[dotnet] Add examples for BiDi W3C Browsing Context

[diemol]

Description

Adds .NET examples for the BiDi Browsing Context module: creating, closing, navigating, and traversing history for browsing contexts, capturing screenshots, printing, setting the viewport, handling user prompts, and subscribing to browsing-context events. Also wires these examples into the webdriver/bidi/w3c/browsing_context docs (a CSharp tab in every code sample, across en, ja, pt-br, and zh-cn — the translated pages mirror the English page's tabpane structure exactly, so the same references apply to each).

Credit to @nvborisenko for the original work in #1940. That PR was written against a pre-release shape of the .NET BiDi API (OpenQA.Selenium.BiDi.Modules.BrowsingContext) that has since been restructured upstream, so it no longer compiles. This PR carries the same set of examples updated for the current OpenQA.Selenium.BiDi.BrowsingContext API (Selenium.WebDriver 4.46.0): flattened namespace, ImmutableArray-backed results, and IEventSource<T>.SubscribeAsync instead of OnXAsync event methods. It also:

• Fixes a syntax error in BaseTest.cs (comma instead of semicolon).
• Moves options.UseWebSocketUrl = true out of the version-gated branch in BaseTest.cs, since BaseChromeTest calls StartDriver() with no version and the flag would otherwise never be set, so BiDi would never actually be enabled.
• Fixes two timing races found only by actually running the tests against real Chrome/Firefox rather than just compiling them: HandleUserPrompt needed UnhandledPromptBehavior.Ignore on Firefox so the BiDi call could handle the prompt instead of the driver auto-dismissing it, and NavigateForward/TraverseHistory needed to poll for the URL since TraverseHistoryAsync doesn't block until the navigation completes.

Note: gh-codeblock resolves against the GitHub API for the trunk branch, so the new doc references will only render real code once this PR itself is merged (a local ./build-site.sh against an unmerged branch can't verify that content, only that the shortcode syntax and paths are well-formed — which I did verify, including brace-matching every referenced line range against the actual method bodies).

Motivation and Context

#1940 has been open since September 2024 and no longer builds against the current Selenium.WebDriver release. This PR gets the same example coverage into a mergeable, passing state, wires it into the docs, and does all of this without touching the original PR.

Types of changes

• Change to the site (I have double-checked the Netlify deployment, and my changes look good)
• Code example added (and I also added the example to all translated languages)

Checklist

• I have read the contributing document.
• I have used hugo to render the site/docs locally and I am sure it works.

All 31 tests in examples/dotnet/SeleniumDocs/BiDi/BrowsingContext pass locally against Selenium.WebDriver 4.46.0 (verified twice to rule out flakiness). ./build-site.sh builds cleanly with Hugo 0.148.2 (matching README.md's recommended version) across all 4 locales.

🤖 Generated with Claude Code

[qodo-code-review]

PR Summary by Qodo

Add .NET BiDi BrowsingContext examples and enable WebSocket BiDi mode

✨ Enhancement 🧪 Tests 🕐 40+ Minutes

AI Description

• Enable Chrome WebSocket URL so .NET BiDi examples can connect reliably.
• Add MSTest examples for BiDi BrowsingContext commands: create, navigate, screenshot, print.
• Add event-subscription examples using SubscribeAsync for browsing-context lifecycle and navigation
 events.

Diagram

graph TD
  A[Test runner] --> B["BaseTest.cs"] --> C[ChromeOptions]
  A --> D["BrowsingContextTest (partial)"] --> E[BiDi session]
  E --> F[BrowsingContext API]
  F --> G[Commands]
  F --> H[Events]

  subgraph Legend
    direction LR
    _t([Test/Example]) ~~~ _cfg[Config] ~~~ _api([API/Module])
  end

Loading

High-Level Assessment

The following are alternative approaches to this PR:

1. Single-file example suite

• ➕ Easier to read top-to-bottom without jumping across partial class files
• ➕ Less file churn when APIs change
• ➖ Becomes a large monolithic file as modules grow
• ➖ Harder to link individual snippets to docs later

2. Shared helpers for event/polling patterns

• ➕ Reduces repeated TaskCompletionSource/WaitAsync boilerplate
• ➕ Centralizes timeouts/polling defaults for less flakiness
• ➖ May obscure the minimal API usage intended for documentation examples
• ➖ Adds indirection that can make snippets less copy/paste friendly

Recommendation: The current approach (small, topic-focused partial test files) is a good fit for documentation-style examples and future snippet extraction. Consider adding a tiny shared helper (optional) only for the repeated event subscription + timeout pattern if flakiness becomes an issue, but keep the examples mostly “direct API calls” for clarity.

Files changed (19) +558 / -0

Tests (18) +557 / -0

BrowsingContextTest.csAdd BrowsingContext test fixture and BiDi initialization +21/-0

Add BrowsingContext test fixture and BiDi initialization

• Introduces the MSTest fixture (partial class) for BrowsingContext examples. Initializes a BiDi session and captures the top-level browsing context from GetTreeAsync() for reuse across tests.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.cs

BrowsingContextTest.Activate.csAdd example for activating a browsing context +13/-0

Add example for activating a browsing context

• Adds a test that calls context.ActivateAsync() to bring the context to the foreground. Serves as a minimal example of the activation command.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Activate.cs

BrowsingContextTest.CaptureScreenshot.csAdd screenshot examples (full, viewport, element clip) +40/-0

Add screenshot examples (full, viewport, element clip)

• Adds examples for CaptureScreenshotAsync including default screenshots and clipped screenshots via box and element clip rectangles. Validates returned data is present and convertible to bytes.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.CaptureScreenshot.cs

BrowsingContextTest.Close.csAdd examples for closing tab and window contexts +24/-0

Add examples for closing tab and window contexts

• Demonstrates creating new tab/window contexts via BiDi and closing them via context.CloseAsync(). Covers both ContextType.Tab and ContextType.Window cases.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Close.cs

BrowsingContextTest.Create.csAdd examples for creating and discovering browsing contexts +57/-0

Add examples for creating and discovering browsing contexts

• Adds examples for creating tab/window contexts (including using a reference context) and for retrieving an existing context via GetTreeAsync(). Demonstrates the primary entrypoints for obtaining BrowsingContext handles.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Create.cs

BrowsingContextTest.Event.BrowsingContextCreated.csAdd ContextCreated event subscription example +24/-0

Add ContextCreated event subscription example

• Subscribes to bidi.BrowsingContext.ContextCreated via SubscribeAsync and triggers a new window via WebDriver. Uses a TaskCompletionSource with timeout to assert an event is received.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Event.BrowsingContextCreated.cs

BrowsingContextTest.Event.BrowsingContextDestroyed.csAdd ContextDestroyed event subscription example +25/-0

Add ContextDestroyed event subscription example

• Subscribes to bidi.BrowsingContext.ContextDestroyed and closes the current context to trigger the event. Asserts the event’s context matches the one that was closed.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Event.BrowsingContextDestroyed.cs

BrowsingContextTest.Event.BrowsingContextLoaded.csAdd Load event subscription example +24/-0

Add Load event subscription example

• Subscribes to context.Load and navigates with Wait=Complete to trigger the event. Uses timeout-based awaiting to keep the example deterministic.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Event.BrowsingContextLoaded.cs

BrowsingContextTest.Event.DomContentLoaded.csAdd DomContentLoaded event subscription example +24/-0

Add DomContentLoaded event subscription example

• Subscribes to context.DomContentLoaded and navigates to a known page to trigger the event. Awaits the resulting event args with a bounded timeout.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Event.DomContentLoaded.cs

BrowsingContextTest.Event.FragmentNavigated.csAdd fragment navigation event example +26/-0

