rack 3.0.0 → 3.2.7

data/lib/rack/query_parser.rb

@@ -1,40 +1,69 @@
 # frozen_string_literal: true
 
+require_relative 'bad_request'
+require 'uri'
+
 module Rack
   class QueryParser
-    DEFAULT_SEP = /[&] */n
-    COMMON_SEP = { ";" => /[;] */n, ";," => /[;,] */n, "&" => /[&] */n }
+    DEFAULT_SEP = /& */n
+    COMMON_SEP = { ";" => /; */n, ";," => /[;,] */n, "&" => /& */n }
 
     # ParameterTypeError is the error that is raised when incoming structural
     # parameters (parsed by parse_nested_query) contain conflicting types.
-    class ParameterTypeError < TypeError; end
+    class ParameterTypeError < TypeError
+      include BadRequest
+    end
 
     # InvalidParameterError is the error that is raised when incoming structural
     # parameters (parsed by parse_nested_query) contain invalid format or byte
     # sequence.
-    class InvalidParameterError < ArgumentError; end
+    class InvalidParameterError < ArgumentError
+      include BadRequest
+    end
 
-    # ParamsTooDeepError is the error that is raised when params are recursively
-    # nested over the specified limit.
-    class ParamsTooDeepError < RangeError; end
+    # QueryLimitError is for errors raised when the query provided exceeds one
+    # of the query parser limits.
+    class QueryLimitError < RangeError
+      include BadRequest
+    end
 
-    def self.make_default(_key_space_limit=(not_deprecated = true; nil), param_depth_limit)
-      unless not_deprecated
-        warn("`first argument `key_space limit` is deprecated and no longer has an effect. Please call with only one argument, which will be required in a future version of Rack", uplevel: 1)
-      end
+    # ParamsTooDeepError is the old name for the error that is raised when params
+    # are recursively nested over the specified limit. Make it the same as
+    # as QueryLimitError, so that code that rescues ParamsTooDeepError error
+    # to handle bad query strings also now handles other limits.
+    ParamsTooDeepError = QueryLimitError
 
-      new Params, param_depth_limit
+    def self.make_default(param_depth_limit, **options)
+      new(Params, param_depth_limit, **options)
     end
 
     attr_reader :param_depth_limit
 
-    def initialize(params_class, _key_space_limit=(not_deprecated = true; nil), param_depth_limit)
-      unless not_deprecated
-        warn("`second argument `key_space limit` is deprecated and no longer has an effect. Please call with only two arguments, which will be required in a future version of Rack", uplevel: 1)
+    env_int = lambda do |key, val|
+      if str_val = ENV[key]
+        begin
+          val = Integer(str_val, 10)
+        rescue ArgumentError
+          raise ArgumentError, "non-integer value provided for environment variable #{key}"
+        end
       end
 
+      val
+    end
+
+    BYTESIZE_LIMIT = env_int.call("RACK_QUERY_PARSER_BYTESIZE_LIMIT", 4194304)
+    private_constant :BYTESIZE_LIMIT
+
+    PARAMS_LIMIT = env_int.call("RACK_QUERY_PARSER_PARAMS_LIMIT", 4096)
+    private_constant :PARAMS_LIMIT
+
+    attr_reader :bytesize_limit
+
+    def initialize(params_class, param_depth_limit, bytesize_limit: BYTESIZE_LIMIT, params_limit: PARAMS_LIMIT)
       @params_class = params_class
       @param_depth_limit = param_depth_limit
+      @bytesize_limit = bytesize_limit
+      @params_limit = params_limit
     end
 
     # Stolen from Mongrel, with some small modifications:
@@ -42,14 +71,9 @@ module Rack
     # to parse cookies by changing the characters used in the second parameter
     # (which defaults to '&').
     def parse_query(qs, separator = nil, &unescaper)
-      unescaper ||= method(:unescape)
-
       params = make_params
 
-      (qs || '').split(separator ? (COMMON_SEP[separator] || /[#{separator}] */n) : DEFAULT_SEP).each do |p|
-        next if p.empty?
-        k, v = p.split('=', 2).map!(&unescaper)
-
+      each_query_pair(qs, separator, unescaper) do |k, v|
         if cur = params[k]
           if cur.class == Array
             params[k] << v
@@ -64,6 +88,19 @@ module Rack
       return params.to_h
     end
 
+    # Parses a query string by breaking it up at the '&', returning all key-value
+    # pairs as an array of [key, value] arrays. Unlike parse_query, this preserves
+    # all duplicate keys rather than collapsing them.
+    def parse_query_pairs(qs, separator = nil)
+      pairs = []
+
+      each_query_pair(qs, separator) do |k, v|
+        pairs << [k, v]
+      end
+
+      pairs
+    end
+
     # parse_nested_query expands a query string into structural types. Supported
     # types are Arrays, Hashes and basic value types. It is possible to supply
     # query strings with parameters of conflicting types, in this case a
@@ -72,17 +109,11 @@ module Rack
     def parse_nested_query(qs, separator = nil)
       params = make_params
 
-      unless qs.nil? || qs.empty?
-        (qs || '').split(separator ? (COMMON_SEP[separator] || /[#{separator}] */n) : DEFAULT_SEP).each do |p|
-          k, v = p.split('=', 2).map! { |s| unescape(s) }
-
-          _normalize_params(params, k, v, 0)
-        end
+      each_query_pair(qs, separator) do |k, v|
+        _normalize_params(params, k, v, 0)
       end
 
       return params.to_h
-    rescue ArgumentError => e
-      raise InvalidParameterError, e.message, e.backtrace
     end
 
     # normalize_params recursively expands parameters into structural types. If
@@ -128,8 +159,6 @@ module Rack
 
       return if k.empty?
 
-      v ||= String.new
-
       if after == ''
         if k == '[]' && depth != 0
           return [v]
@@ -190,63 +219,42 @@ module Rack
       true
     end
 
-    def unescape(s)
-      Utils.unescape(s)
-    end
-
-    class Params
-      def initialize
-        @size   = 0
-        @params = {}
-      end
+    def each_query_pair(qs, separator, unescaper = nil)
+      return if !qs || qs.empty?
 
-      def [](key)
-        @params[key]
+      if qs.bytesize > @bytesize_limit
+        raise QueryLimitError, "total query size exceeds limit (#{@bytesize_limit})"
       end
 
-      def []=(key, value)
-        @params[key] = value
-      end
+      pairs = qs.split(separator ? (COMMON_SEP[separator] || /[#{separator}] */n) : DEFAULT_SEP, @params_limit + 1)
 
-      def key?(key)
-        @params.key?(key)
+      if pairs.size > @params_limit
+        param_count = pairs.size + pairs.last.count(separator || "&")
+        raise QueryLimitError, "total number of query parameters (#{param_count}) exceeds limit (#{@params_limit})"
       end
 
-      # Recursively unwraps nested `Params` objects and constructs an object
-      # of the same shape, but using the objects' internal representations
-      # (Ruby hashes) in place of the objects. The result is a hash consisting
-      # purely of Ruby primitives.
-      #
-      #   Mutation warning!
-      #
-      #   1. This method mutates the internal representation of the `Params`
-      #      objects in order to save object allocations.
-      #
-      #   2. The value you get back is a reference to the internal hash
-      #      representation, not a copy.
-      #
-      #   3. Because the `Params` object's internal representation is mutable
-      #      through the `#[]=` method, it is not thread safe. The result of
-      #      getting the hash representation while another thread is adding a
-      #      key to it is non-deterministic.
-      #
-      def to_h
-        @params.each do |key, value|
-          case value
-          when self
-            # Handle circular references gracefully.
-            @params[key] = @params
-          when Params
-            @params[key] = value.to_h
-          when Array
-            value.map! { |v| v.kind_of?(Params) ? v.to_h : v }
-          else
-            # Ignore anything that is not a `Params` object or
-            # a collection that can contain one.
-          end
+      if unescaper
+        pairs.each do |p|
+          next if p.empty?
+          k, v = p.split('=', 2).map!(&unescaper)
+          yield k, v
+        end
+      else
+        pairs.each do |p|
+          next if p.empty?
+          k, v = p.split('=', 2).map! { |s| unescape(s) }
+          yield k, v
         end
-        @params
       end
+    rescue ArgumentError => e
+      raise InvalidParameterError, e.message, e.backtrace
+    end
+
+    def unescape(string, encoding = Encoding::UTF_8)
+      URI.decode_www_form_component(string, encoding)
+    end
+
+    class Params < Hash
       alias_method :to_params_hash, :to_h
     end
   end

data/lib/rack/request.rb

@@ -44,7 +44,7 @@ module Rack
     @x_forwarded_proto_priority = [:proto, :scheme]
 
     valid_ipv4_octet = /\.(25[0-5]|2[0-4][0-9]|[01]?[0-9]?[0-9])/
-    
+
     trusted_proxies = Regexp.union(
       /\A127#{valid_ipv4_octet}{3}\z/,                          # localhost IPv4 range 127.x.x.x, per RFC-3330
       /\A::1\z/,                                                # localhost IPv6 ::1
@@ -54,16 +54,21 @@ module Rack
       /\A192\.168#{valid_ipv4_octet}{2}\z/,                     # private IPv4 range 192.168.x.x
       /\Alocalhost\z|\Aunix(\z|:)/i,                            # localhost hostname, and unix domain sockets
     )
-    
+
     self.ip_filter = lambda { |ip| trusted_proxies.match?(ip) }
 
     ALLOWED_SCHEMES = %w(https http wss ws).freeze
 
     def initialize(env)
       @env = env
+      @ip = nil
       @params = nil
     end
 
+    def ip
+      @ip ||= super
+    end
+
     def params
       @params ||= super
     end
@@ -398,8 +403,8 @@ module Rack
               return forwarded.last
             end
           when :x_forwarded
-            if value = get_header(HTTP_X_FORWARDED_HOST)
-              return wrap_ipv6(split_header(value).last)
+            if (value = get_header(HTTP_X_FORWARDED_HOST)) && (x_forwarded_host = split_header(value).last)
+              return wrap_ipv6(x_forwarded_host)
             end
           end
         end
@@ -413,10 +418,9 @@ module Rack
 
       def ip
         remote_addresses = split_header(get_header('REMOTE_ADDR'))
-        external_addresses = reject_trusted_ip_addresses(remote_addresses)
 
-        unless external_addresses.empty?
-          return external_addresses.last
+        remote_addresses.reverse_each do |ip|
+          return ip unless trusted_proxy?(ip)
         end
 
         if (forwarded_for = self.forwarded_for) && !forwarded_for.empty?
@@ -424,7 +428,10 @@ module Rack
           # So we reject all the trusted addresses (proxy*) and return the
           # last client. Or if we trust everyone, we just return the first
           # address.
-          return reject_trusted_ip_addresses(forwarded_for).last || forwarded_for.first
+          forwarded_for.reverse_each do |ip|
+            return ip unless trusted_proxy?(ip)
+          end
+          return forwarded_for.first
         end
 
         # If all the addresses are trusted, and we aren't forwarded, just return
@@ -482,12 +489,49 @@ module Rack
 
       # Returns the data received in the query string.
       def GET
-        if get_header(RACK_REQUEST_QUERY_STRING) == query_string
-          get_header(RACK_REQUEST_QUERY_HASH)
-        else
-          query_hash = parse_query(query_string, '&')
-          set_header(RACK_REQUEST_QUERY_STRING, query_string)
-          set_header(RACK_REQUEST_QUERY_HASH, query_hash)
+        get_header(RACK_REQUEST_QUERY_HASH) || set_header(RACK_REQUEST_QUERY_HASH, parse_query(query_string, '&'))
+      end
+
+      # Returns the form data pairs received in the request body.
+      #
+      # This method support both application/x-www-form-urlencoded and
+      # multipart/form-data.
+      def form_pairs
+        if pairs = get_header(RACK_REQUEST_FORM_PAIRS)
+          return pairs
+        elsif error = get_header(RACK_REQUEST_FORM_ERROR)
+          raise error.class, error.message, cause: error.cause
+        end
+
+        begin
+          rack_input = get_header(RACK_INPUT)
+
+          # Otherwise, figure out how to parse the input:
+          if rack_input.nil?
+            set_header(RACK_REQUEST_FORM_PAIRS, [])
+          elsif form_data? || parseable_data?
+            if pairs = Rack::Multipart.parse_multipart(env, Rack::Multipart::ParamList)
+              set_header RACK_REQUEST_FORM_PAIRS, pairs
+            else
+              # Add 2 bytes. One to check whether it is over the limit, and a second
+              # in case the slice! call below removes the last byte
+              # If read returns nil, use the empty string
+              form_vars = get_header(RACK_INPUT).read(query_parser.bytesize_limit + 2) || ''
+
+              # Fix for Safari Ajax postings that always append \0
+              # form_vars.sub!(/\0\z/, '') # performance replacement:
+              form_vars.slice!(-1) if form_vars.end_with?("\0")
+
+              set_header RACK_REQUEST_FORM_VARS, form_vars
+              pairs = query_parser.parse_query_pairs(form_vars, '&')
+              set_header(RACK_REQUEST_FORM_PAIRS, pairs)
+            end
+          else
+            set_header(RACK_REQUEST_FORM_PAIRS, [])
+          end
+        rescue => error
+          set_header(RACK_REQUEST_FORM_ERROR, error)
+          raise
         end
       end
 
@@ -496,27 +540,14 @@ module Rack
       # This method support both application/x-www-form-urlencoded and
       # multipart/form-data.
       def POST
-        if get_header(RACK_INPUT).nil?
-          raise "Missing rack.input"
-        elsif get_header(RACK_REQUEST_FORM_INPUT) == get_header(RACK_INPUT)
-          get_header(RACK_REQUEST_FORM_HASH)
-        elsif form_data? || parseable_data?
-          unless set_header(RACK_REQUEST_FORM_HASH, parse_multipart)
-            form_vars = get_header(RACK_INPUT).read
-
-            # Fix for Safari Ajax postings that always append \0
-            # form_vars.sub!(/\0\z/, '') # performance replacement:
-            form_vars.slice!(-1) if form_vars.end_with?("\0")
-
-            set_header RACK_REQUEST_FORM_VARS, form_vars
-            set_header RACK_REQUEST_FORM_HASH, parse_query(form_vars, '&')
-          end
-          set_header RACK_REQUEST_FORM_INPUT, get_header(RACK_INPUT)
-          get_header RACK_REQUEST_FORM_HASH
-        else
-          set_header RACK_REQUEST_FORM_INPUT, get_header(RACK_INPUT)
-          set_header(RACK_REQUEST_FORM_HASH, {})
+        if form_hash = get_header(RACK_REQUEST_FORM_HASH)
+          return form_hash
+        elsif error = get_header(RACK_REQUEST_FORM_ERROR)
+          raise error.class, error.message, cause: error.cause
         end
+
+        pairs = form_pairs
+        set_header RACK_REQUEST_FORM_HASH, expand_param_pairs(pairs)
       end
 
       # The union of GET and POST data.
@@ -526,6 +557,10 @@ module Rack
         self.GET.merge(self.POST)
       end
 
+      # Allow overriding the query parser that the receiver will use.
+      # By default Rack::Utils.default_query_parser is used.
+      attr_writer :query_parser
+
       # Destructively update a parameter, whether it's in GET and/or POST. Returns nil.
       #
       # The parameter is updated wherever it was previous defined, so GET, POST, or both. If it wasn't previously defined, it's inserted into GET.
@@ -585,27 +620,6 @@ module Rack
         Rack::Request.ip_filter.call(ip)
       end
 
-      # shortcut for <tt>request.params[key]</tt>
-      def [](key)
-        warn("Request#[] is deprecated and will be removed in a future version of Rack. Please use request.params[] instead", uplevel: 1)
-
-        params[key.to_s]
-      end
-
-      # shortcut for <tt>request.params[key] = value</tt>
-      #
-      # Note that modifications will not be persisted in the env. Use update_param or delete_param if you want to destructively modify params.
-      def []=(key, value)
-        warn("Request#[]= is deprecated and will be removed in a future version of Rack. Please use request.params[]= instead", uplevel: 1)
-
-        params[key.to_s] = value
-      end
-
-      # like Hash#values_at
-      def values_at(*keys)
-        keys.map { |key| params[key] }
-      end
-
       private
 
       def default_session; {}; end
@@ -625,14 +639,26 @@ module Rack
       end
 
       def parse_http_accept_header(header)
-        header.to_s.split(/\s*,\s*/).map do |part|
-          attribute, parameters = part.split(/\s*;\s*/, 2)
+        # It would be nice to use filter_map here, but it's Ruby 2.7+
+        parts = header.to_s.split(',')
+
+        parts.map! do |part|
+          part.strip!
+          next if part.empty?
+
+          attribute, parameters = part.split(';', 2)
+          attribute.strip!
+          parameters&.strip!
           quality = 1.0
           if parameters and /\Aq=([\d.]+)/ =~ parameters
             quality = $1.to_f
           end
           [attribute, quality]
         end
+
+        parts.compact!
+
+        parts
       end
 
       # Get an array of values set in the RFC 7239 `Forwarded` request header.
@@ -641,7 +667,7 @@ module Rack
       end
 
       def query_parser
-        Utils.default_query_parser
+        @query_parser || Utils.default_query_parser
       end
 
       def parse_query(qs, d = '&')
@@ -649,11 +675,22 @@ module Rack
       end
 
       def parse_multipart
+        warn "Rack::Request#parse_multipart is deprecated and will be removed in a future version of Rack.", uplevel: 1
         Rack::Multipart.extract_multipart(self, query_parser)
       end
 
+      def expand_param_pairs(pairs, query_parser = query_parser())
+        params = query_parser.make_params
+
+        pairs.each do |k, v|
+          query_parser.normalize_params(params, k, v)
+        end
+
+        params.to_params_hash
+      end
+
       def split_header(value)
-        value ? value.strip.split(/[,\s]+/) : []
+        value ? value.strip.split(/[, \t]+/) : []
       end
 
       # ipv6 extracted from resolv stdlib, simplified
@@ -686,8 +723,8 @@ module Rack
           # Match IPv6 as a string of hex digits and colons in square brackets
           \[(?<address>#{ipv6})\]
           |
-          # Match any other printable string (except square brackets) as a hostname
-          (?<address>[[[:graph:]&&[^\[\]]]]*?)
+          # Match characters allowed by RFC 3986 Section 3.2.2
+          (?<address>[-a-zA-Z0-9._~%!$&'()*+,;=]*?)
         )
         (:(?<port>\d+))?
         \z
@@ -701,10 +738,6 @@ module Rack
         return match[:host], match[:address], match[:port]&.to_i
       end
 
-      def reject_trusted_ip_addresses(ip_addresses)
-        ip_addresses.reject { |ip| trusted_proxy?(ip) }
-      end
-
       FORWARDED_SCHEME_HEADERS = {
         proto: HTTP_X_FORWARDED_PROTO,
         scheme: HTTP_X_FORWARDED_SCHEME

data/lib/rack/response.rb

@@ -31,19 +31,12 @@ module Rack
     attr_accessor :length, :status, :body
     attr_reader :headers
 
-    # Deprecated, use headers instead.
-    def header
-      warn 'Rack::Response#header is deprecated and will be removed in Rack 3.1', uplevel: 1
-
-      headers
-    end
-
     # Initialize the response object with the specified +body+, +status+
     # and +headers+.
     #
     # If the +body+ is +nil+, construct an empty response object with internal
     # buffering.
-    # 
+    #
     # If the +body+ responds to +to_str+, assume it's a string-like object and
     # construct a buffered response object containing using that string as the
     # initial contents of the buffer.
@@ -62,7 +55,7 @@ module Rack
       @status = status.to_i
 
       unless headers.is_a?(Hash)
-        warn "Providing non-hash headers to Rack::Response is deprecated and will be removed in Rack 3.1", uplevel: 1
+        raise ArgumentError, "Headers must be a Hash!"
       end
 
       @headers = Headers.new
@@ -79,7 +72,8 @@ module Rack
       if body.nil?
         @body = []
         @buffered = true
-        @length = 0
+        # Body is unspecified - it may be a buffered response, or it may be a HEAD response.
+        @length = nil
       elsif body.respond_to?(:to_str)
         @body = [body]
         @buffered = true
@@ -87,7 +81,7 @@ module Rack
       else
         @body = body
         @buffered = nil # undetermined as of yet.
-        @length = 0
+        @length = nil
       end
 
       yield self if block_given?
@@ -102,20 +96,30 @@ module Rack
       CHUNKED == get_header(TRANSFER_ENCODING)
     end
 
+    def no_entity_body?
+      # The response body is an enumerable body and it is not allowed to have an entity body.
+      @body.respond_to?(:each) && STATUS_WITH_NO_ENTITY_BODY[@status]
+    end
+
     # Generate a response array consistent with the requirements of the SPEC.
     # @return [Array] a 3-tuple suitable of `[status, headers, body]`
     # which is suitable to be returned from the middleware `#call(env)` method.
     def finish(&block)
-      if STATUS_WITH_NO_ENTITY_BODY[@status]
+      if no_entity_body?
         delete_header CONTENT_TYPE
         delete_header CONTENT_LENGTH
         close
         return [@status, @headers, []]
       else
         if block_given?
+          # We don't add the content-length here as the user has provided a block that can #write additional chunks to the body.
           @block = block
           return [@status, @headers, self]
         else
+          # If we know the length of the body, set the content-length header... except if we are chunked? which is a legacy special case where the body might already be encoded and thus the actual encoded body length and the content-length are likely to be different.
+          if @length && !chunked?
+            @headers[CONTENT_LENGTH] = @length.to_s
+          end
           return [@status, @headers, @body]
         end
       end
@@ -133,7 +137,9 @@ module Rack
       end
     end
 
-    # Append to body and update content-length.
+    # Append a chunk to the response body.
+    #
+    # Converts the response into a buffered response if it wasn't already.
     #
     # NOTE: Do not mix #write and direct #body access!
     #
@@ -315,27 +321,34 @@ module Rack
 
     protected
 
+      # Convert the body of this response into an internally buffered Array if possible.
+      #
+      # `@buffered` is a ternary value which indicates whether the body is buffered. It can be:
+      # * `nil` - The body has not been buffered yet.
+      # * `true` - The body is buffered as an Array instance.
+      # * `false` - The body is not buffered and cannot be buffered.
+      #
+      # @return [Boolean] whether the body is buffered as an Array instance.
       def buffered_body!
         if @buffered.nil?
           if @body.is_a?(Array)
             # The user supplied body was an array:
             @body = @body.compact
-            @body.each do |part|
-              @length += part.to_s.bytesize
-            end
+            @length = @body.sum{|part| part.bytesize}
+            @buffered = true
           elsif @body.respond_to?(:each)
             # Turn the user supplied body into a buffered array:
             body = @body
             @body = Array.new
+            @buffered = true
 
             body.each do |part|
               @writer.call(part.to_s)
             end
 
             body.close if body.respond_to?(:close)
-            
-            @buffered = true
           else
+            # We don't know how to buffer the user-supplied body:
             @buffered = false
           end
         end
@@ -344,11 +357,13 @@ module Rack
       end
 
       def append(chunk)
+        chunk = chunk.dup unless chunk.frozen?
         @body << chunk
 
-        unless chunked?
+        if @length
           @length += chunk.bytesize
-          set_header(CONTENT_LENGTH, @length.to_s)
+        elsif @buffered
+          @length = chunk.bytesize
         end
 
         return chunk

data/lib/rack/rewindable_input.rb

@@ -21,7 +21,10 @@ module Rack
       end
 
       def call(env)
-        env[RACK_INPUT] = RewindableInput.new(env[RACK_INPUT])
+        if (input = env[RACK_INPUT])
+          env[RACK_INPUT] = RewindableInput.new(input)
+        end
+
         @app.call(env)
       end
     end