텍스트 응답 음성 재생 연결

apps/web/e2e/voice-success.spec.ts

@@ -139,6 +139,25 @@ async function startFakeEngine(): Promise<TestServer> {
       );
       return;
     }
+    if (req.method === "POST" && req.url === "/v1/stream") {
+      res.writeHead(200, {
+        "content-type": "text/event-stream",
+        "cache-control": "no-cache",
+      });
+      res.write(`event: token\ndata: ${JSON.stringify({ text: "괜찮아요. " })}\n\n`);
+      res.write(`event: token\ndata: ${JSON.stringify({ text: "천천히 말해볼게요." })}\n\n`);
+      res.end(
+        `event: done\ndata: ${JSON.stringify({
+          provider: "e2e",
+          model: "fake-client",
+          tokens_in: 1,
+          tokens_out: 1,
+          cost_usd: 0,
+          turns: 1,
+        })}\n\n`,
+      );
+      return;
+    }
     res.writeHead(404, { "content-type": "application/json" });
     res.end(JSON.stringify({ error: "not found" }));
   });
@@ -836,7 +855,7 @@ test.describe("voice cascade success path", () => {
   test("drives one voice turn through the Session mic UI with synthetic browser audio @single-run", async ({
     page,
   }, testInfo) => {
-    test.setTimeout(90_000);
+    test.setTimeout(150_000);
 
     const diagnostics: string[] = [];
     page.on("pageerror", (error) => diagnostics.push(`pageerror: ${error.message}`));
@@ -848,7 +867,10 @@ test.describe("voice cascade success path", () => {
     });
     page.on("response", (response) => {
       const url = response.url();
-      if (response.status() >= 400 && (url.includes("/sessions") || url.includes("/voice/ws"))) {
+      if (
+        response.status() >= 400 &&
+        (url.includes("/sessions") || url.includes("/voice/ws") || url.includes("/voice/speech"))
+      ) {
         diagnostics.push(`response: ${response.status()} ${url}`);
       }
     });
@@ -964,6 +986,39 @@ test.describe("voice cascade success path", () => {
         ].join("\n\n"),
       ).toBeVisible({ timeout: 20_000 });
 
+      const textInput = page.getByLabel("학습자 발화 입력");
+      const sendButton = page.getByRole("button", { name: "보내기", exact: true });
+      await textInput.fill("텍스트로 말해도 내담자 음성을 들려주세요.");
+      await sendButton.click();
+      await expect
+        .poll(
+          () => openai.requests().filter((request) => request === "POST /v1/audio/speech").length,
+          { timeout: 30_000 },
+        )
+        .toBeGreaterThan(0);
+      await expect
+        .poll(async () => (await readVoiceUiProbe(page)).audioBufferStarts, { timeout: 30_000 })
+        .toBeGreaterThan(0);
+      await expect(page.locator(".sx-utt").filter({ hasText: "괜찮아요. 천천히 말해볼게요." })).toBeVisible();
+
+      const freshVoiceSession = await page.evaluate(async ({ apiBase, personaCode }) => {
+        const response = await fetch(`${apiBase}/sessions`, {
+          method: "POST",
+          credentials: "include",
+          headers: { "content-type": "application/json" },
+          body: JSON.stringify({ persona_code: personaCode, theory_mode: "humanistic" }),
+        });
+        return {
+          ok: response.ok,
+          status: response.status,
+          body: await response.json() as { session_id?: string },
+        };
+      }, { apiBase: api.baseURL, personaCode: SEEDED_VOICE_PERSONA_CODE });
+      expect(freshVoiceSession, api.logs()).toMatchObject({ ok: true });
+      expect(typeof freshVoiceSession.body.session_id).toBe("string");
+      await page.goto(`${web.baseURL}/learn/session/${freshVoiceSession.body.session_id}`);
+      await expect(page.locator(".sx-page.sx-page--active")).toBeVisible({ timeout: 20_000 });
+
       const mic = page.locator(".sx-mic");
       await expect(mic).toBeEnabled();
       await mic.click();