Add fragment navigation event example

• Navigates to a page, subscribes to FragmentNavigated, then navigates to the same page with a hash fragment. Validates the fragment navigation event is emitted.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Event.FragmentNavigated.cs

BrowsingContextTest.Event.NavigationStarted.csAdd NavigationStarted event subscription example +23/-0

Add NavigationStarted event subscription example

• Subscribes to context.NavigationStarted and triggers navigation to a test page. Asserts that navigation-start event args are received within a timeout.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Event.NavigationStarted.cs

BrowsingContextTest.Event.UserPrompt.csAdd user-prompt open/close event examples +46/-0

Add user-prompt open/close event examples

• Subscribes to BiDi user prompt events (opened/closed) and triggers a prompt via page interaction. Demonstrates SubscribeAsync usage for prompt-related events with timeouts.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Event.UserPrompt.cs

BrowsingContextTest.GetTree.csAdd GetTree examples (depth and top-level contexts) +43/-0

Add GetTree examples (depth and top-level contexts)

• Demonstrates retrieving the browsing context tree, validating iframe children, and limiting depth via MaxDepth. Also shows getting all top-level contexts after opening a new window.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.GetTree.cs

BrowsingContextTest.HandleUserPrompt.csAdd HandleUserPrompt example using Firefox BiDi session +35/-0

Add HandleUserPrompt example using Firefox BiDi session

• Creates a FirefoxDriver configured for WebSocket BiDi and prompt handling, then subscribes to UserPromptOpened and handles the prompt via context.HandleUserPromptAsync(). This provides a working end-to-end example of prompt interaction via BiDi.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.HandleUserPrompt.cs

BrowsingContextTest.Navigate.csAdd navigation and history traversal examples with URL polling +81/-0

Add navigation and history traversal examples with URL polling

• Adds examples for NavigateAsync (with and without readiness state) and for TraverseHistoryAsync back/forward. Includes a helper that polls GetTreeAsync() until the expected URL is observed to avoid timing races.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Navigate.cs

BrowsingContextTest.Print.csAdd PrintAsync example with page ranges +17/-0

Add PrintAsync example with page ranges

• Demonstrates printing via context.PrintAsync with a variety of page range syntaxes. Asserts the returned PDF data is present and convertible to bytes.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Print.cs

BrowsingContextTest.Reload.csAdd ReloadAsync example +20/-0

Add ReloadAsync example

• Navigates to a known page and calls context.ReloadAsync(), asserting navigation info is returned. Provides a minimal reload example for the module.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Reload.cs

BrowsingContextTest.SetViewport.csAdd viewport configuration example +14/-0

Add viewport configuration example

• Demonstrates setting viewport size and device pixel ratio via context.SetViewportAsync(). Serves as a minimal example of viewport control via BiDi.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.SetViewport.cs

Other (1) +1 / -0

BaseTest.csEnable WebSocket URL for Chrome driver startup +1/-0

Enable WebSocket URL for Chrome driver startup

• Sets ChromeOptions.UseWebSocketUrl = true unconditionally when starting the driver. This ensures BiDi connectivity is enabled for tests that rely on driver.AsBiDiAsync().

examples/dotnet/SeleniumDocs/BaseTest.cs

[netlify]

✅ Deploy Preview for selenium-dev ready!

Name	Link
🔨 Latest commit	39ff8cb
🔍 Latest deploy log	https://app.netlify.com/projects/selenium-dev/deploys/6a60bbdcc7f4360008ba2b21
😎 Deploy Preview	https://deploy-preview-2735--selenium-dev.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.
🤖 Make changes	Run an agent on this branch

---

To edit notification comments on pull requests, go to your Netlify project configuration.

