範例
這些整合範例使用公開 API。請將已載入的 OpenAPI 3.2 文件作為 spec,並確保其中包含各範例使用的操作 ID 和路徑。範例 URL、權杖和本機檔案路徑需替換為你自己的值。網路範例需要可存取的服務,並非可獨立執行的 Mock 伺服器。
1. 使用 SendOptions 傳送 HTTP 請求
prepare() 和 send() 都接收包含 spec、target 和 values 的 SendOptions 物件。
import { createClient, createDebugger } from "@powerduck/openapi-request";
// Create client and debugger (options are optional)
const client = createClient();
const dbg = createDebugger();
// Plan a request by operationId
const plan = client.prepare({
spec,
target: { operationId: "getUserById" },
});
console.log(plan.protocol); // "http"
console.log(plan.display.mode); // "response"
console.log(plan.stream.kind); // "none"
// Send the request with concrete values
const result = await dbg.send({
spec,
target: { operationId: "getUserById" },
values: {
path: { id: "42" },
query: { include: "profile" },
},
serverUrl: "https://api.example.com",
auth: { type: "bearer", token: "your-token" },
timeout: 15000,
});
console.log(result.response.status); // e.g. 200
console.log(result.response.body); // body value returned by the adapter
console.log(result.response.headers); // response headers
console.log(result.response.timings.durationMs); // request duration in ms
透過方法與路徑指定目標
const plan2 = client.prepare({
spec,
target: { method: "GET", path: "/users/{id}" },
});
使用 sendMany 批次執行
const batch = await dbg.sendMany(
spec,
[
{ target: { operationId: "listUsers" } },
{ target: { operationId: "getUserById" }, values: { path: { id: "1" } } },
{ target: { operationId: "getUserById" }, values: { path: { id: "2" } } },
],
{
serverUrl: "https://api.example.com",
auth: { type: "bearer", token: "your-token" },
},
);
batch.results.forEach((r, i) => {
if ("error" in r) {
console.log(`Request ${i} failed:`, r.error);
} else {
console.log(`Request ${i} status:`, r.response.status);
}
});
2. SSE 串流回應
當回應宣告 text/event-stream 時,用戶端會自動規劃為 event-list 顯示模式。
const plan = client.prepare({
spec,
target: { operationId: "streamEvents" },
});
console.log(plan.stream.kind); // "sse"
console.log(plan.display.mode); // "event-list"
探測實際端點
import { probeStreamingResponse } from "@powerduck/openapi-request";
const probe = await probeStreamingResponse("https://api.example.com/events");
console.log(probe.kind, probe.status); // "sse" | "ndjson" | "chunked" | "none"
// The probe performs a request; consume or cancel its response body.
await probe.response.body?.cancel();
解析 SSE 串流
import { SseParser } from "@powerduck/openapi-request/http";
const response = await fetch("https://api.example.com/events");
if (!response.ok || !response.body) throw new Error("SSE response unavailable");
const parser = new SseParser();
const reader = response.body.getReader();
try {
while (true) {
const { done, value } = await reader.read();
if (done) break;
for (const event of parser.push(value)) console.log(event.event, event.data);
}
for (const event of parser.flush()) console.log(event.event, event.data);
} finally {
await reader.cancel();
reader.releaseLock();
}
3. WebSocket 工作階段
import { createManualSession } from "@powerduck/openapi-request";
const session = createManualSession({
kind: "websocket",
url: "wss://api.example.com/ws",
});
// Subscribe before opening so connection events are included.
const subscription = session.onEvent((event) => {
console.log("Event:", event.kind, event.data);
});
try {
await session.open();
await session.send({ type: "subscribe", channel: "updates" });
// Keep the session open for as long as your application needs it.
} finally {
await session.close();
subscription.unsubscribe();
}
4. GraphQL 操作
import {
discoverAndWriteGraphQLSchema,
runGraphQL,
} from "@powerduck/openapi-request/graphql";
const endpoint = "https://api.example.com/graphql";
const headers = { Authorization: `Bearer ${token}` };
// Returns a new spec; introspection must be enabled by the server.
const discovered = await discoverAndWriteGraphQLSchema(spec, endpoint, { headers });
console.log(discovered.spec, discovered.warnings);
// Execute an explicit query with a matching server schema.
const result = await runGraphQL({
endpoint,
headers,
query: "query GetUser($id: ID!) { user(id: $id) { id name } }",
operationName: "GetUser",
variables: { id: "42" },
useGet: false,
}, {
spec,
target: { operationId: "getUserById" },
timeout: 15000,
});
console.log(result.response.body);
5. MCP 工具呼叫(HTTP 與 stdio)
import { createManualSession } from "@powerduck/openapi-request";
// Streamable HTTP
const httpSession = createManualSession({
kind: "mcp",
endpoint: "https://mcp.example.com/mcp",
headers: { Authorization: `Bearer ${token}` },
});
// stdio child process
const stdioSession = createManualSession({
kind: "mcp",
transport: "stdio",
command: "node",
args: ["./mcp-server.js"],
});
// Only the HTTP session is opened here. Opening stdioSession starts its process.
try {
await httpSession.open();
const outcome = await httpSession.send({
method: "tools/call",
params: { name: "get_user", arguments: { id: "42" } },
});
console.log(outcome);
} finally {
await httpSession.close();
}
6. gRPC 一元呼叫
import { grpcCall } from "@powerduck/openapi-request/grpc";
const result = await grpcCall({
address: "grpc.example.com:50051",
service: "helloworld.Greeter",
method: "SayHello",
protoPaths: ["./protos/helloworld.proto"],
metadata: { "x-tenant": "acme" },
deadlineMs: 5000,
}, {
messages: [{ name: "World" }],
});
console.log(result.messages); // decoded response(s)
console.log(result.status); // { code, codeName, details }
同一個呼叫也支援伺服器端串流、用戶端串流及雙向串流,只需傳入 options.messages 和 maxMessages、idleTimeoutMs、maxSessionMs 等限制。若需要長時間維持串流,請使用 createGrpcManualSession()。
7. 回寫回應
import { writeBackResponse, toResponseObject, locateOperation } from "@powerduck/openapi-request";
// Locate the operation
const located = locateOperation(spec, { operationId: "getUserById" });
// Build a response object from the result
const fragment = toResponseObject(result, {
// Optional: response normalization options
});
// Write back into the spec
const updated = writeBackResponse(
spec,
located.path,
located.method,
fragment,
{
// Optional: WriteBackOptions
overwriteExamples: true,
},
);
writeBackResponse() 會傳回修改後的副本,不會儲存檔案。發佈前請檢查擷取的範例。與偵錯器的傳送流程不同,這個底層函式不會執行 requirePassingTests 檢查。
從回應推斷 Schema 並產生樣本
import { inferSchema, sampleFromSchema } from "@powerduck/openapi-request";
// Use this for JSON responses; handle parse errors for non-JSON bodies.
const body = result.response.body;
const schema = inferSchema(typeof body === "string" ? JSON.parse(body) : body);
// Generate a realistic sample from the schema
const sample = sampleFromSchema(schema);