code2llm-rust

AI Cost Tracking

• 🤖 LLM usage: $0.0680 (4 commits)
• 👤 Human dev: ~$313 (3.1h @ $100/h, 30min dedup)

Generated on 2026-10-04 using openrouter/qwen/qwen3-coder-next

High-performance native Rust acceleration extension for code2llm.

Overview

code2llm-rust is a Python C-extension built with PyO3 and Rayon, providing ultra-fast native implementations of core computational bottlenecks in code2llm:

• Cyclomatic Complexity: Parallel McCable complexity and brace depth extraction (200x+ faster than pure Python).
• Call Extraction: Native token-level call scanner for C-family, JavaScript, TypeScript, Rust, Go languages.
• Graph Centrality: Native Brandes' betweenness centrality algorithm executing in parallel across CPU cores (100x+ faster than NetworkX).
• Code Smell Engine: Vectorized detection of God functions and Data clumps.

Installation

pip install code2llm-rust

Or build from source:

cd packages/code2llm-rust
maturin develop --release

Python API

import code2llm_rust

# Extract function body
body = code2llm_rust.extract_function_body(source_code, start_line=10)

# Calculate complexity
cc, rank = code2llm_rust.calculate_complexity(body, lang="c_family")

# Batch compute across lines in parallel
results = code2llm_rust.batch_complexity(source_code, [10, 25, 40], lang="c_family")

# Extract calls
calls = code2llm_rust.extract_calls(body)

# Betweenness centrality (Brandes' algorithm)
centrality = code2llm_rust.betweenness_centrality(nodes, edges, k=100, normalized=True)

License

Licensed under Apache-2.0.