mirror of
https://github.com/rails/rails.git
synced 2022-11-09 12:12:34 -05:00
71040173cb
git-svn-id: http://svn-commit.rubyonrails.org/rails/trunk@4688 5ecf4fe2-1ee6-0310-87b1-e25e094e27de
165 lines
4.4 KiB
Ruby
165 lines
4.4 KiB
Ruby
# Donated by Florian Gross
|
|
|
|
require 'webrick'
|
|
require 'cgi'
|
|
require 'stringio'
|
|
|
|
include WEBrick
|
|
|
|
ABSOLUTE_RAILS_ROOT = File.expand_path(RAILS_ROOT)
|
|
|
|
class CGI #:nodoc:
|
|
def stdinput
|
|
@stdin || $stdin
|
|
end
|
|
|
|
def env_table
|
|
@env_table || ENV
|
|
end
|
|
|
|
def initialize(type = "query", table = nil, stdin = nil)
|
|
@env_table, @stdin = table, stdin
|
|
|
|
if defined?(MOD_RUBY) && !ENV.key?("GATEWAY_INTERFACE")
|
|
Apache.request.setup_cgi_env
|
|
end
|
|
|
|
extend QueryExtension
|
|
@multipart = false
|
|
if defined?(CGI_PARAMS)
|
|
warn "do not use CGI_PARAMS and CGI_COOKIES"
|
|
@params = CGI_PARAMS.dup
|
|
@cookies = CGI_COOKIES.dup
|
|
else
|
|
initialize_query() # set @params, @cookies
|
|
end
|
|
@output_cookies = nil
|
|
@output_hidden = nil
|
|
end
|
|
end
|
|
|
|
# A custom dispatch servlet for use with WEBrick. It dispatches requests
|
|
# (using the Rails Dispatcher) to the appropriate controller/action. By default,
|
|
# it restricts WEBrick to a managing a single Rails request at a time, but you
|
|
# can change this behavior by setting ActionController::Base.allow_concurrency
|
|
# to true.
|
|
class DispatchServlet < WEBrick::HTTPServlet::AbstractServlet
|
|
REQUEST_MUTEX = Mutex.new
|
|
|
|
# Start the WEBrick server with the given options, mounting the
|
|
# DispatchServlet at <tt>/</tt>.
|
|
def self.dispatch(options = {})
|
|
Socket.do_not_reverse_lookup = true # patch for OS X
|
|
|
|
params = { :Port => options[:port].to_i,
|
|
:ServerType => options[:server_type],
|
|
:BindAddress => options[:ip] }
|
|
params[:MimeTypes] = options[:mime_types] if options[:mime_types]
|
|
|
|
server = WEBrick::HTTPServer.new(params)
|
|
server.mount('/', DispatchServlet, options)
|
|
|
|
trap("INT") { server.shutdown }
|
|
|
|
server.start
|
|
end
|
|
|
|
def initialize(server, options) #:nodoc:
|
|
@server_options = options
|
|
@file_handler = WEBrick::HTTPServlet::FileHandler.new(server, options[:server_root])
|
|
Dir.chdir(ABSOLUTE_RAILS_ROOT)
|
|
super
|
|
end
|
|
|
|
def service(req, res) #:nodoc:
|
|
unless handle_file(req, res)
|
|
begin
|
|
REQUEST_MUTEX.lock unless ActionController::Base.allow_concurrency
|
|
unless handle_dispatch(req, res)
|
|
raise WEBrick::HTTPStatus::NotFound, "`#{req.path}' not found."
|
|
end
|
|
ensure
|
|
unless ActionController::Base.allow_concurrency
|
|
REQUEST_MUTEX.unlock if REQUEST_MUTEX.locked?
|
|
end
|
|
end
|
|
end
|
|
end
|
|
|
|
def handle_file(req, res) #:nodoc:
|
|
begin
|
|
req = req.dup
|
|
path = req.path.dup
|
|
|
|
# Add .html if the last path piece has no . in it
|
|
path << '.html' if path != '/' && (%r{(^|/)[^./]+$} =~ path)
|
|
path.gsub!('+', ' ') # Unescape + since FileHandler doesn't do so.
|
|
|
|
req.instance_variable_set(:@path_info, path) # Set the modified path...
|
|
|
|
@file_handler.send(:service, req, res)
|
|
return true
|
|
rescue HTTPStatus::PartialContent, HTTPStatus::NotModified => err
|
|
res.set_error(err)
|
|
return true
|
|
rescue => err
|
|
return false
|
|
end
|
|
end
|
|
|
|
def handle_dispatch(req, res, origin = nil) #:nodoc:
|
|
data = StringIO.new
|
|
Dispatcher.dispatch(
|
|
CGI.new("query", create_env_table(req, origin), StringIO.new(req.body || "")),
|
|
ActionController::CgiRequest::DEFAULT_SESSION_OPTIONS,
|
|
data
|
|
)
|
|
|
|
header, body = extract_header_and_body(data)
|
|
|
|
set_charset(header)
|
|
assign_status(res, header)
|
|
res.cookies.concat(header.delete('set-cookie') || [])
|
|
header.each { |key, val| res[key] = val.join(", ") }
|
|
|
|
res.body = body
|
|
return true
|
|
rescue => err
|
|
p err, err.backtrace
|
|
return false
|
|
end
|
|
|
|
private
|
|
def create_env_table(req, origin)
|
|
env = req.meta_vars.clone
|
|
env.delete "SCRIPT_NAME"
|
|
env["QUERY_STRING"] = req.request_uri.query
|
|
env["REQUEST_URI"] = origin if origin
|
|
return env
|
|
end
|
|
|
|
def extract_header_and_body(data)
|
|
data.rewind
|
|
data = data.read
|
|
|
|
raw_header, body = *data.split(/^[\xd\xa]+/on, 2)
|
|
header = WEBrick::HTTPUtils::parse_header(raw_header)
|
|
|
|
return header, body
|
|
end
|
|
|
|
def set_charset(header)
|
|
ct = header["content-type"]
|
|
if ct.any? { |x| x =~ /^text\// } && ! ct.any? { |x| x =~ /charset=/ }
|
|
ch = @server_options[:charset] || "UTF-8"
|
|
ct.find { |x| x =~ /^text\// } << ("; charset=" + ch)
|
|
end
|
|
end
|
|
|
|
def assign_status(res, header)
|
|
if /^(\d+)/ =~ header['status'][0]
|
|
res.status = $1.to_i
|
|
header.delete('status')
|
|
end
|
|
end
|
|
end
|