Skip to content

armgit / bento

Public Imported from GitHub
0

Bento, the office suite that fits in a file

iOS host: decline element fullscreen everywhere, hide the iPad status bar

nyblnete803589098d0
diff --git a/tray/ios/DocumentBrowserViewController.swift b/tray/ios/DocumentBrowserViewController.swift
index c3ec5b69..6db1f3eb 100644
--- a/tray/ios/DocumentBrowserViewController.swift
+++ b/tray/ios/DocumentBrowserViewController.swift
@@ -96,7 +96,7 @@ final class DocumentBrowserViewController: UIDocumentBrowserViewController,
let editor = EditorViewController(document: doc)
// Two extensions: "Q3-board.bento.html" -> "Q3-board".
editor.title = url.deletingPathExtension().deletingPathExtension().lastPathComponent
- let nav = UINavigationController(rootViewController: editor)
+ let nav = DocumentNavigationController(rootViewController: editor)
nav.modalPresentationStyle = .fullScreen
editor.onDone = { [weak self, weak nav] in
// close() flushes and relinquishes file coordination; without it
@@ -113,3 +113,14 @@ final class DocumentBrowserViewController: UIDocumentBrowserViewController,
}
}
}
+
+/// Lets the editor decide whether the status bar is shown.
+///
+/// A UINavigationController answers UIKit's status-bar questions itself unless
+/// it is told to defer, so `prefersStatusBarHidden` on the view controller
+/// inside it is simply never consulted — the editor's request to hide the bar
+/// on iPad would be silently dropped.
+final class DocumentNavigationController: UINavigationController {
+ override var childForStatusBarHidden: UIViewController? { topViewController }
+ override var childForStatusBarStyle: UIViewController? { topViewController }
+}
diff --git a/tray/ios/EditorViewController.swift b/tray/ios/EditorViewController.swift
index d81e02c6..b071ead3 100644
--- a/tray/ios/EditorViewController.swift
+++ b/tray/ios/EditorViewController.swift
@@ -131,8 +131,25 @@ final class EditorViewController: UIViewController, WKScriptMessageHandler, WKUR
view.bringSubviewToFront(floatingExit)
}
- /// Start the page BELOW the status bar wherever iOS is drawing one, so a
- /// document's own toolbar is reachable; go full bleed where it is not.
+ /// Hide the status bar wherever nothing forces it to exist.
+ ///
+ /// On iPad it is pure cost. There is no sensor housing to reserve a band
+ /// for, so the status bar is the ONLY thing keeping the page off the whole
+ /// screen — and it stayed lit over a presenting deck, which is the opposite
+ /// of what presenting is for. Hiding it takes safeAreaInsets.top to 0, and
+ /// the deck then goes edge to edge with even letterboxing.
+ ///
+ /// On iPhone it would buy nothing: landscape hides it already, and portrait
+ /// draws it INSIDE the band the sensor housing reserves regardless, so
+ /// hiding there would blank that band rather than reclaim it — losing the
+ /// clock to gain no pixels.
+ override var prefersStatusBarHidden: Bool {
+ UIDevice.current.userInterfaceIdiom == .pad
+ }
+
+ /// Reserve exactly what the system says is unsafe at the top, and nothing
+ /// else — so a document's own toolbar is reachable but never gives up a
+ /// pixel it did not have to.
///
/// Done NATIVELY rather than by asking the page to pad itself. env() is dead
/// in this WKWebView, and --tray-safe-* only helps a page that has heard of
@@ -140,19 +157,18 @@ final class EditorViewController: UIViewController, WKScriptMessageHandler, WKUR
/// controls ended up under the pill and could not be tapped. Insetting the
/// web view works for every document without any cooperation.
///
- /// A regular vertical size class is the test because it tracks exactly the
- /// thing that matters: iOS hides the status bar in a compact height and
- /// shows it otherwise. On iPhone that reads as portrait-inset /
- /// landscape-full-bleed, which is what a presenter wants. On iPad it is
- /// regular in BOTH orientations — and correctly so, because an iPad keeps
- /// its status bar in both, and a deck presenting there letterboxes anyway.
+ /// `safeAreaInsets.top` is the whole rule now, and it already says the right
+ /// thing everywhere: iPhone portrait reports the sensor housing, so the page
+ /// starts below the pill; iPhone landscape and iPad both report 0 once the
+ /// status bar is gone, so the page gets the entire screen. No orientation or
+ /// device test — the earlier one claimed landscape was always full bleed,
+ /// which iPad quietly disproved.
override func viewDidLayoutSubviews() {
super.viewDidLayoutSubviews()
- let statusBarShowing = traitCollection.verticalSizeClass == .regular
- let top = statusBarShowing ? view.safeAreaInsets.top : 0
+ let top = view.safeAreaInsets.top
webView.frame = CGRect(x: 0, y: top, width: view.bounds.width,
height: max(0, view.bounds.height - top))
- publishSafeArea(topHandledNatively: statusBarShowing)
+ publishSafeArea(topHandledNatively: top > 0)
}
override func viewDidAppear(_ animated: Bool) {
@@ -215,26 +231,24 @@ final class EditorViewController: UIViewController, WKScriptMessageHandler, WKUR
override func viewDidLoad() {
super.viewDidLoad()
let cfg = WKWebViewConfiguration()
- // Element fullscreen: iPad only.
- //
- // WKWebView will grant it on iPhone too, and it looked like a win — the
- // same file in Safari cannot do it. But WebKit's fullscreen view brings
- // its own close button, which no public API can hide or restyle, and
- // insets the content so a 16:9 deck letterboxed asymmetrically instead
- // of filling the screen.
+ // Element fullscreen is DECLINED on every device.
//
- // On a phone that buys nothing: the status bar is ALREADY hidden in
- // landscape, so declining fullscreen costs no screen space at all, and
- // the page then fills the web view edge to edge with its own chrome and
- // no foreign ✕. Declining is simply better here.
+ // WKWebView offers it as an opt-in that mobile Safari never gives a
+ // page, so it looked like free capability. It is not. WebKit's
+ // fullscreen view brings its own close button that no public API can
+ // hide, restyle or move, and it insets the content — so a 16:9 deck
+ // letterboxed asymmetrically, and the foreign ✕ spilled off the
+ // letterbox onto the slide. It does not even hide the status bar on
+ // iPad, so it fails at the one thing fullscreen is for.
//
- // iPad is the opposite case — the status bar IS visible there and there
- // is no notch gutter to fight — so it keeps real fullscreen.
+ // Declining costs nothing, because the host hands the page the whole
+ // screen anyway (see prefersStatusBarHidden and viewDidLayoutSubviews):
+ // the deck then fills the web view edge to edge, letterboxes evenly,
+ // and wears its OWN chrome instead of WebKit's.
//
- // A page that asks for fullscreen and is refused is not broken: that is
- // the same path it takes in mobile Safari, which is well-trodden.
- cfg.preferences.isElementFullscreenEnabled =
- UIDevice.current.userInterfaceIdiom == .pad
+ // A page refused fullscreen is not broken — that is exactly the path it
+ // takes in mobile Safari, which is the well-trodden one.
+ cfg.preferences.isElementFullscreenEnabled = false
cfg.setURLSchemeHandler(self, forURLScheme: "bento-tray")
cfg.userContentController.add(self, name: "bentoFile")