rack 2.2.17 → 3.2.0

data/lib/rack/mock_response.rb

@@ -0,0 +1,147 @@
+# frozen_string_literal: true
+
+require 'time'
+
+require_relative 'response'
+
+module Rack
+  # Rack::MockResponse provides useful helpers for testing your apps.
+  # Usually, you don't create the MockResponse on your own, but use
+  # MockRequest.
+
+  class MockResponse < Rack::Response
+    class Cookie
+      attr_reader :name, :value, :path, :domain, :expires, :secure
+
+      def initialize(args)
+        @name = args["name"]
+        @value = args["value"]
+        @path = args["path"]
+        @domain = args["domain"]
+        @expires = args["expires"]
+        @secure = args["secure"]
+      end
+
+      def method_missing(method_name, *args, &block)
+        @value.send(method_name, *args, &block)
+      end
+      # :nocov:
+      ruby2_keywords(:method_missing) if respond_to?(:ruby2_keywords, true)
+      # :nocov:
+
+      def respond_to_missing?(method_name, include_all = false)
+        @value.respond_to?(method_name, include_all) || super
+      end
+    end
+
+    class << self
+      alias [] new
+    end
+
+    # Headers
+    attr_reader :original_headers, :cookies
+
+    # Errors
+    attr_accessor :errors
+
+    def initialize(status, headers, body, errors = nil)
+      @original_headers = headers
+
+      if errors
+        @errors = errors.string if errors.respond_to?(:string)
+      else
+        @errors = ""
+      end
+
+      super(body, status, headers)
+
+      @cookies = parse_cookies_from_header
+      buffered_body!
+    end
+
+    def =~(other)
+      body =~ other
+    end
+
+    def match(other)
+      body.match other
+    end
+
+    def body
+      return @buffered_body if defined?(@buffered_body)
+
+      # FIXME: apparently users of MockResponse expect the return value of
+      # MockResponse#body to be a string.  However, the real response object
+      # returns the body as a list.
+      #
+      # See spec_showstatus.rb:
+      #
+      #   should "not replace existing messages" do
+      #     ...
+      #     res.body.should == "foo!"
+      #   end
+      buffer = @buffered_body = String.new
+
+      @body.each do |chunk|
+        buffer << chunk
+      end
+
+      return buffer
+    end
+
+    def empty?
+      [201, 204, 304].include? status
+    end
+
+    def cookie(name)
+      cookies.fetch(name, nil)
+    end
+
+    private
+
+    def parse_cookies_from_header
+      cookies = Hash.new
+      set_cookie_header = headers['set-cookie']
+      if set_cookie_header && !set_cookie_header.empty?
+        Array(set_cookie_header).each do |cookie|
+          cookie_name, cookie_filling = cookie.split('=', 2)
+          cookie_attributes = identify_cookie_attributes cookie_filling
+          parsed_cookie = Cookie.new(
+            'name' => cookie_name.strip,
+            'value' => cookie_attributes.fetch('value'),
+            'path' => cookie_attributes.fetch('path', nil),
+            'domain' => cookie_attributes.fetch('domain', nil),
+            'expires' => cookie_attributes.fetch('expires', nil),
+            'secure' => cookie_attributes.fetch('secure', false)
+          )
+          cookies.store(cookie_name, parsed_cookie)
+        end
+      end
+      cookies
+    end
+
+    def identify_cookie_attributes(cookie_filling)
+      cookie_bits = cookie_filling.split(';')
+      cookie_attributes = Hash.new
+      cookie_attributes.store('value', Array(cookie_bits[0].strip))
+      cookie_bits.drop(1).each do |bit|
+        if bit.include? '='
+          cookie_attribute, attribute_value = bit.split('=', 2)
+          cookie_attributes.store(cookie_attribute.strip.downcase, attribute_value.strip)
+        end
+        if bit.include? 'secure'
+          cookie_attributes.store('secure', true)
+        end
+      end
+
+      if cookie_attributes.key? 'max-age'
+        cookie_attributes.store('expires', Time.now + cookie_attributes['max-age'].to_i)
+      elsif cookie_attributes.key? 'expires'
+        cookie_attributes.store('expires', Time.httpdate(cookie_attributes['expires']))
+      end
+
+      cookie_attributes
+    end
+
+  end
+end

