Issue 72349

Created attachment 41192 [details]
demo document

Created attachment 41193 [details]
patch to fix

Reassigned. Please handle.

cl->tl: please evaluate and integrate if this patch is ok

.

The patch works for selected characters but not if one simply travels with the
cursor from char to char.

The following code works well with both cases:

:    for ( USHORT n = 0; n < rTypes.Count(); n++ )
:    {
-/+     if (rTypes[n].nStartPos <= nS  &&  nE <= rTypes[n].nEndPos)
:        {
:            if ( rTypes[n].nScriptType != i18n::ScriptType::WEAK )
:            {
:                nScriptType |= GetItemScriptType ( rTypes[n].nScriptType );
:            }
:            else
:            {
:                if ( !nScriptType && n )
:                {
:                    // #93548# When starting with WEAK, use prev ScriptType...
:                    nScriptType = rTypes[n-1].nScriptType;
:                }
:            }
+            break;
:        }
:    }

Fixed in CWS tl37.

Files changed:
- svx/source/editeng/impedit2.cxx

.

The detection of the scripttype is fixed. 
For example copy the characters from the presentation sample document into a
Writer text box. 

It still does not work as expected in Draw/Impress and Calc though!
This is because sd and sc choose to overwrite that scriptype with the one from
the keyboard input language. Which should not happen. Since I do not want to
mess with the implementation of the use-keyboard-language feature I'm going to
write separate issues to sd and sc. 

I submitted a follow-up for the remaining problem in Draw/Impress and Calc (see
issue 77682).

Created attachment 45557 [details]
document for testing

I'm not sure that the final fix was right either, what I think we want to get
out of this function is the combination of scripts which are active for the
selection.

i.e. the sum of scripts which intersect with the selection. So if we have two
scripts used inside the selection we want to get both of them back. While the
final patch gets the script of the range which completely encloses the selection. 

As a side problem, when we have no selection, just a cursor to get the script of
the previous character. Basically the selection of the previous character. 

cmc->tl/mru: I think the above document and ctrl+a and change fontsize should
show what I mean. In a mixed script selection in the workspace install sets I
think only the western script will change size, not both of them. 

Does the below attached patch seem plausible instead ? or maybe I'm wrong and
the above testcase shows no regression ?

Created attachment 45558 [details]
patch

TL->CMC: I see the problem. Still that problem is a new one entirely different
from the original one described in this issue.
Thus the original problem is fixed and the new found issue should not prevent
this CWS from being integrated.
Please submit the patch to the new problem in an issue of it's own. 
Thanks in advance!

done, issue 78017

Reassigned to SBA.

SBA: Verified in CWS tl37.

closed