rack-proxy 0.7.7 → 2.0.1

data/lib/rack/proxy.rb

@@ -1,45 +1,127 @@
-require "net_http_hacked"
+# frozen_string_literal: true
+
+require "rack"
+require "net/https"
 require "rack/http_streaming_response"
+require "rack/proxy/version"
 
 module Rack
-
   # Subclass and bring your own #rewrite_request and #rewrite_response
   class Proxy
-    VERSION = "0.7.7".freeze
-
     HOP_BY_HOP_HEADERS = {
-      'connection' => true,
-      'keep-alive' => true,
-      'proxy-authenticate' => true,
-      'proxy-authorization' => true,
-      'te' => true,
-      'trailer' => true,
-      'transfer-encoding' => true,
-      'upgrade' => true
+      "connection" => true,
+      "keep-alive" => true,
+      "proxy-authenticate" => true,
+      "proxy-authorization" => true,
+      "te" => true,
+      "trailer" => true,
+      "transfer-encoding" => true,
+      "upgrade" => true
     }.freeze
 
+    # Backend/network failures that must surface as 502 Bad Gateway rather than
+    # crashing the proxy with a raw 500. Construction- and policy-time failures
+    # are mapped separately (400/501/502) in #perform_request.
+    BACKEND_ERRORS = [
+      Errno::ECONNREFUSED, Errno::ECONNRESET, Errno::ECONNABORTED,
+      Errno::EHOSTUNREACH, Errno::ENETUNREACH, Errno::ETIMEDOUT, Errno::EPIPE,
+      SocketError,
+      Timeout::Error,               # includes Net::OpenTimeout
+      Net::ReadTimeout, Net::WriteTimeout,
+      IOError,                      # includes EOFError
+      OpenSSL::SSL::SSLError,
+      Net::ProtocolError,
+      # A malformed status line / header block raises these; they subclass
+      # StandardError directly, NOT Net::ProtocolError, so list them explicitly
+      # or a hostile backend crashes the proxy with a raw 500 instead of a 502.
+      Net::HTTPBadResponse, Net::HTTPHeaderSyntaxError
+    ].freeze
+
+    class InvalidRequest < StandardError; end
+
+    # Net::HTTP copies a body stream until EOF, even when Content-Length is
+    # smaller. Bound the input so extra bytes cannot become a second request.
+    class RequestBodyStream
+      def initialize(input, length)
+        @input, @remaining = input, length
+      end
+
+      def read(length = nil, buffer = nil)
+        return read_remaining(buffer) if length.nil?
+
+        if @remaining.zero?
+          buffer&.clear
+          return nil
+        end
+
+        data = @input.read([length, @remaining].min, buffer)
+        raise InvalidRequest, "request body is shorter than Content-Length" if data.nil? || data.empty?
+
+        @remaining -= data.bytesize
+        data
+      end
+
+      private
+
+      # IO#read without a length returns everything up to EOF ("" once there).
+      # Net::HTTP never calls it that way, but instrumentation layers wrapping
+      # Net::HTTP#request (WebMock's adapter, for one) do.
+      def read_remaining(buffer)
+        data = +"".b
+        while (chunk = read(16_384))
+          data << chunk
+        end
+        buffer ? buffer.replace(data) : data
+      end
+    end
+    private_constant :InvalidRequest, :RequestBodyStream
+
     class << self
       def extract_http_request_headers(env)
         headers = env.reject do |k, v|
-          !(/^HTTP_[A-Z0-9_\.]+$/ === k) || v.nil?
+          !(/^HTTP_[A-Z0-9_.]+$/ === k) || v.nil?
         end.map do |k, v|
           [reconstruct_header_name(k), v]
         end.then { |pairs| build_header_hash(pairs) }
 