[qodo-code-review]

Code Review by Qodo

🐞 Bugs (6) 📘 Rule violations (0) 📜 Skill insights (0)

Context used

✅ Compliance rules (platform): 10 rules

1. Partial accessibility mismatch 🐞 Bug ≡ Correctness

Description

BrowsingContextTest is declared public in one file but the new partial declarations omit an
access modifier (defaulting to internal), which causes a compile-time error due to inconsistent
partial type accessibility.

Code

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Activate.cs[R6-7]

+partial class BrowsingContextTest
+{

Evidence

The primary test class is explicitly public, while the other partial declarations are implicitly
internal due to missing modifiers, which is not allowed for partial type declarations.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.cs[7-10]
examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Activate.cs[4-7]
examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Print.cs[4-7]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
`BrowsingContextTest` is declared as `public partial` in `BrowsingContextTest.cs`, but all the other new `partial class BrowsingContextTest` declarations omit an access modifier (implicitly `internal`). C# requires all partial declarations of a type to have compatible accessibility, so this currently fails compilation.

## Issue Context
This affects every newly-added `BrowsingContextTest.*.cs` partial file.

## Fix Focus Areas
- examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.cs[7-9]
- examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Activate.cs[4-7]

(Apply the same `public partial class BrowsingContextTest` modifier to all other `BrowsingContextTest.*.cs` files in this folder.)

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

2. Print uses unrealistic ranges 🐞 Bug ☼ Reliability ⭐ New

Description

PrintPage prints the initial about:blank browsing context and requests multiple page ranges
(including ranges beyond page 1), which can cause PrintAsync to throw when the printed document
does not have those pages.

Code

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Print.cs[R9-15]

+    public async Task PrintPage()
+    {
+        var pdf = await context.PrintAsync(new() { PageRanges = [1, 2, 3..5, new(3, 5), 7..] });
+
+        Assert.IsNotNull(pdf);
+        Assert.IsNotNull(pdf.Data);
+        Assert.IsNotNull(pdf.ToByteArray());

Evidence

The C# print example prints immediately with complex ranges, while test initialization does not
navigate the context to a page; the existing Java example navigates to a page before printing and
uses default options, demonstrating a more deterministic setup.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Print.cs[8-16]
examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.cs[14-20]
examples/java/src/test/java/dev/selenium/bidirectional/webdriver_bidi/BrowsingContextTest.java[304-314]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

### Issue description
`BrowsingContextTest.PrintPage` calls `context.PrintAsync(...)` without first navigating to a deterministic page, and it requests page ranges that are likely invalid for the initial document (often a single-page `about:blank`). This can cause print to fail depending on how the browser/driver validates page ranges.

### Issue Context
`InitializeBidi` only attaches to the existing top-level context; it does not navigate it to content before tests like `PrintPage` run.

### Fix Focus Areas
- examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Print.cs[8-16]
- examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.cs[14-20]

### Suggested fix
- Navigate to a known page before printing (e.g., `await context.NavigateAsync(..., new(){ Wait = ReadinessState.Complete })`).
- Use default print options (no `PageRanges`) or constrain to a known-valid range like `[1]` unless the example explicitly demonstrates multi-page printing.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

3. Assumes nonempty context tree 🐞 Bug ☼ Reliability

Description

BrowsingContextTest indexes Contexts[0] from GetTreeAsync() without checking Contexts.Length,
which can throw IndexOutOfRangeException and obscure the real setup/navigation failure when the
response is unexpectedly empty. This occurs both during test initialization and during URL polling.

Code

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.cs[19]

+        context = (await bidi.BrowsingContext.GetTreeAsync()).Contexts[0].Context;

Evidence

Both the test setup and the URL polling helper directly index Contexts[0] from a GetTreeAsync()
result without checking whether any contexts were returned, which is a direct
IndexOutOfRangeException risk if the returned array is empty.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.cs[15-20]
examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Navigate.cs[56-63]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

### Issue description
`GetTreeAsync()` results are indexed with `Contexts[0]` without validating the array is non-empty. If the tree is unexpectedly empty, the tests fail with `IndexOutOfRangeException` rather than a descriptive failure.

