In touch exploration mode an accessibility service can move
accessibility focus in response to user gestures. In this case
when the user double-taps the system is sending down and up
events at the center of the acessibility focused view. This
works fine until the clicked view's center is covered by another
clickable view. In such a scenario the user thinks he is clicking
on one view but the click is handled by another. Terrible.
This change solves the problem of clicking on the wrong view
and also solves the problem of clicking on the wrong window.
The key idea is that when the system detects a double tap or
a double tap and hold it asks the accessibility focused node
(if such) to compute a point at which a click can be performed.
In respinse to that the node is asking the source view to
compute this.
If a view is partially covered by siblings or siblings of
predecessors that are clickable, the click point will be
properly computed to ensure the click occurs on the desired
view. The click point is also bounded in the interactive
part of the host window.
The current approach has rare edge cases that may produce false
positives or false negatives. For example, a portion of the
view may be covered by an interactive descendant of a
predecessor, which we do not compute (we check only siblings of
predecessors). Also a view may be handling raw touch events
instead of registering click listeners, which we cannot compute.
Despite these limitations this approach will work most of the
time and it is a huge improvement over just blindly sending
the down and up events in the center of the view.
Note that the additional computational complexity is incurred
only when the user wants to click on the accessibility focused
view which is very a rare event and this is a good tradeoff.
bug:15696993
Change-Id: I85927a77d6c24f7550b0d5f9f762722a8230830f
We get calls for window changes from the window manager and
as a result send accessibility events for these changes. It
is possible that the call for a windows change from the window
manager comes as a result of an unprivileged client calling
into the window manager. However, the event sending code
is performing security checks.
bug:17364244
Change-Id: Ief016f9dafd13ac35418676817848b3ea3ed4d66
The isVisibleToUser property of an AccessibilityNodeInfo specifies
whether the user can see the source view. It is used by accessibility
services to figure out whether to focus on a view. This property
was giving a wrong value if the view is covered by another window
such as the keyboard. As a result the user hears one thing but when
double taps interacts with the overlaid window which is another thing.
bug:15938254
Change-Id: Ib9feb20ea422a24a512c47ed1234961ae0386a7f
1. There was a bug in the touch explorer which was crashing almost
every time after accessibility was enabled via the gesture. The
problem was that in dragging state when a finger goes up we were
not transitioning to touch exploring state.
2. The global actions dialog was not going away after enabling
accessibility while it should as the user brought it up to
turn accessibility on rather to interact with global actions.
bug:15254250
Change-Id: Iaa45f758e09566822775b53e87d2980138e85ef9
1. An external contribution changed the ordering of views for
accessibility. While it attempted to fix a platform issue
for a comparator breaking transitivity, it changed the way
we order views and results in very unnatural accessibility
traversal order. It also broke CTS tets. This change tweaks
the comparator which fixes the tests and improves traversal
order.
2. If there is at least one accessibility service which cares
about windows we register a callback in the window manager
for window change notifications. We are updating the window
list on this callback. There was a case where if the service
requests window updates and immediately asks for the windows
it gets none as we have not received a callback from the
window manager yet. Now this call returns after we get the
callback in a timed fashion. This is consistent with how the
other introspection APIs work.
3. Window info objects are cached in the accessibility service
process. When putting them in the cache a cloning call was
missing resulting in some cases of clobbering windows given
to the client. For example, we get some windows, cache them,
and return these windows to the client. Now a call to clear
the cache arrives while the user processes the windows and
the client windows get clobbered.
4. Added API for checking if a window has accessiblity focus
to be consistent to the API we have to check whether this
window has input focus.
5. Removed some obsolete code.
bug:16402352
Change-Id: Ided6da4a82cc0fc703008c58a2dff0119a3ff317
Fixed a bug caused by a missing temporary when swapping two
matrices. This could produce undefined results in the case where
both display inversion and the greyscale daltonizer were enabled
simultaneously.
Tightened up the code a little to make this kind of error
less likely by removing redundant state.
Removed the native prefix on some methods since they are not
native at all.
Change-Id: I716ffc8fbe76a304c60d45870074340c0121059f
We are caching the window data in the accessibility service process.
When windows change we were sending the dalta of the windows the
service knows about. To make this work when the app asked for all
windows we had to call into the system as new windows may have
appeared. This was slow.
Now we are telling the service some windows change and if it gets
the windows we cache them. We call into the system only on a cache
miss and evict all windows from the cache on window change event.
We do not evict the nodes of the window as the former may have
just moved. If views in a window change they fire accessibility
events that trigger the correct eviction.
Change-Id: I586a72a2497b0d44a75288fa758e7e88817f3300
b/14624452
Adds a feature which draws all text (in the HW accelerated standard
path) in a high contrast mode. Text is drawn at full alpha, and either
white or black (depending on its original color) with a starkly
contrasted outline beneath it.
Change-Id: I943f624b6367de35367cced3b2a8298f2bc62377
Also fixes an infinite recursion bug in the WindowManagerService
implementation of WindowManagerInternal.
BUG: 16129909
Change-Id: I4f9d32f4e6c3ad460652c5e5271540fa5032a1f5
1. The APIs for introspecting interactive windows were reporting only
the touchable windows but were missing the focused window. The user
can interact with the latter by typing, hence it should always be
reported. Also this was breaking backwards compatibility as if the
focused window is covered by a modal one, the focused window was not
reporeted and this was putting the active window in a bad state as
the latter is either the focused window or the one the user is touching.