-        x_forwarded_for = (headers['X-Forwarded-For'].to_s.split(/, +/) << env['REMOTE_ADDR']).join(', ')
+        # Strip hop-by-hop headers before forwarding. Relaying the client's
+        # Connection / TE / Transfer-Encoding (etc.) enables request smuggling
+        # and confuses the backend — these are connection-scoped, not end-to-end.
+        # Per RFC 7230 §6.1 any field named in the inbound Connection header is
+        # itself hop-by-hop for this hop. Use #delete (not #reject!) so the
+        # returned HeaderHash's case-insensitive index stays consistent on Rack 2.
+        connection_named = headers["Connection"].to_s.downcase.split(/,\s*/).map(&:strip)
+        headers.keys.each do |key|
+          headers.delete(key) if HOP_BY_HOP_HEADERS[key.downcase] || connection_named.include?(key.downcase)
+        end
+
+        x_forwarded_for = (headers["X-Forwarded-For"].to_s.split(/, +/) << env["REMOTE_ADDR"]).join(", ")
 
-        headers.merge!('X-Forwarded-For' => x_forwarded_for)
+        headers.merge!("X-Forwarded-For" => x_forwarded_for)
       end
 
       def normalize_headers(headers)
         mapped = headers.map do |k, v|
-          [titleize(k), if v.is_a? Array then v.join("\n") else v end]
+          value = if v.is_a?(Array)
+            if v.length == 1
+              v.first
+            elsif Rack.const_defined?(:Headers, false)
+              v
+            else
+              v.join("\n")
+            end
+          else
+            v
+          end
+          [titleize(k), value]
         end
-        build_header_hash Hash[mapped]
+        build_header_hash mapped.to_h
       end
 
       def build_header_hash(pairs)
-        if Rack.const_defined?(:Headers)
+        # Pass inherit: false so we only check Rack's own constants — otherwise
+        # a top-level ::Headers defined by the host app would falsely match.
+        if Rack.const_defined?(:Headers, false)
           # Rack::Headers is only available from Rack 3 onward
           Headers.new.tap { |headers| pairs.each { |k, v| headers[k] = v } }
         else
@@ -51,7 +133,7 @@ module Rack
       protected
 
       def reconstruct_header_name(name)
-        titleize(name.sub(/^HTTP_/, "").gsub("_", "-"))
+        titleize(name.sub(/^HTTP_/, "").tr("_", "-"))
       end
 
       def titleize(str)
@@ -60,7 +142,7 @@ module Rack
     end
 
     # @option opts [String, URI::HTTP] :backend Backend host to proxy requests to
-    def initialize(app = nil, opts= {})
+    def initialize(app = nil, opts = {})
       if app.is_a?(Hash)
         opts = app
         @app = nil
@@ -69,17 +151,55 @@ module Rack
       end
 
       @streaming = opts.fetch(:streaming, true)
-      @ssl_verify_none = opts.fetch(:ssl_verify_none, false)
       @backend = opts[:backend] ? URI(opts[:backend]) : nil
+      # With no :backend (and no env["rack.backend"]), the destination is
+      # derived from the client-controlled Host header. Since 1.0 that dynamic
+      # mode is refused (502) unless explicitly opted into, because a bare
+      # proxy would otherwise be an open proxy / SSRF pivot (cloud metadata
+      # endpoints, loopback, RFC1918). Combine the opt-in with a
+      # #backend_allowed? allowlist — see the README "Security considerations".
+      @allow_dynamic_backend = opts.fetch(:allow_dynamic_backend, false)
       @read_timeout = opts.fetch(:read_timeout, 60)
+      # Connect and per-write deadlines. Without these a slow/hostile backend can
+      # stall a thread for Net::HTTP's 60s defaults even with a small read_timeout.
+      @open_timeout = opts[:open_timeout]
+      @write_timeout = opts[:write_timeout]
+      # Optional cap (in bytes) on the backend response size, to bound memory
+      # against a hostile/huge backend. Checked before buffering each chunk in
+      # either mode, as well as against declared lengths. Default: no cap.
+      @max_response_length = opts[:max_response_length]
+      # :ssl_version pins an exact protocol and is deprecated (it forbids TLS 1.3);
+      # prefer :min_version / :max_version, which map to Net::HTTP#min_version=/#max_version=.
       @ssl_version = opts[:ssl_version]
+      @min_version = opts[:min_version]
+      @max_version = opts[:max_version]
       @cert = opts[:cert]
       @key = opts[:key]
+      # Trust anchors for VERIFY_PEER: :ca_file is a PEM bundle path, :cert_store
+      # an OpenSSL::X509::Store. Use these for private-CA backends instead of
+      # disabling verification with ssl_verify_none.
+      @ca_file = opts[:ca_file]
+      @cert_store = opts[:cert_store]
+      # SSL verification: defaults to VERIFY_PEER (Ruby's Net::HTTP default).
+      # Pass ssl_verify_none: true to explicitly disable cert verification, or
+      # pass verify_mode: <OpenSSL::SSL::VERIFY_*> for full control.
       @verify_mode = opts[:verify_mode]
+      @verify_mode ||= OpenSSL::SSL::VERIFY_NONE if opts[:ssl_verify_none]
 
