startup menu for ai_settings.yaml to support multiple configurations

[bassie661]

User description

Summary

I created this PR to add support for storing multiple configurations, thus allowing the user to 'create', 'edit', 'view' or 'delete' configurations and have their settings stored in ai_settings.yaml. The ai_settings file now saves configurations as dictionary and if a configuration exists it is showed as a menu option when autogpt is started.

When no configurations are found, or no ai_settings.yaml file, the user get's the usual generate_aiconfig_automatic questions and can choose to either let autogpt create the config from a single input or the user can choose "--manual" to manually input the first configuration. At the end of this process the first configuration is saved in the ai_settings file.

Next time when autogpt starts and/or when configurations are found in ai_settings.yaml, the prompt starts-up with the following menu:

Change

• Support storing multiple AI configurations
• Support choose existing AI configuration(s)
• Support create new AI configuration(s)
• Support edit existing AI configuration(s)
• Support view existing AI configuration(s)
• Support delete existing AI configuration(s)
  -> e.g. ai_name, ai_role, ai_goals[], api_budget, plugins
• Support for plugins per AI configuration (add, delete, change), plugins are expected to land with add plugins list to aiconfig #4089

This PR is big but it was needed to restructure the old and add the new functionality to the '--manual' startup flow while keeping the functionality clear and maintainable.

• Initial plan was to also add support for setting temperature (ai_temp) per AI configuration (add, delete, change) but this clearly needs extra changes, so a separate PR for ai_temp support is needed.
• This is a new approach which rendered Feature/add .env fixed ai_name, ai_role and or extra goals #3711 and new UI for ai_settings yaml to support multiple configurations #4478 obsolete.
• I had to rewrite a lot of parts to keep up with the commits and release of v0.4.0 / v0.4.1 and v0.4.2
• added support for Pass config everywhere in order to get rid of singleton #4666

Test Plan

• integration\test_setup.py
• unit\test_ai_config.py
• unit\test_config.py
• unit\test_prompt.py

PR Quality Checklist

[x] My pull request is atomic and focuses on a single change.
-> mostly true but reworking the old startup prompts was quite complex which inevitably needed more than a single change.

[x] I have thoroughly tested my changes with multiple different prompts.
[x] I have considered potential risks and mitigations for my changes.
[x] I have documented my changes clearly and comprehensively.
[x] I have not snuck in any "extra" small tweaks changes.
[x] I have run the following commands against my code to ensure it passes our linters: black . / isort . / mypy
autoflake --remove-all-unused-imports --recursive --ignore-init-module-imports autogpt tests --in-place

---

PR Type

Enhancement, Tests

---

Description

• Added support for storing and managing multiple AI configurations, including plugins.
• Refactored prompt handling to introduce a new interactive menu for configuration management.
• Integrated automatic configuration generation with the new menu system.
• Added comprehensive unit and integration tests for the new features.

---

Changes walkthrough 📝

	Relevant files
Enhancement	5 files ai_config.py Support for multiple AI configurations with plugins            autogpt/config/ai_config.py Added support for plugins in AI configurations. Introduced methods for loading, saving, and deleting multiple AI configurations. Updated configuration format to support multiple configurations. +234/-41 config.py Added plugin management and ai_settings_filepath                  autogpt/config/config.py Added methods to set plugins allowlist and denylist. Introduced ai_settings_filepath attribute. +13/-1    main.py Main entry point modifications for configuration management autogpt/main.py Modified main entry point to support new configuration management. Added start_agent_directly function for direct agent start from menu. +59/-4    prompt.py Refactored prompt handling and configuration management    autogpt/prompts/prompt.py Refactored prompt handling to support new configuration menu. Added functions for managing AI configurations interactively. +811/-44 setup.py Integrated automatic configuration with new menu system    autogpt/setup.py Removed manual configuration setup. Integrated automatic configuration generation with new menu system. +40/-162
Tests	4 files test_setup.py Integration tests for configuration management                      tests/integration/test_setup.py Added integration tests for new configuration management features. Tested various scenarios including creation, deletion, and editing of configurations. +765/-39 test_ai_config.py Unit tests for AIConfig class                                                        tests/unit/test_ai_config.py Added unit tests for AIConfig class methods. Tested loading, saving, and deleting configurations. +421/-45 test_config.py Unit tests for plugin management in Config class                  tests/unit/test_config.py Added unit tests for new plugin management methods in Config class. +30/-0    test_prompt.py Unit tests for prompt handling functions                                  tests/unit/test_prompt.py Added unit tests for prompt handling functions. Tested validation and management of AI names. +162/-0

---

💡 PR-Agent usage:
Comment /help on the PR to get a list of all available PR-Agent tools and their descriptions

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

1 Ignored Deployment

Name	Status	Preview	Comments	Updated (UTC)
docs	⬜️ Ignored (Inspect)			Jun 12, 2023 5:16pm

[github-actions]

This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR.

