utils/converter.py

"""
Core conversion logic for transforming Python code to Gradio apps
"""

import re
import traceback
from typing import Dict, Any
from langchain_groq import ChatGroq
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser


def convert_code(input_code: str, groq_api_key: str, model_name: str = "llama-3.3-70b-versatile") -> Dict[str, str]:
    """
    Converts input Python/IPYNB code into a Gradio app structure.
    """
    
    # Validate API key
    if not groq_api_key:
        raise ValueError("Groq API Key is required for conversion.")
    
    if not groq_api_key.startswith('gsk_'):
        raise ValueError("Invalid Groq API Key. It should start with 'gsk_'")
    
    try:
        # Initialize Groq client
        chat = ChatGroq(
            temperature=0.1,
            model_name=model_name,
            api_key=groq_api_key,
            max_retries=2,
            timeout=30
        )
    except Exception as e:
        raise ValueError(f"Failed to initialize Groq client: {str(e)}")
    
    # System prompt for conversion - UPDATED with Gradio 6.2.0
    system_prompt = """You are an expert Python developer specializing in Gradio and Hugging Face Spaces. 
    Your task is to convert the provided Python code into a deployable Gradio web application.
    
    Return your response in this exact format:
    
    ===APP_PY===
    [Complete app.py code here - NO markdown backticks or ```python]
    ===APP_PY_END===
    
    ===REQUIREMENTS_TXT===
    [Requirements.txt content here, one dependency per line - NO markdown backticks]
    ===REQUIREMENTS_TXT_END===
    
    ===README_MD===
    [README.md content here with proper YAML frontmatter]
    ===README_MD_END===
    
    CRITICAL RULES:
    1. app.py MUST be valid Python code ONLY
    2. requirements.txt MUST be plain text
    3. README.md MUST have YAML frontmatter at the top
    4. Include all necessary imports in app.py
    5. Always include: if __name__ == "__main__": demo.launch()
    6. requirements.txt must include: gradio>=6.2.0  # CHANGED FROM 4.0.0 TO 6.2.0
    
    YAML frontmatter template:
    ---
    title: Generated Gradio App
    emoji: 🚀
    colorFrom: blue
    colorTo: purple
    sdk: gradio
    sdk_version: "6.2.0"  # CHANGED FROM 4.0.0 TO 6.2.0
    app_file: app.py
    pinned: false
    ---
    
    Do NOT include markdown code blocks (```) in any output.
    """
    
    human_template = """Convert this Python code to a complete Gradio app.
    
    Code to convert:
    ```python
    {code}
    ```"""
    
    # Create the chain
    prompt = ChatPromptTemplate.from_messages([
        ("system", system_prompt),
        ("human", human_template)
    ])
    
    chain = prompt | chat | StrOutputParser()
    
    try:
        # Invoke the chain
        response = chain.invoke({"code": input_code[:50000]})
        
        # Parse the response
        result = parse_llm_response(response)
        
        # Clean and validate outputs
        result = clean_all_outputs(result)
        result = post_process_conversion(result, input_code)
        
        return result
        
    except Exception as e:
        print(f"Conversion error: {e}")
        print(f"Traceback: {traceback.format_exc()}")
        return create_basic_template(input_code)


def parse_llm_response(response: str) -> Dict[str, str]:
    """Parse the LLM response into structured components"""
    
    result = {
        "app_py": "",
        "requirements_txt": "",
        "readme_md": ""
    }
    
    patterns = [
        (r'===APP_PY===(.*?)===APP_PY_END===', "app_py"),
        (r'===REQUIREMENTS_TXT===(.*?)===REQUIREMENTS_TXT_END===', "requirements_txt"),
        (r'===README_MD===(.*?)===README_MD_END===', "readme_md")
    ]
    
    for pattern, key in patterns:
        match = re.search(pattern, response, re.DOTALL)
        if match:
            result[key] = match.group(1).strip()
    
    return result


def clean_all_outputs(result: Dict[str, str]) -> Dict[str, str]:
    """Clean markdown backticks from all outputs"""
    
    for key in ["app_py", "requirements_txt", "readme_md"]:
        if key in result and result[key]:
            content = result[key]
            
            # Remove markdown code blocks
            content = re.sub(r'^```(python|py)?\s*', '', content, flags=re.IGNORECASE)
            content = re.sub(r'\s*```$', '', content)
            content = re.sub(r'```(python|py)?\s*(.*?)\s*```', r'\2', content, flags=re.DOTALL | re.IGNORECASE)
            
            # Remove any remaining stray backticks
            content = content.replace('```', '')
            
            # Strip whitespace
            content = content.strip()
            
            result[key] = content
    
    return result


def post_process_conversion(result: Dict[str, str], original_code: str) -> Dict[str, str]:
    """Post-process and validate the conversion result"""
    