### Issue Context
This happens in test initialization (choosing the baseline `context`) and in `WaitForUrlAsync` (polling the current URL), so the failure mode can appear both at startup and during navigation history examples.

### Fix Focus Areas
- examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.cs[15-20]
- examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Navigate.cs[56-63]

### Suggested fix
- Store the tree result in a local variable and `Assert.IsTrue(tree.Contexts.Length > 0, "...")` (or throw a descriptive exception) before indexing.
- In `WaitForUrlAsync`, similarly guard against empty `Contexts` and fail with a message that includes the last observed state.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

4. Closes last context 🐞 Bug ☼ Reliability

Description

BrowsingContextDestroyedEvent closes context, which is initialized to the first top-level
browsing context in the session; if that is the only window, behavior after closing becomes
browser/driver-dependent and can lead to intermittent timeouts or cleanup failures. The repo’s
existing destroyed-event examples keep the session alive by creating and closing a separate
window/tab instead.

Code

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Event.BrowsingContextDestroyed.cs[R15-19]

+        await bidi.BrowsingContext.ContextDestroyed.SubscribeAsync(args => tcs.TrySetResult(args));
+
+        await context.CloseAsync();
+
+        var contextInfo = await tcs.Task.WaitAsync(TimeSpan.FromSeconds(5));

Evidence

The test’s context field is set from the first entry of bidi.BrowsingContext.GetTreeAsync(), so
BrowsingContextDestroyedEvent is closing the primary/only context for a fresh session. In
contrast, the repo’s Java destroyed-event example creates a new window and closes it before awaiting
the destroyed event, keeping the original context alive during the wait.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.cs[14-20]
examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Event.BrowsingContextDestroyed.cs[10-23]
examples/java/src/test/java/dev/selenium/bidirectional/webdriver_bidi/BrowsingContextInspectorTest.java[169-185]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
`BrowsingContextDestroyedEvent` currently calls `await context.CloseAsync()` on the default session context (initialized from `GetTreeAsync().Contexts[0]`). If that is the only top-level browsing context, closing it can make subsequent behavior (including event delivery timing and test cleanup) browser/driver-dependent.

## Issue Context
Other destroyed-event examples in this repo trigger the event by creating a *new* window and closing that window, leaving the original context open.

## Fix Focus Areas
- examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Event.BrowsingContextDestroyed.cs[10-23]
- examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.cs[14-20]

## Suggested change
Within `BrowsingContextDestroyedEvent`, create a new browsing context (tab/window), subscribe to `ContextDestroyed`, close **that newly created context**, and assert the event corresponds to it (instead of closing the session’s primary `context`).

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

View more (4)

5. Prompt close is implicit ✓ Resolved 🐞 Bug ☼ Reliability

Description

UserPromptClosedEvent waits for UserPromptClosed after triggering a JS prompt but never
explicitly accepts/dismisses it, so the event depends on Chrome/driver auto-handling and can time
out under different prompt-handling behavior. This also weakens the example by not demonstrating a
deterministic “close” lifecycle.

Code

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Event.UserPrompt.cs[R30-41]

+    public async Task UserPromptClosedEvent()
+    {
+        TaskCompletionSource<UserPromptClosedEventArgs> tcs = new();
+
+        await context.NavigateAsync("https://www.selenium.dev/selenium/web/alerts.html", new() { Wait = ReadinessState.Complete });
+
+        // TODO: this event can be a part of context
+        await bidi.BrowsingContext.UserPromptClosed.SubscribeAsync(args => tcs.TrySetResult(args));
+
+        driver.FindElement(By.Id("prompt")).Click();
+
+        var userPromptClosedEventArgs = await tcs.Task.WaitAsync(TimeSpan.FromSeconds(5));

