1 files changed, 112 insertions, 0 deletions
diff --git a/layout/base/ViewportUtils.h b/layout/base/ViewportUtils.h
new file mode 100644
index 0000000000..176ebcb93c
--- /dev/null
+++ b/layout/base/ViewportUtils.h
@@ -0,0 +1,112 @@
+/* This Source Code Form is subject to the terms of the Mozilla Public
+ * License, v. 2.0. If a copy of the MPL was not distributed with this
+ * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
+
+#ifndef mozilla_ViewportUtils_h
+#define mozilla_ViewportUtils_h
+
+#include "Units.h"
+#include "mozilla/layers/ScrollableLayerGuid.h"
+
+class nsIFrame;
+
+namespace mozilla {
+
+class PresShell;
+
+class ViewportUtils {
+ public:
+  /* Return a transform to be applied to the coordinates of input events
+     targeting content inside the scroll frame identified by |aScrollId|, which
+     converts from "visual coordinates" (which are the coordinates events have
+     when they arrive from APZ) to "layout coordinates" (which are the
+     coordinates used in most places by layout code). The transform has two
+     components:
+     1. The pres shell resolution, representing the pinch-zoom scale
+        (if the scroll frame |aScrollId| is inside the resolution, which
+        is most of the time).
+     2. A translation representing async scrolling. This can contain:
+         - For any scroll frame, a scroll component resulting from the main
+           thread incompletely applying an APZ-requested scroll position.
+         - For the RCD-RSF only, a persistent component representing the
+           offset of the visual viewport relative to the layout viewport.
+       The translation is accumulated for all scroll frames form |aScrollId|
+       up to the root, using values populated in
+       APZCCallbackHelper::UpdateCallbackTransform. See that method's
+       documentation for additional details. */
+  template <typename Units = CSSPixel>
+  static gfx::Matrix4x4Typed<Units, Units> GetVisualToLayoutTransform(
+      layers::ScrollableLayerGuid::ViewID aScrollId);
+
+  /* The functions below apply GetVisualToLayoutTransform() or its inverse
+   * to various quantities.
+   *
+   * To determine the appropriate scroll id to pass into
+   * GetVisualToLayoutTransform(), these functions prefer to use the one
+   * from InputAPZContext::GetTargetLayerGuid() if one is available.
+   * If one is not available, they use the scroll id of the root scroll
+   * frame of the pres shell passed in as |aContext| as a fallback.
+   */
+  static nsPoint VisualToLayout(const nsPoint& aPt, PresShell* aContext);
+  static nsRect VisualToLayout(const nsRect& aRect, PresShell* aContext);
+  static nsPoint LayoutToVisual(const nsPoint& aPt, PresShell* aContext);
+
+  /*
+   * These functions convert the point/rect from layout to visual space
+   * by applying the inverse of GetLayoutToVisualTransform() of the root
+   * scrollframe of provided presShell.
+   */
+  static LayoutDevicePoint DocumentRelativeLayoutToVisual(
+      const LayoutDevicePoint& aPoint, PresShell* aShell);
+  static LayoutDeviceRect DocumentRelativeLayoutToVisual(
+      const LayoutDeviceRect& aRect, PresShell* aShell);
+  static LayoutDeviceRect DocumentRelativeLayoutToVisual(
+      const LayoutDeviceIntRect& aRect, PresShell* aShell);
+  static CSSRect DocumentRelativeLayoutToVisual(const CSSRect& aRect,
+                                                PresShell* aShell);
+
+  /*
+   * These convert aPt/aRect, which is the layout space of aCtx, into a
+   * screen-relative visual-space quantity. Formally, we can think of the input
+   * as being in RelativeTo{aCtx->PresShell()->GetRootFrame(),
+   * ViewportType::Layout} space. And then the function iterates up the chain of
+   * presContexts, transforming the point along, until it gets to the
+   * RelativeTo{aCtx->GetRootPresContext()->PresShell()->GetRootFrame(),
+   * ViewportType::Visual} space. Then it further transforms it into a
+   * screen-relative space. Unlike the GetTransformToAncestor family of helper
+   * methods, this transformation does NOT include CSS transforms. Also, unlike
+   * other ViewportUtils methods, the input is relative to `aCtx` which might be
+   * a sub-shell of the presShell with the resolution.
+   */
+  static LayoutDevicePoint ToScreenRelativeVisual(const LayoutDevicePoint& aPt,
+                                                  nsPresContext* aCtx);
+  static LayoutDeviceRect ToScreenRelativeVisual(const LayoutDeviceRect& aRect,
+                                                 nsPresContext* aCtx);
+
+  /**
+   * Returns non-null if |aFrame| is inside the async zoom container but its
+   * parent frame is not, thereby making |aFrame| a root of a subtree of
+   * frames representing content that is zoomed in. The value returned in such
+   * cases is the root scroll frame inside the async zoom container.
+
+   * Callers use this to identify points during frame tree traversal where the
+   * visual-to-layout transform needs to be applied.
+   */
+  static const nsIFrame* IsZoomedContentRoot(const nsIFrame* aFrame);
+};
+
+// Forward declare explicit instantiations of GetVisualToLayoutTransform() for
+// CSSPixel and LayoutDevicePixel, the only two types it gets used with.
+// These declarations promise to callers in any translation unit that _some_
+// translation unit (in this case, ViewportUtils.cpp) will contain the
+// definitions of these instantiations. This allows us to keep the definition
+// out-of-line in the source.
+extern template CSSToCSSMatrix4x4 ViewportUtils::GetVisualToLayoutTransform<
+    CSSPixel>(layers::ScrollableLayerGuid::ViewID);
+extern template LayoutDeviceToLayoutDeviceMatrix4x4
+    ViewportUtils::GetVisualToLayoutTransform<LayoutDevicePixel>(
+        layers::ScrollableLayerGuid::ViewID);
+
+}  // namespace mozilla
+
+#endif /* mozilla_ViewportUtils_h */