prepare_kaggle_dataset.py

#!/usr/bin/env python3
"""
Kaggle Dataset Preparation Script
Packages agent code for upload to Kaggle
"""

import os
import zipfile
import shutil
from pathlib import Path

def create_kaggle_dataset():
    """Create a zip file with all necessary code for Kaggle"""
    
    # Define directories and files to include
    directories = [
        "agent",
        "utils", 
        "data_prep"
    ]
    
    files = [
        "kaggle_llm_engine.py",
        "test_server.py",
        "requirements.txt",
        "README.md"
    ]
    
    # Create output directory
    output_dir = Path("kaggle_dataset")
    output_dir.mkdir(exist_ok=True)
    
    # Create zip file
    zip_path = output_dir / "telco_agent_code.zip"
    
    with zipfile.ZipFile(zip_path, 'w', zipfile.ZIP_DEFLATED) as zipf:
        # Add directories
        for directory in directories:
            if os.path.exists(directory):
                for root, dirs, files_in_dir in os.walk(directory):
                    for file in files_in_dir:
                        if file.endswith('.py'):
                            file_path = os.path.join(root, file)
                            arcname = os.path.relpath(file_path, ".")
                            zipf.write(file_path, arcname)
                            print(f"Added: {arcname}")
        
        # Add individual files
        for file in files:
            if os.path.exists(file):
                zipf.write(file, file)
                print(f"Added: {file}")
    
    # Create dataset metadata
    metadata = {
        "title": "Telco Troubleshooting Agent Code",
        "description": "Complete agent implementation for Telco Troubleshooting Agentic Challenge",
        "keywords": ["telco", "troubleshooting", "agent", "llm", "react"],
        "licenses": ["mit"]
    }
    
    # Save metadata
    import json
    with open(output_dir / "dataset-metadata.json", "w") as f:
        json.dump(metadata, f, indent=2)
    
    print(f"\nKaggle dataset created: {zip_path}")
    print(f"Size: {zip_path.stat().st_size / 1024 / 1024:.1f} MB")
    print(f"Files included: {len(zipf.namelist())}")
    
    # Create upload instructions
    instructions = """
# Kaggle Dataset Upload Instructions

## Files Created:
- telco_agent_code.zip: Contains all agent code
- dataset-metadata.json: Dataset metadata

## Upload Steps:
1. Go to https://www.kaggle.com/datasets
2. Click "New Dataset"
3. Upload telco_agent_code.zip
4. Copy metadata from dataset-metadata.json
5. Set as private dataset
6. Note the dataset path (e.g., /kaggle/input/username/telco-agent-code)

## Usage in Notebook:
```python
# Add dataset to path
import sys
sys.path.append('/kaggle/input/username/telco-agent-code')

# Import modules
from kaggle_llm_engine import create_kaggle_llm_engine
from agent.react_loop import ReActAgent
```
"""
    
    with open(output_dir / "upload_instructions.md", "w") as f:
        f.write(instructions)
    
    print(f"Upload instructions saved: {output_dir / 'upload_instructions.md'}")
    
    return zip_path

if __name__ == "__main__":
    zip_path = create_kaggle_dataset()
    print(f"\nReady for Kaggle upload: {zip_path}")