       @username = opts[:username]
       @password = opts[:password]
 
+      # Opt-in request hardening (see README "Security considerations"):
+      # :strip_credentials drops Cookie/Authorization from the forwarded
+      # request; :replace_x_forwarded_for discards the client-supplied
+      # X-Forwarded-For chain and forwards only this hop's REMOTE_ADDR.
+      @strip_credentials = opts[:strip_credentials]
+      @replace_x_forwarded_for = opts[:replace_x_forwarded_for]
+
+      # Optional logger for Net::HTTP debug output. Accepts anything with a #<< method
+      # (e.g. $stdout, a StringIO, or a Ruby Logger instance).
+      @logger = opts[:logger]
+
       @opts = opts
     end
 
@@ -97,72 +217,238 @@ module Rack
       triplet
     end
 
+    # SSRF guardrail, consulted for EVERY request with the resolved backend
+    # (`backend` responds to #host, #port and #scheme). Return false to refuse,
+    # which makes the proxy respond 502. The default allows the backend, because
+    # by the time this hook runs the destination is either app-configured
+    # (:backend / env["rack.backend"]) or the deployment has explicitly passed
+    # allow_dynamic_backend: true — override it to pin an allowlist on top:
+    #
+    #   def backend_allowed?(backend)
+    #     %w[api.internal.example.com].include?(backend.host)
+    #   end
+    def backend_allowed?(backend)
+      true
+    end
+
     protected
 
     def perform_request(env)
       source_request = Rack::Request.new(env)
 
-      # Initialize request
-      if source_request.fullpath == ""
-        full_path = URI.parse(env['REQUEST_URI']).request_uri
-      else
-        full_path = source_request.fullpath
-      end
+      # Everything that can fail on a hostile/unreachable request or backend is
+      # mapped to a status code here so the proxy never surfaces a raw 500:
+      #   400 malformed request URI, 501 unknown method, 502 backend failure.
+      begin
+        # Initialize request
+        full_path = if source_request.fullpath == ""
+          URI.parse(env["REQUEST_URI"]).request_uri
+        else
+          source_request.fullpath
+        end
 
-      target_request = Net::HTTP.const_get(source_request.request_method.capitalize, false).new(full_path)
+        request_class = net_http_request_class(source_request.request_method)
+        return [501, {}, []] if request_class.nil?
 
-      # Setup headers
-      target_request.initialize_http_header(self.class.extract_http_request_headers(source_request.env))
+        target_request = request_class.new(full_path)
+
+        # Setup headers
+        request_headers = self.class.extract_http_request_headers(source_request.env)
+        if @strip_credentials
+          request_headers.delete("Cookie")
+          request_headers.delete("Authorization")
+        end
+        if @replace_x_forwarded_for
+          if (remote_addr = env["REMOTE_ADDR"])
+            request_headers["X-Forwarded-For"] = remote_addr
+          else
+            request_headers.delete("X-Forwarded-For")
+          end
+        end
+        target_request.initialize_http_header(request_headers)
 
-      # Setup body
-      if target_request.request_body_permitted? && source_request.body
-        target_request.body_stream    = source_request.body
-        target_request.content_length = source_request.content_length.to_i
-        target_request.content_type   = source_request.content_type if source_request.content_type
-        target_request.body_stream.rewind
+        # Forward the backend response verbatim: don't let Net::HTTP transparently
+        # gzip-decode it, which would leave the forwarded Content-Length describing
+        # the compressed size (a body/Content-Length desync). The client decodes
+        # its own content-encoding.
+        target_request.instance_variable_set(:@decode_content, false) if target_request.instance_variable_defined?(:@decode_content)
+
+        # Rack supplies decoded input. Generate framing for this hop instead
+        # of forwarding the client's Transfer-Encoding or guessing a zero size.
+        if target_request.request_body_permitted? && source_request.body
+          input = source_request.body
+          input.rewind if input.respond_to?(:rewind)
+          if (length = source_request.content_length)
+            raise InvalidRequest, "invalid Content-Length" unless /\A[0-9]+\z/.match?(length)
+
+            target_request.content_length = length.to_i
+            target_request.body_stream = RequestBodyStream.new(input, length.to_i)
+          else
+            target_request.delete("Content-Length")
+            target_request["Transfer-Encoding"] = "chunked"
+            target_request.body_stream = input
+          end
+          target_request.content_type = source_request.content_type if source_request.content_type
+        end
+
+        # Use basic auth if we have to
+        target_request.basic_auth(@username, @password) if @username && @password
+
+        backend = env.delete("rack.backend") || @backend
+        # env["rack.backend"] is documented as a URI, but accept a URI-parseable
+        # string too (symmetric with the :backend option) rather than crash on
+        # #scheme later; a malformed string surfaces as 400 via the rescue.
+        backend = URI(backend) if backend.is_a?(String)
+        if backend.nil?
+          # Dynamic mode: the destination would come from the client-controlled
+          # Host header. Refused unless the deployment opted in (SSRF guard).
+          unless @allow_dynamic_backend
+            if @logger.respond_to?(:<<)
+              @logger << "rack-proxy: refusing Host-derived backend #{source_request.host.inspect} " \
+                         "(no :backend configured; pass allow_dynamic_backend: true to opt in)\n"
+            end
+            return [502, {}, []]
+          end
+          backend = source_request
+        end
+        unless backend_allowed?(backend)
+          if @logger.respond_to?(:<<)
+            @logger << "rack-proxy: backend #{backend.host.inspect} refused by backend_allowed?\n"
+          end
+          return [502, {}, []]
+        end
+
+        use_ssl = backend.scheme == "https" || @cert
+        read_timeout = env.delete("http.read_timeout") || @read_timeout
+
+        if @streaming
+          # streaming response (the actual network communication is deferred, a.k.a. streamed)
+          target_response = HttpStreamingResponse.new(target_request, backend.host, backend.port) do |http|
+            configure_backend_connection(http, use_ssl: use_ssl, read_timeout: read_timeout)
+          end
+          target_response.logger = @logger if @logger
+          code = target_response.code
+          headers = prepare_response_headers(target_response.headers, code, target_request)
+          if response_body_permitted?(target_request, code)
+            target_response.max_response_length = @max_response_length
+            body = target_response
+          else
+            target_response.close
+            body = []
+          end
+        else
+          http = Net::HTTP.new(backend.host, backend.port)
+          configure_backend_connection(http, use_ssl: use_ssl, read_timeout: read_timeout)
+
+          http.start do
+            http.request(target_request) do |response|
+              code = response.code.to_i
+              headers = prepare_response_headers(response.to_hash, code, target_request)
+              body = []
+              if response_body_permitted?(target_request, code)
+                buffered = +"".b
+                response.read_body do |chunk|
+                  check_response_length!(buffered.bytesize + chunk.bytesize)
+                  buffered << chunk
+                end
+                if response.content_length && buffered.bytesize != response.content_length
+                  raise EOFError, "backend response is shorter than Content-Length"
+                end
+                body << buffered unless buffered.empty?
+              end
+            end
+          end
+        end
+      rescue URI::InvalidURIError, InvalidRequest
+        target_response&.close
+        return [400, {}, []]
+      rescue *BACKEND_ERRORS, HttpStreamingResponse::ResponseTooLarge => e
+        target_response&.close
+        @logger << "rack-proxy: backend request failed: #{e.class}: #{e.message}\n" if @logger.respond_to?(:<<)
+        return [502, {}, []]
       end
 
-      # Use basic auth if we have to
-      target_request.basic_auth(@username, @password) if @username && @password
+      [code, headers, body]
+    end
 
