From: Michael Glaesemann <grzm@...>
Date: 2007-06-28T00:25:37+09:00
Subject: Re: Encoding of strings received from db


On Jun 27, 2007, at 9:37 , Marko Marjanovic wrote:

> Axel Etzold wrote:
>> Dear mare,
>>
>> does the information here:
>>
>> http://groups.google.ca/group/rubyonrails-core/browse_thread/ 
>> thread/db8d0c594cb4cc73
>>
>> help ?
>
> Unfortunately it doesn't, because rails has some kind of configuration
> (I was only playing with rails, never digging deeper into it, so  
> I'm not
> sure what it looks like). As for my problem, I'm trying to do this:
>
>   db = PGconn.connect('localhost', 5432, '', '', 'isys', 'postgres',
> 'postgres')
>   res = db.exec('select * from "ADDRESS"')
>   puts res[1][1] == "S�dstrasse"
>
> and make the output "true". But i can't, because res[1][1] (second
> column of the second row) is in some other encoding and instead of
> S�dstrasse ("Sudstrasse" where "u" has umlauts - the two dots above  
> the
> letter) I get S��dstrasse ("u" with umlauts is presented as two
> letters).
>
> I've tried using:
>
> Iconv.new('UTF-8', 'ISO-8859-1').iconv(res[1][1])
>
> but now every byte that I have instead of umlauted u is transformed  
> into
> another two characters - instead of u with umlauts, I now have 4
> characters.
> I believe that the problem lies in 'postgres' gem and that it  
> should be
> configured somehow, but I don't have the slightest idea how...

I think your issue is that your Ruby script doesn't know what  
encoding is coming back from the database. Try setting $KCODE = 'u'  
in your script.

$ cat db_encoding_test.rb
#!/usr/local/bin/ruby -w
$KCODE = 'u'
require 'rubygems'
require 'postgres'
require 'dbi'

DBI.connect('dbi:Pg:test:localhost:54824', 'postgres') do |dbh|
   sth = dbh.prepare(<<-EOS)
     SELECT * FROM streets;
   EOS
   sth.execute
   sth.fetch do |row|
     p row
     p row.first == 'S�dstrasse'
   end
end
$ ./db_encoding_test.rb
["S�dstrasse"]
true

Michael Glaesemann
grzm seespotcode net