data/lib/rack/multipart/generator.rb

@@ -1,5 +1,7 @@
 # frozen_string_literal: true
 
+require_relative 'uploaded_file'
+
 module Rack
   module Multipart
     class Generator
@@ -74,12 +76,12 @@ module Rack
 
       def content_for_tempfile(io, file, name)
         length = ::File.stat(file.path).size if file.path
-        filename = "; filename=\"#{Utils.escape(file.original_filename)}\"" if file.original_filename
+        filename = "; filename=\"#{Utils.escape_path(file.original_filename)}\""
 <<-EOF
 --#{MULTIPART_BOUNDARY}\r
-Content-Disposition: form-data; name="#{name}"#{filename}\r
-Content-Type: #{file.content_type}\r
-#{"Content-Length: #{length}\r\n" if length}\r
+content-disposition: form-data; name="#{name}"#{filename}\r
+content-type: #{file.content_type}\r
+#{"content-length: #{length}\r\n" if length}\r
 #{io.read}\r
 EOF
       end
@@ -87,7 +89,7 @@ EOF
       def content_for_other(file, name)
 <<-EOF
 --#{MULTIPART_BOUNDARY}\r
-Content-Disposition: form-data; name="#{name}"\r
+content-disposition: form-data; name="#{name}"\r
 \r
 #{file}\r
 EOF

data/lib/rack/multipart/parser.rb

@@ -2,14 +2,55 @@
 
 require 'strscan'
 
+require_relative '../utils'
+require_relative '../bad_request'
+
 module Rack
   module Multipart
-    class MultipartPartLimitError < Errno::EMFILE; end
-    class MultipartTotalPartLimitError < StandardError; end
+    class MultipartPartLimitError < Errno::EMFILE
+      include BadRequest
+    end
 
-    class Parser
-      (require_relative '../core_ext/regexp'; using ::Rack::RegexpExtensions) if RUBY_VERSION < '2.4'
+    class MultipartTotalPartLimitError < StandardError
+      include BadRequest
+    end
+
+    # Use specific error class when parsing multipart request
+    # that ends early.
+    class EmptyContentError < ::EOFError
+      include BadRequest
+    end
+
+    # Base class for multipart exceptions that do not subclass from
+    # other exception classes for backwards compatibility.
+    class BoundaryTooLongError < StandardError
+      include BadRequest
+    end
 
+    # Prefer to use the BoundaryTooLongError class or Rack::BadRequest.
+    Error = BoundaryTooLongError
+
+    EOL = "\r\n"
+    FWS = /[ \t]+(?:\r\n[ \t]+)?/ # whitespace with optional folding
+    HEADER_VALUE = "(?:[^\r\n]|\r\n[ \t])*" # anything but a non-folding CRLF
+    MULTIPART = %r|\Amultipart/.*boundary=\"?([^\";,]+)\"?|ni
+    MULTIPART_CONTENT_TYPE = /^Content-Type:#{FWS}?(#{HEADER_VALUE})/ni
+    MULTIPART_CONTENT_DISPOSITION = /^Content-Disposition:#{FWS}?(#{HEADER_VALUE})/ni
+    MULTIPART_CONTENT_ID = /^Content-ID:#{FWS}?(#{HEADER_VALUE})/ni
+
+    # Rack::Multipart::Parser handles parsing of multipart/form-data requests.
+    #
+    # File Parameter Contents
+    #
+    # When processing file uploads, the parser returns a hash containing
+    # information about uploaded files. For +file+ parameters, the hash includes:
+    #
+    # * +:filename+ - The original filename, already URL decoded by the parser
+    # * +:type+ - The content type of the uploaded file  
+    # * +:name+ - The parameter name from the form
+    # * +:tempfile+ - A Tempfile object containing the uploaded data
+    # * +:head+ - The raw header content for this part
+    class Parser
       BUFSIZE = 1_048_576
       TEXT_PLAIN = "text/plain"
       TEMPFILE_FACTORY = lambda { |filename, content_type|
@@ -18,8 +59,6 @@ module Rack
         Tempfile.new(["RackMultipart", extension])
       }
 
-      BOUNDARY_REGEX = /\A([^\n]*(?:\n|\Z))/
-
       class BoundedIO # :nodoc:
         def initialize(io, content_length)
           @io             = io
@@ -41,16 +80,12 @@ module Rack
           if str
             @cursor += str.bytesize
           else
-            # Raise an error for mismatching Content-Length and actual contents
+            # Raise an error for mismatching content-length and actual contents
             raise EOFError, "bad content body"
           end
 
           str
         end
-
-        def rewind
-          @io.rewind
-        end
       end
 
