# Web Directory Refactoring Plan

## Current Structure Issues

1. **Flat structure**: All files in one directory (20 Python files)
2. **Naming inconsistency**: Mix of `admin_*`, `async_*`, `batch_*` prefixes
3. **Mixed concerns**: Routes, schemas, services, and workers in same directory
4. **Poor scalability**: Hard to navigate and maintain as project grows

## Proposed Structure (Best Practices)

```
src/web/
├── __init__.py                  # Main exports
├── app.py                       # FastAPI app factory
├── config.py                    # App configuration
├── dependencies.py              # Global dependencies
│
├── api/                         # API Routes Layer
│   ├── __init__.py
│   └── v1/                      # API version 1
│       ├── __init__.py
│       ├── routes.py            # Public API routes (inference)
│       ├── admin/               # Admin API routes
│       │   ├── __init__.py
│       │   ├── documents.py     # admin_routes.py → documents.py
│       │   ├── annotations.py   # admin_annotation_routes.py → annotations.py
│       │   ├── training.py      # admin_training_routes.py → training.py
│       │   └── auth.py          # admin_auth.py → auth.py (routes only)
│       ├── async_api/           # Async processing API
│       │   ├── __init__.py
│       │   └── routes.py        # async_routes.py → routes.py
│       └── batch/               # Batch upload API
│           ├── __init__.py
│           └── routes.py        # batch_upload_routes.py → routes.py
│
├── schemas/                     # Pydantic Models
│   ├── __init__.py
│   ├── common.py                # Shared schemas (ErrorResponse, etc.)
│   ├── inference.py             # schemas.py → inference.py
│   ├── admin.py                 # admin_schemas.py → admin.py
│   ├── async_api.py             # New: async API schemas
│   └── batch.py                 # New: batch upload schemas
│
├── services/                    # Business Logic Layer
│   ├── __init__.py
│   ├── inference.py             # services.py → inference.py
│   ├── autolabel.py             # admin_autolabel.py → autolabel.py
│   ├── async_processing.py      # async_service.py → async_processing.py
│   └── batch_upload.py          # batch_upload_service.py → batch_upload.py
│
├── core/                        # Core Components
│   ├── __init__.py
│   ├── auth.py                  # admin_auth.py → auth.py (logic only)
│   ├── rate_limiter.py          # rate_limiter.py → rate_limiter.py
│   └── scheduler.py             # admin_scheduler.py → scheduler.py
│
└── workers/                     # Background Task Queues
    ├── __init__.py
    ├── async_queue.py           # async_queue.py → async_queue.py
    └── batch_queue.py           # batch_queue.py → batch_queue.py
```

## File Mapping

### Current → New Location

| Current File | New Location | Purpose |
|--------------|--------------|---------|
| `admin_routes.py` | `api/v1/admin/documents.py` | Document management routes |
| `admin_annotation_routes.py` | `api/v1/admin/annotations.py` | Annotation routes |
| `admin_training_routes.py` | `api/v1/admin/training.py` | Training routes |
| `admin_auth.py` | Split: `api/v1/admin/auth.py` + `core/auth.py` | Auth routes + logic |
| `admin_schemas.py` | `schemas/admin.py` | Admin Pydantic models |
| `admin_autolabel.py` | `services/autolabel.py` | Auto-label service |
| `admin_scheduler.py` | `core/scheduler.py` | Training scheduler |
| `routes.py` | `api/v1/routes.py` | Public inference API |
| `schemas.py` | `schemas/inference.py` | Inference models |
| `services.py` | `services/inference.py` | Inference service |
| `async_routes.py` | `api/v1/async_api/routes.py` | Async API routes |
| `async_service.py` | `services/async_processing.py` | Async processing service |
| `async_queue.py` | `workers/async_queue.py` | Async task queue |
| `batch_upload_routes.py` | `api/v1/batch/routes.py` | Batch upload routes |
| `batch_upload_service.py` | `services/batch_upload.py` | Batch upload service |
| `batch_queue.py` | `workers/batch_queue.py` | Batch task queue |
| `rate_limiter.py` | `core/rate_limiter.py` | Rate limiting logic |
| `config.py` | `config.py` | Keep as-is |
| `dependencies.py` | `dependencies.py` | Keep as-is |
| `app.py` | `app.py` | Keep as-is (update imports) |

## Benefits

### 1. Clear Separation of Concerns
- **Routes**: API endpoint definitions
- **Schemas**: Data validation models
- **Services**: Business logic
- **Core**: Reusable components
- **Workers**: Background processing

### 2. Better Scalability
- Easy to add new API versions (`v2/`)
- Clear namespace for each domain
- Reduced file size (no 800+ line files)

### 3. Improved Maintainability
- Find files by function, not by prefix
- Each module has single responsibility
- Easier to write focused tests

### 4. Standard Python Patterns
- Package-based organization
- Follows FastAPI best practices
- Similar to Django/Flask structures

## Implementation Steps

### Phase 1: Create New Structure (No Breaking Changes)
1. Create new directories: `api/`, `schemas/`, `services/`, `core/`, `workers/`
2. Copy files to new locations (don't delete originals yet)
3. Update imports in new files
4. Add `__init__.py` with proper exports

### Phase 2: Update Tests
5. Update test imports to use new structure
6. Run tests to verify nothing breaks
7. Fix any import issues

### Phase 3: Update Main App
8. Update `app.py` to import from new locations
9. Run full test suite
10. Verify all endpoints work

### Phase 4: Cleanup
11. Delete old files
12. Update documentation
13. Final test run

## Migration Priority

**High Priority** (Most used):
- Routes and schemas (user-facing APIs)
- Services (core business logic)

**Medium Priority**:
- Core components (auth, rate limiter)
- Workers (background tasks)

**Low Priority**:
- Config and dependencies (already well-located)

## Backwards Compatibility

During migration, maintain backwards compatibility:

```python
# src/web/__init__.py
# Old imports still work
from src.web.api.v1.admin.documents import router as admin_router
from src.web.schemas.admin import AdminDocument

# Keep old names for compatibility (temporary)
admin_routes = admin_router  # Deprecated alias
```

## Testing Strategy

1. **Unit Tests**: Test each module independently
2. **Integration Tests**: Test API endpoints still work
3. **Import Tests**: Verify all old imports still work
4. **Coverage**: Maintain current 23% coverage minimum

## Rollback Plan

If issues arise:
1. Keep old files until fully migrated
2. Git allows easy revert
3. Tests catch breaking changes early

---

## Next Steps

Would you like me to:
1. **Start Phase 1**: Create new directory structure and move files?
2. **Create migration script**: Automate the file moves and import updates?
3. **Focus on specific area**: Start with admin API or async API first?