Digital Credentials Browser-Side Implementation

This directory contains the browser-side implementation of the Digital Credentials API.

Testing Strategy

To test the Digital Credentials API (e.g., via Web Platform Tests), we need to simulate different user behaviors (e.g., successful selection, dismissal, aborting).

Since the real UI is platform-specific and requires user interaction, we use fake implementations for testing.

1. Default Behavior (Hanging)

By default in test environments, we want the API calls (navigator.credentials.get() and navigator.credentials.create()) to hang (i.e., remain pending). This simulates the browser waiting indefinitely for the user to interact with the UI.

This hanging behavior is crucial for testing:

Aborts: Testing that AbortSignal can cancel a pending request.
Concurrency: Testing that a second request is rejected if a first request is already pending.

To achieve this:

Chrome: The desktop provider (DigitalIdentityProviderDesktop) naturally hangs because it attempts to show a UI and waits for user interaction (which never happens in automated tests).
Content Shell: ShellContentBrowserClient overrides CreateDigitalIdentityProvider to return ShellDigitalIdentityProvider, which hangs by default (does not invoke the callback).
Headless Shell: HeadlessContentBrowserClient overrides CreateDigitalIdentityProvider to return HeadlessDigitalIdentityProvider, which also hangs by default.

2. Auto-Resolve Behavior (Fake UI)

For tests that expect a successful token retrieval (e.g., wpt_internal tests), we can configure the fake UI to auto-resolve with a fake token after 1ms.

This is enabled by passing the command-line flag: --use-fake-ui-for-digital-identity

When this flag is present:

DigitalIdentityRequestImpl bypasses the embedder's provider.
It immediately posts a task to resolve the request with a fake token.

This flag is used in the digital-credentials-fake-ui virtual test suite (defined in third_party/blink/web_tests/VirtualTestSuites) for running success path tests.