orca r4227 - in trunk: . src/orca/scripts/apps

[joanied svn gnome org]

Author: joanied
Date: Mon Sep 15 17:41:15 2008
New Revision: 4227
URL: http://svn.gnome.org/viewvc/orca?rev=4227&view=rev

Log:
* src/orca/scripts/apps/yelp.py:
  More work toward the fix for bug #356041 - GNOME Help (yelp) is
  inaccessible. Ginn Chen created another patch for Yelp which 
  makes things work even better. Thanks Ginn! To use Orca with
  Yelp, you just need the latest Orca and Yelp from trunk -- or
  wait for the GNOME 2.24 release. 



Modified:
   trunk/ChangeLog
   trunk/src/orca/scripts/apps/yelp.py

Modified: trunk/src/orca/scripts/apps/yelp.py
==============================================================================
--- trunk/src/orca/scripts/apps/yelp.py	(original)
+++ trunk/src/orca/scripts/apps/yelp.py	Mon Sep 15 17:41:15 2008
@@ -19,9 +19,9 @@
 
 """Custom script for Yelp."""
 
-__id__        = "$Id:$"
-__version__   = "$Revision:$"
-__date__      = "$Date:$"
+__id__        = "$Id$"
+__version__   = "$Revision$"
+__date__      = "$Date$"
 __copyright__ = "Copyright (c) 2005-2008 Sun Microsystems Inc."
 __license__   = "LGPL"
 
@@ -43,21 +43,13 @@
         #
         self.presentIfInactive = False
 
-    def onDocumentLoadComplete(self, event):
-        """Called when a web page load is completed."""
-
-        # We can't count on this event, but often when the first Yelp
-        # window is opened, this event is emitted. If we silently set
-        # the locusOfFocus here, the frame won't be the locusOfFocus
-        # for the initial call to getDocumentFrame(). As an added
-        # bonus, when we detect this condition, we can place the user
-        # at the top of the content. Yelp seems to prefer the end,
-        # which is confusing non-visually.
+        # Each time the user presses Return on a link, we get all the events
+        # associated with a page loading -- even if the link is to an anchor
+        # on the same page. We don't want to move to the top of the page and
+        # do a SayAll unless we have a new page. We'll use the frame name as
+        # a way to identify this condition.
         #
-        if event.source.getRole() == pyatspi.ROLE_DOCUMENT_FRAME:
-            orca.setLocusOfFocus(event, event.source, False)
-
-        return Gecko.Script.onDocumentLoadComplete(self, event)
+        self._currentFrameName = ""
 
     def getDocumentFrame(self):
         """Returns the document frame that holds the content being shown."""
@@ -103,33 +95,71 @@
         # the locusOfFocus here before sending this event on.
         #
         if self.inDocumentContent(event.source):
-            orca.setLocusOfFocus(event, event.source)
+            obj = event.source
+            characterOffset = event.detail1
+
+            # If the event is for an anchor with no text, try to find
+            # something more meaningful.
+            #
+            while obj and obj.getRole() == pyatspi.ROLE_LINK \
+                  and not self.queryNonEmptyText(obj):
+                [obj, characterOffset] = \
+                    self.findNextCaretInOrder(obj, characterOffset)
+
+            # We need to notify the presentation managers because we don't
+            # seem to get any focus/focused events for links. We seem to
+            # get caret-moved events for anchors only when the user has
+            # followed a link. Changes in the content being displayed is
+            # handled by onStateChanged(). Therefore, we need to notify the
+            # presentation managers if this event is not for an empty anchor.
+            #
+            notify = self.isSameObject(event.source, obj)
+            orca.setLocusOfFocus(event, obj, notify)
+            self.setCaretPosition(obj, characterOffset)
 
         return Gecko.Script.onCaretMoved(self, event)
 
     def onChildrenChanged(self, event):
-        """Called when a child node has changed. When the user follows a
-        link in an open Yelp window, the document frame content changes,
-        but we don't get any events to tell us something has loaded.
-        STILL INVESTIGATING, but it seems like we get a reliable
-        object:children-changed:add event for detail1 == -1 for the
-        document frame. Let's go with that for now....
+        """Called when a child node has changed.  The document frame emits
+        these events quite a bit, each time the page changes. We want to
+        ignore those events.
         """
 
-        if event.type.startswith("object:children-changed:add") \
-           and event.detail1 == -1 and event.source \
-           and event.source.getRole() == pyatspi.ROLE_DOCUMENT_FRAME:
-
-            #print "Yelp object:children-changed:add", obj, characterOffset
-
-            # If the frame is the locusOfFocus, getCaretContext() will fail
-            # because we don't want to look from the top down to find the
-            # document frame and cause Yelp to have an identity crisis. So
-            # let's just set it here and avoid that whole mess.
+        if event.source.getRole() != pyatspi.ROLE_DOCUMENT_FRAME:
+            Gecko.Script.onChildrenChanged(self, event)
+
+    def onStateChanged(self, event):
+        """Called whenever an object's state changes.
+
+        Arguments:
+        - event: the Event
+        """
+
+        if event.type.startswith("object:state-changed:busy") \
+           and event.source.getRole() == pyatspi.ROLE_DOCUMENT_FRAME \
+           and not event.detail1:
+
+            # We need to set the locusOfFocus to the document frame in
+            # order to reliably get the caret context.
             #
             orca.setLocusOfFocus(event, event.source, False)
             [obj, characterOffset] = self.getCaretContext()
-            if obj:
+
+            # Often the first object is an anchor with no text. Try to
+            # find something more meaningful and set the caret there.
+            #
+            while obj and obj.getRole() == pyatspi.ROLE_LINK \
+                  and not self.queryNonEmptyText(obj):
+                [obj, characterOffset] = \
+                    self.findNextCaretInOrder(obj, characterOffset)
+
+            if not obj:
+                return
+
+            if event.source.name == self._currentFrameName:
+                orca.setLocusOfFocus(event, obj)
+            else:
+                self._currentFrameName = event.source.name
                 self.setCaretPosition(obj, characterOffset)
                 if obj.getState().contains(pyatspi.STATE_FOCUSED):
                     speech.speakUtterances(\
@@ -140,5 +170,6 @@
                 elif settings.enableSpeech:
                     self.sayAll(None)
 
-        else:
-            Gecko.Script.onChildrenChanged(self, event)
+            return
+
+        Gecko.Script.onStateChanged(self, event)