This is a sample Pet Store Server based on the OpenAPI 3.0 specification. You can find out more about Swagger at https://swagger.io. In the third iteration of the pet store, we've switched to the design first approach! You can now help us improve the API whether it's by making changes to the definition itself or to the code. That way, with time, we can improve the API in general, and expose some of the new features in OAS3.
Some useful links:
Package
octri_petstore· Version1.0.27· 29 operations
# In the consumer crate
cargo add --path /path/to/generated-sdk
# After this version is published
cargo add octri_petstore@1.0.27The example calls loginUser (GET /user/login), a low-friction operation that requires no request arguments.
use octri_petstore::{ClientConfig, ClientAuthConfig, SwaggerPetstoreOpenAPI30};
use octri_petstore::client::OAuth2Auth;
fn main() -> Result<(), Box<dyn std::error::Error>> {
let config = ClientConfig {
base_url: Some("https://petstore3.swagger.io/api/v3".to_string()),
auth: Some(ClientAuthConfig { petstore_auth: Some(OAuth2Auth { access_token: Some(std::env::var("API_TOKEN")?), ..Default::default() }), ..Default::default() }),
..Default::default()
};
let client = SwaggerPetstoreOpenAPI30::new(config);
let result = client.user.login()?;
println!("{result:#?}");
Ok(())
}Keep credentials outside source control. The quickstart reads them from the environment and the client applies them to every request.
| Scheme | ClientAuthConfig field | Sent as |
|---|---|---|
| petstore_auth | petstore_auth |
OAuth 2 access token |
| api_key | api_key |
api_key in header |
- Base URL:
https://petstore3.swagger.io/api/v3. - Transport: ureq.
- Timeout: 30,000 ms per attempt.
- Retries: up to 3 attempts for status codes
408,425,429,500,502,503,504, with 500–8,000 ms backoff. - Idempotency: disabled.
- Error telemetry is disabled by default, even when a reporting endpoint is baked into the build. Consumers must opt in explicitly.
- Telemetry PII filtering is enabled by default: common credentials and direct identifiers are recursively replaced with
[REDACTED]before reports are sent. Disable it only through the generated logging config'sfilterPii(or language-native equivalent) for a trusted private sink.
High-level operation methods return the typed response body directly. The low-level request layer returns an SdkResponse<T> envelope containing data, status, headers, request ID, latency, and attempt count.
All failure paths use a small, predictable hierarchy:
| Error | Meaning |
|---|---|
SdkValidationError |
A request argument failed an OpenAPI constraint before network I/O. |
SdkHttpError |
The server returned a non-2xx response. |
SdkNetworkError |
DNS, connection, TLS, or socket failure. |
SdkTimeoutError |
The configured per-attempt timeout elapsed. |
HTTP errors expose status_code, the response body and headers, plus request_id when the server supplies one. Preserve the request ID in support logs; it is the fastest way to correlate a failed SDK call with server-side traces.
- Operation implementations are grouped under
src/methods/. - 6 component models are split by API domain under
src/models/<domain>.rsorsrc/models/<tag path>/models.rs, re-exported through the compatibilitytypesmodule. - Component schemas can choose a nested model folder with
x-octri-sdk-tags: ["Billing/Invoices"]; the first tag owns the model and/creates nesting. sdk-manifest.jsonis the language-neutral public API index: operations, request/response modes, model properties, enum values, and generation settings.- Public barrel/module exports are the compatibility boundary. Import public model names from those exports; internal domain filenames may evolve without changing model names.
Generated SDK includes schema-derived, zero-dependency mock server and network contract suite. Node.js 20+ required. Contract probes use authored response examples only; schema-synthesized routes remain available to the local server.
./scripts/mock --port 4010 starts server. ./scripts/test runs the mock contract suite, then native SDK tests. A zero-authored-example contract run succeeds with an explicit zero-test
summary; mismatches in authored examples still fail.