ghostest 0.1.0

data/lib/ghostest/test_condition.rb

@@ -0,0 +1,24 @@
+module Ghostest
+  class TestCondition
+    def initialize(language_klass)
+      @language_klass = language_klass
+      unless File.exist?(@language_klass.test_condition_yml_path)
+        FileUtils.mkdir_p(File.dirname(@language_klass.test_condition_yml_path))
+        File.write(@language_klass.test_condition_yml_path, YAML.dump({}))
+      end
+      @test_condition = YAML.load(File.read(@language_klass.test_condition_yml_path)) || {}
+    end
+
+    def save_as_updated!(source_path)
+      source_md5 = Digest::MD5.hexdigest(File.read(source_path))
+      @test_condition[source_path] = { source_md5: }
+
+      File.write(@language_klass.test_condition_yml_path, YAML.dump(@test_condition))
+    end
+
+    def should_update_test?(source_path)
+      source_md5 = Digest::MD5.hexdigest(File.read(source_path))
+      @test_condition[source_path].nil? || @test_condition[source_path][:source_md5] != source_md5
+    end
+  end
+end

data/lib/ghostest/version.rb

@@ -0,0 +1,3 @@
+module Ghostest
+  VERSION = "0.1.0"
+end

data/lib/ghostest.rb

@@ -0,0 +1,58 @@
+# don't need to test this file
+require_relative "ghostest/version"
+
+require 'digest'
+
+require 'openai'
+require 'html2markdown'
+require 'addressable'
+require 'baran'
+require 'tiktoken_ruby'
+require 'google-apis-customsearch_v1'
+require 'colorize'
+require 'pry'
+require 'i18n'
+unless defined?(HashWithIndifferentAccess)
+  require 'indifference'
+end
+require 'erb'
+require 'bundler'
+
+require "ghostest/attr_reader"
+require "ghostest/config/agent"
+require "ghostest/config"
+require "ghostest/config_error"
+require "ghostest/error"
+require "ghostest/languages/ruby"
+require "ghostest/logger"
+require "ghostest/manager"
+require "ghostest/test_condition"
+require "ghostest/version"
+require "ghostest"
+require "google_custom_search"
+require "i18n_translator"
+require "initializers/i18n"
+
+require "llm/message_container"
+require "llm/agents/base"
+require "llm/agents/reviewer"
+require "llm/agents/test_designer"
+require "llm/agents/test_programmer"
+require "llm/clients/base"
+require "llm/clients/azure_open_ai"
+require "llm/clients/open_ai"
+require "llm/functions/base"
+require "llm/functions/add_to_memory"
+require "llm/functions/exec_rspec_test"
+require "llm/functions/fix_one_rspec_test"
+require "llm/functions/get_files_list"
+require "llm/functions/get_gem_files_list"
+require "llm/functions/make_new_file"
+require "llm/functions/overwrite_file"
+require "llm/functions/read_file"
+require "llm/functions/record_lgtm"
+require "llm/functions/report_bug"
+require "llm/functions/switch_assignee"
+
+module Ghostest
+end

data/lib/google_custom_search.rb

@@ -0,0 +1,30 @@
+class GoogleCustomSearch
+  def search(query, args = {})
+    service.list_cses(cx: ENV['GOOGLE_CUSTOM_SEARCH_CSE_ID'], q: query, **args)
+  end
+
+  def service
+    @service ||= begin
+      service = Google::Apis::CustomsearchV1::CustomSearchAPIService.new
+      authorizer = make_authorizer
+      authorizer.fetch_access_token!
+      service.authorization = authorizer
+      service
+    end
+  end
+
+  def make_authorizer
+    sa_key = ENV.fetch("GOOGLE_SA_PRIVATE_KEY")
+    key = ::OpenSSL::PKey::RSA.new(sa_key)
+    cred = ::Signet::OAuth2::Client.new(
+      token_credential_uri: "https://oauth2.googleapis.com/token",
+      audience: "https://oauth2.googleapis.com/token",
+      scope: %w[
+        https://www.googleapis.com/auth/cse
+      ],
+      issuer: ENV.fetch("GOOGLE_SA_CLIENT_EMAIL"),
+      signing_key: key
+    )
+    cred.configure_connection({})
+  end
+end

