From ff0a2678c4bce9da348e1263915558795e3a3640 Mon Sep 17 00:00:00 2001 From: Joshua Peek Date: Sat, 17 Jan 2009 20:29:50 -0600 Subject: Build query string and POST params parser on top of Rack::Request. Also switch our multipart parser to use Racks. Moved XML, JSON, and YAML parsers into ActionController::ParamsParser middleware [#1661 state:resolved] --- actionpack/lib/action_controller/base.rb | 5 +- actionpack/lib/action_controller/middlewares.rb | 1 + actionpack/lib/action_controller/params_parser.rb | 71 +++++ actionpack/lib/action_controller/rack_ext.rb | 1 + .../lib/action_controller/rack_ext/parse_query.rb | 18 ++ actionpack/lib/action_controller/request.rb | 34 ++- actionpack/lib/action_controller/request_parser.rb | 315 --------------------- actionpack/lib/action_controller/uploaded_file.rb | 7 + .../action_controller/url_encoded_pair_parser.rb | 61 ++++ 9 files changed, 184 insertions(+), 329 deletions(-) create mode 100644 actionpack/lib/action_controller/params_parser.rb create mode 100644 actionpack/lib/action_controller/rack_ext/parse_query.rb delete mode 100644 actionpack/lib/action_controller/request_parser.rb (limited to 'actionpack/lib/action_controller') diff --git a/actionpack/lib/action_controller/base.rb b/actionpack/lib/action_controller/base.rb index e22114195c..7a380bd1fb 100644 --- a/actionpack/lib/action_controller/base.rb +++ b/actionpack/lib/action_controller/base.rb @@ -301,10 +301,7 @@ module ActionController #:nodoc: # A YAML parser is also available and can be turned on with: # # ActionController::Base.param_parsers[Mime::YAML] = :yaml - @@param_parsers = { Mime::MULTIPART_FORM => :multipart_form, - Mime::URL_ENCODED_FORM => :url_encoded_form, - Mime::XML => :xml_simple, - Mime::JSON => :json } + @@param_parsers = {} cattr_accessor :param_parsers # Controls the default charset for all renders. diff --git a/actionpack/lib/action_controller/middlewares.rb b/actionpack/lib/action_controller/middlewares.rb index cbcb5cb3e4..30dbc26f11 100644 --- a/actionpack/lib/action_controller/middlewares.rb +++ b/actionpack/lib/action_controller/middlewares.rb @@ -19,4 +19,5 @@ use "ActiveRecord::QueryCache", :if => lambda { defined?(ActiveRecord) } end use ActionController::RewindableInput +use ActionController::ParamsParser use Rack::MethodOverride diff --git a/actionpack/lib/action_controller/params_parser.rb b/actionpack/lib/action_controller/params_parser.rb new file mode 100644 index 0000000000..d269fe07fa --- /dev/null +++ b/actionpack/lib/action_controller/params_parser.rb @@ -0,0 +1,71 @@ +module ActionController + class ParamsParser + ActionController::Base.param_parsers[Mime::XML] = :xml_simple + ActionController::Base.param_parsers[Mime::JSON] = :json + + def initialize(app) + @app = app + end + + def call(env) + if params = parse_formatted_parameters(env) + env["action_controller.request.request_parameters"] = params + end + + @app.call(env) + end + + private + def parse_formatted_parameters(env) + request = Request.new(env) + + return false if request.content_length.zero? + + mime_type = content_type_from_legacy_post_data_format_header(env) || request.content_type + strategy = ActionController::Base.param_parsers[mime_type] + + return false unless strategy + + case strategy + when Proc + strategy.call(request.raw_post) + when :xml_simple, :xml_node + body = request.raw_post + body.blank? ? {} : Hash.from_xml(body).with_indifferent_access + when :yaml + YAML.load(request.raw_post) + when :json + body = request.raw_post + if body.blank? + {} + else + data = ActiveSupport::JSON.decode(body) + data = {:_json => data} unless data.is_a?(Hash) + data.with_indifferent_access + end + else + false + end + rescue Exception => e # YAML, XML or Ruby code block errors + raise + { "body" => request.raw_post, + "content_type" => request.content_type, + "content_length" => request.content_length, + "exception" => "#{e.message} (#{e.class})", + "backtrace" => e.backtrace } + end + + def content_type_from_legacy_post_data_format_header(env) + if x_post_format = env['HTTP_X_POST_DATA_FORMAT'] + case x_post_format.to_s.downcase + when 'yaml' + return Mime::YAML + when 'xml' + return Mime::XML + end + end + + nil + end + end +end diff --git a/actionpack/lib/action_controller/rack_ext.rb b/actionpack/lib/action_controller/rack_ext.rb index 17cd08f39a..2ba6654e3d 100644 --- a/actionpack/lib/action_controller/rack_ext.rb +++ b/actionpack/lib/action_controller/rack_ext.rb @@ -1,2 +1,3 @@ require 'action_controller/rack_ext/lock' require 'action_controller/rack_ext/multipart' +require 'action_controller/rack_ext/parse_query' diff --git a/actionpack/lib/action_controller/rack_ext/parse_query.rb b/actionpack/lib/action_controller/rack_ext/parse_query.rb new file mode 100644 index 0000000000..2f21a57770 --- /dev/null +++ b/actionpack/lib/action_controller/rack_ext/parse_query.rb @@ -0,0 +1,18 @@ +# Rack does not automatically cleanup Safari 2 AJAX POST body +# This has not yet been commited to Rack, please +1 this ticket: +# http://rack.lighthouseapp.com/projects/22435/tickets/19 + +module Rack + module Utils + alias_method :parse_query_without_ajax_body_cleanup, :parse_query + module_function :parse_query_without_ajax_body_cleanup + + def parse_query(qs, d = '&;') + qs = qs.dup + qs.chop! if qs[-1] == 0 + qs.gsub!(/&_=$/, '') + parse_query_without_ajax_body_cleanup(qs, d) + end + module_function :parse_query + end +end diff --git a/actionpack/lib/action_controller/request.rb b/actionpack/lib/action_controller/request.rb index b4ab1ccda1..cbbfca41f6 100755 --- a/actionpack/lib/action_controller/request.rb +++ b/actionpack/lib/action_controller/request.rb @@ -9,11 +9,6 @@ module ActionController class Request < Rack::Request extend ActiveSupport::Memoizable - def initialize(env) - super - @parser = ActionController::RequestParser.new(env) - end - %w[ AUTH_TYPE GATEWAY_INTERFACE PATH_TRANSLATED REMOTE_HOST REMOTE_IDENT REMOTE_USER REMOTE_ADDR @@ -92,7 +87,11 @@ module ActionController # For backward compatibility, the post \format is extracted from the # X-Post-Data-Format HTTP header if present. def content_type - Mime::Type.lookup(@parser.content_type_without_parameters) + if @env['CONTENT_TYPE'] =~ /^([^,\;]*)/ + Mime::Type.lookup($1.strip.downcase) + else + nil + end end memoize :content_type @@ -380,7 +379,11 @@ EOM # Read the request \body. This is useful for web services that need to # work with raw requests directly. def raw_post - @parser.raw_post + unless @env.include? 'RAW_POST_DATA' + @env['RAW_POST_DATA'] = body.read(@env['CONTENT_LENGTH'].to_i) + body.rewind if body.respond_to?(:rewind) + end + @env['RAW_POST_DATA'] end # Returns both GET and POST \parameters in a single hash. @@ -409,19 +412,30 @@ EOM @env["rack.routing_args"] ||= {} end + # The request body is an IO input stream. If the RAW_POST_DATA environment + # variable is already set, wrap it in a StringIO. def body - @parser.body + if raw_post = @env['RAW_POST_DATA'] + raw_post.force_encoding(Encoding::BINARY) if raw_post.respond_to?(:force_encoding) + StringIO.new(raw_post) + else + @env['rack.input'] + end + end + + def form_data? + FORM_DATA_MEDIA_TYPES.include?