Text 36917, 112 rader
Skriven 2016-10-03 01:39:58 av FidoNews Robot (2:2/2.0)
Ärende: FidoNews 33:40 [01/06]: General Articles
================================================
=================================================================
                        GENERAL ARTICLES
=================================================================

                An UTF-8 nodelist. And now what?
                By Michiel van der Vlist, 2:280/5555


As you may or may not know, ZC2 in cooperation with some Z2 RCs,
distributes an UTF-8 encoded nodelist in addition to the weekly and
daily ASCII encoded lists. It is distributed on a daily basis in the
file area DAILYUTF.

The project started a couple of years ago and took a slow start.
Presently only R28 and R56 participate actively by submitting an UTF-8
encoded segment. The number of regions participating passively is
unknown.

The UTF-8 nodelist offers sysops the opportunity to have their names,
their loaction and the name of their system listed as spelled in the
alfabet of their native language.

I am often asked "what use is an UTF-8 encoded nodelist, when there is
hardly any FTN sofware around that can handle it?" This article is
about how I deal with the UTF-8 nodelist in my system. I hope it will
be a guideline for others.

Of the the software I use: binkd, Fmail and Golded, only Golded is
relevant. Binkd and Fmail are character encoding agnostic with regard
to the nodelist. But Golded uses the nodelist as a lookup data base
and it it looks at the names of the sysops.

Golded can not deal directly with UTF-8. Even worse, it has the nasty
habit - at least the Windows version - of disregarding the active
codepage  and always forcing the code page that was in effect at
system start up. In my case CP850. So I have no choice but - for
better or worse - to convert to CP850 whatever I offer to Golded's
nodelist compiler.

For this I use two commonly available utilities: sed and iconv. Sed is
a so called serial or stream editor. It converts streams of bytes into
something else. I this case I use it to convert characters not present
in CP850 to something else. Sed is driven bij a script in this case
called to850.scr

Here is the line in the batch file that is called  when DILYUTF.ddd
comes in:

Daynbr Sed -f to850.scr dailyutf.@### >dailyutf.998

Daynbr is the win32 version of the famous daynbr.com by Ben Baker.

Here is the content of to850.scr:

s/ßÖ·â·ßÖ·â¶/ij/g
s/ßÖ·â·ßÖ·Öà/IJ/g
s/âéü/EUR/g

Oops, that looks nasty. It looks OK when viewed in in UTF-8
environment, but unfortunately Fidonews can not deal with UTF-8 yet,
so it does no look so good. What is betweem the first slashes in the
first line is the two byte UTF-8 code for the Dutch concatenated 'ij'
as viewed in CP850. This translates the Dutch concatenated 'ij'into
the diftong 'ij'. The second line does the same for the comapnion
capital.

The third line translates the Euro currency symbol into "EUR". Those
are all the characters that I presently expect in the UTF-8 nodelist
that do not fit into CP850.

The next line in the batch file calls iconv. Iconv is a character code
conversion utility. It converts files from one character encoding
scheme to another. Of course it will only work properly if the
characters to bve converted are present in the target encoding set.

Oh, wait, I first introduce a 2 second delay. To make sure that the
next file, DAILYUTF.999 has a later time stamp then DAILYUTF.998.
That way Golded's nodelist compiler sees it as "the latest".

ping -n 2 loopback
Iconv -c -f utf-8 -t cp850 dailyutf.998 >dailyutf.999
cd \fido\golded
gncyg -f -d

The -f tells golded to do a forced compile and the -d tells it to
remove duplicate entries. There will be a lot as I compile both the
ASCII nodelist and the UTF-8 noelist

In golded's config I have this for the nodelist section:

NODEPATH d:\fido\nodelist\
NODELIST d:\fido\\nodelist\dailyutf.*
NODELIST d:\fido\nodelist\dailylst.*
NODELIST d:\fido\z2pnt\z2pnt.*


As a result I can now type either "schroeter" or "schröter" in the To:
field of a message in Golded and the nodelist lookup of Golded will
give me the node numbers of Ullrich Schroeter or Ullrich Schröter.

This is an ongoing project. When our friends in Eastern Europe join
the project, things get more complicated. I do not yet know how to
deal with that. If and when it happens and I find a workaround, that
may trigger a follow up article.


(C) 2016, Michiel van der Vlist.

-----------------------------------------------------------------

--- Azure/NewsPrep 3.0
 * Origin: Home of the Fidonews (2:2/2.0)