code-transformer-platform 1.0.0

AI-powered code migration platform for transforming legacy codebases to modern frameworks

Code Transformer Platform

Prerequisites

Environment Variables

Before running the platform, set the following environment variables for Azure OpenAI:

$env:AZURE_OPENAI_API_KEY = "your-api-key-here"
$env:AZURE_OPENAI_ENDPOINT = "https://your-endpoint.openai.azure.com/"

Or in bash:

export AZURE_OPENAI_API_KEY="your-api-key-here"
export AZURE_OPENAI_ENDPOINT="https://your-endpoint.openai.azure.com/"

Features

• Multi-Agent Pipeline: Orchestrates 8 specialized agents for code transformation
• Automated Documentation: Every agent logs its analysis to migration_analysis.md
• Dependency Analysis: Detects cycles, entry points, and external dependencies
• Risk Assessment: Categorizes components by migration risk level
• LLM-Powered Generation: Uses Azure OpenAI for intelligent code generation
• Separate File Output: Each component generates its own file

Usage

Basic Command

python -m cli.main --input-path ./legacy --output-path ./output --target-config config/targets/angular18.yaml --source-stack extjs+xhtml --target-stack angular18 --llm-config config/llm/azure_openai.yaml --policy-config config/llm/generation_policy.yaml

With Custom Output Directory

python -m cli.main --input-path ./legacy --output-path ./migrated-code --target-config config/targets/angular18.yaml --llm-config config/llm/azure_openai.yaml --policy-config config/llm/generation_policy.yaml

Alternative Command

python -m cli.main ./legacy --target-config config/target/angular18.yaml --llm-config config/llm/azure_openai.yaml --policy-config config/llm/generation_policy.yaml

Output Files

After running the pipeline, you'll find in your output directory:

Directory Structure

output/
├── generated/                    # Clean, formatted generated code
│   ├── ComponentName1/
│   │   ├── component.component.ts
│   │   ├── component.component.html
│   │   ├── component.component.css
│   │   └── component.service.ts
│   └── ComponentName2/
│       └── ...
├── prompts_log/                  # LLM interaction logs
│   ├── prompt_Component1_*.txt   # Prompts sent to LLM
│   ├── prompt_Component2_*.txt
│   └── responses/                # Raw LLM responses
│       ├── response_Component1_*.txt
│       └── response_Component2_*.txt
└── migration_analysis.md         # Comprehensive migration report

Files Explained

1. generated/ - Production-ready migrated code files

   • Each component in its own subdirectory
   • Properly formatted TypeScript/HTML/CSS
   • Ready for integration into your Angular project

2. prompts_log/ - Debugging and analysis

   • Shows exactly what was sent to the LLM
   • Contains original source code, dependencies, and instructions
   • Helps understand and improve the migration process

3. prompts_log/responses/ - Raw LLM output

   • Unprocessed responses from Azure OpenAI
   • Useful for debugging parsing issues
   • Contains metadata like token usage

4. migration_analysis.md - Complete audit trail documenting:

   • Files scanned and components discovered
   • Dependency resolution and cycles detected
   • AST parsing results
   • IR transformation details
   • Migration plan with risk assessment
   • Code generation metadata

This comprehensive output provides full traceability and easy access to both the migrated code and the migration process details.

Note: If --output-path is not specified, the default output directory is ./output