Evidence

The UserPromptClosedEvent test subscribes to the closed event and clicks a prompt-triggering
element, but contains no explicit accept/dismiss call (BiDi or classic alert handling), so
WaitAsync will only complete if something else closes the prompt. The test class is Chrome-based
by default, and the separate HandleUserPrompt example explicitly documents relying on Chrome
auto-handling prompts, reinforcing that this closed-event example is currently
implicit/behavior-dependent.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Event.UserPrompt.cs[30-45]
examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.cs[7-20]
examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.HandleUserPrompt.cs[14-18]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
`UserPromptClosedEvent` triggers a prompt and then awaits the `UserPromptClosed` event without any explicit action to close the prompt. This makes the example reliant on browser/driver defaults (e.g., Chrome auto-dismiss), which can turn into a timeout/flaky test and is not a deterministic demonstration.

## Issue Context
The class runs under Chrome by default (`BrowsingContextTest : BaseChromeTest`), while another example explicitly notes Chrome auto-handles prompts.

## Fix Focus Areas
- Add deterministic prompt close (e.g., subscribe to `UserPromptOpened`, then call `context.HandleUserPromptAsync(...)`, then await `UserPromptClosed`).
- Keep the example robust if Chrome auto-closes (e.g., tolerate “already closed” errors if needed).

### Code references
- examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Event.UserPrompt.cs[30-45]
- examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.cs[7-20]
- examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.HandleUserPrompt.cs[14-18]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

6. Two browsers started 🐞 Bug ☼ Reliability

Description

HandleUserPrompt creates a FirefoxDriver inside a class that always auto-starts a Chrome driver
via BaseChromeTest, so the test requires both browsers and starts two browser sessions per run.

Code

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.HandleUserPrompt.cs[R16-18]

+        // temporary use firefox because of chrome automatically handle prompts
+        using var driver = new FirefoxDriver(new FirefoxOptions() { UseWebSocketUrl = true, UnhandledPromptBehavior = UnhandledPromptBehavior.Ignore });
+

Evidence

The base class unconditionally starts Chrome for every test, while HandleUserPrompt additionally
creates a Firefox driver, resulting in two browser sessions for the same test.

examples/dotnet/SeleniumDocs/BaseChromeTest.cs[6-13]
examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.HandleUserPrompt.cs[14-23]
examples/dotnet/SeleniumDocs/BaseTest.cs[25-34]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
`BrowsingContextTest` derives from `BaseChromeTest`, which always starts Chrome in `[TestInitialize]`. `HandleUserPrompt` then starts Firefox in the test body. This makes the single test depend on both Chrome and Firefox availability and adds unnecessary overhead.

## Issue Context
The base `Cleanup` will quit only the inherited Chrome `driver` field; the Firefox instance is separately disposed via `using`, so both browsers are active concurrently.

## Fix Focus Areas
- examples/dotnet/SeleniumDocs/BaseChromeTest.cs[6-13]
- examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.HandleUserPrompt.cs[13-34]

## Suggested change
Move `HandleUserPrompt` into a separate `[TestClass]` that derives from `BaseTest` or `BaseFirefoxTest` (and does not inherit the Chrome auto-start), so only Firefox is started for that test.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

7. Non-idempotent event completion ✓ Resolved 🐞 Bug ☼ Reliability

Description

The event examples pass TaskCompletionSource.SetResult directly to SubscribeAsync; if multiple
matching events are delivered during the subscription window, SetResult will throw and can break
the test run.

Code

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Event.NavigationStarted.cs[R13-16]

+        TaskCompletionSource<NavigationStartedEventArgs> tcs = new();
+
+        await context.NavigationStarted.SubscribeAsync(tcs.SetResult);
+

Evidence

The code wires BiDi events to TaskCompletionSource.SetResult without guarding against repeated
invocations of the handler.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Event.NavigationStarted.cs[11-17]
examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Event.BrowsingContextCreated.cs[13-17]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
Several event tests subscribe with `SubscribeAsync(tcs.SetResult)`. `SetResult` is not safe if the callback can be invoked more than once; a second invocation throws and can surface as intermittent failures.