-      backend = env.delete('rack.backend') || @backend || source_request
-      use_ssl = backend.scheme == "https" || @cert
-      read_timeout = env.delete('http.read_timeout') || @read_timeout
+    private
 
-      # Create the response
-      if @streaming
-        # streaming response (the actual network communication is deferred, a.k.a. streamed)
-        target_response = HttpStreamingResponse.new(target_request, backend.host, backend.port)
-        target_response.use_ssl = use_ssl
-        target_response.read_timeout = read_timeout
-        target_response.ssl_version = @ssl_version if @ssl_version
-        target_response.verify_mode = (@verify_mode || OpenSSL::SSL::VERIFY_NONE) if use_ssl
-        target_response.cert = @cert if @cert
-        target_response.key = @key if @key
-      else
-        http = Net::HTTP.new(backend.host, backend.port)
-        http.use_ssl = use_ssl if use_ssl
-        http.read_timeout = read_timeout
-        http.ssl_version = @ssl_version if @ssl_version
-        http.verify_mode = (@verify_mode || OpenSSL::SSL::VERIFY_NONE if use_ssl) if use_ssl
-        http.cert = @cert if @cert
-        http.key = @key if @key
-
-        target_response = http.start do
-          http.request(target_request)
+    def response_body_permitted?(request, code)
+      request.response_body_permitted? && !Rack::Utils::STATUS_WITH_NO_ENTITY_BODY[code] && code != 205
+    end
+
+    def check_response_length!(length)
+      if @max_response_length && length && length > @max_response_length
+        raise HttpStreamingResponse::ResponseTooLarge, "backend response exceeded max_response_length=#{@max_response_length}"
+      end
+    end
+
+    # Validate framing before dropping hop-by-hop fields or reading a body.
+    # Net::HTTP dechunks responses but otherwise preserves their headers.
+    def prepare_response_headers(raw_headers, code, request)
+      # Rack 2 HeaderHash#each joins arrays with newlines. Read values directly
+      # so repeated framing/Connection fields stay distinct during validation.
+      headers = self.class.build_header_hash(raw_headers.keys.map { |key| [key, raw_headers[key]] })
+      transfer_encoding = headers["Transfer-Encoding"]
+      content_length = headers["Content-Length"]
+      if transfer_encoding
+        if content_length || Array(transfer_encoding).join(",").strip.downcase != "chunked"
+          raise Net::HTTPBadResponse, "ambiguous or unsupported backend transfer encoding"
+        end
+      end
+      if content_length
+        lengths = Array(content_length).flat_map { |value| value.split(",", -1).map(&:strip) }
+        unless lengths.all? { |value| /\A[0-9]+\z/.match?(value) } && lengths.map(&:to_i).uniq.length == 1
+          raise Net::HTTPBadResponse, "invalid backend Content-Length"
         end
+        headers["Content-Length"] = lengths.first.to_i.to_s
+        check_response_length!(lengths.first.to_i) if response_body_permitted?(request, code)
       end
 
-      code    = target_response.code
-      headers = self.class.normalize_headers(target_response.respond_to?(:headers) ? target_response.headers : target_response.to_hash)
-      body    = target_response.body || [""]
-      body    = [body] unless body.respond_to?(:each)
+      connection_named = Array(headers["Connection"]).join(",").downcase.split(",").map(&:strip)
+      headers.keys.each do |key|
+        headers.delete(key) if HOP_BY_HOP_HEADERS[key.downcase] || connection_named.include?(key.downcase)
+      end
+      if Rack::Utils::STATUS_WITH_NO_ENTITY_BODY[code]
+        headers.delete("Content-Length")
+        headers.delete("Content-Type")
+      end
+      self.class.normalize_headers(headers)
+    end
 
-      # According to https://tools.ietf.org/html/draft-ietf-httpbis-p1-messaging-14#section-7.1.3.1Acc
-      # should remove hop-by-hop header fields
-      headers.reject! { |k| HOP_BY_HOP_HEADERS[k.downcase] }
+    # Resolve the Net::HTTP request class for an HTTP method, or nil if there is
+    # no matching Net::HTTP::<Verb> (unknown/unsupported method -> 501).
+    def net_http_request_class(method)
+      Net::HTTP.const_get(method.capitalize, false)
+    rescue NameError
+      nil
+    end
 