(content_type.to_s) end # Override Rack's GET method to support nested query strings def GET - @parser.query_parameters + @env["action_controller.request.query_parameters"] ||= UrlEncodedPairParser.parse_query_parameters(query_string) end alias_method :query_parameters, :GET # Override Rack's POST method to support nested query strings def POST - @parser.request_parameters + @env["action_controller.request.request_parameters"] ||= UrlEncodedPairParser.parse_hash_parameters(super) end alias_method :request_parameters, :POST diff --git a/actionpack/lib/action_controller/request_parser.rb b/actionpack/lib/action_controller/request_parser.rb deleted file mode 100644 index d1739ef4d0..0000000000 --- a/actionpack/lib/action_controller/request_parser.rb +++ /dev/null @@ -1,315 +0,0 @@ -module ActionController - class RequestParser - def initialize(env) - @env = env - freeze - end - - def request_parameters - @env["action_controller.request_parser.request_parameters"] ||= parse_formatted_request_parameters - end - - def query_parameters - @env["action_controller.request_parser.query_parameters"] ||= self.class.parse_query_parameters(query_string) - end - - # Returns the query string, accounting for server idiosyncrasies. - def query_string - @env['QUERY_STRING'].present? ? @env['QUERY_STRING'] : (@env['REQUEST_URI'].split('?', 2)[1] || '') - end - - # The request body is an IO input stream. If the RAW_POST_DATA environment - # variable is already set, wrap it in a StringIO. - def body - if raw_post = @env['RAW_POST_DATA'] - raw_post.force_encoding(Encoding::BINARY) if raw_post.respond_to?(:force_encoding) - StringIO.new(raw_post) - else - @env['rack.input'] - end - end - - # The raw content type string with its parameters stripped off. - def content_type_without_parameters - self.class.extract_content_type_without_parameters(content_type_with_parameters) - end - - def raw_post - unless @env.include? 'RAW_POST_DATA' - @env['RAW_POST_DATA'] = body.read(content_length) - body.rewind if body.respond_to?(:rewind) - end - @env['RAW_POST_DATA'] - end - - private - - def parse_formatted_request_parameters - return {} if content_length.zero? - - content_type, boundary = self.class.extract_multipart_boundary(content_type_with_parameters) - - # Don't parse params for unknown requests. - return {} if content_type.blank? - - mime_type = Mime::Type.lookup(content_type) - strategy = ActionController::Base.param_parsers[mime_type] - - # Only multipart form parsing expects a stream. - body = (strategy && strategy != :multipart_form) ? raw_post : self.body - - case strategy - when Proc - strategy.call(body) - when :url_encoded_form - self.class.clean_up_ajax_request_body! body - self.class.parse_query_parameters(body) - when :multipart_form - self.class.parse_multipart_form_parameters(body, boundary, content_length, @env) - when :xml_simple, :xml_node - body.blank? ? {} : Hash.from_xml(body).with_indifferent_access - when :yaml - YAML.load(body) - when :json - if body.blank? - {} - else - data = ActiveSupport::JSON.decode(body) - data = {:_json => data} unless data.is_a?(Hash) - data.with_indifferent_access - end - else - {} - end - rescue Exception => e # YAML, XML or Ruby code block errors - raise - { "body" => body, - "content_type" => content_type_with_parameters, - "content_length" => content_length, - "exception" => "#{e.message} (#{e.class})", - "backtrace" => e.backtrace } - end - - def content_length - @env['CONTENT_LENGTH'].to_i - end - - # The raw content type string. Use when you need parameters such as - # charset or boundary which aren't included in the content_type MIME type. - # Overridden by the X-POST_DATA_FORMAT header for backward compatibility. - def content_type_with_parameters - content_type_from_legacy_post_data_format_header || @env['CONTENT_TYPE'].to_s - end - - def content_type_from_legacy_post_data_format_header - if x_post_format = @env['HTTP_X_POST_DATA_FORMAT'] - case x_post_format.to_s.downcase - when 'yaml'; 'application/x-yaml' - when 'xml'; 'application/xml' - end - end - end - - class << self - def parse_query_parameters(query_string) - return {} if query_string.blank? - - pairs = query_string.split('&').collect do |chunk| - next if chunk.empty? - key, value = chunk.split('=', 2) - next if key.empty? - value = value.nil? ? nil : CGI.unescape(value) - [ CGI.unescape(key), value ] - end.compact - - UrlEncodedPairParser.new(pairs).result - end - - def parse_request_parameters(params) - parser = UrlEncodedPairParser.new - - params = params.dup - until params.empty? - for key, value in params - if key.blank? - params.delete key - elsif !key.include?('[') - # much faster to test for the most common case first (GET) - # and avoid the call to build_deep_hash - parser.result[key] = get_typed_value(value[0]) - params.delete key - elsif value.is_a?(Array) - parser.parse(key, get_typed_value(value.shift)) - params.delete key if value.empty? - else - raise TypeError, "Expected array, found #{value.inspect}" - end - end - end - - parser.result - end - - def parse_multipart_form_parameters(body, boundary, body_size, env) - parse_request_parameters(read_multipart(body, boundary, body_size, env)) - end - - def extract_multipart_boundary(content_type_with_parameters) - if content_type_with_parameters =~ MULTIPART_BOUNDARY - ['multipart/form-data', $1.dup] - else - extract_content_type_without_parameters(content_type_with_parameters) - end - end - - def extract_content_type_without_parameters(content_type_with_parameters) - $1.strip.downcase if content_type_with_parameters =~ /^([^,\;]*)/ - end - - def clean_up_ajax_request_body!(body) - body.chop! if body[-1] == 0 - body.gsub!(/&_=$/, '') - end - - - private - def get_typed_value(value) - case value - when String - value - when NilClass - '' - when Array - value.map { |v| get_typed_value(v) } - else - if value.respond_to? :original_filename - # Uploaded file - if value.original_filename - value - # Multipart param - else - result = value.read - value.rewind - result - end - # Unknown value, neither string nor multipart. - else - raise "Unknown form value: #{value.inspect}" - end - end - end - - MULTIPART_BOUNDARY = %r|\Amultipart/form-data.*boundary=\"?([^\";,]+)\"?|n - - EOL = "\015\012" - - def read_multipart(body, boundary, body_size, env) - params = Hash.new([]) - boundary = "--" + boundary - quoted_boundary = Regexp.quote(boundary) - buf = "" - bufsize = 10 * 1024 - boundary_end="" - - # start multipart/form-data - body.binmode if defined? body.binmode - case body - when File - body.set_encoding(Encoding::BINARY) if body.respond_to?(:set_encoding) - when StringIO - body.string.force_encoding(Encoding::BINARY) if body.string.respond_to?(:force_encoding) - end - boundary_size = boundary.size + EOL.size - body_size -= boundary_size - status = body.read(boundary_size) - if nil == status - raise EOFError, "no content body" - elsif boundary + EOL != status - raise EOFError, "bad content body" - end - - loop do - head = nil - content = - if 10240 < body_size - UploadedTempfile.new("CGI") - else - UploadedStringIO.new - end - content.binmode if defined? content.binmode - - until head and /#{quoted_boundary}(?:#{EOL}|--)/n.match(buf) - - if (not head) and /#{EOL}#{EOL}/n.match(buf) - buf = buf.sub(/\A((?:.|\n)*?