    # Ensure app.py is valid Python
    if not result.get("app_py"):
        result["app_py"] = create_basic_app_template(original_code)
    
    # Ensure requirements.txt is clean
    result["requirements_txt"] = clean_requirements(result.get("requirements_txt", ""))
    
    # Ensure README has proper format
    result["readme_md"] = clean_readme(result.get("readme_md", ""))
    
    return result


def clean_requirements(req_text: str) -> str:
    """Clean and format requirements.txt content"""
    if not req_text:
        return "gradio>=6.2.0\nlangchain-groq>=0.1.0\nhuggingface-hub>=0.20.0"  # UPDATED
    
    # Remove any markdown
    req_text = re.sub(r'```.*?\n', '', req_text, flags=re.DOTALL)
    req_text = req_text.replace('```', '')
    
    lines = []
    for line in req_text.strip().split('\n'):
        line = line.strip()
        if line and not line.startswith('#') and not line.startswith('==='):
            line = re.sub(r'[\[\]"\']', '', line)
            if line:
                lines.append(line)
    
    # Ensure essential packages are included - UPDATED VERSION
    essential_packages = ["gradio>=6.2.0", "langchain-groq>=0.1.0", "huggingface-hub>=0.20.0"]  # UPDATED
    for package in essential_packages:
        if not any(package.split('>=')[0] in line.lower() for line in lines):
            lines.insert(0, package)
    
    # Remove duplicates and sort
    return "\n".join(sorted(set(lines)))


def clean_readme(readme_text: str) -> str:
    """Clean README.md to ensure proper format"""
    if not readme_text:
        return create_default_readme()
    
    # Remove markdown code blocks
    readme_text = re.sub(r'```.*?\n', '', readme_text, flags=re.DOTALL)
    readme_text = readme_text.replace('```', '')
    
    # Check for proper YAML frontmatter
    if not re.search(r'^---\s*$', readme_text.strip(), re.MULTILINE):
        return create_default_readme()
    
    # Update sdk_version in frontmatter if it's 4.0.0
    readme_text = re.sub(
        r'sdk_version:\s*["\']4\.0\.0["\']',
        'sdk_version: "6.2.0"',
        readme_text
    )
    
    # Update gradio version in requirements if mentioned
    readme_text = re.sub(
        r'gradio>=4\.0\.0',
        'gradio>=6.2.0',
        readme_text
    )
    
    return readme_text.strip()


def create_default_readme() -> str:
    """Create a default README with proper YAML frontmatter"""
    return """---
title: Generated Gradio App
emoji: 🚀
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: "6.2.0"  # UPDATED
app_file: app.py
pinned: false
---

# Generated Gradio Application

This application was automatically generated using the Hugging Face Space Creator.

## Features

- User-friendly interface built with Gradio 6.2.0
- Ready for deployment on Hugging Face Spaces
- Includes all necessary dependencies

## How to Use

1. Launch the application
2. Enter your inputs in the provided fields
3. Click submit to see results

## Deployment

To deploy this application on Hugging Face Spaces:

1. Create a new Space on Hugging Face
2. Select Gradio as the SDK
3. Upload the generated files
4. Your app will be live automatically

## Requirements

All dependencies are listed in requirements.txt
"""


def create_basic_app_template(input_code: str) -> str:
    """Create a basic, guaranteed-valid app.py template"""
    code_snippet = input_code[:1000]
    
    return f'''import gradio as gr

# Original code snippet
{code_snippet}

# Gradio interface
def process_input(input_text):
    try:
        # Process the input
        return f"Processed: {{input_text}}"
    except Exception as e:
        return f"Error: {{str(e)}}"

# Create the Gradio app
with gr.Blocks(title="Generated Application", theme=gr.themes.Soft()) as demo:
    gr.Markdown("# Generated Application")
    
    with gr.Row():
        with gr.Column():
            input_box = gr.Textbox(
                label="Input Text",
                placeholder="Enter text to process...",
                lines=3
            )
            submit_btn = gr.Button("Process", variant="primary")
    
    with gr.Row():
        output_box = gr.Textbox(
            label="Output",
            interactive=False,
            lines=5
        )
    
    submit_btn.click(
        fn=process_input,
        inputs=[input_box],
        outputs=[output_box]
    )

if __name__ == "__main__":
    demo.launch(share=False)
'''


def create_basic_template(input_code: str) -> Dict[str, str]:
    """Create a basic template when conversion fails"""
    return {
        "app_py": create_basic_app_template(input_code),
        "requirements_txt": "gradio>=6.2.0\nlangchain-groq>=0.1.0\nhuggingface-hub>=0.20.0",  # UPDATED
        "readme_md": create_default_readme()
    }