fix: fix bugs and implemented end to end tests

fix multiple bugs: patternGetValue missing return, clipboard PS quoting, regex group accessor, command binding to prototype vs instance, isolated PS shared buffers, mouseScroll event not enqueued, getDeviceTime signature mismatch; add windows: getWindowElement command;
Implement end to end tests with real calculator, notepad and todo apps

Author: yuri-schwab

README.md

@@ -545,57 +545,28 @@ recursive | boolean | no | If true (default), delete contents recursively. If fa
 
 ### windows: launchApp
 
-Launches an application and waits for it to start. Supports both classic Win32 apps (by path) and UWP apps (by App User Model ID).
+Re-launches the application configured in the `app` session capability. The app path or App User Model ID (AUMID) must have been set when the session was created. Typically used to reopen an app after it has been closed with `windows: closeApp`.
 
-#### Arguments
-
-Name | Type | Required | Description | Example
---- | --- | --- | --- | ---
-app | string | yes | Path to the executable or UWP App User Model ID (AUMID). Classic format: `C:\Path\To\app.exe`. UWP format: `Microsoft.WindowsCalculator_8wekyb3d8bbwe!App` | notepad.exe
-appArguments | string | no | Command-line arguments to pass to the application. | --some-flag
+This command takes no arguments.
 
 #### Example
 
-```csharp
-// Launch Notepad
-driver.ExecuteScript("windows: launchApp", new Dictionary<string, object> { { "app", "notepad.exe" } });
-
-// Launch Calculator (UWP)
-driver.ExecuteScript("windows: launchApp", new Dictionary<string, object> {
-    { "app", "Microsoft.WindowsCalculator_8wekyb3d8bbwe!App" }
-});
-
-// Launch with arguments
-driver.ExecuteScript("windows: launchApp", new Dictionary<string, object> {
-    { "app", "notepad.exe" },
-    { "appArguments", "C:\\path\\to\\file.txt" }
-});
+```javascript
+// Re-launch the app set in the session capability
+await driver.executeScript('windows: launchApp', []);
 ```
 
 ### windows: closeApp
 
-Terminates a running application by process ID, process name, or window handle. All three methods force-kill the process (same outcome). For graceful window close, use `windows: close` with an element. Exactly one identifier must be provided.
-
-#### Arguments
-
-Name | Type | Required | Description | Example
---- | --- | --- | --- | ---
-processId | number | no | Process ID (PID) of the application to terminate. Uses `Stop-Process -Id`. | 12345
-processName | string | no | Process name (e.g. executable name without extension). Terminates all processes with that name. Uses `Stop-Process -Name`. | notepad
-windowHandle | string or number | no | Native window handle of the application window. Resolves the window to its process ID, then terminates it via `Stop-Process -Id`. Accepts hex string (e.g. `"0x12345678"`) or number. | 0x12345678
+Closes the current root application window by sending a close command via the Windows UI Automation WindowPattern. Clears the root element reference in the session afterward. Throws a `NoSuchWindowError` if no active window is found.
 
+This command takes no arguments.
 
 #### Example
 
-```csharp
-// Close by process ID
-driver.ExecuteScript("windows: closeApp", new Dictionary<string, object> { { "processId", 12345 } });
-
-// Close by process name (e.g. all Notepad instances)
-driver.ExecuteScript("windows: closeApp", new Dictionary<string, object> { { "processName", "notepad" } });
-
-// Close by window handle (e.g. from getWindowHandle)
-driver.ExecuteScript("windows: closeApp", new Dictionary<string, object> { { "windowHandle", "0x12345678" } });
+```javascript
+// Close the current app window
+await driver.executeScript('windows: closeApp', []);
 ```
 
 ### windows: clickAndDrag

eslint.config.mjs

@@ -8,4 +8,7 @@ import appiumConfig from '@appium/eslint-config-appium-ts';
 export default defineConfig(
   eslint.configs.recommended,
   ...appiumConfig,
+  {
+    files: ['test/e2e/**/*.ts'],
+  },
 );

