Merge #540: run the Playwright browser suite nightly, and fix the stale browser tests it trips on

.github/workflows/browser-suite.yml

@@ -0,0 +1,61 @@
+name: Browser suite
+
+# The per-push CI gate deliberately skips the Playwright-driven suite (config/test-suites.ts),
+# which has twice let two PRs that conflict semantically merge green. This job runs it on a
+# schedule and on demand, so a browser-only regression (the Shift+Enter keypress bug was one)
+# is caught within a day instead of by a user. It is NOT a merge gate: a red run means "look",
+# and it never blocks a push or a PR.
+#
+# Needs: chromium (installed below), tmux, and the live server the tests start themselves.
+# Not run here: test:mobile (per-machine PNG baselines), test:perf (wall-clock), and
+# codex-predictive-echo (needs a real, authenticated codex binary).
+
+on:
+  schedule:
+    - cron: '17 3 * * *'
+  workflow_dispatch:
+
+permissions:
+  contents: read
+
+jobs:
+  browser:
+    name: Playwright browser suite
+    runs-on: ubuntu-latest
+    timeout-minutes: 60
+
+    steps:
+      - uses: actions/checkout@v6
+
+      - name: Setup Node.js
+        uses: actions/setup-node@v6
+        with:
+          node-version: 22
+          cache: 'npm'
+
+      - name: Install dependencies
+        run: npm ci
+
+      - name: Install tmux
+        run: |
+          if ! command -v tmux >/dev/null; then
+            sudo apt-get update -qq
+            sudo apt-get install -y tmux
+          fi
+
+      - name: Install chromium
+        run: npx playwright install --with-deps chromium
+
+      - name: Run the browser suite
+        run: npm run test:browser -- --exclude test/codex-predictive-echo.test.ts
+
+      - name: Keep the failure artifacts
+        if: failure()
+        uses: actions/upload-artifact@v4
+        with:
+          name: browser-suite-results
+          path: |
+            test-results/
+            screenshots-echo-diag/
+          if-no-files-found: ignore
+          retention-days: 7

.github/workflows/ci.yml

@@ -219,6 +219,8 @@ jobs:
         run: npx vitest run
         working-directory: packages/xterm-zerolag-input
 
+# The browser suite also runs nightly (and on demand) in .github/workflows/browser-suite.yml;
+# that job is informational and never gates a push or a PR.
 # Note: three suites are excluded from CI, each with its own local runner:
 #   npm run test:browser  Playwright + chromium (+ a live server, and a real
 #                         codex binary for codex-predictive-echo)

test/opencode-resize.test.ts

@@ -11,6 +11,7 @@
  * Run: npx vitest run test/opencode-resize.test.ts
  */
 
+import { execSync } from 'node:child_process';
 import { describe, it, expect, beforeAll, afterAll } from 'vitest';
 import { chromium, type Browser, type BrowserContext, type Page } from 'playwright';
 import { WebServer } from '../src/web/server.js';
@@ -18,6 +19,15 @@ import { WebServer } from '../src/web/server.js';
 const PORT = 3211;
 const BASE_URL = `http://localhost:${PORT}`;
 
+const HAS_OPENCODE = (() => {
+  try {
+    execSync('command -v opencode', { stdio: 'ignore', shell: '/bin/bash' });
+    return true;
+  } catch {
+    return false;
+  }
+})();
+
 let server: WebServer;
 let browser: Browser;
 
@@ -103,6 +113,19 @@ describe('OpenCode session initial resize', () => {
 
     // Intercept resize API calls to track when they happen
     const resizeCalls: Array<{ url: string; cols: number; rows: number }> = [];
+    // While the WebSocket is connected, resizes go out as {t:'z',c,r} frames
+    // instead of POST /resize, so record both transports.
+    page.on('websocket', (ws) => {
+      ws.on('framesent', (frame) => {
+        try {
+          const msg = JSON.parse(String(frame.payload));
+          if (msg.t === 'z') resizeCalls.push({ url: ws.url() + '#' + sessionIdForWs, cols: msg.c, rows: msg.r });
+        } catch {
+          /* not JSON */
+        }
+      });
+    });
+    let sessionIdForWs = '';
     await page.route('**/api/sessions/*/resize', async (route) => {
       const request = route.request();
       const body = request.postDataJSON();
@@ -123,10 +146,12 @@ describe('OpenCode session initial resize', () => {
         body: JSON.stringify({ workingDir: '/tmp', name: 'oc-resize-test' }),
       });
       const data = await res.json();
-      return data.id ?? data.session?.id;
+      // POST /api/sessions answers in the { success, data: { session } } envelope.
+      return data.data?.session?.id ?? data.id ?? data.session?.id;
     });
 
     expect(sessionId).toBeTruthy();
+    sessionIdForWs = sessionId;
 
     // Call selectSession (which is what runOpenCode does after fix)
     await page.evaluate(async (sid: string) => {
@@ -168,7 +193,8 @@ describe('OpenCode session initial resize', () => {
         body: JSON.stringify({ workingDir: '/tmp', name: 'oc-earlyret-test' }),
       });
       const data = await res.json();
-      return data.id ?? data.session?.id;
+      // POST /api/sessions answers in the { success, data: { session } } envelope.
+      return data.data?.session?.id ?? data.id ?? data.session?.id;
     });
 
     expect(sessionId).toBeTruthy();
@@ -219,7 +245,8 @@ describe('OpenCode session initial resize', () => {
         body: JSON.stringify({ workingDir: '/tmp', name: 'oc-refresh-test' }),
       });
       const data = await res.json();
-      return data.id ?? data.session?.id;
+      // POST /api/sessions answers in the { success, data: { session } } envelope.
+      return data.data?.session?.id ?? data.id ?? data.session?.id;
     });
 
     expect(sessionId).toBeTruthy();
@@ -230,7 +257,23 @@ describe('OpenCode session initial resize', () => {
       await app.selectSession(sid);
     }, sessionId);
 