2. Window change events are too frequent as on window transition things
are chanign a lot. Now we are trottling the windows changed events
at the standard recurring accessibility event interval.
3. Fixed a wrong flag comparison and removed some unneded code.
buy:15434666
bug:15432989
Change-Id: I825b33067e8cbf26396a4d38642bde4907b6427a
When the position and size of a window changes we have to report that
to the accessibility layer if the window introspection is enabled.
bug:15569915
Change-Id: I3f869e0a582592bfa5f3743d5c2133ee8cb39b34
1. AccessibiltiyAction was incorectly throwing an exception when
a custom action was constructed.
2. AccessibilityManagerService should no longer enforce only standard
actions as we allow custom ones too.
bug:15110963
Change-Id: Iea57e0a6449b87bd8d103c55ca255e80705f2565
Makes the inversion bit public so apps can be smarter about rendering
images non-inverted (if they are so inclined).
BUG: 14680114
Change-Id: I8ca2b2517af2ffc13446d7e0b4e859c1171e2d0e
For user profiles the accessibility layer will use the accessibility
state of the their owner. A profile cannot have its own plugins or
settings. The rationale behind this is that if a user with impaired
vision is using the device with different profiles linked to his
main one, then he/she should be able to interact with the device no
matter what profile the app is being run with. Hence, The behavior
of the accessibility features should be consistent across apps run
with different profile.
bug:14564491
Change-Id: Id19ce81e1a2dfa48015c7ab3053d66c058be3e46
Now that we have APIs to query all interactive windows and allow
an accessibility service to put accessibility focus in each of
them we have to guarantee that there is a single accessibility
focus. This is required for correct operation of the touch
explorer as on double tap in clicks in the center of the focused
area, hence having more that one focus is an issue. Also the
system is maintaining a single input focus so now accessibility
focus behaves consistently with that.
bug:13965563
Change-Id: I0b5c26dadfabbf80dbed8dc4602073aa575ac179
The accessibility manager service was not handling the focused
window properly which as a result was breaking the active window
tracking.
bug:13788022
Change-Id: Iaa3fadfbf7b9545f5151e3603beeda2662fe7618
This change achieves a couple of things:
- Let Keyguard be a library, so we can use it in SystemUI.
- Introduce FLAG_KEYGUARD for windows and deprecate TYPE_KEYGUARD. Make
all the TYPE_KEYGUARD behaviour dependant on the flag.
- Implement a new KeyguardService in SystemUI, and bind that service
from PhoneWindowManager.
- Introduce BaseStatusBar.setKeyguardState and inflate
KeyguardSimpleHostView there and use FLAG_KEYGUARD for the window, such
that the status bar window essentially gets the Keyguard.
Bug: 13635952
Change-Id: I059d80d8b9b9818a778ab685f4672ea2694def63
1. The old introspection model was allowing querying only the active window
which is the one the user is touching or the focused one if no window is
touched. This was limiting as auto completion drop downs were not inspectable,
there was not way to know when the IME toggles, non-focusable windows were
not inspectable if the user taps them as until a screen-reader starts
introspecting the users finger is up, accessibility focus was limited to
only one window and the user couldn't use gestures to visit the whole UI,
and other things I can't remember right now.
The new APIs allow getting all interactive windows, i.e. ones that a
sighted user can interact with. This prevents an accessibility service
from interacting with content a sighter user cannot. The list of windows
can be obtained from an accessibility service or the host window from an
accessibility node info. Introspecting windows obey the same rules for
introspecting node, i.e. the service has to declare this capability
in its manifest.
When some windows change accessibility services receive a new type
of event. Initially the types of windows is very limited. We provide
the bounds in screen, layer, and some other properties which are
enough for a client to determined the spacial and hierarchical
relationship of the windows.
2. Update the documentation in AccessibilityService for newer event types.
3. LongArray was not removing elements properly.
4. Composite accessibility node ids were not properly constructed as they
are composed of two ints, each taking 32 bits. However, the values for
undefined were -1 so composing a 64 long from -1, -1 prevents from getting
back these values when unpacking.
5. Some apps were generating inconsistent AccessibilityNodeInfo tree. Added
a check that enforces such trees to be well formed on dev builds.
6. Removed an necessary code for piping the touch exploration state to
the policy as it should just use the AccessibilityManager from context.
7. When view's visibility changed it was not firing an event to notify
clients it disappeared/appeared. Also ViewGroup was sending accessibility
events for changes if the view is included for accessibility but this is
wrong as there may be a service that want all nodes, hence events from them.
The accessibility manager service takes care of delivering events from
not important for accessibility nodes only to services that want such.
8. Several places were asking for prefetching of sibling but not predecessor
nodes which resulted in prefetching of unconnected subtrees.
9. The local AccessibilityManager implementation was relying on the backing
service being ready when it is created but it can be fetched from a context
before that. If that happens the local manager was in a broken state forever.
Now it is more robust and starts working properly once the backing service
is up. Several places were lacking locking.
bug:13331285
Change-Id: Ie51166d4875d5f3def8d29d77973da4b9251f5c8
At startup, we check with PackageManager whether a system service is
available before attempting to load it. A system service is available
if its associated feature (similar to hardware features) is present.
This does not remove unavailable services from the compiled jar.
Change-Id: I13571805083aa4e65519a74acb52efd17b9fb3d7
Refactored the directory structure so that services can be optionally
excluded. This is step 1. Will be followed by another change that makes
it possible to remove services from the build.
Change-Id: Ideacedfd34b5e213217ad3ff4ebb21c4a8e73f85