data/lib/i18n_translator.rb

@@ -0,0 +1,66 @@
+require 'digest'
+class I18nTranslator
+  def self.update_dictionary!(from_locale, to_locales)
+    from_file_paths = I18n.load_path.select { |path| path.match(/#{from_locale}\.yml$/) }
+    from_file_paths.each do |from_file_path|
+      to_locales.each do |to_locale|
+        from_hash = {
+          to_locale.to_s => YAML.load(File.read(from_file_path))[from_locale.to_s],
+        }
+        to_file_path = from_file_path.gsub(/#{from_locale}\.yml$/, "#{to_locale}.yml")
+        to_hash = File.exist?(to_file_path) ? YAML.load(File.read(to_file_path)) : {}
+        to_hash = deep_translate(from_hash, to_hash, from_locale, to_locale)
+        File.write(to_file_path, YAML.dump(to_hash))
+      end
+    end
+  end
+
+  def self.deep_translate(from_hash, to_hash, from_locale, to_locale)
+    ret = {}
+    from_hash.each do |key, val|
+      if val.is_a?(Hash)
+        h = deep_translate(val, to_hash[key] || {}, from_locale, to_locale)
+        ret[key] = Hash[h.sort]
+      else
+        md5_key = "#{key}_md5"
+        md5 = ::Digest::MD5.hexdigest(val)
+        if to_hash[md5_key].nil? || to_hash[md5_key] != md5
+          ret[md5_key] = md5
+          ret[key] = translate(val)
+        else
+          ret[md5_key] = to_hash[md5_key]
+          ret[key] = to_hash[key]
+        end
+      end
+    end
+    ret
+  end
+
+  def self.translator
+    @translator ||= ::Llm::Clients::AzureOpenAi.new
+  end
+
+  # don't need to test this method
+  def self.translate(str)
+    # DeepLのほうが望ましいがLLMで代用
+    ret = self.translator.chat(parameters: {
+      messages: [
+        {
+          role: "system",
+          content: "You are an excellent translator. We translate strings sent by users into accurate English.\n" + \
+            "We do not output any content other than the translation.\n" + \
+            "Please keep the position and number of the new line code(\\n).\n" + \
+            "Never omit the line feed code at the end of a sentence.",
+        },
+        {
+          role: "user",
+          content: str,
+        },
+      ],
+    })
+    translated = ret.dig("choices", 0, "message", "content")
+    puts("#{str} => #{translated.green}")
+    sleep(1)
+    translated
+  end
+end

data/lib/initializers/i18n.rb

@@ -0,0 +1,9 @@
+
+I18n.load_path += Dir[File.expand_path("config/locales") + "/**/*.yml"]
+
+module I18n
+  def self.just_raise_that_exception(*args)
+    raise "i18n: #{args.first}"
+  end
+end
+I18n.exception_handler = :just_raise_that_exception

data/lib/llm/agents/base.rb

@@ -0,0 +1,31 @@
+module Llm
+  module Agents
+    class Base
+      include AttrReader
+      attr_reader :name, :config, :agent_config
+
+      def initialize(name, config, logger)
+        @name = name
+        @config = config
+        @agent_config = config.agents[name]
+        @logger = logger
+      end
+
+      def name_with_type
+        "#{self.name}(#{self.agent_config.role})"
+      end
+
+      def say(message, color: nil)
+        if color.nil?
+          @logger.info("#{self.name}: #{message}".send(self.agent_config.color))
+        else
+          if color
+            @logger.info("#{self.name}: #{message}".send(color))
+          else
+            @logger.info("#{self.name}: #{message}")
+          end
+        end
+      end
+    end
+  end
+end

data/lib/llm/agents/reviewer.rb

@@ -0,0 +1,50 @@
+module Llm
+  module Agents
+    class Reviewer < Base
+      def initialize(name, config, logger)
+        super(name, config, logger)
+        @record_lgtm_function = Llm::Functions::RecordLgtm.new
+      end
+
+      def lgtm?
+        @record_lgtm_function.lgtm?
+      end
+
+      # skip test for this method
+      def work(
+        source_path: raise("source_path is required"),
+        test_path: nil,
+        switch_assignee_function: raise("switch_assignee_function is required"))
+        say("start to work for #{source_path}")
+
+        message_container = Llm::MessageContainer.new
+        test_path ||= self.config.language_klass.convert_source_path_to_test_path(source_path)
+        message_container.add_system_message(self.agent_config.system_prompt.gsub("%{source_path}", source_path).gsub("%{test_path}", test_path))
+
+        if switch_assignee_function.messages.size > 0
+          message_container.add_system_message(I18n.t('ghostest.agents.reviewer.last_assignee_comment',
+                                                      last_assignee: switch_assignee_function.last_assignee,
+                                                      comment: switch_assignee_function.last_message))
+        end
+
+        azure_open_ai = Llm::Clients::AzureOpenAi.new
+        io = azure_open_ai.chat_with_function_calling_loop(
+          messages: message_container,
+          functions: [
+            @record_lgtm_function,
+            Llm::Functions::GetFilesList.new,
+            Llm::Functions::ReadFile.new,
+            Llm::Functions::AddToMemory.new(message_container),
+            switch_assignee_function,
+
+          ] + self.config.language_klass.create_functions,
+          agent: self,
+        )
+
+        comment = io.rewind && io.read
+        say(comment)
+        comment
+      end
+    end
+  end
+end