-      [code, headers, body]
+    # Single source of truth for TLS/timeout setup, applied to the (real
+    # Net::HTTP) connection on both the streaming and non-streaming paths so a
+    # TLS option — notably the VERIFY_PEER default — can never land on only one.
+    def configure_backend_connection(conn, use_ssl:, read_timeout:)
+      conn.use_ssl = use_ssl
+      # Request input need not be rewindable. A transport retry could replay an
+      # operation with an empty or partial body, so neither path retries it.
+      conn.max_retries = 0
+      conn.read_timeout = read_timeout
+      conn.open_timeout = @open_timeout if @open_timeout
+      conn.write_timeout = @write_timeout if @write_timeout
+
+      if use_ssl
+        conn.verify_mode = @verify_mode || OpenSSL::SSL::VERIFY_PEER
+        conn.ca_file = @ca_file if @ca_file
+        conn.cert_store = @cert_store if @cert_store
+        conn.min_version = @min_version if @min_version
+        conn.max_version = @max_version if @max_version
+        conn.ssl_version = @ssl_version if @ssl_version # deprecated; prefer min/max_version
+        conn.cert = @cert if @cert
+        conn.key = @key if @key
+      end
+
+      conn.set_debug_output(@logger) if @logger
+      conn
     end
   end
 end

data/lib/rack-proxy.rb

@@ -1 +1,3 @@
-require "rack/proxy"
+# frozen_string_literal: true
+
+require "rack/proxy"

data/rack-proxy.gemspec

@@ -1,25 +1,30 @@
-# -*- encoding: utf-8 -*-
 $:.push File.expand_path("../lib", __FILE__)
-require "rack-proxy"
+require "rack/proxy/version"
 
 Gem::Specification.new do |s|
-  s.name        = "rack-proxy"
-  s.version     = Rack::Proxy::VERSION
-  s.platform    = Gem::Platform::RUBY
-  s.license     = 'MIT'
-  s.authors     = ["Jacek Becela"]
-  s.email       = ["jacek.becela@gmail.com"]
-  s.homepage    = "https://github.com/ncr/rack-proxy"
-  s.summary     = %q{A request/response rewriting HTTP proxy. A Rack app.}
-  s.description = %q{A Rack app that provides request/response rewriting proxy capabilities with streaming.}
-  s.required_ruby_version = '>= 2.6'
+  s.name = "rack-proxy"
+  s.version = Rack::Proxy::VERSION
+  s.platform = Gem::Platform::RUBY
+  s.license = "MIT"
+  s.authors = ["Jacek Becela"]
+  s.email = ["jacek.becela@gmail.com"]
+  s.homepage = "https://github.com/ncr/rack-proxy"
+  s.summary = "A request/response rewriting HTTP proxy. A Rack app."
+  s.description = "A Rack app that provides request/response rewriting proxy capabilities with streaming."
+  s.required_ruby_version = ">= 3.0"
 
-  s.files         = `git ls-files`.split("\n")
-  s.test_files    = `git ls-files -- {test,spec,features}/*`.split("\n")
-  s.executables   = `git ls-files -- bin/*`.split("\n").map{ |f| File.basename(f) }
+  s.metadata = {
+    "source_code_uri" => "https://github.com/ncr/rack-proxy",
+    "changelog_uri" => "https://github.com/ncr/rack-proxy/blob/master/CHANGELOG.md",
+    "bug_tracker_uri" => "https://github.com/ncr/rack-proxy/issues",
+    "rubygems_mfa_required" => "true"
+  }
+
+  s.files = Dir["lib/**/*.rb"] + %w[README.md LICENSE CHANGELOG.md SECURITY.md rack-proxy.gemspec]
   s.require_paths = ["lib"]
 
-  s.add_dependency("rack")
+  s.add_dependency("rack", ">= 2.0", "< 4")
   s.add_development_dependency("rack-test")
   s.add_development_dependency("test-unit")
+  s.add_development_dependency("webrick")
 end