## Issue Context
This pattern appears across multiple event tests (NavigationStarted, Load, DomContentLoaded, ContextCreated/Destroyed, UserPrompt*).

## Fix Focus Areas
- examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Event.NavigationStarted.cs[13-16]

(Replicate the same fix in the other `BrowsingContextTest.Event.*.cs` files using the same subscription pattern.)

## Suggested change
Replace `tcs.SetResult` with an idempotent handler, e.g.:
```csharp
await context.NavigationStarted.SubscribeAsync(args => tcs.TrySetResult(args));
```
Optionally, if `SubscribeAsync` returns a subscription/disposable, dispose/unsubscribe after the first completion.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

8. BaseTest.cs gh-codeblock line drift ✓ Resolved 📘 Rule violation ≡ Correctness

Description

The PR adds a line in examples/dotnet/SeleniumDocs/BaseTest.cs, but docs still reference
BaseTest.cs#L48, which now points to } instead of the intended driver-start line. This causes
the rendered documentation snippet to be incorrect/out-of-date.

Code

examples/dotnet/SeleniumDocs/BaseTest.cs[39]

+            options.UseWebSocketUrl = true;

Evidence

The compliance rule requires updating gh-codeblock line references when example files change. After
the new line in BaseTest.cs, line 48 is now } while the driver initialization moved to line
49, but the docs still reference #L48.

Rule 2141349: Update gh-codeblock line ranges when examples change
website_and_docs/content/documentation/webdriver/drivers/_index.en.md[30-40]
examples/dotnet/SeleniumDocs/BaseTest.cs[36-50]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
Docs reference `BaseTest.cs#L48`, but this PR inserted a new line above, shifting the intended snippet line(s). As a result, the gh-codeblock now renders the wrong line (`}`) instead of the intended code.

## Issue Context
`website_and_docs/content/documentation/webdriver/drivers/_index.*.md` uses gh-codeblock pointers into `examples/dotnet/SeleniumDocs/BaseTest.cs`. After adding `options.UseWebSocketUrl = true;`, the referenced line number should be updated.

## Fix Focus Areas
- examples/dotnet/SeleniumDocs/BaseTest.cs[36-50]
- website_and_docs/content/documentation/webdriver/drivers/_index.en.md[30-40]
- website_and_docs/content/documentation/webdriver/drivers/_index.ja.md[30-40]
- website_and_docs/content/documentation/webdriver/drivers/_index.pt-br.md[30-40]
- website_and_docs/content/documentation/webdriver/drivers/_index.zh-cn.md[30-40]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

9. URL wait lacks timeout signal 🐞 Bug ◔ Observability

Description

WaitForUrlAsync has a deadline but returns the last observed URL on expiry instead of signaling a
timeout, so callers only see a generic assertion mismatch rather than an explicit wait failure. This
makes navigation/timing problems harder to diagnose from test output.

Code

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Navigate.cs[R56-68]

+    private async Task<string> WaitForUrlAsync(string expectedUrl)
+    {
+        var deadline = DateTime.UtcNow.AddSeconds(5);
+        string url;
+        do
+        {
+            url = (await context.GetTreeAsync()).Contexts[0].Url;
+            if (url == expectedUrl) return url;
+            await Task.Delay(100);
+        } while (DateTime.UtcNow < deadline);
+
+        return url;
+    }

Evidence

The helper computes a deadline and polls, but after the deadline it executes return url; rather
than throwing/indicating a timeout condition.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Navigate.cs[56-68]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

### Issue description
`WaitForUrlAsync` polls until a deadline, but on expiry it returns normally with the last observed URL. That makes timeouts indistinguishable from “wrong URL” failures and produces less actionable test output.

