Merge pull request #111 from Pythagora-io/fix/handle-step_check.prompt

handle step check.prompt

pilot/helpers/Project.py

@@ -4,6 +4,7 @@ from utils.style import green_bold, yellow_bold, cyan, white_bold
 from const.common import IGNORE_FOLDERS, STEPS
 from database.database import delete_unconnected_steps_from, delete_all_app_development_data
 from const.ipc import MESSAGE_TYPE
+from prompts.prompts import ask_user
 from helpers.exceptions.TokenLimitError import TokenLimitError
 from utils.questionary import styled_text
 from helpers.files import get_files_content, clear_directory, update_file
@@ -302,16 +303,14 @@ class Project:
             reset_branch_id = convo.save_branch()
 
         while answer != 'continue':
-            print(yellow_bold(message))
             if description is not None:
                 print('\n' + '-'*100 + '\n' +
                     white_bold(description) +
                     '\n' + '-'*100 + '\n')
 
-            answer = styled_text(
-                self,
-                'If something is wrong, tell me or type "continue" to continue.',
-            )
+            answer = ask_user(self, yellow_bold(message),
+                              require_some_input=False,
+                              hint='If something is wrong, tell me or type "continue" to continue.')
 
             try:
                 if answer in cbs:

pilot/helpers/agents/Developer.py

@@ -382,26 +382,23 @@ class Developer(Agent):
         return llm_response
 
     def test_code_changes(self, code_monkey, convo):
-        (test_type, command, automated_test_description, manual_test_description) = convo.send_message(
-            'development/task/step_check.prompt',
-            {},
-            GET_TEST_TYPE)
+        test_type, description = convo.send_message('development/task/step_check.prompt', {}, GET_TEST_TYPE)
 
         if test_type == 'command_test':
-            return run_command_until_success(command['command'], command['timeout'], convo)
+            return run_command_until_success(description['command'], description['timeout'], convo)
         elif test_type == 'automated_test':
             # TODO get code monkey to implement the automated test
             pass
         elif test_type == 'manual_test':
             # TODO make the message better
             response = self.project.ask_for_human_intervention(
-                'Message from Pilot: I need your help. Can you please test if this was successful?',
-                manual_test_description
+                'I need your help. Can you please test if this was successful?',
+                description,
             )
 
             user_feedback = response['user_input']
             if user_feedback is not None and user_feedback != 'continue':
-                return_value = self.debugger.debug(convo, user_input=user_feedback, issue_description=manual_test_description)
+                return_value = self.debugger.debug(convo, user_input=user_feedback, issue_description=description)
                 return_value['user_input'] = user_feedback
                 return return_value
             else:

pilot/helpers/agents/test_Developer.py

@@ -51,3 +51,46 @@ class TestDeveloper:
         # Then
         assert llm_response == 'DONE'
         mock_execute_command.assert_called_once_with(self.project, 'python --version', 10)
+
+    @patch('helpers.AgentConvo.get_saved_development_step')
+    @patch('helpers.AgentConvo.save_development_step')
+    # GET_TEST_TYPE has optional properties, so we need to be able to handle missing args.
+    @patch('helpers.AgentConvo.create_gpt_chat_completion',
+           return_value={'text': '{"type": "command_test", "command": {"command": "npm run test", "timeout": 3000}}'})
+    # 2nd arg of return_value: `None` to debug, 'DONE' if successful
+    @patch('helpers.cli.execute_command', return_value=('stdout:\n```\n\n```', 'DONE'))
+    # @patch('helpers.cli.ask_user', return_value='yes')
+    # @patch('helpers.cli.get_saved_command_run')
+    def test_code_changes_command_test(self, mock_get_saved_step, mock_save, mock_chat_completion,
+                               # Note: the 2nd line below will use the LLM to debug, uncomment the @patches accordingly
+                               mock_execute_command):
+                               # mock_ask_user, mock_get_saved_command_run):
+        # Given
+        monkey = None
+        convo = AgentConvo(self.developer)
+        convo.save_branch = lambda branch_name=None: branch_name
+
+        # When
+        # "Now, we need to verify if this change was successfully implemented...
+        result = self.developer.test_code_changes(monkey, convo)
+
+        # Then
+        assert result == {'success': True, 'cli_response': 'stdout:\n```\n\n```'}
+
+    @patch('helpers.AgentConvo.get_saved_development_step')
+    @patch('helpers.AgentConvo.save_development_step')
+    # GET_TEST_TYPE has optional properties, so we need to be able to handle missing args.
+    @patch('helpers.AgentConvo.create_gpt_chat_completion',
+           return_value={'text': '{"type": "manual_test", "manual_test_description": "Does it look good?"}'})
+    @patch('helpers.Project.ask_user', return_value='continue')
+    def test_code_changes_manual_test(self, mock_get_saved_step, mock_save, mock_chat_completion, mock_ask_user):
+        # Given
+        monkey = None
+        convo = AgentConvo(self.developer)
+        convo.save_branch = lambda branch_name=None: branch_name
+
+        # When
+        result = self.developer.test_code_changes(monkey, convo)
+
+        # Then
+        assert result == {'success': True, 'user_input': 'continue'}