> I'm doing something like:
> 
> File.open("target","w") do |target|
>     File.open("source","r") do |source|
>         source.each_line do |line|
>             ... some processing ...
>             target.write(line)
>         end
>      end
> end

Have you looked at 'iconv' in the standard library?

http://www.ruby-doc.org/stdlib/libdoc/iconv/rdoc/classes/Iconv.html

Assuming all your input files were ISO-8859-1, and you wanted your output file in UTF-8, your example might look something like (untested):

File.open("target","w") do |target|
  Iconv.open('UTF-8', 'ISO-8859-1') do | converter |
    File.open("source","r") do |source|
      source.each_line do |line|
        # ... some processing ...
        target.write( converter.iconv(line) )
      end
    end
    target << converter.iconv(nil)
  end
end

Iconv should deal with BOMs, stripping them out or adding them in where necessary. I'm not sure if it will complain if it finds a BOM mid-stream (as you open your second and subsequent input file) - if so you could just instantiate a new Iconv to deal with each input.

HTH
alex