Question

I have a problem running a Ruby on Rails application using Passenger. My application resides in the /Users/Glutexo/Práce/Bytek/Bytek directory, which you can see has a non-ASCII character in in (á). Even when I symlink it into another directory, e.g. /Library/WebServer/Documents/Bytek with all characters being US-ASCII, the problem is still there.

But when I create another Rails application physically in all US-ASCII path, like /Users/Glutexo/rails/pokus, it works: The application starts and runs normally through Passenger.

The error page I get when trying to run a Rails application residing in non-ASCII path says the following:

Error message:

invalid byte sequence in US-ASCII

Exception class:

ArgumentError

Backtrace:

0 | /Users/Glutexo/.rvm/rubies/ruby-1.9.3-p125/lib/ruby/1.9.1/pathname.rb | 45 | in `=~’

The line in pathname.rb mentioned is in the following function:

# chop_basename(path) -> [pre-basename, basename] or nil
def chop_basename(path)
  base = File.basename(path)
  if /\A#{SEPARATOR_PAT}?\z/o =~ base # This is the line no. 45
    return nil
  else
    return path[0, path.rindex(base)], base
  end
end
private :chop_basename

Does anyone have any suggestion, how to convince Passenger to be able to run Ruby on Rails application even from a path containing non-ASCII characters?

Rails is version 3.2.2, Ruby is version 1.9.3-p125, Apache is version 2.2.21, Passenger is version 3.0.12.

Thanks in advance for any help!

Answer 1

The solution to this problem is to add a ‘magic comment’ to your rails app source files. For example, at the beginning of a file that will encounter a non-ascii character, add the following:

# encoding: utf-8

There’s a useful gem out there to help you do this: https://github.com/m-ryan/magic_encoding

To confirm this solution, I setup passenger with a rails app in a directory containing ‘Práce’ and got a similar error to you. I ran the magic_encoding gem’s ‘magic_encoding’ command in the root of my rails app. I restarted passenger and it seems to have done the trick. Hope that helps!

See also: invalid multibyte char (US-ASCII) with Rails and Ruby 1.9