[github-actions]

This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR.

[github-actions]

This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR.

[github-actions]

This pull request has conflicts with the base branch, please resolve those so we can evaluate the pull request.

[github-actions]

This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR.

[github-actions]

Conflicts have been resolved! 🎉 A maintainer will review the pull request shortly.

[github-actions]

This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR.

[github-actions]

This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR.

[github-actions]

This PR exceeds the recommended size of 500 lines. Please make sure you are NOT addressing multiple issues with one PR.

[github-actions]

This pull request has conflicts with the base branch, please resolve those so we can evaluate the pull request.

[bassie661]

same, thanks for nothing

[Swiftyos]

@bassie661 the PR has had no activity since June 2023 and looked stale. Apologies, I've reopened.

Have you tested this with the latest version of AutoGPT?

I'll close again if there has been no activity in 2 weeks.

[qodo-merge-pro]

PR Description updated to latest commit (b077bce)

• Copy walkthrough table to "Files Changed" Tab

[qodo-merge-pro]

PR Reviewer Guide 🔍

⏱️ Estimated effort to review [1-5]	5
🧪 Relevant tests	No
🔒 Security concerns	No
⚡ Key issues to review	Possible Bug: The AIConfig class has a method to_dict where the comment mentions that plugins are a list of objectives, which seems incorrect. This should be a list of plugins or extensions, not objectives.
	Code Duplication: The methods load and load_all in AIConfig class have overlapping functionality with slight variations for loading configurations. Consider refactoring to combine these methods to reduce redundancy and improve maintainability.
	Error Handling: In the save method of AIConfig, there is a potential unhandled exception if the YAML file becomes corrupted or has invalid format. This should be handled gracefully to inform the user of the issue without crashing the application.
	Performance Concern: The update_old_config method in AIConfig reads the entire configuration file into memory, which might not be efficient for very large configuration files. Consider streaming the file or processing it in chunks.

[qodo-merge-pro]

PR Code Suggestions ✨

Category	Suggestion	Score
Best practice	Use context managers for file operations to ensure proper file handling Use a context manager for file operations to ensure the file is properly closed after writing, which is more reliable and Pythonic. tests/integration/test_setup.py [115-116] +with open(temp_config_file, "w") as temp_file: + temp_file.write(config_content) - Apply this suggestion Suggestion importance[1-10]: 10 Why: The suggestion to use a context manager for file operations is correct and follows best practices in Python. It ensures that the file is properly closed after writing, which is more reliable and Pythonic.	10
	Use the clean_input utility for consistent input handling Replace the direct use of the input function with the clean_input utility function from the autogpt.utils module to ensure consistent input handling and sanitization across the application. autogpt/prompts/prompt.py [186] -user_input = input(prompt_text).strip() +user_input = utils.clean_input(config, prompt_text).strip() Apply this suggestion Suggestion importance[1-10]: 9 Why: This suggestion improves input sanitization and consistency across the application by using the existing clean_input utility function. This is a best practice for maintaining code consistency and security.	9
	Use a context manager for file operations to ensure proper resource management Consider using a context manager for file operations to ensure that resources are properly managed and files are closed after operations are completed. This can prevent potential file locking or resource leakage issues. tests/unit/test_ai_config.py [46-47] -config_file = config.ai_settings_filepath -config_file.write_text(old_yaml_content) +with open(config.ai_settings_filepath, 'w') as config_file: + config_file.write(old_yaml_content) Apply this suggestion Suggestion importance[1-10]: 8 Why: Using a context manager for file operations is a best practice that ensures files are properly closed, preventing potential file locking or resource leakage issues. This is a significant improvement for resource management.	8
Maintainability	Use a pytest fixture to manage the configuration file path across tests To avoid hardcoding the file path in multiple test cases, consider using a fixture to manage the configuration file path. This can make the tests easier to maintain and modify. tests/unit/test_ai_config.py [46] -config_file = config.ai_settings_filepath +@pytest.fixture +def config_file_path(tmp_path): + return tmp_path / "ai_settings.yaml" +def test_some_function(config_file_path): + # Use config_file_path in your test + Apply this suggestion Suggestion importance[1-10]: 9 Why: Using a pytest fixture to manage the configuration file path improves maintainability by avoiding hardcoding the file path in multiple test cases. This makes the tests easier to maintain and modify, which is highly beneficial for long-term code maintenance.	9
Possible bug	Add missing import for the List type from the typing module The plugins parameter in the AIConfig class constructor is defined as Optional[List], but it is not imported from the typing module. This might lead to a runtime error if the type is not recognized. It is recommended to import List from the typing module explicitly. autogpt/config/ai_config.py [37] +from typing import List plugins: Optional[List] = None, Apply this suggestion Suggestion importance[1-10]: 9 Why: The suggestion correctly identifies a missing import for the List type, which could lead to runtime errors. This is a crucial fix to ensure the code runs without issues.	9
Possible issue	Ensure num_goals does not exceed the maximum limit Add a check to ensure that the num_goals variable does not exceed a predefined maximum limit to prevent potential misuse or errors in goal setting. autogpt/prompts/prompt.py [146] num_goals = int(num_goals_input) +if num_goals > MAX_GOALS: # Assuming MAX_GOALS is defined elsewhere as the maximum allowed goals + logger.typewriter_log("Number of goals exceeds the maximum limit.", Fore.RED) + continue Apply this suggestion Suggestion importance[1-10]: 8 Why: Adding a check to ensure num_goals does not exceed a predefined maximum limit is crucial for preventing potential misuse or errors in goal setting. This enhances the robustness of the code.	8
Enhancement	Simplify mocking user inputs in tests using MagicMock Instead of manually iterating through user_inputs with next(), consider using unittest.mock.MagicMock to automatically handle input sequences. This simplifies the test setup and makes it more readable. tests/integration/test_setup.py [41-44] -user_inputs_iterator = iter(user_inputs) -def mock_input(_prompt: str) -> str: - return next(user_inputs_iterator) +mock_input = MagicMock(side_effect=user_inputs) Apply this suggestion Suggestion importance[1-10]: 8 Why: Using MagicMock to handle input sequences simplifies the test setup and makes the code more readable. This is a good enhancement for the test code, although the existing approach is also functional.	8
	Correct the description of the plugins attribute in the AIConfig class The description for the plugins attribute in the AIConfig class is incorrect. It is currently duplicated as "The list of objectives the AI is supposed to complete," which is the same as for ai_goals. This should be corrected to reflect the actual purpose of the plugins attribute. autogpt/config/ai_config.py [28] -plugins (list): The list of objectives the AI is supposed to complete. +plugins (list): The list of plugins configured for the AI. Apply this suggestion Suggestion importance[1-10]: 8 Why: The suggestion accurately points out a misleading description for the plugins attribute. Correcting this improves code documentation and clarity.	8
	Ensure the AI name is not an empty string in the delete method The delete method in the AIConfig class raises a ValueError if the AI name is not provided, but it does not check if the AI name is a non-empty string. It is recommended to add a check to ensure that the AI name is not just an empty string. autogpt/config/ai_config.py [282-285] -if ai_name == "": +if not ai_name.strip(): raise ValueError( - "No AI name provided. Please provide an AI name to delete its configuration." + "No AI name provided or AI name is empty. Please provide a valid AI name to delete its configuration." ) Apply this suggestion Suggestion importance[1-10]: 8 Why: The suggestion enhances the robustness of the delete method by ensuring the AI name is not just an empty string, which is important for preventing potential errors.	8

[Swiftyos]

@bassie661 are you planning on picking this up again?

[bassie661]

Nope sorry. Greets Rob

…

________________________________ Van: Swifty ***@***.***> Verzonden: vrijdag 5 juli 2024 10:21 Aan: Significant-Gravitas/AutoGPT ***@***.***> CC: Rob Jonas ***@***.***>; Mention ***@***.***> Onderwerp: Re: [Significant-Gravitas/AutoGPT] startup menu for ai_settings.yaml to support multiple configurations (PR #4646) @bassie661<https://github.com/bassie661> are you planning on picking this up again? — Reply to this email directly, view it on GitHub<#4646 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AQLKIYGPECZPNTTU7JSTUODZKZJQLAVCNFSM6AAAAABKBWDWOOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEMJQGQZDGMRSGI>. You are receiving this because you were mentioned.Message ID: ***@***.***>

[Swiftyos]

Okay no worries, closing now then.