       MultipartInfo = Struct.new :params, :tmp_files
@@ -69,18 +104,17 @@ module Rack
         boundary = parse_boundary content_type
         return EMPTY unless boundary
 
+        if boundary.length > 70
+          # RFC 1521 Section 7.2.1 imposes a 70 character maximum for the boundary.
+          # Most clients use no more than 55 characters.
+          raise BoundaryTooLongError, "multipart boundary size too large (#{boundary.length} characters)"
+        end
+
         io = BoundedIO.new(io, content_length) if content_length
-        outbuf = String.new
 
         parser = new(boundary, tmpfile, bufsize, qp)
-        parser.on_read io.read(bufsize, outbuf)
+        parser.parse(io)
 
-        loop do
-          break if parser.state == :DONE
-          parser.on_read io.read(bufsize, outbuf)
-        end
-
-        io.rewind
         parser.result
       end
 
@@ -180,33 +214,49 @@ module Rack
       def initialize(boundary, tempfile, bufsize, query_parser)
         @query_parser   = query_parser
         @params         = query_parser.make_params
-        @boundary       = "--#{boundary}"
         @bufsize        = bufsize
 
-        @full_boundary = @boundary
-        @end_boundary = @boundary + '--'
         @state = :FAST_FORWARD
         @mime_index = 0
         @collector = Collector.new tempfile
 
         @sbuf = StringScanner.new("".dup)
-        @body_regex = /(?:#{EOL})?#{Regexp.quote(@boundary)}(?:#{EOL}|--)/m
-        @end_boundary_size = boundary.bytesize + 6 # (-- at start, -- at finish, EOL at end)
-        @rx_max_size = EOL.size + @boundary.bytesize + [EOL.size, '--'.size].max
+        @body_regex = /(?:#{EOL}|\A)--#{Regexp.quote(boundary)}(?:#{EOL}|--)/m
+        @body_regex_at_end = /#{@body_regex}\z/m
+        @end_boundary_size = boundary.bytesize + 4 # (-- at start, -- at finish)
+        @rx_max_size = boundary.bytesize + 6 # (\r\n-- at start, either \r\n or -- at finish)
         @head_regex = /(.*?#{EOL})#{EOL}/m
       end
 
-      def on_read(content)
-        handle_empty_content!(content)
-        @sbuf.concat content
-        run_parser
+      def parse(io)
+        outbuf = String.new
+        read_data(io, outbuf)
+
+        loop do
+          status =
+            case @state
+            when :FAST_FORWARD
+              handle_fast_forward
+            when :CONSUME_TOKEN
+              handle_consume_token
+            when :MIME_HEAD
+              handle_mime_head
+            when :MIME_BODY
+              handle_mime_body
+            else # when :DONE
+              return
+            end
+
+          read_data(io, outbuf) if status == :want_read
+        end
       end
 
       def result
         @collector.each do |part|
           part.get_data do |data|
             tag_multipart_encoding(part.filename, part.content_type, part.name, data)
-            @query_parser.normalize_params(@params, part.name, data, @query_parser.param_depth_limit)
+            name, data = handle_dummy_encoding(part.name, data)
+            @query_parser.normalize_params(@params, name, data)
           end
         end
         MultipartInfo.new @params.to_params_hash, @collector.find_all(&:file?).map(&:body)
@@ -214,34 +264,39 @@ module Rack
 
       private
 
-      def run_parser
-        loop do
-          case @state
-          when :FAST_FORWARD
-            break if handle_fast_forward == :want_read
-          when :CONSUME_TOKEN
-            break if handle_consume_token == :want_read
-          when :MIME_HEAD
-            break if handle_mime_head == :want_read
-          when :MIME_BODY
-            break if handle_mime_body == :want_read
-          when :DONE
-            break
-          end
-        end
+      def read_data(io, outbuf)
+        content = io.read(@bufsize, outbuf)
+        handle_empty_content!(content)
+        @sbuf.concat(content)
       end
 
