[poppler] Vertical or horizontal writing?

Tue Jul 27 01:10:40 PDT 2010

Hi,

   I've patched and rebuilt the poppler-qt4 binding so far, and it's ok. 
   Now, I'm not sure whether to design a new search function by considering both horizontal and vertical modes at the same time by wMode() directly or just do the regular horizontal search first and vertical one afterwards benefits. The former seems a bit of complicated but more useful. :-( 
   In fact, I wrote a vertical search function by reconstructing the TEXTBOX list in every page by the top-to-bottom and right-to-left order. So, it seems the latter would reduce the coding work.
    I'll finish the evaluation ASAP. Thx.

         Cobra

-----Original message-----
From:mpsuzuki at hiroshima-u.ac.jp
To:cobra.yu at hyweb.com.tw
Cc:poppler at lists.freedesktop.org
Date:Mon, 26 Jul 2010 18:57:14 +0900
Subject:Re: [poppler] Vertical or horizontal writing?

Hi,

Attached is a draft patch to do:

1) extend poppler's FontInfo object to hold writing mode info.

2) extend poppler-qt4's FontInfo object to provide an API to
   return writing mode info copied from poppler's FontInfo.

3) extend qt4/tests/poppler-fonts to show writing mode info.

Also I attached a sample PDF including vertical/horizontal
Japanese text, generated by MS Word + Adobe Acrobat.
poppler-fonts detects the writing mode aslike:

$ ./qt4/tests/poppler-fonts vert-horiz-ipa-std.pdf 
name                                 type         emb sub font file wmode
------------------------------------ ------------ --- --- --------- -----
LDGIFP+IPAMincho                     CID TrueType      yes yes  Vert
LDGIFP+IPAMincho                     CID TrueType      yes yes  Hori

I think the diff for poppler-fonts.cpp shows how to use the
extended API. Cobra, please check if the extended API can help
your application.

The extention for poppler-glib is work in progress.

Regards,
mpsuzuki

On Mon, 26 Jul 2010 16:31:04 +0800 (CST)
"cobra.yu" <cobra.yu at hyweb.com.tw> wrote:

>Hi,
>
>Hmm, it seems that... 
>1) I'm only familiar with the poppler-qt4 binding for the present time. But, for my application, I really needs some CJK-related information through poppler.
>2) I can't exactly figure out all the necessary steps to accomplish what you said. But, I think, I could try, of course, by your help.
>3) So, let's work it out. :-)
>
>       Cobra
>-----Original message-----
>From:mpsuzuki at hiroshima-u.ac.jp
>To:cobra.yu at hyweb.com.tw
>Cc:poppler at lists.freedesktop.org
>Date:Mon, 26 Jul 2010 17:07:30 +0900
>Subject:Re: [poppler] Vertical or horizontal writing?
>
>Hi,
>
>Hmm, it seems that...
>
>1) poppler-qt4 can access FontInfo object for a font used in PDF,
>but cannot access GfxFont object that poppler creates before FontInfo
>object internally.
>
>2) either poppler-glib can access FontInfo object for a font used
>in PDF, but cannot access GfxFont object that poppler creates before
>FontInfo object internally.
>
>3) according to FontInfo.h (and FontInfo.cc), FontInfo object does
>not have a method to obtain the writing mode of its source GfxFont
>object.
>
>Thus, glib binding does not help your work. If we are forced to use
>current poppler without modification, we will have to use both of
>poppler-qt4 and raw poppler library. I'm afraid it makes your software
>difficult to maintain. If we have sufficient time, I think, making
>a proposal patch to add writing mode info to FontInfo object, and
>extend poppler-{qt4,glib} to access the info is better.
>
>I think adding writing mode to FontInfo is not difficult (I can
>draft), but making poppler-qt4 to access it is not so easy for
>me because I have no experience in Qt4 development. If you have
>sufficient time, I will ask you to draft the patch for poppler-qt4.
>How do you think of?
>
>Regards,
>mpsuzuki
>
>On Mon, 26 Jul 2010 15:46:43 +0800 (CST)
>"cobra.yu" <cobra.yu at hyweb.com.tw> wrote:
>
>>Hi,
>>
>>   Thanks a lot! I think it would be a great help to me. I'm only using poppler-qt4 binding for my application now. So, I guess, I would have to use both Qt4 and cpp/glib bindings in my application, right?
>>
>>         Cobra
>>-----Original message-----
>>From:mpsuzuki at hiroshima-u.ac.jp
>>To:cobra.yu at hyweb.com.tw
>>Cc:poppler at lists.freedesktop.org
>>Date:Mon, 26 Jul 2010 16:24:40 +0900
>>Subject:Re: [poppler] Vertical or horizontal writing?
>>
>>Hi,
>>
>>Yet I'm not sure how you use poppler in your application,
>>but GfxFont class has a method to obtain "writing mode"
>>in PDF terminology. Please see GfxFont.h, you can find:
>>
>>   /* this is default */
>>
>>   106  class GfxFont {
>>   107  public:
>>   108  
>>...
>>   206    // Return the writing mode (0=horizontal, 1=vertical).
>>   207    virtual int getWMode() { return 0; }
>>
>>   /* this is for CID-keyed font */
>>   321  class GfxCIDFont: public GfxFont {
>>   322  public:
>>...
>>   333    // Return the writing mode (0=horizontal, 1=vertical).
>>   334    virtual int getWMode();
>>
>>And, you can find how poppler identify the writing mode
>>in GfxFont.cc:
>>
>>  1763  int GfxCIDFont::getWMode() {
>>  1764    return cMap ? cMap->getWMode() : 0;
>>  1765  }
>>
>>Is this what you want?
>>
>>Regards,
>>mpsuzuki
>>
>>On Mon, 26 Jul 2010 14:49:38 +0800 (CST)
>>"cobra.yu" <cobra.yu at hyweb.com.tw> wrote:
>>
>>>Hi,
>>>
>>>     Is there any way by poppler to acquire the correct writing mode for any Document/Page/Paragraph? I'm writing some code for the vertical Chinese writing mode text search, but finding that it seems no way to know the exact mode by poppler.
>>>     Any suggestions are welcome! Thx.
>>>
>>>         Cobra
>>>_______________________________________________
>>>poppler mailing list
>>>poppler at lists.freedesktop.org
>>>http://lists.freedesktop.org/mailman/listinfo/poppler