.com/ws/google.cloud.aiplatform.v1.LlmBidiService/BidiGenerateContent and model URIs are projects/{project}/locations/{location}/publishers/google/models/{model}.

let google_ai = ApiEndpoint::google_ai("MY_API_KEY");
let with_token = ApiEndpoint::google_ai_token("ya29.ACCESS_TOKEN");
let vertex = ApiEndpoint::vertex("my-project", "us-central1", "ACCESS_TOKEN");

Variants

GoogleAI

Google AI Studio – API-key authentication.

Fields

api_key: String

The API key.

GoogleAIToken

Google AI with OAuth2 access token (e.g. from gcloud).

Fields

access_token: AccessToken

The OAuth2 access token.

VertexAI(VertexConfig)

Vertex AI — project + location + OAuth2 bearer token.

Implementations

impl ApiEndpoint

pub fn google_ai(api_key: impl Into<String>) -> ApiEndpoint

Shorthand for Google AI endpoint with API key.

pub fn google_ai_token(access_token: impl Into<AccessToken>) -> ApiEndpoint

Google AI endpoint with an OAuth2 access token.

Use this when authenticating with gcloud auth print-access-token or any other OAuth2 flow instead of an API key.

pub fn vertex( project: impl Into<String>, location: impl Into<String>, access_token: impl Into<AccessToken>, ) -> ApiEndpoint

Shorthand for Vertex AI endpoint.

access_token is a fixed string or an AccessToken; for a token that refreshes across reconnects see ApiEndpoint::vertex_refreshing.

pub fn vertex_refreshing( project: impl Into<String>, location: impl Into<String>, token: impl Fn() -> String + Send + Sync + 'static, ) -> ApiEndpoint

Vertex AI endpoint whose bearer token is fetched on every connection attempt, so a reconnect after the token’s ~1 h lifetime carries a live credential.

let endpoint = ApiEndpoint::vertex_refreshing("my-project", "us-central1", || {
    std::env::var("GOOGLE_ACCESS_TOKEN").unwrap_or_default()
});

pub fn vertex_with_host( project: impl Into<String>, location: impl Into<String>, access_token: impl Into<AccessToken>, api_host: impl Into<String>, ) -> ApiEndpoint

Vertex AI endpoint with a custom API host (for private endpoints, VPC-SC, or testing).

pub fn from_env() -> Result<ApiEndpoint, EndpointEnvError>

Resolve an endpoint from standard environment variables.

Selects the platform from GOOGLE_GENAI_USE_VERTEXAI (true/1):

• Vertex AI (GOOGLE_GENAI_USE_VERTEXAI=true): reads GOOGLE_CLOUD_PROJECT (required), GOOGLE_CLOUD_LOCATION (default us-central1), and GOOGLE_ACCESS_TOKEN (required — higher layers may fall back to gcloud auth print-access-token).
• Google AI (default): reads the first set of GEMINI_API_KEY, GOOGLE_GENAI_API_KEY, or GOOGLE_API_KEY.

Returns a descriptive error naming the missing variable so the failure is actionable.

Trait Implementations

impl Clone for ApiEndpoint

fn clone(&self) -> ApiEndpoint

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ApiEndpoint

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Credentials never appear in Debug output: a config logged at debug! level must not leak the key that authenticates it.