Imprement yaml parser written by pure ruby instead of Psych

bundler/spec/runtime/inline_spec.rb

@@ -678,8 +678,10 @@ def confirm(msg, newline = nil)
 
     expect(out).to include("Installing psych 999")
     expect(out).to include("Installing stringio 999")
-    expect(out).to include("The psych gem was resolved to 999")
-    expect(out).to include("The stringio gem was resolved to 999")
+    if Gem.respond_to?(:use_psych?) && Gem.use_psych?
+      expect(out).to include("The psych gem was resolved to 999")
+      expect(out).to include("The stringio gem was resolved to 999")
+    end
   end
 
   it "leaves a lockfile in the same directory as the inline script alone" do

lib/rubygems.rb

@@ -640,16 +640,30 @@ def self.add_to_load_path(*paths)
   end
 
   @yaml_loaded = false
+  @use_psych = nil
+
+  ##
+  # Returns true if the Psych YAML parser is enabled via configuration.
+
+  def self.use_psych?
+    @use_psych || false
+  end
 
   ##
   # Loads YAML, preferring Psych
 
   def self.load_yaml
     return if @yaml_loaded
 
-    require "psych"
-    require_relative "rubygems/psych_tree"
+    @use_psych = ENV["RUBYGEMS_USE_PSYCH"] == "true" ||
+                 (defined?(@configuration) && @configuration && !@configuration[:use_psych].nil?)
+
+    if @use_psych
+      require "psych"
+      require_relative "rubygems/psych_tree"
+    end
 
+    require_relative "rubygems/yaml_serializer"
     require_relative "rubygems/safe_yaml"
 
     @yaml_loaded = true

lib/rubygems/commands/specification_command.rb

@@ -147,7 +147,7 @@ def execute
       say case options[:format]
           when :ruby then s.to_ruby
           when :marshal then Marshal.dump s
-          else s.to_yaml
+          else Gem.use_psych? ? s.to_yaml : Gem::YAMLSerializer.dump(s)
       end
 
       say "\n"

lib/rubygems/config_file.rb

@@ -49,6 +49,7 @@ class Gem::ConfigFile
   DEFAULT_IPV4_FALLBACK_ENABLED = false
   DEFAULT_INSTALL_EXTENSION_IN_LIB = false
   DEFAULT_GLOBAL_GEM_CACHE = false
+  DEFAULT_USE_PSYCH = false
 
   ##
   # For Ruby packagers to set configuration defaults.  Set in
@@ -161,6 +162,11 @@ class Gem::ConfigFile
 
   attr_accessor :global_gem_cache
 
+  ##
+  # Use Psych (C extension YAML parser) instead of the pure Ruby YAMLSerializer.
+
+  attr_accessor :use_psych
+
   ##
   # Path name of directory or file of openssl client certificate, used for remote https connection with client authentication
 
@@ -199,6 +205,7 @@ def initialize(args)
     @install_extension_in_lib = DEFAULT_INSTALL_EXTENSION_IN_LIB
     @ipv4_fallback_enabled = ENV["IPV4_FALLBACK_ENABLED"] == "true" || DEFAULT_IPV4_FALLBACK_ENABLED
     @global_gem_cache = ENV["RUBYGEMS_GLOBAL_GEM_CACHE"] == "true" || DEFAULT_GLOBAL_GEM_CACHE
+    @use_psych = ENV["RUBYGEMS_USE_PSYCH"] == "true" || DEFAULT_USE_PSYCH
 
     operating_system_config = Marshal.load Marshal.dump(OPERATING_SYSTEM_DEFAULTS)
     platform_config = Marshal.load Marshal.dump(PLATFORM_DEFAULTS)
@@ -221,7 +228,7 @@ def initialize(args)
       # gemhome and gempath are not working with symbol keys
       if %w[backtrace bulk_threshold verbose update_sources cert_expiration_length_days
             concurrent_downloads install_extension_in_lib ipv4_fallback_enabled
-            global_gem_cache sources
+            global_gem_cache use_psych sources
             disable_default_gem_server ssl_verify_mode ssl_ca_cert ssl_client_cert].include?(k)
         k.to_sym
       else
@@ -239,6 +246,7 @@ def initialize(args)
     @install_extension_in_lib    = @hash[:install_extension_in_lib]    if @hash.key? :install_extension_in_lib
     @ipv4_fallback_enabled       = @hash[:ipv4_fallback_enabled]       if @hash.key? :ipv4_fallback_enabled
     @global_gem_cache            = @hash[:global_gem_cache]            if @hash.key? :global_gem_cache
+    @use_psych                   = @hash[:use_psych]                   if @hash.key? :use_psych
 
     @home                        = @hash[:gemhome]                     if @hash.key? :gemhome
     @path                        = @hash[:gempath]                     if @hash.key? :gempath
@@ -378,7 +386,9 @@ def load_file(filename)
 
     begin
       config = self.class.load_with_rubygems_config_hash(File.read(filename))
-      if config.keys.any? {|k| k.to_s.gsub(%r{https?:\/\/}, "").include?(": ") }
+      has_invalid_keys = config.keys.any? {|k| k.to_s.gsub(%r{https?:\/\/}, "").include?(": ") }
+      has_invalid_values = config.values.any? {|v| v.is_a?(String) && v.gsub(%r{https?:\/\/}, "").match?(/\A\S+: /) }
+      if has_invalid_keys || has_invalid_values
         warn "Failed to load #{filename} because it doesn't contain valid YAML hash"
         return {}
       else
@@ -563,7 +573,9 @@ def self.dump_with_rubygems_yaml(content)
   def self.load_with_rubygems_config_hash(yaml)
     require_relative "yaml_serializer"
 
