Bug#500132: [release-notes] [l10n] CJK users need to install poppler-data (non-free) package to view PDF files with evince or so
Hi,
2008/9/26 Jens Seidel <jensseidel@users.sf.net>:
> On Fri, Sep 26, 2008 at 04:58:13PM +0900, Noritada Kobayashi wrote:
>> 2008/9/26 Hideki Yamane <henrich@debian.or.jp>:
>> > On Thu, 25 Sep 2008 17:57:07 +0200
>> > Jens Seidel <jensseidel@users.sf.net> wrote:
>> >> Please explain in more detail. I have no problem viewing
>> >> /usr/share/doc/Debian/quick-reference/quick-reference.ja.pdf.gz
>> >> using evince 2.22.2-2 without poppler-data. Also xpdf, ... works well.
>> >
>> > Most of PDF files are not able to view without poppler-data.
>
>> To explain clearly, we cannot display fonts that are not embedded in
>
> we == evince
> or
> we == Asian people???
"Debian users" or "Free software users". ;-)
(And, "fonts" means "CJK fonts".)
>> the displayed PDF file, although we have no problems viewing PDF files
>> in which CJK fonts are embedded.
>
>> This is basically due to the Adobe CMAP issue that CMAP files for
>> PS/PDF are not approved to be modified (and thus cannot be included in
>> the "main" section in Debian). This issue is problematic for (at
>
> So there is a license problem. Isn't it strange that xpdf is able to
> display the mentioned PDF file? Yep, it outputs
> Error: Unknown character collection 'Adobe-Japan1'
> but most (all?) glyphs are readable!
Hmm..., I confirmed that Xpdf can render Japanese text in that
document without xpdf-japanese. But it seems to be a correct
behaviour. I checked fonts inside that PDF with pdffonts (included in
xpdf-utils/poppler-utils packages) and found that fonts are embedded
in that PDF. So, what is curious is that Poppler cannot display fonts
that are embedded inside the PDF. Gee... I cannot explain now.
Instead, let's try to display documentations for Tokyo Debian study
meetings, such as:
http://tokyodebian.alioth.debian.org/pdf/debianmeetingresume200806.pdf
In this PDF, fonts are not embedded. With Evince and Inkscape
(without poppler-data), Japanese text only does not show up. With
Xpdf (without xpdf-japanese), Japanese text looks garbled. This is a
behaviour that I expected.
>> least) Japanese users for around ten years (sigh...) and we work
>> around the issue by additionally installing
>> cmap-adobe-{cns1,gb1,japan1,japan2,korea1}, gs-cjk-resource, and
>> xpdf-japanese for viewing PS/PDF files.
>
> I don't have any of these packages installed in my Lenny system ...
> (But I have KDE4 packages from experimental, maybe this includes free
> font replacements?)
No, CMap is not a font; it is only a mapping. CJK-related
documentation[1] of Ghostscript explains that "A CMap provides a map
from the multibyte character code to the CID of corresponding glyph,
under a specified encoding system." It is only a data like a map of
Unicode, but OTOH it is also a PostScript program (so, it is
impossible to create a free alternative by copying and pasting
uncopyrighted data...).
[1] http://pages.cs.wisc.edu/~ghost/doc/gnu/7.05/CJK.htm
>> Poppler, which is based on Xpdf, used to read CMAP files from the same
>> place as Xpdf did. So, we need only above packages to display CJK
>> fonts using Poppler. In recent versions, however, Poppler reads CMAPs
>> from its own path. So, we must additionally install poppler-data
>> which installs CMAPs for Poppler.
>
> I also have poppler-data not installed.
>
>> # I think it would be better to make Xpdf and Poppler share CMAP files
>> # again, but at least we have no time for lenny. ;-)
>
> Where does the the fonts used by xpdf come from? Does xpdf replace not found
> commercial fonts with free ones? If yes, than evince should do it as
> well.
>
> According to strace xpdf opens /usr/lib/libfreetype.so.6, all other
> opened files are probably less relevant.
>
>> Is this explanation comprehensible to you, Jens?
>
> Related to this bug report, yes, it is. Thanks for the explanations.
>
> I'm nevertheless curious why xpdf has no (obvious) problems.
So, from an inspection with pdffonts, what is curious is Poppler,
which can display some PDF files with fonts embedded but cannot do so
other ones. Thank you for noticing this fact. I'll check more.
Thanks,
-nori
Reply to: