feat: add OpenAI responses wrapper

2026-06-11 10:32:13 +01:00
parent ae61b4ce72
commit 189b3372a4
4 changed files with 283 additions and 4 deletions
@@ -7,11 +7,11 @@ ChatGPT MCP Server
 ## Status

 Planning complete.
-Phase 0 complete. Phase 1 in progress.
+Phase 0 complete. Phase 1 complete. Phase 2 in progress.

 ## Current Phase

-Phase 1 - Core Utilities
+Phase 2 - OpenAI Integration

 ## Completed Tasks

@@ -21,7 +21,13 @@ Phase 1 - Core Utilities
 - Task 1.2 — Secret redaction utility (`src/utils/redact.js`) ✅
 - Task 1.3 — Context budget utility (`src/utils/context-budget.js`) ✅
 - Task 1.4 — Safe logging helper (`src/utils/logging.js`) ✅
+- Task 2.1 — OpenAI client wrapper (`src/openai/client.js`) ✅
+- Task 2.2 — Response builder (`src/openai/responses.js`) ✅

 ## Next Phase

 Phase 2 - OpenAI Integration
+
+## Next Pending Task
+
+Task 2.3 — Error handling and edge cases for OpenAI integration
@@ -101,7 +101,7 @@ Requirements:
 - Support configurable model and temperature.
 - Implement streaming via the Responses API.

-Status: ⏳ Pending
+Status: ✅ Complete

 ### Task 2.2 - Response builder

@@ -115,4 +115,16 @@ Requirements:
 - Handle OpenAI API errors gracefully (authentication, rate limits, timeouts).
 - Return safe advisory-formatted text responses.

