diff options
Diffstat (limited to 'lib/unicorn/http_request.rb')
| -rw-r--r-- | lib/unicorn/http_request.rb | 118 |
1 files changed, 60 insertions, 58 deletions
diff --git a/lib/unicorn/http_request.rb b/lib/unicorn/http_request.rb index ee407ab..7106f62 100644 --- a/lib/unicorn/http_request.rb +++ b/lib/unicorn/http_request.rb @@ -13,6 +13,20 @@ module Unicorn # class HttpRequest + # default parameters we merge into the request env for Rack handlers + DEF_PARAMS = { + "rack.errors" => $stderr, + "rack.multiprocess" => true, + "rack.multithread" => false, + "rack.run_once" => false, + "rack.url_scheme" => "http", + "rack.version" => [0, 1], + "SCRIPT_NAME" => "", + + # this is not in the Rack spec, but some apps may rely on it + "SERVER_SOFTWARE" => "Unicorn #{Const::UNICORN_VERSION}" + }.freeze + def initialize(logger) @logger = logger @body = nil @@ -29,59 +43,39 @@ module Unicorn @body = nil end - # # Does the majority of the IO processing. It has been written in - # Ruby using about 7 different IO processing strategies and no - # matter how it's done the performance just does not improve. It is - # currently carefully constructed to make sure that it gets the best - # possible performance, but anyone who thinks they can make it - # faster is more than welcome to take a crack at it. + # Ruby using about 8 different IO processing strategies. + # + # It is currently carefully constructed to make sure that it gets + # the best possible performance for the common case: GET requests + # that are fully complete after a single read(2) + # + # Anyone who thinks they can make it faster is more than welcome to + # take a crack at it. # # returns an environment hash suitable for Rack if successful # This does minimal exception trapping and it is up to the caller # to handle any socket errors (e.g. user aborted upload). def read(socket) - data = String.new(read_socket(socket)) - nparsed = 0 - - # Assumption: nparsed will always be less since data will get - # filled with more after each parsing. If it doesn't get more - # then there was a problem with the read operation on the client - # socket. Effect is to stop processing when the socket can't - # fill the buffer for further parsing. - while nparsed < data.length - nparsed = @parser.execute(@params, data, nparsed) - - if @parser.finished? - # From http://www.ietf.org/rfc/rfc3875: - # "Script authors should be aware that the REMOTE_ADDR and - # REMOTE_HOST meta-variables (see sections 4.1.8 and 4.1.9) - # may not identify the ultimate source of the request. They - # identify the client for the immediate request to the server; - # that client may be a proxy, gateway, or other intermediary - # acting on behalf of the actual source client." - @params[Const::REMOTE_ADDR] = socket.unicorn_peeraddr - - handle_body(socket) and return rack_env # success! - return nil # fail - else - # Parser is not done, queue up more data to read and continue - # parsing - data << read_socket(socket) - if data.length >= Const::MAX_HEADER - raise HttpParserError.new("HEADER is longer than allowed, " \ - "aborting client early.") - end - end + # short circuit the common case with small GET requests first + @parser.execute(@params, read_socket(socket)) and + return handle_body(socket) + + data = @buffer.dup # read_socket will clobber @buffer + + # Parser is not done, queue up more data to read and continue parsing + # an Exception thrown from the @parser will throw us out of the loop + loop do + data << read_socket(socket) + @parser.execute(@params, data) and + return handle_body(socket) end - nil # XXX bug? rescue HttpParserError => e @logger.error "HTTP parse error, malformed request " \ "(#{@params[Const::HTTP_X_FORWARDED_FOR] || socket.unicorn_peeraddr}): #{e.inspect}" @logger.error "REQUEST DATA: #{data.inspect}\n---\n" \ "PARAMS: #{@params.inspect}\n---\n" - socket.closed? or socket.close rescue nil nil end @@ -109,7 +103,7 @@ module Unicorn # This will probably truncate them but at least the request goes through # usually. if remain > 0 - read_body(socket, remain) or return false # fail! + read_body(socket, remain) or return nil # fail! end @body.rewind @body.sysseek(0) if @body.respond_to?(:sysseek) @@ -118,29 +112,37 @@ module Unicorn # another request, we'll truncate it. Again, we don't do pipelining # or keepalive @body.truncate(content_length) - true + rack_env(socket) end # Returns an environment which is rackable: # http://rack.rubyforge.org/doc/files/SPEC.html # Based on Rack's old Mongrel handler. - def rack_env + def rack_env(socket) + # I'm considering enabling "unicorn.client". It gives + # applications some rope to do some "interesting" things like + # replacing a worker with another process that has full control + # over the HTTP response. + # @params["unicorn.client"] = socket + + # From http://www.ietf.org/rfc/rfc3875: + # "Script authors should be aware that the REMOTE_ADDR and + # REMOTE_HOST meta-variables (see sections 4.1.8 and 4.1.9) + # may not identify the ultimate source of the request. They + # identify the client for the immediate request to the server; + # that client may be a proxy, gateway, or other intermediary + # acting on behalf of the actual source client." + @params[Const::REMOTE_ADDR] = socket.unicorn_peeraddr + # It might be a dumbass full host request header - @params[Const::REQUEST_PATH] ||= - URI.parse(@params[Const::REQUEST_URI]).path - raise "No REQUEST PATH" unless @params[Const::REQUEST_PATH] - - @params["QUERY_STRING"] ||= '' - @params.update({ "rack.version" => [0,1], - "rack.input" => @body, - "rack.errors" => $stderr, - "rack.multithread" => false, - "rack.multiprocess" => true, - "rack.run_once" => false, - "rack.url_scheme" => "http", - Const::PATH_INFO => @params[Const::REQUEST_PATH], - Const::SCRIPT_NAME => "", - }) + @params[Const::PATH_INFO] = ( + @params[Const::REQUEST_PATH] ||= + URI.parse(@params[Const::REQUEST_URI]).path) or + raise "No REQUEST_PATH" + + @params[Const::QUERY_STRING] ||= '' + @params[Const::RACK_INPUT] = @body + @params.update(DEF_PARAMS) end # Does the heavy lifting of properly reading the larger body requests in |