data/lib/llm/agents/test_designer.rb

@@ -0,0 +1,43 @@
+module Llm
+  module Agents
+    class TestDesigner < Base
+
+      # skip test for this method
+      def work(
+        source_path: raise("source_path is required"),
+        test_path: nil,
+        switch_assignee_function: raise("switch_assignee_function is required"))
+
+        say("start to work for #{source_path}")
+
+        message_container = Llm::MessageContainer.new
+        test_path ||= self.config.language_klass.convert_source_path_to_test_path(source_path)
+        message_container.add_system_message(self.agent_config.system_prompt.gsub("%{source_path}", source_path).gsub("%{test_path}", test_path))
+
+        if switch_assignee_function.messages.size > 0
+          message_container.add_system_message(I18n.t('ghostest.agents.test_designer.last_assignee_comment',
+                                                      last_assignee: switch_assignee_function.last_assignee,
+                                                      comment: switch_assignee_function.last_message))
+        end
+
+        azure_open_ai = Llm::Clients::AzureOpenAi.new
+        io = azure_open_ai.chat_with_function_calling_loop(
+          messages: message_container,
+          functions: [
+            Llm::Functions::GetFilesList.new,
+            Llm::Functions::ReadFile.new,
+            Llm::Functions::AddToMemory.new(message_container),
+            Llm::Functions::ReportBug.new,
+            switch_assignee_function,
+
+          ] + self.config.language_klass.create_functions,
+          agent: self,
+        )
+
+        comment = io.rewind && io.read
+        say(comment)
+        comment
+      end
+    end
+  end
+end

data/lib/llm/agents/test_programmer.rb

@@ -0,0 +1,45 @@
+module Llm
+  module Agents
+    class TestProgrammer < Base
+
+      # skip test for this method
+      def work(
+        source_path: raise("source_path is required"),
+        test_path: nil,
+        switch_assignee_function: raise("switch_assignee_function is required"))
+
+        say("start to work for #{source_path}")
+
+        message_container = Llm::MessageContainer.new
+        test_path ||= self.config.language_klass.convert_source_path_to_test_path(source_path)
+        message_container.add_system_message(self.agent_config.system_prompt.gsub("%{source_path}", source_path).gsub("%{test_path}", test_path))
+
+        if switch_assignee_function.messages.size > 0
+          message_container.add_system_message(I18n.t('ghostest.agents.test_programmer.last_assignee_comment',
+                                                      last_assignee: switch_assignee_function.last_assignee,
+                                                      comment: switch_assignee_function.last_message))
+        end
+
+        azure_open_ai = Llm::Clients::AzureOpenAi.new
+        io = azure_open_ai.chat_with_function_calling_loop(
+          messages: message_container,
+          functions: [
+            Llm::Functions::GetFilesList.new,
+            Llm::Functions::ReadFile.new,
+            Llm::Functions::OverwriteFile.new,
+            Llm::Functions::MakeNewFile.new,
+            Llm::Functions::AddToMemory.new(message_container),
+            Llm::Functions::ReportBug.new,
+            switch_assignee_function,
+
+          ] + self.config.language_klass.create_functions,
+          agent: self,
+        )
+
+        comment = io.rewind && io.read
+        say(comment)
+        comment
+      end
+    end
+  end
+end

data/lib/llm/clients/azure_open_ai.rb

@@ -0,0 +1,15 @@
+module Llm
+  module Clients
+    class AzureOpenAi < Llm::Clients::Base
+      def initialize(timeout: 30000)
+        @client = OpenAI::Client.new(
+          api_type: :azure,
+          api_version: ENV.fetch("AZURE_API_VERSION"),
+          access_token: ENV.fetch("AZURE_OPENAI_API_KEY"),
+          uri_base: "#{ENV.fetch("AZURE_API_BASE")}/openai/deployments/#{ENV.fetch("AZURE_DEPLOYMENT_NAME")}",
+          request_timeout: timeout,
+        )
+      end
+    end
+  end
+end

data/lib/llm/clients/base.rb

@@ -0,0 +1,88 @@
+require "ghostest/attr_reader"
+module Llm
+  module Clients
+    class Base
+      include AttrReader
+      attr_reader :client
+
+      def chat(parameters: {})
+        parameters = parameters.with_indifferent_access
+        if parameters[:messages].nil? || parameters[:messages].empty?
+          raise 'messages is required.'
+        end
+        parameters[:temperature] ||= 0.5
+        parameters[:top_p] ||= 1
+        parameters[:frequency_penalty] ||= 0
+        parameters[:presence_penalty] ||= 0
+
+        self.client.chat(parameters:)
+      end
+
+      def chat_with_function_calling_loop(**args)
+        agent = args.delete(:agent) || (raise 'agent is required.')
+        chat_message_io = StringIO.new
+
+        if args[:messages].is_a?(Llm::MessageContainer)
+          message_container = args[:messages]
+        else
+          message_container = Llm::MessageContainer.new
+          message_container.add_raw_messages(args[:messages])
+        end
+
+        i = 0
+        while (i += 1) < 20
+          ret = self.chat(parameters: args.merge({
+                                                   messages: message_container.to_capped_messages,
+                                                   functions: args[:functions].map { |f| f.definition },
+                                                 }))
+          if ret.dig("choices", 0, "finish_reason") != 'function_call'
+            break
+          end
+
+          # Function calling
+          message = ret.dig("choices", 0, "message")
+          function = args[:functions].detect { |f| f.function_name == message['function_call']['name'].to_sym }
+          message_container.add_raw_message(message.merge({ content: nil }))
+
+          function_args = (JSON.parse(message.dig('function_call', 'arguments')) || {}).with_indifferent_access
+          agent.say(function.function_name)
+          agent.say(function_args, color: false)
+
+          function_result = function.execute_and_generate_message(function_args)
+          message_container.add_raw_message({
+                                              role: "function",
+                                              name: function.function_name,
+                                              content: JSON.dump(function_result),
+                                            })
+
+          if function.stop_llm_call?
+            chat_message_io.write(function_result)
+            return chat_message_io
+          end
+        end
+
+        # メッセージ表示
+        if content = ret.dig("choices", 0, "message", "content")
+          chat_message_io.write(content)
+        else
+          # エラー発生か、function callingの回数が多すぎる場合は、他エージェントに相談する
+          functions = args[:functions].select do |f| %w[
+            switch_assignee
+            report_bug
+          ].include?(f.function_name.to_s) end
+          ret = self.chat(parameters: args.merge({
+                                                   messages: message_container.to_capped_messages,
+                                                   functions: functions.map { |f| f.definition },
+                                                 }))
+          if content = ret.dig("choices", 0, "message", "content")
+            chat_message_io.write(content)
+          else
+            puts ret
+            exit 1
+          end
+        end
+        chat_message_io
+      end
+    end
+  end
+end

data/lib/llm/clients/open_ai.rb

@@ -0,0 +1,14 @@
+module Llm
+  module Clients
+    class OpenAi < Llm::Clients::Base
+      def initialize(timeout: 300)
+        @client = OpenAI::Client.new(
+          api_version: ENV.fetch("OPENAI_API_VERSION"),
+          access_token: ENV.fetch("OPENAI_API_KEY"),
+          uri_base: "https://openai.com/openai/deployments/chat/completions",
+          request_timeout: timeout,
+        )
+      end
+    end
+  end
+end

