mad-professor

The project is organized into several functional modules:

• Core Modules (核心模块/): Contains the main logic including AI_manager.py (AI orchestration), AI_professor_chat.py (chat logic), rag_retriever.py (RAG system), and TTS_manager.py (voice synthesis).
• UI Components (ui/): PyQt6 components like chat_widget.py, markdown_view.py, and sidebar_widget.py.
• Processors (processor/): Handles the data pipeline, including pdf_processor.py (PDF to MD), translate_processor.py (translation), and rag_processor.py (vector indexing).
• Prompts (prompt/): Contains text files defining AI personalities and task-specific instructions (e.g., ai_explain_prompt.txt).
• Data & Output: data/ stores source PDFs, and output/ stores processed results.

Launch the application

Run the following command from the project root:

python main.py

Importing Papers

1. Click the "导入论文" (Import Paper) button in the sidebar.
2. Select a PDF file.
3. Click "继续" (Continue) and wait for processing (translation and indexing).
4. Note: PDFs are stored in the data/ folder. Adding multiple PDFs to this folder allows for batch processing.

Reading and Interaction

• Reading: Select a processed paper from the sidebar. Use the toggle in the top-right corner to switch between Chinese and English views.
• Voice Interaction: Select your input device in the chat window. Click the microphone button; wait for the indicator to turn green before speaking.
• Tip: Use headphones to prevent the AI's voice from being re-recorded as user input.

Enable CUDA

To enable CUDA support, locate the magic-pdf.json file in your user directory and set device-mode to "cuda":

{
    "device-mode":"cuda"
}

Configure LLM and TTS APIs

Edit config.py to provide your API credentials for DeepSeek (LLM) and MiniMax (TTS):

# DeepSeek Configuration
API_BASE_URL = "YOUR_API_URL"
API_KEY = "YOUR_API_KEY"   

# MiniMax Configuration
TTS_GROUP_ID = "YOUR_MINIMAX_GROUP_ID"
TTS_API_KEY = "YOUR_MINIMAX_API_KEY"

{
    "device-mode":"cuda"
}

# In config.py
API_BASE_URL = "YOUR_API_URL"
API_KEY = "YOUR_API_KEY"   
TTS_GROUP_ID = "YOUR_MINIMAX_GROUP_ID"
TTS_API_KEY = "YOUR_MINIMAX_API_KEY"

The application is a PyQt6-based desktop application. To start the application, run the main.py script. The entrypoint initializes the QApplication with the 'Fusion' style, sets up custom fonts (Source Han Sans and Source Han Serif), configures a modern color palette, and launches the AIProfessorUI window.

Note: On Windows, the application sets a specific AppUserModelID (ai.professor.paperassistant.1.0) to ensure the taskbar icon is handled correctly.

Use katex.render to render a TeX expression directly into a specific DOM element.

Note: Ensure you have included the KaTeX CSS and font files in your project. You can use the throwOnError: false option to render invalid TeX as the original source code in red with the error message provided as hover text.

katex.render("c = \\pm\\sqrt{a^2 + b^2}", element, {
    throwOnError: false
});

Use katex.renderToString to generate an HTML string of the rendered math. This is primarily used for server-side rendering (SSR) where you want to pre-render expressions using Node.js and send them as plain HTML to the client.

Note: If you are performing all rendering on the server, you do not need to include the KaTeX JavaScript on the client side, but you must still include the CSS.

var html = katex.renderToString("c = \\pm\\sqrt{a^2 + b^2}", {
    throwOnError: false
});
// '<span class="katex">...</span>'

The generate_app_icon() function programmatically creates a QIcon using QPainter. The icon features a circular gradient background (deep blue/indigo), a book pattern with white pages and spine lines, and a simplified "AI" text overlay in the center.

from main import generate_app_icon
from PyQt6.QtWidgets import QApplication
import sys

app = QApplication(sys.argv)
app_icon = generate_app_icon()
app.setWindowIcon(app_icon)