[Cjk] Fwd: CJKutf8 in TeXLive2011 cannot generate copy-and-paste PDF with other CJK fonts

[Nien-Po Chen]

Dear Listers,

The following is the message from Werner.
------
Nien-Po Chen

Begin forwarded message:

> From: Werner LEMBERG <wl at gnu.org>
> Date: July 21, 2012 12:24:09 AM GMT+08:00
> To: npchen at saturn.yzu.edu.tw
> Cc: cjk at ffii.org
> Subject: Re: [Cjk] CJKutf8 in TeXLive2011 cannot generate copy-and-paste PDF with other CJK fonts
> 
> 
> Nien-Po,
> 
> 
> for reasons which neither I nor the FFII maintainers have been able to
> resolve, I no longer can post to the CJK mailing lists since a few
> months.  My plan is to eventually move the lists to the GNU site, but
> I haven't found the time yet to do that.  Please forward this my
> private reply to the list, thanks.
> 
>> I read it from
>> 	Werner's <Unicode Support in the CJK Package> in
>> 	The Asian Journal of TEX, Volume 2, No. 1, April 2008
>> that using CJKutf8 and cmap packages along with pdflatex can
>> generate a PDF where we can successfully copy texts and paste them
>> somewhere else.  [...]
>> 
>> I would like to make other CJK fonts to have the same capability,
>> but I don't know where to start.
>> 
>> If I attempt to use Type 1 fonts of cw family (available at
>> <ftp://cle.linux.org.tw/tex/cjk/fonts/cwt1/>), say 'cwkb', in the
>> example above, the resultant PDF looks fine but its copy-and-paste
>> text has garbled Chinese text. This is obviously a sympton that
>> ToUnicode map is not set properly for this font.
> 
> Yes.
> 
>> I notice that the file c70bsmi.fdx has some code to deal with
>> ToUnicode stuff, whereas cwkb does not even have c70cwkb.fdx. Is
>> this the right place to work with?
> 
> Right now the ToUnicode maps are added automatically by CJKutf8.sty
> (since they are really trivial and just a few code lines), and this
> assumes that the font is available in Unicode encoding.  In other
> words, if a file c70cwkb.fd (not c70cwkb.fdx) exists, you should have
> proper cut and paste without any addition!
> 
> So it seems that the font encoding of the cw family is Big5 (I have
> bad internet access currently so I can't download the package).
> Adding ToUnicode maps for non-Unicode font encodings is not supported
> by the CJK package.  It might be theoretically possible to extend the
> internal LaTeX command \pickup at font in a similar way as done in
> CJKutf8.sty.  However, a Big5->Unicode ToUnicode mapping table
> consists of more than 11000 lines...
> 
> The right way IMHO is to reencode the cw Type 1 subfonts from Big5 to
> Unicode.  Writing a FontForge script, this shouldn't be too difficult.
> 
> 
>    Werner
>