🚀 Static Code Analyzer & CFG Optimizer

A compiler middle-end optimization framework that parses C programs into Abstract Syntax Trees (AST), constructs Control Flow Graphs (CFG), performs classical compiler optimizations, and visualizes the optimized execution flow through an interactive web dashboard.

---

📌 Overview

This project implements a simplified compiler middle-end in Python, inspired by modern compiler optimization pipelines.

Starting from raw C source code, it:

• Parses the code into an Abstract Syntax Tree (AST) using pycparser
• Constructs a mathematical Control Flow Graph (CFG)
• Performs multiple static analysis passes
• Applies classical compiler optimizations
• Visualizes the optimized CFG using Graphviz
• Benchmarks optimization performance on large datasets such as CodeNet and SV-COMP

The project also includes an interactive Streamlit dashboard for visual exploration of CFG transformations.

---

📸 Demo

---

✨ Features

🔹 AST Generation

• Parses ANSI C programs using pycparser
• Converts source code into an Abstract Syntax Tree

🔹 Control Flow Graph (CFG) Construction

• Builds CFG nodes and directed edges
• Models execution paths using graph theory

🔹 Compiler Optimizations

✅ Constant Folding

Evaluates constant expressions during compile time.

Example:

int x = 5 * 4;

↓

int x = 20;

---

✅ Constant Propagation

Propagates known constant values through subsequent statements.

---

✅ Live Variable Analysis

Computes:

• live_in
• live_out

sets for every CFG node.

---

✅ Dead Code Elimination

Removes assignments whose values never contribute to future computation.

Example:

int x = 10;
x = 20;
return x;

↓

x = 20;
return x;

---

✅ Unreachable Code Removal

Uses CFG traversal to identify and eliminate execution blocks that cannot be reached from the function entry point.

---

📊 Interactive Dashboard

The Streamlit application allows users to:

• Paste custom C programs
• Generate AST
• Visualize the CFG
• Observe optimization passes
• Inspect optimized execution flow

---

📁 Project Structure

CFG_OPTIMIZER/
│
├── datasets/                  # CodeNet & SV-COMP benchmark datasets
│
├── app.py                     # Streamlit dashboard
├── batch_runner.py            # Batch benchmarking framework
├── main.py                    # CFG builder & optimization engine
│
├── codenet_report.json        # Generated benchmark report
├── sv_comp_report.json        # Generated benchmark report
│
├── cfg_demo.png               # Dashboard preview
└── README.md

---

🛠️ Tech Stack

Programming Language

• Python

Parsing

• pycparser

Graph Modeling

• NetworkX

Graph Visualization

• Graphviz
• pydot

Web Interface

• Streamlit

---

⚙️ Installation

1. Clone the Repository

git clone https://github.com/adi-2254/cfg_optimizer.git
cd cfg_optimizer

---

2. Create a Virtual Environment

Linux / macOS

python -m venv env
source env/bin/activate

Windows

python -m venv env
.\env\Scripts\activate

---

3. Install Dependencies

pip install streamlit networkx pycparser pydot

---

4. Install Graphviz

Graphviz is required for CFG rendering.

Windows

Download from:

https://graphviz.org/download/

During installation, enable "Add Graphviz to PATH".

macOS

brew install graphviz

Ubuntu / Debian

sudo apt-get install graphviz

---

🚀 Usage

Launch Interactive Dashboard

streamlit run app.py

Open your browser and paste any C program to visualize:

• AST
• CFG
• Optimized CFG

---

Run Batch Benchmarking

Evaluate the optimizer on large benchmark datasets.

python batch_runner.py

The generated reports include:

• Optimization statistics
• Number of optimized CFG nodes
• Removed dead code
• Reachability metrics

Reports are stored as:

codenet_report.json
sv_comp_report.json

---

📈 Supported Optimizations

Optimization	Implemented
AST Generation	✅
CFG Construction	✅
Constant Folding	✅
Constant Propagation	✅
Live Variable Analysis	✅
Dead Code Elimination	✅
Unreachable Code Removal	✅
CFG Visualization	✅
Dataset Benchmarking	✅

---

👨‍💻 Authors

• Anurag Raj (2024CSB1101)
• Aditya Raj (2024CSB1003)
• Aakash Jaisinghani (2024CSB1092)

If you found this project useful, consider ⭐ starring the repository.