lib/commands/actions.ts

@@ -227,3 +227,33 @@ export async function handleKeyAction(this: NovaWindowsDriver, action: KeyAction
             }
     }
 }
+
+export async function releaseActions(this: NovaWindowsDriver): Promise<void> {
+    if (this.keyboardState.shift) {
+        keyUp(Key.SHIFT);
+        keyUp(Key.R_SHIFT);
+        this.keyboardState.shift = false;
+    }
+    if (this.keyboardState.ctrl) {
+        keyUp(Key.CONTROL);
+        keyUp(Key.R_CONTROL);
+        this.keyboardState.ctrl = false;
+    }
+    if (this.keyboardState.meta) {
+        keyUp(Key.META);
+        keyUp(Key.R_META);
+        this.keyboardState.meta = false;
+    }
+    if (this.keyboardState.alt) {
+        keyUp(Key.ALT);
+        keyUp(Key.R_ALT);
+        this.keyboardState.alt = false;
+    }
+    for (const key of this.keyboardState.pressed) {
+        keyUp(key);
+    }
+    this.keyboardState.pressed.clear();
+    mouseUp(0);
+    mouseUp(1);
+    mouseUp(2);
+}

lib/commands/device.ts

@@ -4,9 +4,9 @@ import { PSString, pwsh$ } from '../powershell';
 const GET_SYSTEM_TIME_COMMAND = pwsh$ /* ps1 */ `(Get-Date).ToString(${0})`;
 const ISO_8061_FORMAT = 'yyyy-MM-ddTHH:mm:sszzz';
 
-export async function getDeviceTime(this: NovaWindowsDriver, format?: string): Promise<string> {
-    format = format ? new PSString(format).toString() : `'${ISO_8061_FORMAT}'`;
-    return await this.sendPowerShellCommand(GET_SYSTEM_TIME_COMMAND.format(format));
+export async function getDeviceTime(this: NovaWindowsDriver, _sessionId?: string, format?: string): Promise<string> {
+    const fmt = format ? new PSString(format).toString() : `'${ISO_8061_FORMAT}'`;
+    return await this.sendPowerShellCommand(GET_SYSTEM_TIME_COMMAND.format(fmt));
 }
 
 // command: 'hideKeyboard'

lib/commands/extension.ts

@@ -65,6 +65,8 @@ const EXTENSION_COMMANDS = Object.freeze({
     deleteFile: 'deleteFile',
     deleteFolder: 'deleteFolder',
     clickAndDrag: 'executeClickAndDrag',
+    getDeviceTime: 'windowsGetDeviceTime',
+    getWindowElement: 'getWindowElement',
 } as const);
 
 const ContentType = Object.freeze({
@@ -78,7 +80,7 @@ const TREE_FILTER_COMMAND = $ /* ps1 */ `$cacheRequest.Pop(); $cacheRequest.Tree
 const TREE_SCOPE_COMMAND = $ /* ps1 */ `$cacheRequest.Pop(); $cacheRequest.TreeScope = ${0}; $cacheRequest.Push()`;
 const AUTOMATION_ELEMENT_MODE = $ /* ps1 */ `$cacheRequest.Pop(); $cacheRequest.AutomationElementMode = ${0}; $cacheRequest.Push()`;
 
-const SET_PLAINTEXT_CLIPBOARD_FROM_BASE64 = $ /* ps1 */ `Set-Clipboard -Value [System.Text.Encoding]::UTF8.GetString([Convert]::FromBase64String(${0}))`;
+const SET_PLAINTEXT_CLIPBOARD_FROM_BASE64 = $ /* ps1 */ `Set-Clipboard -Value ([System.Text.Encoding]::UTF8.GetString([Convert]::FromBase64String(${0})))`;
 const GET_PLAINTEXT_CLIPBOARD_BASE64 = /* ps1 */ `[Convert]::ToBase64String([System.Text.Encoding]::UTF8.GetBytes((Get-Clipboard)))`;
 
 const SET_IMAGE_CLIPBOARD_FROM_BASE64 = $ /* ps1 */ `$b = [Convert]::FromBase64String(${0}); $s = New-Object IO.MemoryStream; $s.Write($b, 0, $b.Length); $s.Position = 0; $i = [System.Windows.Media.Imaging.BitmapFrame]::Create($s); [Windows.Clipboard]::SetImage($i); $s.Close()`;
@@ -148,7 +150,7 @@ export async function pushCacheRequest(this: NovaWindowsDriver, cacheRequest: Ca
     }
 
