README.md

metadata

title: Code Graph
emoji: 💻
colorFrom: red
colorTo: green
sdk: docker
pinned: false

💻 Code Graph

Code Graph is a visual tool that analyzes and displays function dependencies in Python code using AST-based static parsing.

It’s ideal for developers, data scientists, and Streamlit users who want to understand how their functions interact — without needing to execute any code.

---

📌 What It Does

Upload one or more .py files, and this tool will:

• Parse the source code using Python’s built-in ast module
• Identify function definitions and their relationships
• Show which functions:
  • Are called from others
  • Read or write to st.session_state
  • Take specific arguments and return specific values
• Display all this in a fully interactive graph and table

---

🧠 What is AST?

AST stands for Abstract Syntax Tree — a structured representation of your source code as a tree. Each node represents a language construct like a function, loop, or assignment.

Code Graph uses ASTs to analyze code structure without running the code, making it safe and fast to use.

---

🚀 Features

✅ Upload one or more Python files
✅ Auto-detect function calls and dependencies
✅ Highlights:

• Function arguments and return values with inferred types (e.g. df: pd.DataFrame)
• Streamlit session_state reads/writes
• Call order labels on edges (1, 2, 3, …)
• Thicker edges for functions called multiple times

✅ Export the graph as downloadable JSON
✅ View everything in an interactive graph and sortable table

---

📂 How to Use

1. Launch the app or clone this repo
2. Upload one or more .py files
3. Select the "main" or top-level function to explore
4. Interact with the graph and table to trace logic

---

🛠 Tech Stack

• 🐍 Python
• 📦 Dash & Plotly
• 🧠 ast module (for static code parsing)
• 🌐 Cytoscape.js (for interactive graph rendering)

---

🙋‍♂️ About the Author

Created by Tomas Larsson, a data scientist and creator of
my.moneytoolbox.com — a blog covering:

• Tax-optimized investing
• Retirement modeling
• Financial tools for DIY investors

---

📤 Deployment (Docker-based)

If you're running locally:

docker build -t code-graph .
docker run -p 7860:7860 code-graph