Hebrew search in PDFs is backwards?

[Gadi Cohen]

Diego, thanks for taking the time to check this out.  Appreciate it.


I will play around and see what I can do, but I'll only have a chance
after exams.

I'll also take a look at the Evince/GNOME/Pango situation as well.


Dotan - it's not strictly a bug.  Diego has pointed out that qt (the
underlying library used by KDE and hence Okular) has exactly the
function we need - one that can convert logical Hebrew to visual
Hebrew.  The "bug" is that this is probably a private function
(accessible only from within qt, and not from within applications that
are based on qt) - although there's a chance we can still use it.  If
not we can file a bug report and try get the function made public in the
next big qt update.


In short - I will play around after exams, and file the relevant bug
report when I know what's going on.


Thanks guys... this has been quite productive, it will be really nice if
we can have open source Hebrew searching in PDFs :))


Gadi


Dotan Cohen wrote:

> 2010/1/19 Diego Iastrubni <elcuco at kde.org>:
>   
>> This is really, really bad.... from qt/src/gui/painting/qpainter.cpp, inside
>> void QPainter::drawText(const QPointF &p, const QString &str, int tf, int
>> justificationPadding)
>>
>> I see this code:
>>    QStackTextEngine engine(str, d->state->font);
>>    engine.option.setTextDirection(d->state->layoutDirection);
>>    engine.itemize();
>>    QScriptLine line;
>>    line.length = str.length();
>>    engine.shapeLine(line);
>>    int nItems = engine.layoutData->items.size();
>>    QVarLengthArray<int> visualOrder(nItems);
>>    QVarLengthArray<uchar> levels(nItems);
>>    for (int i = 0; i < nItems; ++i)
>>        levels[i] = engine.layoutData->items[i].analysis.bidiLevel;
>>    QTextEngine::bidiReorder(nItems, levels.data(), visualOrder.data());
>>
>> Now... it seems like QTextEngine::bidiReorder()  does exactly what you want
>> logical->visual. However... this class is defined inside qtextengine_p.h...
>> which means this is not official API.
>>
>> But, IMHO, even if it's not marked, you can open a bug report on this at
>> Nokia's bug tracker, and they will handle it for Qt 4.8, maybe 4.7. Meanwhile,
>> if the header is installed - abuse it. If it's not included - hack it inside
>> your application
>>
>> (*) I know that this will not be accepted at KDE's svn ...
>>
>>     
>
> Could you file that bug, Diego? It is far above my understanding or
> level. Thanks!
>
> Here is the Qt bug tracker:
> http://bugreports.qt.nokia.com/secure/CreateIssue!default.jspa
>
>
>   


-- 
Gadi Cohen aka Kinslayer <dragon at wastelands.net> www.wastelands.net
Freelance admin/coding/design HABONIM DROR linux/fantasy enthusiast
KeyID 0x93F26EF5: 256A 1FC7 AA2B 6A8F 1D9B 6A5A 4403 F34B 93F2 6EF5

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cs.huji.ac.il/pipermail/linux-il/attachments/20100120/8de1f179/attachment.html>