data/lib/llm/functions/add_to_memory.rb

@@ -0,0 +1,41 @@
+module Llm
+  module Functions
+    class AddToMemory < Base
+      def function_name
+        :add_to_memory
+      end
+
+      def initialize(message_container)
+        @message_container = message_container
+      end
+
+      def definition
+        return @definition unless @definition.nil?
+
+        @definition = {
+          name: self.function_name,
+          description: I18n.t("ghostest.functions.#{self.function_name}.description"),
+          parameters: {
+            type: :object,
+            properties: {
+              contents_to_memory: {
+                type: :string,
+                description: I18n.t("ghostest.functions.#{self.function_name}.parameters.contents_to_memory"),
+              },
+            },
+            required: [:contents_to_memory],
+          },
+        }
+        @definition
+      end
+
+      def execute_and_generate_message(args)
+        if args[:contents_to_memory].nil? || args[:contents_to_memory].empty?
+          raise "contents_to_memory is required"
+        end
+        @message_container.add_system_message(I18n.t("ghostest.functions.#{self.function_name}.system_message_prefix", contents_to_memory: args[:contents_to_memory]))
+        { result: 'success' }
+      end
+    end
+  end
+end

data/lib/llm/functions/base.rb

@@ -0,0 +1,13 @@
+module Llm
+  module Functions
+    class Base
+      def function_name
+        nil
+      end
+
+      def stop_llm_call?
+        false
+      end
+    end
+  end
+end

data/lib/llm/functions/exec_rspec_test.rb

@@ -0,0 +1,39 @@
+module Llm
+  module Functions
+    class ExecRspecTest < Base
+      def function_name
+        :exec_rspec_test
+      end
+
+      def definition
+        return @definition unless @definition.nil?
+
+        @definition = {
+          name: self.function_name,
+          description: I18n.t("ghostest.functions.#{self.function_name}.description"),
+          parameters: {
+            type: :object,
+            properties: {
+              file_or_dir_path: {
+                type: :string,
+                description: I18n.t("ghostest.functions.#{self.function_name}.parameters.file_or_dir_path"),
+              },
+            },
+            required: [:file_or_dir_path],
+          },
+        }
+        @definition
+      end
+
+      def execute_and_generate_message(args)
+        if args[:file_or_dir_path].nil? || args[:file_or_dir_path].empty?
+          raise Ghostest::Error.new("Please specify the file or directory path.")
+        end
+        script = "bundle exec rspec '#{args['file_or_dir_path']}'"
+        stdout, stderr, status = Open3.capture3(script)
+
+        { stdout:, stderr:, exit_status: status.exitstatus }
+      end
+    end
+  end
+end

data/lib/llm/functions/fix_one_rspec_test.rb

@@ -0,0 +1,55 @@
+module Llm
+  module Functions
+    class FixOneRspecTest < Base
+      def function_name
+        :fix_one_rspec_test
+      end
+
+      def definition
+        return @definition unless @definition.nil?
+
+        @definition = {
+          name: self.function_name,
+          description: I18n.t("ghostest.functions.#{self.function_name}.description"),
+          parameters: {
+            type: :object,
+            properties: {
+              file_path: {
+                type: :string,
+                description: I18n.t("ghostest.functions.#{self.function_name}.parameters.file_path"),
+              },
+              line_num: {
+                type: :string,
+                description: I18n.t("ghostest.functions.#{self.function_name}.parameters.line_num"),
+              },
+            },
+            required: [:file_path, :line_num],
+          },
+        }
+        @definition
+      end
+
+      def execute_and_generate_message(args)
+        if args[:file_path].nil? || args[:file_path].empty? || !File.exist?(args[:file_path])
+          raise Ghostest::Error.new("Please specify the file path.")
+        end
+        line_num = args[:line_num].to_i
+        if line_num < 1
+          raise Ghostest::Error.new("Please specify the line num. #{args[:line_num]}")
+        end
+
+        n = 0
+        while n < 5
+          n += 1
+          script = "bundle exec rspec '#{args['file_path']}:#{args['line_num']}'"
+          stdout, stderr, status = Open3.capture3(script)
+          if status.exitstatus != 0
+
+          end
+        end
+
+        { stdout:, stderr:, exit_status: status.exitstatus }
+      end
+    end
+  end
+end

data/lib/llm/functions/get_files_list.rb

@@ -0,0 +1,29 @@
+module Llm
+  module Functions
+    class GetFilesList < Base
+      def function_name
+        :get_files_list
+      end
+
+      def definition
+        return @definition unless @definition.nil?
+
+        @definition = {
+          name: self.function_name,
+          description: I18n.t("ghostest.functions.#{self.function_name}.description"),
+          parameters: {
+            type: :object,
+            properties: {},
+          },
+        }
+        @definition
+      end
+
+      def execute_and_generate_message(args)
+        files_list = Dir.glob("**/*.{rb,yml}")
+
+        { files_list: }
+      end
+    end
+  end
+end