Apache OpenOffice (AOO) Bugzilla – Issue 19430
Rotated content in spreadsheet cells are cut off
Last modified: 2007-02-14 15:22:05 UTC
Content that is rotated in a spreadsheet cell is truncated. Gnopernicus only reports the part that is visible when the content would have alligned horizontally.
Created attachment 9190 [details] Output of JUAB with XAccessibleText logging enabled
The code that calculates the text segment for getTextAtIndex() with text type LINE seems not to take rotation into account (see log file).
Sascha, please have a look. Rotated text is not an issue for draw/impress, since we don't make text accessible in non-edit mode.
KA: I don't consider this a relevant task for OOo 2.0
.
We don't want to make too many changes for Accessibility in OOo 2.0, but maybe this can be important if many spread sheets contain rotated cells. Other problem: UAA/JAA/ATK+ don't have information about rotated text, so this can not become fixed correctly, only thing that we can fix is the bounding rectangle, and only for 90/270°. Rotation is not an information in the EditEngine model, it's only passed to the drawing methods, so maybe the Accessibility implementation currently doesn't have that information?
Actually this is not about providing information about the rotation itself. The point is, the text is cut off on XAccessibleText interface. Stating from the log file: [label] Cell A1: getTextAtIndex(LINE,0) returns (0,12,Abcdefghijkl) [label] Cell A1: getTextAtIndex(WORD,0) returns (0,23,Abcdefghijklmnopqrstuvw) So somewhere in the office code, the wrong coordinates are used to calculate the LINE bounderies.
Yeah. Of course, for line calculations, formatting has to be performed, which is done by the EditEngine in this case. If the EditEngine has no clue about the rotation, the result does not surprise me. But before we invest time and energy into this bug, maybe we should discuss what is needed here. Is it only the correct part of the visible content, or should rotated text in general work? If the latter, IMHO lots more is lacking, first of all the fact that bounding boxes cannot really be rotated (and therefore, character bounds will grossly overlap for multiples of pi/4).
As discussed with TZ target goes to OOo2.0.
removed SO9 Flag according to latest comment from RU.
I take it.
Due to limited time ressources I re-Target this one to OOo Later
re-targeted to OOo 3.0
reassigning
started
MD: re-targeting to 2.2 since it's an important a11y issue
dr: can you please update the status of this issue? Can you make it into 2.2?
this issue depends on http://bugzilla.gnome.org/buglist.cgi?query=306902
fixed in CWS src680/dr53
back to QA
seen ok into CWS dr53
Seen ok into the master -> closed