ucm<=>charmapml conversion scripts?

[Autrijus Tang]

Greetings.

On http://www-124.ibm.com/icu/charset/index.html you have said:

We are going to add into the same repository the code to
generate the mapping table files.

As I'm writing a Convert::CharMap module in Perl that performs
conversion between UCM, CharMapML, Tcl(.enc) and libiconv Map
formats, it would be wonderful if you could make available
the current code.

Also, you might be interested in additional Chinese-related maps
I have compiled, available at:

http://search.cpan.org/src/AUTRIJUS/Encode-HanExtra-0.06/

Which includes Big5e, Big5plus, as well as CCCII (the later adapted
from the Unihan.txt).

Thanks,
/Autrijus/

[George Rhoten]

There is only code to convert UCM files to the UTR 22 XML format. We have
no other code. Did you want that code available?

Regarding the mappings, we don't take mappings that people modify or hand
edit. We only post the files that we collect from various platforms. That
way we only get the 100% true implementation for each platform, and not a
new intermediate implementation that isn't really used in the real world.
If you collected the mapping information algorithmically (with fallbacks)
from a specific platform, then we would be interested.

George Rhoten
IBM Globalization Center of Competency/ICU San Jose, CA, USA




Autrijus Tang <autr...@autrijus.org>
Sent by: icu-chars...@oss.software.ibm.com
09/18/2002 06:49 PM


To: icu-ch...@oss.software.ibm.com
cc:
Subject: ucm<=>charmapml conversion scripts?

[Autrijus Tang]

On Thu, Sep 19, 2002 at 12:11:21PM -0700, George Rhoten wrote:
> There is only code to convert UCM files to the UTR 22 XML format. We have
> no other code. Did you want that code available?

Yes. Thank you.

> Regarding the mappings, we don't take mappings that people modify or hand
> edit. We only post the files that we collect from various platforms. That
> way we only get the 100% true implementation for each platform, and not a
> new intermediate implementation that isn't really used in the real world.
> If you collected the mapping information algorithmically (with fallbacks)
> from a specific platform, then we would be interested.

Hrm, define 'platform'? My mappings are those which are used in the Perl
language, which could be thought as a platform in itself. :-)

The implementation was algorithmically derived from the material made
available on Windows platform by the canonical source for Big5 extensions,
http://www.cmex.org.tw/, as well as the Li18nux-Big5 working group's
documents.

I shall collect a list of sources, descriptions and CharMappingAlias data
for each encoding, and keep you posted.

Thanks,
/Autrijus/