-    await page.waitForTimeout(300);
+    // The handler only resizes after it has replayed a NON-EMPTY terminal
+    // buffer, so give the session a real PTY with some output first.
+    await page.evaluate(async (sid: string) => {
+      await fetch(`/api/sessions/${sid}/shell`, { method: 'POST' });
+      const deadline = Date.now() + 5000;
+      for (;;) {
+        await fetch(`/api/sessions/${sid}/input`, {
+          method: 'POST',
+          headers: { 'Content-Type': 'application/json' },
+          body: JSON.stringify({ input: 'echo needs-refresh-seed\r', useMux: false }),
+        });
+        await new Promise((r) => setTimeout(r, 400));
+        const res = await fetch(`/api/sessions/${sid}/terminal?full=1`);
+        if ((((await res.json())?.data?.terminalBuffer as string) ?? '').includes('needs-refresh-seed')) break;
+        if (Date.now() > deadline) throw new Error('seed output never appeared');
+      }
+    }, sessionId);
 
     // Intercept resize calls
     const resizeCalls: Array<{ url: string }> = [];
@@ -275,7 +318,7 @@ describe('OpenCode close modal text', () => {
     await context?.close();
   });
 
-  it('shows "Kill Tmux & OpenCode" for opencode sessions', async () => {
+  it.skipIf(!HAS_OPENCODE)('shows "Kill Tmux & OpenCode" for opencode sessions', async () => {
     ({ context, page } = await freshPage());
     await navigateAndWait(page);
 
@@ -287,7 +330,8 @@ describe('OpenCode close modal text', () => {
         body: JSON.stringify({ workingDir: '/tmp', name: 'oc-close-test', mode: 'opencode' }),
       });
       const data = await res.json();
-      return data.id ?? data.session?.id;
+      // POST /api/sessions answers in the { success, data: { session } } envelope.
+      return data.data?.session?.id ?? data.id ?? data.session?.id;
     });
 
     expect(sessionId).toBeTruthy();
@@ -329,7 +373,8 @@ describe('OpenCode close modal text', () => {
         body: JSON.stringify({ workingDir: '/tmp', name: 'cc-close-test' }),
       });
       const data = await res.json();
-      return data.id ?? data.session?.id;
+      // POST /api/sessions answers in the { success, data: { session } } envelope.
+      return data.data?.session?.id ?? data.id ?? data.session?.id;
     });
 
     expect(sessionId).toBeTruthy();

test/perf-browser.test.ts

@@ -69,7 +69,8 @@ async function createSession(page: Page, name: string): Promise<string> {
       body: JSON.stringify({ workingDir: '/tmp', name: n }),
     });
     const data = await res.json();
-    return data.id ?? data.session?.id;
+    // POST /api/sessions answers in the { success, data: { session } } envelope.
+    return data.data?.session?.id ?? data.id ?? data.session?.id;
   }, name);
   return result as string;
 }

test/split-pane-terminal.browser.test.ts

@@ -113,18 +113,21 @@ describe('TerminalTile in a real browser', () => {
       // own startup can race an early write and, on this box, a startup
       // script issues a `clear` that erases scrollback (modern ncurses
       // `clear` emits \x1b[3J) if the input lands before the shell is ready.
-      await fetch(`/api/sessions/${id}/input`, {
+      // Codeman itself writes `clear` into a NEW shell session ~100ms after
-        method: 'POST',
+      // creating it, which can erase an early marker, so re-send until the
-        headers: { 'Content-Type': 'application/json' },
+      // marker is present in the capture rather than writing once.
-        body: JSON.stringify({ input: 'PRE_EXISTING_MARKER\r' }),
+      const deadline = Date.now() + 8000;
-      });
-      const deadline = Date.now() + 5000;
       for (;;) {
+        await fetch(`/api/sessions/${id}/input`, {
+          method: 'POST',
+          headers: { 'Content-Type': 'application/json' },
+          body: JSON.stringify({ input: 'PRE_EXISTING_MARKER\r', useMux: false }),
+        });
+        await new Promise((r) => setTimeout(r, 400));
         const res2 = await fetch(`/api/sessions/${id}/terminal?full=1`);
         const buffer = (await res2.json())?.data?.terminalBuffer ?? '';
         if (buffer.includes('PRE_EXISTING_MARKER')) break;
         if (Date.now() > deadline) throw new Error('marker never landed in ?full=1 capture: ' + JSON.stringify(buffer));
-        await new Promise((r) => setTimeout(r, 200));
       }
       return id;
     });
@@ -166,7 +169,7 @@ describe('TerminalTile in a real browser', () => {
     await page.evaluate(async (id) => {
       await fetch(`/api/sessions/${id}`, { method: 'DELETE' });
     }, sessionId);
-  });
+  }, 20000);
 
   it('gates app-level chords out of Pane B instead of forwarding their raw bytes', async () => {
     // Regression guard for PR #453's Ctrl+K/Alt+1/Alt+B leak: Pane B had no