Encoding: nvalid byte sequence in US-ASCII on really UTF string

[kalabiyau]

Invoice.last
Invoice Load (0.4ms)  SELECT "invoices".* FROM "invoices" ORDER BY "invoices"."id" DESC LIMIT 1
Hirb Error: invalid byte sequence in US-ASCII
   /.rvm/gems/ruby-1.9.3-p0/gems/hirb-0.5.0/lib/hirb/helpers/table.rb:333:in `gsub'

but

Invoice.last.attributes.each {|k,v| p "#{k} => #{v ? v : "nil"} => #{v.encoding if v.is_a?(String)}"}
  Invoice Load (0.3ms)  SELECT "invoices".* FROM "invoices" ORDER BY "invoices"."id" DESC LIMIT 1
"id => 143 => "
"name => nil => "
"document => B7_1_Территория_игр_20111107.pdf => UTF-8"
"checked_at => nil => "
"checked_by => nil => "
"upload_date => 2011-01-23 15:26:08 +0300 => "
"tenant_id => nil => "
"created_at => 2011-11-23 15:26:08 +0400 => "
"updated_at => 2011-11-23 15:26:08 +0400 => "
=> {"id"=>143, "name"=>nil, "document"=>"B7_1_Территория_игр_20111107.pdf", "checked_at"=>nil, "checked_by"=>nil, "upload_date"=>Sun, 23 Jan 2011 15:26:08 MSK +03:00, "tenant_id"=>nil, "created_at"=>Wed, 23 Nov 2011 15:26:08 MSK +04:00, "updated_at"=>Wed, 23 Nov 2011 15:26:08 MSK +04:00}

[kalabiyau]

With plain irb all is ok.

[cldwalker]

Don't have time or knowledge to look into this. Leaving open for anyone willing and able.