08_code2llm_integration

Example 8: code2llm Integration

This example demonstrates integrating vallm with code2llm for comprehensive code analysis and validation.

Overview

• code2llm: Analyzes code structure, extracts insights, generates TOON format reports
• vallm: Validates code quality through the 4-tier pipeline
• Combined: Full project analysis with quality gates

Prerequisites

pip install code2llm vallm

What It Demonstrates

1. Project Structure Analysis — code2llm analyzes file structure and language distribution
2. Quality Validation — vallm validates each file through syntax, imports, complexity checks
3. Combined Reporting — unified report with both structural and quality metrics
4. Batch Processing — process entire project directories

Running the Example

cd examples/08_code2llm_integration
python main.py

Expected Output

🚀 code2llm + vallm Integration Example
============================================================

📁 Creating sample project...
   Created at: sample_project

============================================================
Analyzing with code2llm...
============================================================
Analyzing project structure...
Found 3 Python files

============================================================
Validating with vallm...
============================================================
✓ math_utils.py: pass (score: 1.00)
✓ string_utils.py: pass (score: 1.00)
✓ main.py: pass (score: 1.00)

📊 Report saved to .vallm/code2llm_integration_report.json

============================================================
INTEGRATION SUMMARY
============================================================
Total files analyzed: 3
Passed: 3
Failed: 0

🎉 All files passed quality checks!

How It Works

1. Create Sample Project

SAMPLE_PROJECT = {
    "math_utils.py": '''...''',
    "string_utils.py": '''...''',
    "main.py": '''...''',
}

2. Analyze Structure

from code2llm import analyze_directory

# Get project structure
result = analyze_directory(project_path)
print(f"Found {len(result['files'])} files")

3. Validate Quality

from vallm import Proposal, validate

# Validate each file
for file_path in project_path.rglob("*.py"):
    code = file_path.read_text()
    proposal = Proposal(code=code, language="python")
    result = validate(proposal)
    print(f"{file_path}: {result.verdict.value}")

4. Generate Report

{
  "analysis_tools": {
    "code2llm": true,
    "vallm": true
  },
  "project_structure": {
    "files": 3,
    "languages": {"Python": 3}
  },
  "quality_validation": {
    "files": [
      {"file": "math_utils.py", "verdict": "pass", "score": 1.0}
    ]
  },
  "summary": {
    "total_files": 3,
    "passed": 3,
    "failed": 0
  }
}

Use Cases

CI/CD Pipeline

# Analyze and validate entire codebase
python -m vallm batch src/ --recursive --format json > validation.json
code2llm analyze ./ --format toon > analysis.toon

# Combine reports
python combine_reports.py

Pre-commit Hook

#!/usr/bin/env python
"""Pre-commit hook using code2llm + vallm."""

import subprocess
import sys

# Analyze staged files
result = subprocess.run(
    ["code2llm", "analyze", "--staged"],
    capture_output=True
)

# Validate staged files
result = subprocess.run(
    ["vallm", "batch", "--staged", "--fail-fast"],
    capture_output=True
)

if result.returncode != 0:
    print("❌ Validation failed")
    sys.exit(1)

IDE Integration

# VS Code extension using both tools
from code2llm import get_file_info
from vallm import validate_file

def on_file_save(file_path):
    # Get structural info
    info = get_file_info(file_path)
    
    # Validate quality
    result = validate_file(file_path)
    
    # Show inline diagnostics
    show_diagnostics(file_path, result.issues)

API Reference

code2llm Functions

• analyze_directory(path) — Analyze entire directory
• analyze_file(path) — Analyze single file
• TOONFormat.parse(path) — Parse TOON format output

vallm Integration

from vallm import Proposal, validate, VallmSettings
from vallm.core.languages import detect_language

# Auto-detect language
lang = detect_language(file_path)

# Create proposal with detected language
proposal = Proposal(
    code=code,
    language=lang.tree_sitter_id if lang else "python"
)

# Validate
result = validate(proposal)

Troubleshooting

code2llm not found

pip install code2llm

Import errors

# Ensure both packages are in same environment
pip install code2llm vallm

Further Reading

• code2llm documentation
• vallm documentation
• Batch validation