Add music-gen-api console script entrypoint for FastAPI server

[Copilot]

The FastAPI server existed but lacked a proper entrypoint. Users had to invoke uvicorn directly with the full module path.

Changes

• api.py: Added main() function that reads PORT from environment (defaults to 8080) and runs uvicorn
• setup.py: Added music-gen-api console script
• Dockerfile: Updated CMD to use music-gen-api entrypoint
• Documentation: Updated README and run-api.sh to use new entrypoint

Usage

# Start server (binds to 0.0.0.0:8080)
music-gen-api

# Custom port
PORT=3000 music-gen-api

# Cloud Run compatible (reads $PORT)
docker run -e PORT=8080 -e MUSIC_GEN_MODE=simulate music-track-generator

The server runs in simulate mode by default (no GCP credentials required). All existing functionality unchanged.

Original prompt

Context

You are working on branch copilot/add-song-track-generation-api in kngms/github-dev-sandbox.

Already implemented:

• Python package under src/music_generator with:
  • Click CLI music-gen (generate/list-presets/show-preset/save-preset/delete-preset/setup commands)
  • Pydantic models (TrackConfig, PresetConfig, SongStructure, StyleReference) with duration validation (60–240s)
  • PresetManager storing YAML presets in ./presets with built-ins
  • MusicGenerator that builds prompts and returns simulated generation results, but currently initializes Vertex AI and requires GOOGLE_CLOUD_PROJECT
• Good README, but it claims "API" exists even though there is no HTTP server yet
• Files: src/music_generator/{__init__.py, cli.py, generator.py, models.py, presets.py}, requirements.txt, setup.py, README.md, examples/, presets/

Goal:
Make this easily runnable locally and in Cloud Run by adding a FastAPI HTTP server, making GCP truly optional (simulate mode), and adding deployment support.

---

Requirements

A) Make GCP optional (must be done first)

1. Introduce a mode flag:

   • Environment variable: MUSIC_GEN_MODE=simulate|gcp (default: simulate)

2. In simulate mode:

   • MusicGenerator must NOT require GOOGLE_CLOUD_PROJECT
   • Must NOT call aiplatform.init() at startup
   • Must still build prompts and return the simulated result as it does now
   • All CLI commands must work without any GCP credentials

3. In gcp mode:

   • Require project id (from constructor parameter or GOOGLE_CLOUD_PROJECT env var)
   • Initialize Vertex AI as needed
   • Support ADC (Application Default Credentials) for Cloud Run - do not require GOOGLE_APPLICATION_CREDENTIALS file path

4. Update existing code:

   • Refactor MusicGenerator.__init__ to check mode and conditionally initialize GCP
   • Update CLI to work without errors when GOOGLE_CLOUD_PROJECT is missing in simulate mode
   • Keep all existing functionality working

---

B) Add FastAPI HTTP server (new)

1. Add dependencies to requirements.txt:

   • fastapi>=0.104.0
   • uvicorn[standard]>=0.24.0
   • For tests: pytest>=7.4.0, httpx>=0.25.0

2. Create FastAPI app at src/music_generator/api.py:

   Endpoints:

   • POST /tracks/generate

     • Request body supports both direct config and preset-based generation:

       {
         "text_input": "string (required)",
         "genre": "string (optional if preset_name provided)",
         "duration_seconds": 180,
         "preset_name": "string (optional)",
         "structure": {...},
         "style_references": [...],
         "temperature": 0.7
       }

     • If preset_name is provided: load preset, then merge/override with any provided fields (text_input, duration_seconds, structure, style_references, temperature)
     • Validate duration 60–240 (use Pydantic models)
     • Call MusicGenerator.generate_track() and return JSON result
     • Return 404 if preset not found
     • Return 422 for validation errors

   • GET /presets

     • List all presets with metadata (name, genre, description)
     • Response: [{"name": "...", "genre": "...", "description": "..."}]

   • GET /presets/{name}

     • Return full preset configuration
     • Return 404 if not found

   • POST /presets

     • Body: PresetConfig
     • Create or update preset
     • Return created preset

   • DELETE /presets/{name}

     • Delete preset
     • Return 404 if not found
     • Return success message

   • GET /prompt-tips

     • Optional query param: preset_name (string)
     • If preset_name provided: return tips for that preset only
     • Otherwise: return aggregated tips from all presets
     • Response: [{"preset": "name", "tips": "..."}] or {"tips": "..."}

   • GET /health

     • Simple health check endpoint
     • Return: {"status": "healthy", "mode": "simulate|gcp"}

3. Add optional API authentication:

   • If env var MUSIC_GEN_API_KEY is set, require it on all endpoints except /health
   • Accept either:
     • Header: X-API-Key: <key>
     • Header: Authorization: Bearer <key>
   • Return 401 if key is required but missing or incorrect
   • If MUSIC_GEN_API_KEY is not set, run without auth (useful for local dev)

4. Server entrypoint:

   • Add main() function in api.py that runs uvicorn.run(app, host="0.0.0.0", port=port)
   • Port should read from $PORT env var (for Cloud Run), default to 8080
   • Add console_scripts entry to setup.py: music-gen-api=music_generator.api:main

---

C) Cloud Run readiness

1. Add Dockerfile:
   • Use Python 3.11 slim base image
   • Copy and install requirements
   • Copy src/
   • Install package with pip install -e .
   • Set ENV MUSIC_GEN_MODE=simulate as default
   • Expose $PORT (Cloud Run requirement)
   • CMD to run `music-gen-ap...

This pull request was created from Copilot chat.

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.