[Root ↑]

src \ endpoints \ hypo69

Русский

Вот обновленный текст файла README.md для модуля code_assistant, который описывает все скрипты (code_assistant.py, make_summary.py, onela_bot.py, bot_handlers.py):

---

.. module:: src.endpoints.hypo69.code_assistant

[Root ↑]

src \ endpoints \ hypo69

English

Code Assistant Module

Overview

The Code Assistant module is a set of tools designed to interact with Gemini and OpenAI models for processing project source code. It performs tasks such as generating documentation, code review, and test generation based on specified files. Additionally, it includes scripts for creating a SUMMARY.md file for documentation compilation and a Telegram bot for handling code-related tasks.

Key Features

code_assistant.py

• File Reading: Reads code from .py and README.MD files in specified directories.
• Model Interaction: Sends code to models for tasks like documentation generation or error checking.
• Result Generation: Saves model responses in designated directories for each role.

make_summary.py

• SUMMARY.md Generation: Recursively traverses a directory to create a SUMMARY.md file for documentation compilation.
• Language Filtering: Supports filtering files by language (ru or en).

onela_bot.py and bot_handlers.py

• Telegram Bot: A bot for handling code-related tasks, such as sending code snippets for review or generating documentation.
• Bot Handlers: Contains handlers for processing bot commands and messages.

Project Structure

• Models: Uses Gemini and OpenAI models for processing requests.
• Prompts: Reads prompts from files in src/ai/prompts/developer/ (e.g., doc_writer_en.md).
• Files: Processes .py and README.MD files in specified directories.

Usage Examples

Running with JSON Settings:

python code_assistant.py --settings settings.json

Running with Explicit Parameters:

python code_assistant.py --role doc_writer --lang ru --models gemini openai --start_dirs /path/to/dir1 /path/to/dir2

Example for Role code_checker:

python code_assistant.py --role code_checker --lang en --models gemini --start_dirs /path/to/dir

Example for Model openai:

python code_assistant.py --role doc_writer --lang en --models openai

Command-Line Parameters

• --settings: Path to a JSON file with settings. Loads parameters from the file.
• --role: The role of the model for the task (e.g., doc_writer, code_checker).
• --lang: The language for the task (e.g., ru or en).
• --models: List of models to initialize (e.g., gemini, openai).
• --start_dirs: List of directories to process (e.g., /path/to/dir1).

Workflow

1. File Reading: Searches for .py and README.MD files in specified directories.
2. Prompt Loading: Loads role-specific prompts from src/ai/prompts/developer/.
3. Request Processing: Forms requests based on loaded files and sends them to models.
4. Response Saving: Saves model responses in directories corresponding to the role and model (e.g., docs/raw_rst_from_<model>/<lang>/).

Exceptions

Configure exceptions for files and directories using parameters:

• exclude_file_patterns: List of regex patterns to exclude files.
• exclude_dirs: List of directories to exclude.
• exclude_files: List of files to exclude.

Logging

Logs are saved using the logger library and contain information about the file processing and model responses.

Dependencies

• Gemini API: Requires an API key for the Gemini model.
• OpenAI API: Requires an API key for the OpenAI model.

Creating a New Role for AI Models

1. Update code_assistant.json:

   • Add the new role to the list of roles:

     "roles": [
       "code_checker",
       ...
     ]

   • Alternatively, exclude it in "exclude-roles".

2. Add Role to Translations:

   • Update the translations/translations.json file with the new role.

3. Create a System Prompt:

   • Add a new system prompt in ai/prompts/developer/.

4. Create a Command Instruction:

   • Add a new command instruction in instructions/.

---

This README.md provides a comprehensive overview of the code_assistant module, its scripts, and how to use them effectively.