     if (cacheRequest.treeScope) {
-        const treeScope = TREE_SCOPE_REGEX.exec(cacheRequest.treeScope)?.groups?.[0];
+        const treeScope = TREE_SCOPE_REGEX.exec(cacheRequest.treeScope)?.[1];
         if (!treeScope || (Number(cacheRequest.treeScope) < 1 && Number(cacheRequest.treeScope) > 16)) {
             throw new errors.InvalidArgumentError(`Invalid value '${cacheRequest.treeScope}' passed to TreeScope for cache request.`);
         }
@@ -157,7 +159,7 @@ export async function pushCacheRequest(this: NovaWindowsDriver, cacheRequest: Ca
     }
 
     if (cacheRequest.automationElementMode) {
-        const treeScope = AUTOMATION_ELEMENT_MODE_REGEX.exec(cacheRequest.automationElementMode)?.groups?.[0];
+        const treeScope = AUTOMATION_ELEMENT_MODE_REGEX.exec(cacheRequest.automationElementMode)?.[1];
 
         if (!treeScope || (Number(cacheRequest.automationElementMode) < 0 && Number(cacheRequest.automationElementMode) > 1)) {
             throw new errors.InvalidArgumentError(`Invalid value '${cacheRequest.automationElementMode}' passed to AutomationElementMode for cache request.`);
@@ -235,8 +237,8 @@ export async function patternSetValue(this: NovaWindowsDriver, element: Element,
     }
 }
 
-export async function patternGetValue(this: NovaWindowsDriver, element: Element): Promise<void> {
-    await this.sendPowerShellCommand(new FoundAutomationElement(element[W3C_ELEMENT_KEY]).buildGetValueCommand());
+export async function patternGetValue(this: NovaWindowsDriver, element: Element): Promise<string> {
+    return await this.sendPowerShellCommand(new FoundAutomationElement(element[W3C_ELEMENT_KEY]).buildGetValueCommand());
 }
 
 export async function patternMaximize(this: NovaWindowsDriver, element: Element): Promise<void> {
@@ -291,9 +293,9 @@ export async function setClipboardFromBase64(this: NovaWindowsDriver, args: { co
 
     switch (contentType.toLowerCase()) {
         case ContentType.PLAINTEXT:
-            return await this.sendPowerShellCommand(SET_PLAINTEXT_CLIPBOARD_FROM_BASE64.format(args.b64Content));
+            return await this.sendPowerShellCommand(SET_PLAINTEXT_CLIPBOARD_FROM_BASE64.format(`'${args.b64Content}'`));
         case ContentType.IMAGE:
-            return await this.sendPowerShellCommand(SET_IMAGE_CLIPBOARD_FROM_BASE64.format(args.b64Content));
+            return await this.sendPowerShellCommand(SET_IMAGE_CLIPBOARD_FROM_BASE64.format(`'${args.b64Content}'`));
         default:
             throw new errors.InvalidArgumentError(`Unsupported content type '${contentType}'.`);
     }
@@ -867,3 +869,16 @@ export async function executeClickAndDrag(this: NovaWindowsDriver, dragArgs: {
         keyUp(Key.META);
     }
 }
+
+export async function windowsGetDeviceTime(this: NovaWindowsDriver, args?: { format?: string }): Promise<string> {
+    return this.getDeviceTime(undefined, args?.format);
+}
+
+export async function getWindowElement(this: NovaWindowsDriver): Promise<Element> {
+    const result = await this.sendPowerShellCommand(AutomationElement.automationRoot.buildCommand());
+    const elementId = result.split('\n').map((id) => id.trim()).filter(Boolean)[0];
+    if (!elementId) {
+        throw new errors.NoSuchWindowError('No active app window is found for this session.');
+    }
+    return { [W3C_ELEMENT_KEY]: elementId };
+}

lib/commands/powershell.ts

@@ -98,20 +98,20 @@ export async function sendIsolatedPowerShellCommand(this: NovaWindowsDriver, com
         powerShell.stdout.setEncoding('utf8');
         powerShell.stdout.setEncoding('utf8');
 