-    content = Gem::YAMLSerializer.load(yaml)
+    content = Gem::YAMLSerializer.load(yaml, permitted_classes: [])
+    return {} unless content.is_a?(Hash)
+
     deep_transform_config_keys!(content)
   end
 
@@ -597,7 +609,7 @@ def self.deep_transform_config_keys!(config)
         else
           v
         end
-      elsif v.empty?
+      elsif v.respond_to?(:empty?) && v.empty?
         nil
       elsif v.is_a?(Hash)
         deep_transform_config_keys!(v)

lib/rubygems/package.rb

@@ -232,7 +232,11 @@ def add_checksums(tar)
 
     tar.add_file_signed "checksums.yaml.gz", 0o444, @signer do |io|
       gzip_to io do |gz_io|
-        Psych.dump checksums_by_algorithm, gz_io
+        if Gem.use_psych?
+          Psych.dump checksums_by_algorithm, gz_io
+        else
+          gz_io.write Gem::YAMLSerializer.dump(checksums_by_algorithm)
+        end
       end
     end
   end

lib/rubygems/safe_marshal.rb

@@ -54,6 +54,7 @@ module SafeMarshal
       "Gem::NameTuple" => %w[@name @version @platform],
       "Gem::Platform" => %w[@os @cpu @version],
       "Psych::PrivateType" => %w[@value @type_id],
+      "YAML::PrivateType" => %w[@value @type_id],
     }.freeze
     private_constant :PERMITTED_IVARS
 

lib/rubygems/safe_yaml.rb

@@ -35,11 +35,34 @@ def self.aliases_enabled? # :nodoc:
     end
 
     def self.safe_load(input)
-      ::Psych.safe_load(input, permitted_classes: PERMITTED_CLASSES, permitted_symbols: PERMITTED_SYMBOLS, aliases: @aliases_enabled)
+      if Gem.use_psych?
+        ::Psych.safe_load(input, permitted_classes: PERMITTED_CLASSES,
+                                 permitted_symbols: PERMITTED_SYMBOLS, aliases: @aliases_enabled)
+      else
+        Gem::YAMLSerializer.load(
+          input,
+          permitted_classes: PERMITTED_CLASSES,
+          permitted_symbols: PERMITTED_SYMBOLS,
+          aliases: aliases_enabled?
+        )
+      end
     end
 
     def self.load(input)
-      ::Psych.safe_load(input, permitted_classes: [::Symbol])
+      if Gem.use_psych?
+        if ::Psych.respond_to?(:unsafe_load)
+          ::Psych.unsafe_load(input)
+        else
+          ::Psych.load(input)
+        end
+      else
+        Gem::YAMLSerializer.load(
+          input,
+          permitted_classes: PERMITTED_CLASSES,
+          permitted_symbols: PERMITTED_SYMBOLS,
+          aliases: aliases_enabled?
+        )
+      end
     end
   end
 end

lib/rubygems/specification.rb

@@ -1275,7 +1275,7 @@ def self._load(str)
       raise unless message.include?("YAML::")
 
       unless Object.const_defined?(:YAML)
-        Object.const_set "YAML", Psych
+        Object.const_set "YAML", Module.new
         yaml_set = true
       end
 
@@ -1284,7 +1284,7 @@ def self._load(str)
 
         YAML::Syck.const_set "DefaultKey", Class.new if message.include?("YAML::Syck::DefaultKey") && !YAML::Syck.const_defined?(:DefaultKey)
       elsif message.include?("YAML::PrivateType") && !YAML.const_defined?(:PrivateType)
-        YAML.const_set "PrivateType", Class.new
+        YAML.const_set "PrivateType", Class.new { attr_accessor :type_id, :value }
       end
 
       retry_count += 1
@@ -2455,24 +2455,28 @@ def to_spec
   def to_yaml(opts = {}) # :nodoc:
     Gem.load_yaml
 
-    # Because the user can switch the YAML engine behind our
-    # back, we have to check again here to make sure that our
-    # psych code was properly loaded, and load it if not.
-    unless Gem.const_defined?(:NoAliasYAMLTree)
-      require_relative "psych_tree"
-    end
+    if Gem.use_psych?
+      # Because the user can switch the YAML engine behind our
+      # back, we have to check again here to make sure that our
+      # psych code was properly loaded, and load it if not.
+      unless Gem.const_defined?(:NoAliasYAMLTree)
+        require_relative "psych_tree"
+      end
 
-    builder = Gem::NoAliasYAMLTree.create
-    builder << self
-    ast = builder.tree
+      builder = Gem::NoAliasYAMLTree.create
+      builder << self
+      ast = builder.tree
 
-    require "stringio"
-    io = StringIO.new
-    io.set_encoding Encoding::UTF_8
+      require "stringio"
+      io = StringIO.new
+      io.set_encoding Encoding::UTF_8
 
-    Psych::Visitors::Emitter.new(io).accept(ast)
+      Psych::Visitors::Emitter.new(io).accept(ast)
 
-    io.string.gsub(/ !!null \n/, " \n")
+      io.string.gsub(/ !!null \n/, " \n")
+    else
+      Gem::YAMLSerializer.dump(self)
+    end
   end
 
   ##