Closed Bug 670087 Opened 13 years ago Closed 11 years ago

AccessibleObjectFromPoint returns incorrect accessible for popup menus

Categories

(Core :: Disability Access APIs, defect)

x86
Windows 7
defect
Not set
normal

Tracking

()

RESOLVED FIXED
mozilla25

People

(Reporter: Jamie, Assigned: surkov)

References

(Blocks 1 open bug)

Details

Attachments

(1 file)

Str: 1. Press alt+f to activate the File menu. 2. Get the location of the "New Tab" menu item. 3. Call AccessibleObjectFromPoint() with a point covered by the location retrieved in step 2. Expected: The accessible for the "New tab" menu item should be returned. Actual: Some other accessible outside the menu is returned. (I'm guessing this accessible is overlapped by the menu item.) The popup window containing the menu (window class name: MozillaDropShadowWindowClass) returns the Firefox top level frame when queried for its client (OBJID_CLIENT). IAccessible::accHitTest on the Firefox top level frame doesn't seem to know about popup menu items. The MozillaDropShadowWindowClass window should return the popup menu accessible as its client. This menu accessible should answer IAccessible::accHitTest correctly for the menu and its items. If that isn't an option, you can probably get away with having IAccessible::accHitTest on the top level frame answer for the menu and its items.
Impact: This breaks mouse tracking for Firefox menu items with NVDA. Related NVDA ticket: http://www.nvda-project.org/ticket/1647
Blocks: boundsa11y
the same issue as in bug 480347. I think it's reasonable to follow Jamie's suggestion in any case so keeping open.
Robert, what's the best way to find popup content from window (nsWindow::ProcessMessage) created for popup?
Flags: needinfo?(roc)
get the window's view's frame.
Flags: needinfo?(roc)
Hi, are there any news on this bug? We are developing a GUI automation tool and are having trouble with menus, particular in Thunderbird.
nominating as q2 target
Blocks: 2013q2a11y
Robert, one more question please. We use nsLayoutUtils::GetFrameForPoint on document root frame (http://mxr.mozilla.org/mozilla-central/source/accessible/src/generic/Accessible.cpp#838) but it seems doesn't take into account any popups shown above the document. How can I detect that the given point points to an element inside popup (originally the point is in screen coordinates)?
Flags: needinfo?(roc)
Use nsLayoutUtils::GetPopupFrameForEventCoordinates(rootPresContext, aEvent). You can look at how PresShell::HandleEvent does this.
Flags: needinfo?(roc)
Attached patch patchSplinter Review
Assignee: nobody → surkov.alexander
Status: NEW → ASSIGNED
Attachment #761495 - Flags: superreview?(roc)
Attachment #761495 - Flags: review?(trev.saunders)
Attachment #761495 - Flags: superreview?(roc) → superreview+
Comment on attachment 761495 [details] [diff] [review] patch > void* uniqueID = reinterpret_cast<void*>(-aVarChild.lVal); > > // Document. > if (IsDoc()) >- return AsDoc()->GetAccessibleByUniqueIDInSubtree(uniqueID); >+ return Document()->GetAccessibleByUniqueIDInSubtree(uniqueID); what is reason for change? >+#define NS_LOG_WMGETOBJECT(aWnd, aHwnd, aAcc) \ >+ PR_LOG(gWindowsLog, PR_LOG_ALWAYS, \ >+ ("Get the window:\n {\n HWND: %d, parent HWND: %d, wndobj: %p,\n",\ >+ aHwnd, ::GetParent(aHwnd), aWnd)); \ >+ PR_LOG(gWindowsLog, PR_LOG_ALWAYS, (" acc: %p", aAcc)); \ >+ if (aAcc) { \ > nsAutoString name; \ >- acc->GetName(name); \ >+ aAcc->GetName(name); \ use Name()? >-nsWindow::GetRootAccessible() >+nsWindow::GetAccessible() > { >+ nsView* view = nsView::GetViewFor(this); >+ if (view) { >+ nsIFrame* frame = view->GetFrame(); >+ if (frame && nsLayoutUtils::IsPopup(frame)) { >+ nsCOMPtr<nsIAccessibilityService> accService = >+ services::GetAccessibilityService(); >+ if (accService) { >+ a11y::DocAccessible* docAcc = >+ GetAccService()->GetDocAccessible(frame->PresContext()->PresShell()); >+ if (docAcc) { >+ NS_LOG_WMGETOBJECT(this, mWnd, >+ docAcc->GetAccessible(frame->GetContent())); >+ return docAcc->GetAccessible(frame->GetContent()); roc I'm assuming this stuff seems reasonable to you. I'd worry a little this might slow down the no a11y case some, but but what becomes GetRootAccessible() has already been pretty slow and could be easily speed up if someone cared I suspect so I'm not really concerned.
Attachment #761495 - Flags: review?(trev.saunders) → review+
Status: ASSIGNED → RESOLVED
Closed: 11 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla25
Depends on: 891338
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: