🥘 Recipe Chat Assistant

A modern, chat‑style desktop application that lets you converse naturally with an AI about cooking, recipes, and food preparation. Built with Python3.8+ and PyQt5, it supports both OpenAI and LM Studio models and renders beautifully parsed, interactive answers.

---

✨ Key Features

Category	Highlights
Conversational UI	Sleek chat interface with message bubbles, typing indicators, and rich recipe widgets
Dual Model Support	Seamlessly switch between OpenAI LMStudio — or type any custom LM Studio model name
Structured JSON Replies	Prompts request recipes in JSON; the parser extracts & validates even if the LLM adds extra commentary
Smart Parsing Engine	Handles wildly formatted ingredient lines, creates tags automatically, and falls back to plain‑text when JSON is missing
Theme System	Centralised AppTheme guarantees a consistent dark/light look across all widgets
Threaded API Calls	Background threads keep the UI responsive while the LLM thinks
Robust Error Handling	Connection checks, detailed pop‑ups, and graceful fallbacks throughout the stack

Module Breakdown

File	Responsibility
models.py	• APIConfig, ParsedRecipe, and other data models • Central AppTheme (fonts, colours, spacing) • Enums / constants for message roles, cuisines, etc.
parser.py	• Extracts JSON embedded in text • Validates & converts to ParsedRecipe • Accepts messy inputs ("2 Tbsp olive‑oil", "½ cup milk") • Auto‑generates keyword tags
api_client.py	• One class, two engines: OpenAI/LM Studio • Adds system prompts requesting strict JSON • Timeout & retry logic, HTTP/WS support • verify_connection() utility
widgets.py	• ModernCard base with rounded corners & shadow • Recipe widgets: ingredients grid, step‑by‑step view, chef‑tips • Chat components: user/assistant bubbles, loading dots, error boxes • SettingsPanel with free‑text model field
main.py	• Builds the window & navbar • Manages chat history & scrollbar • Spawns worker threads for each API call • Handles drag‑drop of ingredient lists

---

🚀 Quick Start

# 1. Clone the repository
$ git clone https://github.com/Roialfassi/Recipe-Chat-Assistant-Application.git
$ cd recipe‑chat‑assistant

# 2. Create & activate a virtual environment
$ python -m venv .venv && source .venv/bin/activate  # Windows: .venv\Scripts\activate

# 3. Install dependencies
$ pip install -r requirements.txt

# 4. Run the app
$ python main.py

# add the api key the model and provider you want to use and start getting recipes

Tip: In SettingsModels you can pick gpt‑4o, phi‑3‑mini‑128k, or type any custom LM Studio model identifier.

The UI’s Settings Panel updates these values live and writes back to the file.

---

🤖 Prompt Format

The assistant requests recipes in a strict JSON envelope to simplify parsing:

{
  "title": "Spaghetti Carbonara",
  "servings": 4,
  "ingredients": [
    "200 g spaghetti",
    "100 g pancetta, diced",
    "2 eggs",
    "50 g grated parmesan",
    "1 Tbsp olive oil",
    "Salt & freshly ground pepper"
  ],
  "instructions": [
    "Cook spaghetti until al dente.",
    "Fry pancetta until crispy.",
    "Whisk eggs and parmesan together.",
    "Combine drained pasta with pancetta, remove heat, stir in egg mixture.",
    "Season and serve immediately."
  ],
  "tags": ["italian", "quick", "comfort‑food"]
}

The parser falls back to free‑text rendering if valid JSON is absent.

---

🛣️ Roadmap

• Image support (inline step photos)
• Voice input & TTS cooking mode
• Cloud‑sync of favourite recipes
• Plugin system for grocery‑list export

---

👫 Contributing

1. Fork the repository and create a branch (feat/awesome‑thing).
2. Ensure code passes ruff / black style checks: make format.
3. Submit a pull request – describe what your change improves.

All contributors agree to follow the Contributor Covenant

---

📜 License

Distributed under the MIT License. See LICENSE for details.

---

🙏 Acknowledgements

• PyQt5 for the GUI framework.
• OpenAI Python SDK for API access.
• LM Studio for local model serving.

Bon appétit