Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/ayushpai/AI-Math-Notes/llms.txt

Use this file to discover all available pages before exploring further.

​
Overview

AI Math Notes is built around a single DrawingApp class that manages the Tkinter-based canvas interface, PIL image processing, and OpenAI API integration. The architecture follows a monolithic design pattern with event-driven user interactions.

​
DrawingApp Class

The DrawingApp class (main.py:8) is the core component that handles all application functionality.

​
Initialization

​
root
tk.Tk
The Tkinter root window instance
class DrawingApp:
    def __init__(self, root):
        self.root = root
        self.root.title("AI Math Notes")

        self.canvas_width = 1200
        self.canvas_height = 800

        self.canvas = tk.Canvas(root, bg='black', width=self.canvas_width, height=self.canvas_height)
        self.canvas.pack()

        self.image = Image.new("RGB", (self.canvas_width, self.canvas_height), (0, 0, 0))
        self.draw = ImageDraw.Draw(self.image)

        self.client = OpenAI()
The constructor initializes:
  • A 1200x800 black canvas for drawing
  • A PIL Image instance for image processing
  • An ImageDraw object for rendering
  • Event bindings for mouse and keyboard interactions
  • Action history for undo functionality
  • OpenAI client for API calls

​
Canvas System

The application uses a dual-rendering approach:
  1. Tkinter Canvas (main.py:16) - Provides real-time visual feedback
  2. PIL Image (main.py:19-20) - Creates the actual image sent to the API
This architecture ensures that user drawings are both displayed interactively and captured for image analysis.

​
Event Bindings

The application binds several events (main.py:22-26): 
self.canvas.bind("<Button-1>", self.start_draw)
self.canvas.bind("<B1-Motion>", self.paint)
self.canvas.bind("<ButtonRelease-1>", self.reset)
self.root.bind("<Command-z>", self.command_undo)
self.root.bind("<Return>", self.command_calculate)

​
Core Methods

​
Drawing Methods

​
start_draw()

​
event
tk.Event
Mouse click event containing x, y coordinates
Initializes a new drawing action when the mouse button is pressed (main.py:45-47). 
def start_draw(self, event):
    self.current_action = []
    self.last_x, self.last_y = event.x, event.y

​
paint()

​
event
tk.Event
Mouse motion event with current cursor position
Draws lines on both the Tkinter canvas and PIL image as the mouse moves (main.py:49-55). 
def paint(self, event):
    x, y = event.x, event.y
    if self.last_x and self.last_y:
        line_id = self.canvas.create_line((self.last_x, self.last_y, x, y), fill='white', width=5)
        self.draw.line((self.last_x, self.last_y, x, y), fill='white', width=5)
        self.current_action.append((line_id, (self.last_x, self.last_y, x, y)))
    self.last_x, self.last_y = x, y
Each stroke is stored with its line ID and coordinates for undo functionality.

​
reset()

​
event
tk.Event
Mouse release event
Completes the current drawing action and adds it to the action history (main.py:57-60).

​
Action Management

​
undo()

Reverts the last drawing action by removing lines from both canvas and image (main.py:68-73). 
def undo(self):
    if self.actions:
        last_action = self.actions.pop()
        for line_id, coords in last_action:
            self.canvas.delete(line_id)
        self.redraw_all()

​
redraw_all()

Recreates the entire canvas and PIL image from the action history (main.py:78-85). 
def redraw_all(self):
    self.image = Image.new("RGB", (self.canvas_width, self.canvas_height), (0, 0, 0))
    self.draw = ImageDraw.Draw(self.image)
    self.canvas.delete("all")
    for action in self.actions:
        for _, coords in action:
            self.draw.line(coords, fill='white', width=5)
            self.canvas.create_line(coords, fill='white', width=5)

​
clear()

Clears the entire canvas and resets all state (main.py:62-66).

​
PIL Integration

The application uses Pillow (PIL) for image processing:
  • Image Creation: Creates a blank RGB image matching canvas dimensions
  • Drawing: Uses ImageDraw to render white lines (width=5) on black background
  • Font Rendering: Loads default font at size 100 for displaying answers
  • Export: Converts images to PNG format for base64 encoding

​
Image Flow

  1. User draws on Tkinter canvas
  2. Each stroke is simultaneously drawn on PIL Image
  3. Image is encoded to base64 PNG (main.py:88-91)
  4. Base64 string is sent to OpenAI API
  5. Answer is rendered back onto both canvas and image

​
Answer Display

The draw_answer() method (main.py:119-138) positions the calculated result: 
def draw_answer(self, answer):
    if not self.actions:
        return
    
    last_action = self.actions[-1]
    last_coords = last_action[-1][-1]

    equals_x = last_coords[2]
    equals_y = last_coords[3]

    x_start = equals_x + 70
    y_start = equals_y - 20

    self.canvas.create_text(x_start, y_start, text=answer, font=self.custom_font, fill="#FF9500")

    font = ImageFont.load_default(size=100)
    self.draw.text((x_start, y_start - 50), answer, font=font, fill="#FF9500")
The answer appears 70 pixels to the right of the last drawn element in orange (#FF9500).

​
Dependencies

From requirements.txt:
  • pillow==10.2.0: Image processing and rendering
  • openai==1.14.2: GPT-4o API client
  • tkinter: GUI framework (built into Python)

Build docs developers (and LLMs) love

Get started for freeTalk to us