+        let localStdOut = '';
+        let localStdErr = '';
+
         powerShell.stdout.on('data', (chunk: any) => {
-            this.powerShellStdOut += chunk.toString();
+            localStdOut += chunk.toString();
         });
 
         powerShell.stderr.on('data', (chunk: any) => {
-            this.powerShellStdErr += chunk.toString();
+            localStdErr += chunk.toString();
         });
 
         const result = await new Promise<string>((resolve, reject) => {
-            // eslint-disable-next-line @typescript-eslint/no-non-null-assertion
-            const powerShell = this.powerShell!;
-
-            this.powerShellStdOut = '';
-            this.powerShellStdErr = '';
+            localStdOut = '';
+            localStdErr = '';
 
             powerShell.stdin.write(`${SET_UTF8_ENCODING}\n`);
             if (this.caps.appWorkingDir) {
@@ -130,17 +130,17 @@ export async function sendIsolatedPowerShellCommand(this: NovaWindowsDriver, com
             powerShell.stdin.write(`${command}\n`);
             powerShell.stdin.write(/* ps1 */ `Write-Output $([char]0x${magicNumber.toString(16)})\n`);
 
-            const onData: Parameters<typeof powerShell.stdout.on>[1] = ((chunk: any) => {
+            const onData: Parameters<typeof powerShell.stdout.on>[1] = (chunk: any) => {
                 const magicChar = String.fromCharCode(magicNumber);
                 if (chunk.toString().includes(magicChar)) {
                     powerShell.stdout.off('data', onData);
-                    if (this.powerShellStdErr) {
-                        reject(new errors.UnknownError(this.powerShellStdErr));
+                    if (localStdErr) {
+                        reject(new errors.UnknownError(localStdErr));
                     } else {
-                        resolve(this.powerShellStdOut.replace(`${magicChar}`, '').trim());
+                        resolve(localStdOut.replace(`${magicChar}`, '').trim());
                     }
                 }
-            }).bind(this);
+            };
 
             powerShell.stdout.on('data', onData);
         });

lib/driver.ts

@@ -75,8 +75,10 @@ export class NovaWindowsDriver extends BaseDriver<NovaWindowsDriverConstraints,
         this.locatorStrategies = [...LOCATION_STRATEGIES];
         this.desiredCapConstraints = UI_AUTOMATION_DRIVER_CONSTRAINTS;
 
+        // Bind commands to this instance (not prototype) so each driver instance uses its own
+        // PowerShell session and state when multiple sessions exist
         for (const key in commands) { // TODO: create a decorator that will do that for the class
-            NovaWindowsDriver.prototype[key] = commands[key].bind(this);
+            (this as any)[key] = commands[key].bind(this);
         }
     }
 

lib/winapi/user32.ts

@@ -449,6 +449,7 @@ function makeMouseMoveEvents(args: {
             const verticalScrollEvent = makeEmptyMouseEvent();
             verticalScrollEvent.u.mi.dwFlags = MouseEventFlags.MOUSEEVENTF_WHEEL;
             verticalScrollEvent.u.mi.mouseData = y;
+            mouseEvents.push(verticalScrollEvent);
         }
 
         return mouseEvents;

package.json

@@ -16,7 +16,7 @@
     "watch": "tsc -b --watch",
     "lint": "eslint .",
     "test": "vitest run",
-    "test:watch": "vitest"
+    "test:e2e": "vitest run --config vitest.e2e.config.ts"
   },
   "author": "Automate The Planet",
   "license": "Apache-2.0",
@@ -58,6 +58,7 @@
     "semantic-release": "^25.0.1",
     "typescript": "^5.9.3",
     "typescript-eslint": "^8.46.1",
-    "vitest": "^2.1.0"
+    "vitest": "^2.1.0",
+    "webdriverio": "^9.0.0"
   }
 }

test/commands/device.test.ts

@@ -32,14 +32,13 @@ describe('getDeviceTime', () => {
         expect(cmd).toContain('yyyy-MM-ddTHH:mm:sszzz');
     });
 
-    it('uses custom format when provided', async () => {
+    it('uses custom format when provided as second argument', async () => {
         const driver = createMockDriver() as any;
         driver.sendPowerShellCommand.mockResolvedValue('25/02/2026');
-        const result = await getDeviceTime.call(driver, 'dd/MM/yyyy');
+        const result = await getDeviceTime.call(driver, undefined, 'dd/MM/yyyy');
         expect(result).toBe('25/02/2026');
         const cmd = decodeCommand(driver.sendPowerShellCommand.mock.calls[0][0]);
         expect(cmd).toContain('Get-Date');
-        // The custom format is embedded as a PSString (unicode-escaped)
         expect(cmd).toContain('ToString');
     });
 });