Detect browser language in Rails

Last night I did this tiny gem: accept_language

It can be integrated in a Rails app like that:

# app/controllers/application_controller.rb
class ApplicationController < ActionController::Base
  before_action :best_locale_from_request!

  def best_locale_from_request!
    I18n.locale = best_locale_from_request
  end

  def best_locale_from_request
    return I18n.default_locale unless request.headers.key?("HTTP_ACCEPT_LANGUAGE")

    string = request.headers.fetch("HTTP_ACCEPT_LANGUAGE")
    locale = AcceptLanguage.parse(string).match(*I18n.available_locales)

    # If the server cannot serve any matching language,
    # it can theoretically send back a 406 (Not Acceptable) error code.
    # But, for a better user experience, this is rarely done and more
    # common way is to ignore the Accept-Language header in this case.
    return I18n.default_locale if locale.nil?

    locale
  end
end

I hope it can help.

Here's the solution:

2.6 Setting the Locale from the Client Supplied Information

In specific cases, it would make sense to set the locale from client-supplied information, i.e. not from the URL. This information may come for example from the users’ preferred language (set in their browser), can be based on the users’ geographical location inferred from their IP, or users can provide it simply by choosing the locale in your application interface and saving it to their profile. This approach is more suitable for web-based applications or services, not for websites — see the box about sessions, cookies and RESTful architecture above...

One source of client supplied information would be an Accept-Language HTTP header. People may set this in their browser or other clients (such as curl)...

Another way of choosing the locale from client information would be to use a database for mapping the client IP to the region, such as GeoIP Lite Country...

You can also provide users of your application with means to set (and possibly over-ride) the locale in your application interface, as well...

This is an old question, but I came across it as a soul in search of answers and the only available answer was a link without any context. So here's a bit more depth, based on my subsequent digging.

Accessing the Accept-Language header

Query the request object in the relevant controller:

request.env['HTTP_ACCEPT_LANGUAGE'] #returns nil or a string, e.g.:
# => "en-AU,en-US;q=0.7,en;q=0.3"

That's the easy part though. To make good use of it in Rails, you'll need to parse that string (or maybe it's nil?).

Sidebar: A primer on Accept-* headers

So, looking at the example string above, some languages have "q-values" - the relative quality factor, between 0 and 1. Higher q-values mean that language is preferred by the client. Lack of a q-value is really the highest q-value - an implicit 1.0 (as with en-AU in the above string). A slight complication though - the browser may send you languages with q-values that are, say, 0 - and I gather this means you should reject those languages if possible.

Parsing the Accept-Language header

Bearing that in mind, here's a couple of different yet similar approaches I've looked at for parsing such a string into a list of languages, ordered by their q-values. With straightforward Ruby:

# to_s to deal with possible nil value, since nil.to_s => ""
langs = request.env['HTTP_ACCEPT_LANGUAGE'].to_s.split(",").map do |lang| 
  l, q = lang.split(";q=")
  [l, (q || '1').to_f]
end
# => [["en-AU", 1.0], ["en-US", 0.7], ["en", 0.3]]

Or if you're proficient at regular expressions, you can achieve the same as above, and probably improve upon my butchered regex at the same time:

rx = /([A-Za-z]{2}(?:-[A-Za-z]{2})?)(?:;q=(1|0?\.[0-9]{1,3}))?/
langs = request.env['HTTP_ACCEPT_LANGUAGE'].to_s.scan(rx).map do |lang, q|
  [lang, (q || '1').to_f]
end

Either way, you can follow up as needed with something like:

# return array of just languages, ordered by q-value
langs.sort_by(&:last).map(&:first).reverse
# => ["en-AU", "en-US", "en"]

I started out my parsing by looking at this gist, but ended up modifying it fairly significantly. The regex especially was throwing away perfectly good secondary locales, e.g. en-AU became en, zh-TW became zh. I've attempted to rectify this with my modifications to that regex.