fix: ensure history navigations are sandboxed-iframe-aware

[jeremyspiegel]

Description of Change

This fixes #35391

When a sandboxed iframe calls history.go(), the renderer calls into the browser at RenderFrameHostImpl::GoToEntryAtOffset:

  if (delegate_->IsAllowedToGoToEntryAtOffset(offset)) {
    if (IsSandboxed(network::mojom::WebSandboxFlags::kTopNavigation)) {
      // Keep track of whether this is a session history from a sandboxed iframe
      // with top level navigation disallowed.
      frame_tree_->controller().GoToOffsetInSandboxedFrame(
          offset, GetFrameTreeNodeId());
    } else {
      frame_tree_->controller().GoToOffsetFromRenderer(offset);
    }
  }

The call to delegate_->IsAllowedToGoToEntryAtOffset(offset) goes to WebContentsImpl::IsAllowedToGoToEntryAtOffset:

bool WebContentsImpl::IsAllowedToGoToEntryAtOffset(int32_t offset) {
  // TODO(https://crbug.com/1170277): This should probably be renamed to
  // WebContentsDelegate::IsAllowedToGoToEntryAtOffset or
  // ShouldGoToEntryAtOffset
  return !delegate_ || delegate_->OnGoToEntryOffset(offset);
}

And Electron's implementation of delegate_->OnGoToEntryOffset(offset) in electron::api::WebContents::OnGoToEntryOffset handles the navigation itself, rather than allowing RenderFrameHostImpl::GoToEntryAtOffset's default behavior of navigating in the frame instead of the parent context:

bool WebContents::OnGoToEntryOffset(int offset) {
  GoToOffset(offset);
  return false;
}

This change removes the override of OnGoToEntryOffset in electron::api::WebContents (the default implementation of WebContentsDelegate::OnGoToEntryOffset is just return true).

The override was added by @deepak1556 in #3875 to fix #3734. I think that the calling Chrome code may have changed since then, since it was so long ago (2015) and because of the comment in WebContentsImpl::IsAllowedToGoToEntryAtOffset about needing to rename WebContentsDelegate::OnGoToEntryOffset to WebContentsDelegate::IsAllowedToGoToEntryAtOffset or ShouldGoToEntryAtOffset.

Checklist

• PR description included and stakeholders cc'd
• npm test passes
• tests are changed or added
• relevant documentation is changed or added
• PR release notes describe the change in a way relevant to app developers, and are capitalized, punctuated, and past tense.

Release Notes

Notes: Fixed issue with history.back() in sandboxed iframes affecting parent browsing context.

[deepak1556]

Thanks for working on this issue!

Electron had the override before for an older process architecture where Renderer process would always restart expect for history navigations. But we have since then aligned our process architecture to that of Chromium and this hack is not required.

[jeremyspiegel]

Great, thank you for the reviews @deepak1556 and @codebytere!

[zcbenz]

@jeremyspiegel Can you rebase this PR on latest main? It should be able to fix the CI failure.

[jeremyspiegel]

@zcbenz I've rebased on latest main and the checks are passing now.

[release-clerk]

Release Notes Persisted

Fixed issue with history.back() in sandboxed iframes affecting parent browsing context.

[trop]

I have automatically backported this PR to "19-x-y", please check out #35621

[trop]

I have automatically backported this PR to "20-x-y", please check out #35622

[trop]

I have automatically backported this PR to "21-x-y", please check out #35623