@RendijsSmukulis Thanks for the post. We are tracking extension related feature request under #191.

@novac42 , there's a big difference between extensions with UI support, and extensions without. Does it make sense to track both feature requests under one ticket?

Related: #3259

@MarkIngramUK 191 is the overall post for extensions requests but I understand your point. I will add tracked label to generate a separate item in backlog for UI-less extensions specifically.

Thanks @novac42 , I appreciate the consideration.

As the Extensions API proposal is now published, as discussed previously we're happy to provide some background how we intend to use the feature.

Background

We are currently looking to migrate a subset of our Privacy Essentials Extension functionality to the our WebView2-based browser using the proposed WebView2 Extensions API.

As the initial version of this API is understood to be "UI-less", that is, will not provide ways to surface UI elements to the user (such as the extension popup window or the extension's status icon on the address bar), we are looking for ways to host these elements in our application's native components.

After the review of the proposed Extensions API, we have two high-level questions to judge the feasibility of what we're trying to achieve:

1. How will communications between the native app an extension be supported?
2. What set of Extension APIs will be expected to work on initial release?

1. Extension-To-Native Communications

To be able to host the extension's UI in a native control, separate from WebView2, the state of the extension needs to be communicated to the native code in some way. For example, when user modifies a setting in the native app, the setting change needs to be sent to the extension to modify its behaviour. Similarly, when some event happens in the extension (e.g. it has blocked a network request), this information needs to flow through the native code to be presented to the user.

Similarly, the two-way communication is also required to fill in API gaps for concepts that exist outside a webview, such as "tabs" and "windows". For example, there is no way for WebView2 to correctly respond to the extensions query of "What tabs exist in the active window" if it has no concept of:

• Current collection of windows,
• Currently active window
• Tabs in windows
• Information specific to those tabs, such as sizing (or at least what WebView2 instance is associated with said tab).

As no API is currently presented to convey these concepts from native app to webview2, a method of communicating between the extension and native code is required to create "polyfill" that can allow the app's native code side to respond to the queries (even if this requires changes to the extension's code to use the polyfill version rather than the "chrome.tabs", "chrome.window" or similar APIs)

Apple is currently in the process of adding extension support to their WKWebView, and has published some interfaces that, implemented by the host app, allows to communicate the required information to the WebView:

• Their API is also allowing the host application to communicate such info as tab and window data to the extensions by providing interfaces that can be implemented in the host app to represent concepts that the extensions "understand", e.g.:
• _WKWebExtensionTab.h interface to communicate a tab state to the extension and accept commands from the extension, e.g.:
  • (BOOL)isSelectedForWebExtensionContext:(_WKWebExtensionContext *)context; to indicate if the tab is the currently selected tab
  • (void)reloadForWebExtensionContext:(_WKWebExtensionContext *)context; to force a reload of the specific frame
• and _WKWebExtensionWindow.h to achieve the same, but for a window, e.g.:
  • NSArray<id <_WKWebExtensionTab>> *)tabsForWebExtensionContext to get the tabs in a specific window
  • (BOOL)isFocusedForWebExtensionContext:(_WKWebExtensionContext *)context; to indicate if the window is currently focused
  • (NSRect)frameForWebExtensionContext:(_WKWebExtensionContext *)context; to get the window size/location

2. What Extension APIs Will Be Supported?

The second open question is what is the set of supported APIs for extensions in the initial release. It's clear that such apis as "chrome.tabs" and "chrome.windows" can't currently be supported, but it is not clear what other APIs might be expected to not be functional. We provide a list of APIs we currently rely on in Appendix A.

Appendix A

What APIs the Privacy Extension currently relies on

Note: this list skips UI-specific apis, as the current iteration of WebView2 is not expected to support any UI elements for extensions.

webRequest

chrome.webRequest.onBeforeRequest.addListener - multiple features
chrome.webRequest.onBeforeSendHeaders.addListener - 3rd party cookie protection, GPC header protection, referrer trimming feature
chrome.webRequest.onHeadersReceived.addListener - 3rd party cookie protection, ad click detection

webNavigation

chrome.webNavigation.onCommitted.addListener
chrome.webNavigation.onCommitted.removeListener
chrome.webNavigation.onCreatedNavigationTarget.addListener
chrome.webNavigation.onBeforeNavigate.addListener
chrome.webNavigation.onCompleted.addListener

declarativeNetRequest

Same functionality as the webRequest API above, required for when MV2 is no longer supported
chrome.declarativeNetRequest.updateSessionRules
chrome.declarativeNetRequest.updateDynamicRules

runtime

chrome.runtime.onMessage.addListener - multiple features (e.g. injected content scripts communicating with the extension)
chrome.runtime.sendMessage - as above

tabs

chrome.tabs.update(thisTab.id, { url: newUrl }): for AMP protections
chrome.tabs.sendMessage(tab.id, {..}): autofill
chrome.tabs.onCreated.addListener: keeping track of current tabs, multiple features
chrome.tabs.onUpdated.addListener((id, info) ..: keeping track of current tabs, multiple features
chrome.tabs.onRemoved.addListener((id, info) ..: keeping track of current tabs, multiple features
chrome.tabs.query({}) - autofill

scripting

chrome.scripting.insertCSS
chrome.scripting.registerContentScripts

storage

chrome.storage.local.set
chrome.storage.local.get
chrome.storage.session.set
chrome.storage.session.get
chrome.storage.session.remove