QuickSheet.AI 📚

Turn your course materials into an exam-ready cheatsheet in minutes.

QuickSheet.AI analyzes your past exams, lecture notes, slides, and assignments to automatically generate a dense, prioritized cheatsheet as a LaTeX PDF. Topics that appear most frequently on past exams always come first. You set the page limit — it fills it optimally.

---

How it works

Upload PDFs (exams, notes, slides, assignments)
         ↓
   Extract text + images
         ↓
   LLM analyzes each file type
         ↓
   Topics ranked by exam frequency
         ↓
   Content ordered by importance
         ↓
   LaTeX generated + compiled to PDF

Each file type has its own pipeline:

• Past exams → extract question topics and count frequency
• Lecture notes / slides → extract formulas, definitions, key concepts
• Assignments → identify solution patterns and methods
• Syllabus → extract topic weights and assessment emphasis
• Prof hints → boost importance of explicitly flagged topics

All signals are merged into a single importance score per topic. High-frequency exam topics always appear first, with formulas before definitions before concepts within each topic.

---

Features

• 🎯 Exam-frequency ordering — the most tested topics always appear first
• 📄 Strict page limits — content is truncated to fit exactly N pages
• 🎨 Color-coded type tags — instantly distinguish formulas, definitions, concepts, examples, diagrams
• 🔍 OCR support — scanned PDFs and handwritten notes handled via Tesseract
• 🖼️ Diagram extraction — embedded images from slides are pulled into the cheatsheet
• ✏️ Editable output — get both the compiled PDF and the raw .tex source
• ⚙️ Config-driven — save settings per course, override with CLI flags anytime

---

Setup

Prerequisites

• Python 3.10+
• An OpenAI-compatible API key
• (Optional) Tesseract for OCR on scanned PDFs
• (Optional) pdflatex for compiling the .tex output

Install

git clone https://github.com/your-username/quick-sheet.git
cd quick-sheet

python -m venv .venv
.venv\Scripts\activate        # Windows
# source .venv/bin/activate   # macOS / Linux

pip install -r requirements.txt

Environment

Create a .env file in the project root:

OPENAI_API_KEY=your_key_here

Optional: OCR (for scanned PDFs)

pip install pytesseract Pillow

Download the Tesseract binary:

• Windows: github.com/UB-Mannheim/tesseract/wiki
• macOS: brew install tesseract
• Linux: sudo apt install tesseract-ocr

Optional: PDF compilation

• Windows: MiKTeX or TeX Live
• macOS: brew install --cask mactex
• Linux: sudo apt install texlive-full

Without pdflatex the .tex file is still generated and can be compiled manually or via Overleaf.

---

Running

Web UI

python -m streamlit run ui/streamlit_app.py

1. Enter your course code in the sidebar
2. Upload PDFs into the appropriate type slots
3. Adjust page limit, columns, and color settings
4. Click Generate Final Cheatsheet
5. Download the PDF and/or .tex source

Command line

# Step 1 — parse all PDFs in a course folder
python app/processing/run_all_pipelines.py data/uploads/ECE342

# Step 2 — generate the cheatsheet
python app/processing/generate_sheet.py data/uploads/ECE342 --pages 2 --columns 4 --compile

Run either script with --help for all options.

---

Project structure

quick-sheet/
├── ui/
│   └── streamlit_app.py          # Streamlit frontend
├── app/
│   ├── main.py                   # Pipeline orchestrator (called by UI)
│   └── processing/
│       ├── run_all_pipelines.py  # PDF parsing → JSON
│       └── generate_sheet.py    # JSON → LaTeX → PDF
├── data/
│   ├── uploads/
│   │   └── <course_code>/
│   │       ├── exam/             # Past exam PDFs
│   │       ├── notes/            # Lecture note PDFs
│   │       ├── slide_deck/       # Slide PDFs
│   │       ├── assignment/       # Assignment PDFs
│   │       ├── syllabus/         # Syllabus PDFs
│   │       ├── misc/             # Prof hints, other
│   │       ├── parsed/           # Auto-generated JSON results
│   │       └── generated/        # Output .tex and .pdf files
│   └── configs/
│       └── <course_code>.json    # Per-course settings
├── .env                          # API key (not committed)
└── requirements.txt

---

Content priority

Within the page budget, blocks are ordered:

Priority	Type
1	High-frequency exam topics
2	Formulas
3	Definitions
4	Diagrams
5	Examples
6	Concepts

When the page limit is hit, the lowest-priority content is dropped first — so you always get the most important content.

---

Color tags

When color mode is on, each block gets a small colored label:

Tag	Color	Meaning
F	Blue	Formula
D	Orange	Definition
C	Green	Concept
E	Purple	Example
X	Grey	Diagram

---

Tech stack

Component	Technology
Frontend	Streamlit
PDF extraction	PyMuPDF (fitz)
OCR fallback	Tesseract + pytesseract
Image processing	Pillow
LLM	OpenAI-compatible API
Output format	LaTeX (pdflatex / TeX Live)
Language	Python 3.10+

---