RepIt

RepIt is an experimental AI “class representative” that uses Google Gemini tool-calling to answer questions and trigger helper functions like class scheduling, canceling, polling, and quiz creation. The codebase currently contains a simple CLI loop plus several service stubs for future integrations (MongoDB, PostgreSQL, Excel-based schedules).

Quick start

• Python 3.11+ recommended.
• Create and activate a virtual environment, then install the deps (rough set):

   pip install google-genai google-auth-httplib2 google-api-python-client pymongo psycopg2-binary python-dotenv pandas openpyxl

• Set your Gemini API key (the repo currently contains placeholder keys that should be removed):

   set GEMINI_API_KEY=your-key-here   # Windows PowerShell: $env:GEMINI_API_KEY="your-key-here"

• Run the CLI demo (note: core/ai_client.py still needs a GeminiClient implementation for this to work):

   python main.py

Project layout

• main.py – CLI loop that prompts the user and calls RepItAgent.
• core/agents.py – RepItAgent orchestrates prompt construction and routes Gemini function calls vs. text.
• core/tools.py – Tool schema definitions (CheckUser, CancelClass, SceduleClass, ShowSchedule, ShowClassForToday, MakeQuiz, MakePoll).
• core/ai_client.py – Placeholder Gemini client script; missing the GeminiClient class expected by main.py.
• services/scedule_service.py – String-format stubs for scheduling/canceling/showing classes.
• services/poll_service.py – Empty stub for poll creation.
• rule.py – Hard-coded role lookup for Admin/Teacher/Student.
• db.py – MongoDB helper stub (hard-coded URI; read not yet implemented).
• supa.py – Example PostgreSQL connection using .env variables.
• scedule.py – Reads Book1.xlsx into a DataFrame; prints schedule data.
• ai.py and agent.py – Earlier prototypes of the Gemini interaction loop.

How it works (intended)

1. main.py builds a context with today’s date/day and instantiates RepItAgent with an AI client plus the tool schema.
2. User input is sent to Gemini with tool definitions; responses are split into plain text vs. function calls.
3. Function calls are currently just printed; wiring to real services (e.g., services/ functions or DB calls) is pending.

Configuration

• GEMINI_API_KEY must be set for any Gemini calls. Remove hard-coded keys from the codebase before deploying.
• Database examples:
  • Mongo: replace the connection string in db.py with MONGO_URI from env, and implement read/queries.
  • Postgres: supply user, password, host, port, dbname in a .env file for supa.py.
• If you rely on Excel schedules, keep Book1.xlsx in the project root for scedule.py.

Caveats / TODO

• Implement GeminiClient in core/ai_client.py and replace hard-coded API keys with env-based config.
• Wire tool calls to services/ implementations and expand them beyond string responses.
• Add error handling, logging, and tests.
• Add a requirements.txt or pyproject.toml to lock dependencies.