Commits nvaccess/nvda - SemanticDiff

Add virtual buffer backend which allows for experimentation with new browsers supporting MSAA such as Google Chrome and Opera.

jcsteh committed 15 years ago

2fda9341

MSHTML NVDAObject: fix firstChild and lastChild properties so that they don't bail out if they hit an object that should be skipped (e.g. script, comment). Instead of just returning None, firstChild should call obj.next, and lastChild should call obj.previous, to to skip past one or more of those objects. Also fix a typo in lastChild, at the very bottom it called super's firstChild, should be lastChild. This should fix some non-semetrical navigation in Intrnet Explorer when viewing complex pages such as Google.

michaelDCurran committed 15 years ago

68edc43e

changes: Fix encoding.

jcsteh committed 15 years ago

17d2795a

ExcelWorksheetObject: implement firstChild property so it returns the A1 cell

michaelDCurran committed 15 years ago

eea6b868

Slight reordering of the English user guide:

michaelDCurran committed 15 years ago

0d87b929

Update changes document

michaelDCurran committed 15 years ago

e2c7f73a

Merged editableTextBehavior branch, which moves all scripts for text editing in to some edtiableText behavior NVDAObject classes, which means that such things as script key binding for editable text scripts is no longer duplicated everywhere in NVDA. Also included indirectly are some changes to NVDAObject api classes and some changes to Microsoft Excel support (e.g. being able to read cells with the mouse).

michaelDCurran committed 15 years ago

8a9b0eca

Merged NVDAObjectAPIClassRewrite

michaelDCurran committed 15 years ago

c088377c

Rewrite support for Microsoft Excel so that it again actually works (NVDAObjectAPIClassRewrite slightly broke it) and so that NVDA can announce cells as you move the mouse over them. When focus first hits an Excel sptreadsheet, apart from announcing the cell, it also now announces the worksheet name, as the worksheet is now represented in the focus ancestry.

michaelDCurran committed 15 years ago

dff85b48

as some of the API class changes some what broke the way we handle SDM controls (in Microsoft Word etc), rewrite the support a bit.

michaelDCurran committed 15 years ago

1b52eafe

Make sure that the new NVDAObjects.objectInForeground returns the same object as in main. A part from being more correct, this also allows NVDA to find the statisBar in applications that it used to be able to find it in.

michaelDCurran committed 15 years ago

595d7934

base NVDAObject's presentationType property: rather than just classing staticText as layout if it has no name and no description, instead class it as layout if its text (from its textInfo) is empty or only contains whitespace. Note that objects with no specific textInfo have an NVDAObjectTextInfo who's text is made up of name, description and value, so this works quite well.

michaelDCurran committed 15 years ago

7f739a97

IAccessible NVDAObject: name property: Woops. Using simpleParent causes infinite recursion because calling simpleParent indirectly checks name. Therefore, just use the grandparent if the parent is a window object.

jcsteh committed 15 years ago

3588fa71

IAccessible NVDAObject: name property: When checking whether to scrap the label on edit fields inside combo boxes, use simpleParent instead of parent. This simplifies the code and allows it to handle native combo boxes (which have a window object between the objects we're interested in) and combo boxes in toolkits such as Mozilla (where the combo box is a direct parent).

jcsteh committed 15 years ago

6f43a913

MSHTML NVDAObject: implement a table property. Indirectly, this allows the presentationType of tableRows to be classed as layout, thus bing ignored in review and focus ancestry.

michaelDCurran committed 15 years ago

792b7f0c

base NVDAObject's presentationType property: only class tableBody as content if it has a name or description.

michaelDCurran committed 15 years ago

01cc9e5e

Support for formatting in Java Access Bridge text controls, including OpenOffice.org.

jcsteh committed 15 years ago

797976f9

Update changes.

jcsteh committed 15 years ago

98e5eefb

MSHTML NVDAObject: findOverlayClasses(): Don't use IAccessible overlay classes if the IAccessible is for an ancestor node.

jcsteh committed 15 years ago

4549062d

MSHTML NVDAObject: constructor: optimisation: If an IAccessibleObject is supplied, assume that it is for the HTMLNode, not for one of its ancestors; i.e. HTMLNodeHasAncestorIAccessible will be False. This avoids a pointless QueryService when instantiating from a superclass.

jcsteh committed 15 years ago

19e0d701

MSHTML NVDAObject: Convert to use EditableText behavior.

jcsteh committed 15 years ago

2ab0ce86

NVDAObjects.IAccessible.MSHTML.MSHTML's next, previous, firstChild and lastChild properties: properly handle the case where the HTMLNode we get is None -- just return None, rather than trying to instanciate an object with no HTMLNode. This may have previously worked in main as the NVDAObject instanciation would have returned None, but now this does not happen so we can not rely on that.

michaelDCurran committed 15 years ago

63211573

Merge NVDAObjectAPIClassRewrite.

jcsteh committed 15 years ago

20c03e20

MSHTML NVDAObject: Remove unneeded code and fix comment.

jcsteh committed 15 years ago

a86a4e2a

Use relative import instead of unqualified import.

jcsteh committed 15 years ago

a8c61963

Merged pending

michaelDCurran committed 15 years ago

4d98331e

*Convert MVDAObjects.IAccessible.MSHTML.MSHTML in to a real APIClass by implementing kwargsFromSuper.

michaelDCurran committed 15 years ago

6eab89b9

*Improve general performance by making findBestAPIClass, getPossibleAPIClasses and kwargsFromSuper take kwargs as a dictionary, rather than packing the keyword arguments itself.

michaelDCurran committed 15 years ago

a6521d68

Initial support for text attributes for Java Access Bridge.

jcsteh committed 15 years ago

41caff07

Unmerge revision 3398 (cleanups to nvdaHelperLocal beep code). Trying it out again it *really* was stuffed up. The sampleRate was just ... wrong 4100 not 44100. Even with that corrected there was terible aliasing going on. And I was rightly told that there is no need for #define in c++. Hmmm... not really sure what I was thinking about when doing that commit... gone now.

michaelDCurran committed 15 years ago

1f6e9fbd