#{EOL})#{EOL}/n) do - head = $1.dup - "" - end - next - end - - if head and ( (EOL + boundary + EOL).size < buf.size ) - content.print buf[0 ... (buf.size - (EOL + boundary + EOL).size)] - buf[0 ... (buf.size - (EOL + boundary + EOL).size)] = "" - end - - c = if bufsize < body_size - body.read(bufsize) - else - body.read(body_size) - end - if c.nil? || c.empty? - raise EOFError, "bad content body" - end - buf.concat(c) - body_size -= c.size - end - - buf = buf.sub(/\A((?:.|\n)*?)(?:[\r\n]{1,2})?#{quoted_boundary}([\r\n]{1,2}|--)/n) do - content.print $1 - if "--" == $2 - body_size = -1 - end - boundary_end = $2.dup - "" - end - - content.rewind - - head =~ /Content-Disposition:.* filename=(?:"((?:\\.|[^\"])*)"|([^;]*))/ni - if filename = $1 || $2 - if /Mac/ni.match(env['HTTP_USER_AGENT']) and - /Mozilla/ni.match(env['HTTP_USER_AGENT']) and - (not /MSIE/ni.match(env['HTTP_USER_AGENT'])) - filename = CGI.unescape(filename) - end - content.original_path = filename.dup - end - - head =~ /Content-Type: ([^\r]*)/ni - content.content_type = $1.dup if $1 - - head =~ /Content-Disposition:.* name="?([^\";]*)"?/ni - name = $1.dup if $1 - - if params.has_key?(name) - params[name].push(content) - else - params[name] = [content] - end - break if body_size == -1 - end - raise EOFError, "bad boundary end of body part" unless boundary_end=~/--/ - - begin - body.rewind if body.respond_to?(:rewind) - rescue Errno::ESPIPE - # Handles exceptions raised by input streams that cannot be rewound - # such as when using plain CGI under Apache - end - - params - end - end # class << self - end -end diff --git a/actionpack/lib/action_controller/uploaded_file.rb b/actionpack/lib/action_controller/uploaded_file.rb index ea4845c68f..376ba3621a 100644 --- a/actionpack/lib/action_controller/uploaded_file.rb +++ b/actionpack/lib/action_controller/uploaded_file.rb @@ -7,6 +7,13 @@ module ActionController end end + def self.extended(object) + object.class_eval do + attr_accessor :original_path, :content_type + alias_method :local_path, :path + end + end + # Take the basename of the upload's original filename. # This handles the full Windows paths given by Internet Explorer # (and perhaps other broken user agents) without affecting diff --git a/actionpack/lib/action_controller/url_encoded_pair_parser.rb b/actionpack/lib/action_controller/url_encoded_pair_parser.rb index 9883ad0d85..b63dca987d 100644 --- a/actionpack/lib/action_controller/url_encoded_pair_parser.rb +++ b/actionpack/lib/action_controller/url_encoded_pair_parser.rb @@ -1,5 +1,66 @@ module ActionController class UrlEncodedPairParser < StringScanner #:nodoc: + class << self + def parse_query_parameters(query_string) + return {} if query_string.blank? + + pairs = query_string.split('&').collect do |chunk| + next if chunk.empty? + key, value = chunk.split('=', 2) + next if key.empty? + value = value.nil? ? nil : CGI.unescape(value) + [ CGI.unescape(key), value ] + end.compact + + new(pairs).result + end + + def parse_hash_parameters(params) + parser = new + + params = params.dup + until params.empty? + for key, value in params + if key.blank? + params.delete(key) + elsif value.is_a?(Array) + parser.parse(key, get_typed_value(value.shift)) + params.delete(key) if value.empty? + else + parser.parse(key, get_typed_value(value)) + params.delete(key) + end + end + end + + parser.result + end + + private + def get_typed_value(value) + case value + when String + value + when NilClass + '' + when Array + value.map { |v| get_typed_value(v) } + when Hash + if value.has_key?(:tempfile) + upload = value[:tempfile] + upload.extend(UploadedFile) + upload.original_path = value[:filename] + upload.content_type = value[:type] + upload + else + nil + end + else + raise "Unknown form value: #{value.inspect}" + end + end + end + attr_reader :top, :parent, :result def initialize(pairs = []) -- cgit v1.2.3