### Issue Context
This helper is used by `NavigateBack`, `NavigateForward`, and `TraverseHistory`.

### Fix Focus Areas
- examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.Navigate.cs[56-68]

### Suggested fix
- After the loop, throw a `TimeoutException` (or assert-fail) that includes both the expected URL and the last observed URL.
- (Optional) Accept a timeout parameter to make the helper reusable across environments.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

10. Unchecked node indexing ✓ Resolved 🐞 Bug ☼ Reliability

Description

CaptureElementScreenshot indexes Nodes[0] from LocateNodesAsync without validating that any
nodes were returned, which can throw IndexOutOfRangeException and produce a poor failure mode if
the locator/page changes.

Code

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.CaptureScreenshot.cs[R31-35]

+        driver.Url = "https://www.selenium.dev/selenium/web/formPage.html";
+
+        var element = (await context.LocateNodesAsync(new CssLocator("#checky"))).Nodes[0];
+
+        var screenshot = await context.CaptureScreenshotAsync(new() { Clip = new ElementClipRectangle(element) });

Evidence

The code directly accesses Nodes[0] with no guard, so an empty result will throw immediately.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.CaptureScreenshot.cs[28-35]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
The test assumes `LocateNodesAsync` always finds `#checky` and immediately indexes `Nodes[0]`. If the node list is empty, the test throws before producing a clear assertion message.

## Issue Context
This is an example test; a clearer assert makes failures easier to diagnose when the page/locator changes.

## Fix Focus Areas
- examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.CaptureScreenshot.cs[31-35]

## Suggested change
Capture the nodes result and assert it is non-empty before indexing, e.g.:
```csharp
var result = await context.LocateNodesAsync(new CssLocator("#checky"));
Assert.IsTrue(result.Nodes.Length > 0, "Expected to locate #checky");
var element = result.Nodes[0];
```

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

11. Assumed context ordering ✓ Resolved 🐞 Bug ⚙ Maintainability

Description

GetAllTopLevelBrowsingContexts assumes the newly created window will be at contexts[1], which is
a brittle positional assertion and can fail if the returned context array ordering differs.

Code

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.GetTree.cs[R34-42]

+    public async Task GetAllTopLevelBrowsingContexts()
+    {
+        var window = (await bidi.BrowsingContext.CreateAsync(ContextType.Window)).Context;
+
+        var contexts = (await bidi.BrowsingContext.GetTreeAsync()).Contexts;
+
+        Assert.AreEqual(2, contexts.Length);
+        Assert.AreEqual(contexts[1].Context, window);
+    }

Evidence

The test currently relies on contexts[1] to refer to the created window without validating the
ordering contract for the returned array.

examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.GetTree.cs[33-42]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
The test asserts `contexts[1].Context == window`, assuming a specific ordering from `GetTreeAsync()`. If ordering changes, the test fails despite correct behavior.

## Issue Context
A more robust assertion is to verify that the created `window` exists somewhere in the returned contexts.

## Fix Focus Areas
- examples/dotnet/SeleniumDocs/BiDi/BrowsingContext/BrowsingContextTest.GetTree.cs[34-42]

## Suggested change
Replace the positional assertion with a containment check, e.g.:
```csharp
Assert.IsTrue(contexts.Any(c => c.Context == window));
```

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

[qodo-code-review]

Code review by qodo was updated up to the latest commit 2cdf2b0

[qodo-code-review]

Code review by qodo was updated up to the latest commit 550631c

[qodo-code-review]

Code review by qodo was updated up to the latest commit 1d4d679

[qodo-code-review]

Code review by qodo was updated up to the latest commit 5976832

[qodo-code-review]

Code review by qodo was updated up to the latest commit cb2c9ee

[qodo-code-review]

Code review by qodo was updated up to the latest commit cd561d6

[qodo-code-review]

Code review by qodo was updated up to the latest commit 39ff8cb