www: Serve the current directory via HTTP.

www.rb

#!/usr/bin/env ruby

# Serve the current directory via HTTP.
# Like Python's SimpleHTTPServer, but with no-cache headers.
# Default port 8000, specify alternate port as first parameter:
#   www 3000
#   sudo www 80 # (probably a bad idea)

# Inspired by http://chrismdp.github.com/2011/12/cache-busting-ruby-http-server/

require "webrick"

class NonCachingFileHandler < WEBrick::HTTPServlet::FileHandler

  DAY = 86400

  def do_GET(request, response)
    super
    set_no_cache(response)
    set_content_type(response)
  end

  private

  def set_no_cache(response)
    response["Content-Type"]
    response["ETag"]          = nil
    response["Last-Modified"] = Time.now + DAY
    response["Cache-Control"] = "no-store, no-cache, must-revalidate, post-check=0, pre-check=0"
    response["Pragma"]        = "no-cache"
    response["Expires"]       = Time.now - DAY
  end

  def set_content_type(response)
    response["Content-Type"] = content_type(response.filename)
  end

  def content_type(path)
    case path.match(%r{\.(\w+)\z})[1]
    when "html" then "text/html"
    when "js" then "text/javascript"
    when "css" then "text/css"
    else `/usr/bin/file --brief --mime-type #{Shellwords.escape(path)}`.chomp
    end
  end

end

WEBrick::HTTPServer.new(Port: ARGV.first || 8000).tap do |server|
  server.mount "/", NonCachingFileHandler , Dir.pwd
  trap("INT") { server.stop }
  server.start
end

The calls to IO.popen for Content-Type detection were leaving behind dead processes.

Instead of adding the verbosity of pid collection, Process.wait calls etc, I've used backticks instead of IO.popen, with Shellwords.escape protecting against shell injection via path names.

Here's an even simpler solution for always serving fresh files (if you're willing to use Node) https://github.com/curran/dashboardScaffold/blob/4ce4da6d9b79431570ff152620198f4a6612e70d/server.js

You could simply run ruby -run -e httpd . -p 8000.