+Status: ✅ Complete
+
+### Task 2.3 - Error handling and edge cases for OpenAI integration
+
+Update `src/openai/responses.js` test coverage.
+
+Requirements:
+- Add tests for additional error kinds (timeout, unknown API status, empty response body).
+- Test boundary conditions for maxOutputTokens validation (non-positive floats, non-integers, zero).
+- Mock OpenAI SDK network timeout errors separately from rate-limit errors.
+- Verify that no API keys or secrets are ever leaked in error messages across all tested paths.
+
 Status: ⏳ Pending
@@ -1 +1,92 @@
-// OpenAI Responses API interface.
+// OpenAI Responses API wrapper.
+
+/**
+ * Send a request to the OpenAI Responses API and return the text result.
+ * @param {import("openai").OpenAI} client - Configured OpenAI SDK client.
+ * @param {{ input: unknown, model?: string, temperature?: number, maxOutputTokens?: number }} params - Request parameters.
+ * @returns {Promise<{ content: string, model?: string, usage?: object }>} Structured result.
+ */
+export async function sendOpenAIResponse(client, params) {
+  // --- Validation ---
+
+  if (!client || typeof client.responses?.create !== "function") {
+    const err = new Error("Invalid client: client.responses.create is required.");
+    err.kind = "ValidationError";
+    throw err;
+  }
+
+  if (!params || !params.input || !params.model) {
+    const err = new Error("Validation error: params.input and params.model are required.");
+    err.kind = "ValidationError";
+    throw err;
+  }
+
+  if (params.maxOutputTokens != null) {
+    const n = params.maxOutputTokens;
+    if (typeof n !== "number" || !Number.isInteger(n) || n <= 0) {
+      const err = new Error("Validation error: maxOutputTokens must be a positive integer.");
+      err.kind = "ValidationError";
+      throw err;
+    }
+  }
+
+  // --- Build request ---
+
+  const createParams = {
+    model: params.model,
+    input: params.input,
+    temperature: params.temperature,
+  };
+
+  if (params.maxOutputTokens != null) {
+    createParams.max_output_tokens = params.maxOutputTokens;
+  }
+
+  // --- Call API ---
+
+  let response;
+  try {
+    response = await client.responses.create(createParams);
+  } catch (error) {
+    return handleOpenAIError(error);
+  }
+
+  // --- Extract text ---
+
+  const content = response.output_text ?? "";
+
+  return {
+    content,
+    model: response.model ?? params.model,
+    usage: response.usage ?? null,
+  };
+}
+
+/**
+ * Map OpenAI SDK errors to safe, typed Error objects.
+ * @param {Error} error - The caught error from the SDK.
+ */
+function handleOpenAIError(error) {
+  const kind = getErrorKind(error);
+  const safeMessage = `OpenAI API error (${kind}): ${safeErrorMessage(error)}`;
+
+  const err = new Error(safeMessage);
+  err.kind = kind;
+  throw err;
+}
+
+function getErrorKind(error) {
+  if (error.status === 401 || /api[_-]?key/i.test(error.message ?? "")) {
+    return "OpenAIAuthError";
+  }
+  if (error.status === 429) {
+    return "OpenAIRateLimitError";
+  }
+  return "OpenAIRequestError";
+}
+
+function safeErrorMessage(error) {
+  const msg = String(error.message ?? "");
+  // Never leak API keys in error messages.
+  return msg.replace(/(api[_-]?key\s*[=:]\s*)\S+/gi, "$1[REDACTED]");
+}
@@ -0,0 +1,170 @@
+import { describe, it, expect, vi, beforeEach } from "vitest";
+
+// --- Helpers ---
+
+function createMockClient(responseBody) {
+  const mockCreate = vi.fn(async () => ({
+    output_text: responseBody?.output_text ?? "OK response text",
+    model: responseBody?.model ?? "gpt-5.1",
+    usage: responseBody?.usage ?? { inputTokens: 10, outputTokens: 20 },
+  }));
+
+  return { responses: { create: mockCreate } };
+}
+
+function makeAuthError() {
+  const err = new Error("Authentication failed. Provide a valid OPENAI_API_KEY.");
+  err.status = 401;
+  return err;
+}
+
+function makeRateLimitError() {
+  const err = new Error("Rate limit exceeded.");
+  err.status = 429;
+  return err;
+}
+
+function makeNetworkError() {
+  const err = new Error("fetch failed");
+  err.status = undefined;
+  return err;
+}
+
+// --- Import ---
+
+const { sendOpenAIResponse } = await import("../../src/openai/responses.js");
+
+describe("sendOpenAIResponse", () => {
+  let mockClient;
+  let mockCreate;
+
+  beforeEach(() => {
+    mockClient = createMockClient();
+    ({ responses: { create: mockCreate } } = mockClient);
+  });
+
+  // ===== Success (3) =====
+
+  it("passes correct OpenAI params to client.responses.create() including max_output_tokens", async () => {
+    const input = [{ type: "input_text", text: "Hello" }];
+    await sendOpenAIResponse(mockClient, {
+      input,
+      model: "gpt-5.1",
+      temperature: 0.2,
+      maxOutputTokens: 2000,
+    });
+
+    expect(mockCreate).toHaveBeenCalledTimes(1);
+    const callArgs = mockCreate.mock.calls[0][0];
+    expect(callArgs.model).toBe("gpt-5.1");
+    expect(callArgs.input).toBe(input);
+    expect(callArgs.temperature).toBe(0.2);
+    expect(callArgs.max_output_tokens).toBe(2000);
+  });
+
+  it("omits max_output_tokens when maxOutputTokens is not provided", async () => {
+    const input = [{ type: "input_text", text: "Hello" }];
+    await sendOpenAIResponse(mockClient, {
+      input,
+      model: "gpt-5.1",
+      temperature: 0.2,
+    });
+
+    const callArgs = mockCreate.mock.calls[0][0];
+    expect(callArgs).not.toHaveProperty("max_output_tokens");
+  });
+
+  it("returns response.output_text as content", async () => {
+    const client = createMockClient({ output_text: "Custom text" });
+    const result = await sendOpenAIResponse(client, {
+      input: [{ type: "input_text", text: "x" }],
+      model: "gpt-5.1",
+    });
+    expect(result.content).toBe("Custom text");
+  });
+
+  // ===== Validation (4) =====
+
+  it("throws ValidationError when client is missing", async () => {
+    await expect(
+      sendOpenAIResponse(null, { input: ["x"], model: "m" })
+    ).rejects.toHaveProperty("kind", "ValidationError");
+  });
+
+  it("throws ValidationError when client.responses.create is missing", async () => {
+    const badClient = { responses: {} };
+    await expect(
+      sendOpenAIResponse(badClient, { input: ["x"], model: "m" })
+    ).rejects.toHaveProperty("kind", "ValidationError");
+  });
+
+  it("throws ValidationError when params.input is missing", async () => {
+    await expect(
+      sendOpenAIResponse(mockClient, { model: "m" })
+    ).rejects.toHaveProperty("kind", "ValidationError");
+  });
+
+  it("throws ValidationError when params.model is missing", async () => {
+    await expect(
+      sendOpenAIResponse(mockClient, { input: ["x"] })
+    ).rejects.toHaveProperty("kind", "ValidationError");
+  });
+
+  // ===== Error mapping (3) =====
+
+  it("maps 401 to OpenAIAuthError", async () => {
+    mockCreate.mockRejectedValueOnce(makeAuthError());
+    await expect(
+      sendOpenAIResponse(mockClient, { input: ["x"], model: "m" })
+    ).rejects.toHaveProperty("kind", "OpenAIAuthError");
+  });
+
+  it("maps 429 to OpenAIRateLimitError", async () => {
+    mockCreate.mockRejectedValueOnce(makeRateLimitError());
+    await expect(
+      sendOpenAIResponse(mockClient, { input: ["x"], model: "m" })
+    ).rejects.toHaveProperty("kind", "OpenAIRateLimitError");
+  });
+
+  it("maps generic error to OpenAIRequestError", async () => {
+    mockCreate.mockRejectedValueOnce(makeNetworkError());
+    await expect(
+      sendOpenAIResponse(mockClient, { input: ["x"], model: "m" })
+    ).rejects.toHaveProperty("kind", "OpenAIRequestError");
+  });
+
+  // ===== Security (1) =====
+
+  it("does not leak API key in error message", async () => {
+    const authErr = makeAuthError();
+    authErr.message = "Invalid OPENAI_API_KEY: sk-my-secret-key-12345.";
+    mockCreate.mockRejectedValueOnce(authErr);
+
+    try {
+      await sendOpenAIResponse(mockClient, { input: ["x"], model: "m" });
+      expect.fail("Should have thrown");
+    } catch (err) {
+      expect(err.message).not.toContain("sk-my-secret-key-12345");
+      expect(err.kind).toBe("OpenAIAuthError");
+    }
+  });
+
+  // ===== Empty content (1) =====
+
+  it("returns empty content when output_text is empty", async () => {
+    const client = createMockClient({ output_text: "" });
+    const result = await sendOpenAIResponse(client, {
+      input: ["x"],
+      model: "gpt-5.1",
+    });
+    expect(result.content).toBe("");
+  });
+
+  // ===== Validation: invalid maxOutputTokens (1) =====
+
+  it("throws ValidationError when maxOutputTokens is negative", async () => {
+    await expect(
+      sendOpenAIResponse(mockClient, { input: ["x"], model: "m", maxOutputTokens: -5 })
+    ).rejects.toHaveProperty("kind", "ValidationError");
+  });
+});