+      # This handles the initial parser state.  We read until we find the starting
+      # boundary, then we can transition to the next state. If we find the ending
+      # boundary, this is an invalid multipart upload, but keep scanning for opening
+      # boundary in that case. If no boundary found, we need to keep reading data
+      # and retry. It's highly unlikely the initial read will not consume the
+      # boundary.  The client would have to deliberately craft a response
+      # with the opening boundary beyond the buffer size for that to happen.
       def handle_fast_forward
-        tok = consume_boundary
+        while true
+          case consume_boundary
+          when :BOUNDARY
+            # found opening boundary, transition to next state
+            @state = :MIME_HEAD
+            return
+          when :END_BOUNDARY
+            # invalid multipart upload
+            if @sbuf.pos == @end_boundary_size && @sbuf.rest == EOL
+              # stop parsing a buffer if a buffer is only an end boundary.
+              @state = :DONE
+              return
+            end
 
-        if tok == :END_BOUNDARY && @sbuf.pos == @end_boundary_size && @sbuf.eos?
-          # stop parsing a buffer if a buffer is only an end boundary.
-          @state = :DONE
-        elsif tok
-          @state = :MIME_HEAD
-        else
-          raise EOFError, "bad content body" if @sbuf.rest_size >= @bufsize
-          :want_read
+            # retry for opening boundary
+          else
+            # no boundary found, keep reading data
+            return :want_read
+          end
         end
       end
 
@@ -255,17 +310,101 @@ module Rack
         end
       end
 
+      CONTENT_DISPOSITION_MAX_PARAMS = 16
+      CONTENT_DISPOSITION_MAX_BYTES = 1536
       def handle_mime_head
         if @sbuf.scan_until(@head_regex)
           head = @sbuf[1]
           content_type = head[MULTIPART_CONTENT_TYPE, 1]
-          if name = head[MULTIPART_CONTENT_DISPOSITION, 1]
-            name = Rack::Auth::Digest::Params::dequote(name)
+          if (disposition = head[MULTIPART_CONTENT_DISPOSITION, 1]) &&
+              disposition.bytesize <= CONTENT_DISPOSITION_MAX_BYTES
+
+            # ignore actual content-disposition value (should always be form-data)
+            i = disposition.index(';')
+            disposition.slice!(0, i+1)
+            param = nil
+            num_params = 0
+
+            # Parse parameter list
+            while i = disposition.index('=')
+              # Only parse up to max parameters, to avoid potential denial of service
+              num_params += 1
+              break if num_params > CONTENT_DISPOSITION_MAX_PARAMS
+
+              # Found end of parameter name, ensure forward progress in loop
+              param = disposition.slice!(0, i+1)
+
+              # Remove ending equals and preceding whitespace from parameter name
+              param.chomp!('=')
+              param.lstrip!
+
+              if disposition[0] == '"'
+                # Parameter value is quoted, parse it, handling backslash escapes
+                disposition.slice!(0, 1)
+                value = String.new
+
+                while i = disposition.index(/(["\\])/)
+                  c = $1
+
+                  # Append all content until ending quote or escape
+                  value << disposition.slice!(0, i)
+
+                  # Remove either backslash or ending quote,
+                  # ensures forward progress in loop
+                  disposition.slice!(0, 1)
+
+                  # stop parsing parameter value if found ending quote
+                  break if c == '"'
+
+                  escaped_char = disposition.slice!(0, 1)
+                  if param == 'filename' && escaped_char != '"'
+                    # Possible IE uploaded filename, append both escape backslash and value
+                    value << c << escaped_char
+                  else
+                    # Other only append escaped value
+                    value << escaped_char
+                  end
+                end
+              else
+                if i = disposition.index(';')
+                  # Parameter value unquoted (which may be invalid), value ends at semicolon
+                  value = disposition.slice!(0, i)
+                else
+                  # If no ending semicolon, assume remainder of line is value and stop
+                  # parsing
+                  disposition.strip!
+                  value = disposition
+                  disposition = ''
+                end
+              end
+
+              case param
+              when 'name'
+                name = value
+              when 'filename'
+                filename = value
+              when 'filename*'
+                filename_star = value
+              # else
+              # ignore other parameters
+              end
+
+              # skip trailing semicolon, to proceed to next parameter
+              if i = disposition.index(';')
+                disposition.slice!(0, i+1)
+              end
+            end
           else
             name = head[MULTIPART_CONTENT_ID, 1]
           end
 
-          filename = get_filename(head)
+          if filename_star
+            encoding, _, filename = filename_star.split("'", 3)
+            filename = normalize_filename(filename || '')
+            filename.force_encoding(find_encoding(encoding))
+          elsif filename
+            filename = normalize_filename(filename)
+          end
 
           if name.nil? || name.empty?
             name = filename || "#{content_type || TEXT_PLAIN}[]".dup
@@ -280,7 +419,7 @@ module Rack
 
       def handle_mime_body
         if (body_with_boundary = @sbuf.check_until(@body_regex)) # check but do not advance the pointer yet
-          body = body_with_boundary.sub(/#{@body_regex}\z/m, '') # remove the boundary from the string
+          body = body_with_boundary.sub(@body_regex_at_end, '') # remove the boundary from the string
           @collector.on_mime_body @mime_index, body
           @sbuf.pos += body.length + 2 # skip \r\n after the content
           @state = :CONSUME_TOKEN
@@ -297,54 +436,31 @@ module Rack
         end
       end
 
-      def full_boundary; @full_boundary; end
-
+      # Scan until the we find the start or end of the boundary.
+      # If we find it, return the appropriate symbol for the start or
+      # end of the boundary.  If we don't find the start or end of the
+      # boundary, clear the buffer and return nil.
       def consume_boundary
-        while read_buffer = @sbuf.scan_until(BOUNDARY_REGEX)
-          case read_buffer.strip
-          when full_boundary then return :BOUNDARY
-          when @end_boundary then return :END_BOUNDARY
-          end
-          return if @sbuf.eos?
+        if read_buffer = @sbuf.scan_until(@body_regex)
+          read_buffer.end_with?(EOL) ? :BOUNDARY : :END_BOUNDARY
+        else
+          @sbuf.terminate
+          nil
         end
       end
 
-      def get_filename(head)
-        filename = nil
-        case head
-        when RFC2183
-          params = Hash[*head.scan(DISPPARM).flat_map(&:compact)]
-
-          if filename = params['filename']
-            filename = $1 if filename =~ /^"(.*)"$/
-          elsif filename = params['filename*']
-            encoding, _, filename = filename.split("'", 3)
-          end
-        when BROKEN
-          filename = $1
-          filename = $1 if filename =~ /^"(.*)"$/
-        end
-
-        return unless filename
-
+      def normalize_filename(filename)
         if filename.scan(/%.?.?/).all? { |s| /%[0-9a-fA-F]{2}/.match?(s) }
           filename = Utils.unescape_path(filename)
         end
 
         filename.scrub!
 
-        if filename !~ /\\[^\\"]/
-          filename = filename.gsub(/\\(.)/, '\1')
-        end
-
-        if encoding
-          filename.force_encoding ::Encoding.find(encoding)
-        end
-
-        filename
+        filename.split(/[\/\\]/).last || String.new
       end
 
       CHARSET = "charset"
+      deprecate_constant :CHARSET
 
       def tag_multipart_encoding(filename, content_type, name, body)
         name = name.to_s
@@ -365,7 +481,9 @@ module Rack
               k.strip!
               v.strip!
               v = v[1..-2] if v.start_with?('"') && v.end_with?('"')
-              encoding = Encoding.find v if k == CHARSET
+              if k == "charset"
+                encoding = find_encoding(v)
+              end
             end
           end
         end
@@ -374,9 +492,37 @@ module Rack
         body.force_encoding(encoding)
       end
 
+      # Return the related Encoding object. However, because
+      # enc is submitted by the user, it may be invalid, so
+      # use a binary encoding in that case.
+      def find_encoding(enc)
+        Encoding.find enc
+      rescue ArgumentError
+        Encoding::BINARY
+      end
+
+      REENCODE_DUMMY_ENCODINGS = {
+        # ISO-2022-JP is a legacy but still widely used encoding in Japan
+        # Here we convert ISO-2022-JP to UTF-8 so that it can be handled.
+        Encoding::ISO_2022_JP => true
+
+        # Other dummy encodings are rarely used and have not been supported yet.
+        # Adding support for them will require careful considerations.
+      }
+
+      def handle_dummy_encoding(name, body)
+        # A string object with a 'dummy' encoding does not have full functionality and can cause errors.
+        # So here we covert it to UTF-8 so that it can be handled properly.
+        if name.encoding.dummy? && REENCODE_DUMMY_ENCODINGS[name.encoding]
+          name = name.encode(Encoding::UTF_8)
+          body = body.encode(Encoding::UTF_8)
+        end
+        return name, body
+      end
+
       def handle_empty_content!(content)
         if content.nil? || content.empty?
-          raise EOFError
+          raise EmptyContentError
         end
       end
     end