Wan2.1/wan-pwa/apps/api/routes/webhooks.py
Claude 51c5837c43
feat: Implement Phase 3 - Backend Integration & Critical Polish
This commit implements Phase 3 of the Wan2.1 PWA, closing all critical
integration gaps between frontend, backend, database, and Replicate API.

## Backend Integration 

### Database Writes
- Create generation records BEFORE calling Replicate
- Store job_id for tracking Replicate predictions
- Track progress, status, and completion timestamps
- Save video URLs and error messages

### Credit System
- Atomic credit deduction using database function deduct_credits()
- Automatic refunds on generation failures via refund_credits()
- Complete audit trail in credit_transactions table
- Transaction logging for all credit operations

### Webhook Handler
- Created /api/webhooks/replicate endpoint
- HMAC signature verification for security
- Automatic status updates from Replicate push notifications
- Maps Replicate statuses to application statuses
- Triggers refunds for failed generations

### Updated Generation Flow
1. Check user credits before starting
2. Create generation record (status: queued)
3. Start Replicate job and get job_id
4. Update record with job_id (status: processing)
5. Deduct credits atomically
6. Webhook updates status when complete
7. Polling fallback if webhook fails

## Frontend Enhancements 

### Error Handling
- Added sonner for beautiful toast notifications
- Success/error/loading states with retry actions
- User-friendly error messages
- Providers component wraps app with Toaster

### Form Validation
- Zod schemas for T2V and I2V inputs
- Prompt length validation (10-500 chars)
- Model and resolution validation
- Credit cost calculator

### Credit Management
- useCredits hook for real-time credit fetching
- Optimistic updates on generation start
- Credit refresh functionality
- Loading and error states

### Image Upload
- Drag-and-drop ImageUpload component
- Client-side validation (file type, size)
- Image preview functionality
- Max 10MB size limit with user feedback
- Ready for I2V integration

### Settings Page
- Basic settings page structure
- Placeholders for Profile, Billing, API Keys
- Ready for Phase 4 enhancements

## Database Changes 

### New Migration: 002_credit_system.sql
- credit_transactions table with audit trail
- deduct_credits() function for atomic operations
- add_credits() function for purchases/bonuses
- refund_credits() function for failed generations
- Added job_id, progress, error_message columns to generations

## Documentation 

### PHASE_3_IMPLEMENTATION.md
- Complete implementation guide
- Testing checklist (backend, frontend, E2E)
- Deployment steps with webhook registration
- Known issues and limitations
- Metrics to monitor
- Phase 4 roadmap

## Files Changed

### Backend (4 files)
- apps/api/main.py - Added webhooks router
- apps/api/routes/generation.py - Complete rewrite with DB integration
- apps/api/routes/webhooks.py - NEW webhook handler
- packages/db/migrations/002_credit_system.sql - NEW credit system

### Frontend (7 files)
- apps/web/package.json - Added sonner
- apps/web/src/app/layout.tsx - Added Providers wrapper
- apps/web/src/app/dashboard/settings/page.tsx - NEW settings page
- apps/web/src/components/providers.tsx - NEW toast provider
- apps/web/src/components/generation/image-upload.tsx - NEW upload component
- apps/web/src/lib/hooks/use-credits.ts - NEW credit management hook
- apps/web/src/lib/validation/generation.ts - NEW Zod schemas

### Documentation (1 file)
- PHASE_3_IMPLEMENTATION.md - NEW comprehensive guide

## Testing Required

### Backend
- [ ] Database writes on generation start
- [ ] Credit deduction accuracy
- [ ] Webhook updates from Replicate
- [ ] Refunds on failures

### Frontend
- [ ] Toast notifications
- [ ] Form validation
- [ ] Credit display and warnings
- [ ] Image upload

### Integration
- [ ] End-to-end generation flow
- [ ] Credit deduction → generation → completion
- [ ] Webhook vs polling updates

## Next Steps (Phase 4)

1. Payment integration with Stripe
2. Retry logic for failed generations
3. Cancel in-progress generations
4. In-app video player
5. Batch operations
6. Admin panel

## Environment Variables

### New Required Variables
- REPLICATE_WEBHOOK_SECRET - For webhook signature verification

See PHASE_3_IMPLEMENTATION.md for complete setup instructions.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>
2025-10-24 14:36:07 +00:00

125 lines
3.6 KiB
Python

from fastapi import APIRouter, HTTPException, Header, Request
import hmac
import hashlib
import json
import os
from core.supabase import get_supabase
from datetime import datetime
router = APIRouter()
@router.post("/replicate")
async def replicate_webhook(request: Request, webhook_signature: str = Header(None, alias="Webhook-Signature")):
"""
Handle Replicate completion webhook
This endpoint receives push notifications from Replicate when predictions complete,
eliminating the need for constant polling.
"""
# Read raw body for signature verification
body = await request.body()
# Verify webhook signature
secret = os.getenv("REPLICATE_WEBHOOK_SECRET")
if secret:
expected_signature = hmac.new(
secret.encode(),
body,
hashlib.sha256
).hexdigest()
if not webhook_signature or not hmac.compare_digest(webhook_signature, expected_signature):
raise HTTPException(status_code=401, detail="Invalid webhook signature")
# Parse payload
try:
payload = json.loads(body)
except json.JSONDecodeError:
raise HTTPException(status_code=400, detail="Invalid JSON payload")
# Extract prediction data
job_id = payload.get("id")
status = payload.get("status")
output = payload.get("output")
error = payload.get("error")
if not job_id:
raise HTTPException(status_code=400, detail="Missing prediction ID")
# Update database
supabase = get_supabase()
# Find generation by job_id
generation_result = (
supabase.table("generations")
.select("id, user_id")
.eq("job_id", job_id)
.single()
.execute()
)
if not generation_result.data:
# Generation not found - this is expected for non-Wan predictions
return {"status": "ignored", "message": "Generation not found"}
generation_id = generation_result.data["id"]
# Prepare update data
update_data = {}
# Map Replicate status to our status
status_map = {
"starting": "processing",
"processing": "processing",
"succeeded": "completed",
"failed": "failed",
"canceled": "failed",
}
new_status = status_map.get(status, "processing")
update_data["status"] = new_status
# Update progress
if new_status == "processing":
update_data["progress"] = 50
elif new_status == "completed":
update_data["progress"] = 100
update_data["completed_at"] = datetime.utcnow().isoformat()
elif new_status == "failed":
update_data["progress"] = 0
# Save video URL if completed
if status == "succeeded" and output:
video_url = output
if isinstance(video_url, list):
video_url = video_url[0]
update_data["video_url"] = video_url
# Save error if failed
if error:
update_data["error_message"] = str(error)
# Update database
supabase.table("generations").update(update_data).eq("id", generation_id).execute()
# If failed, trigger refund
if new_status == "failed":
try:
supabase.rpc("refund_credits", {"p_gen_id": generation_id}).execute()
except Exception as refund_error:
print(f"Failed to refund credits for generation {generation_id}: {refund_error}")
return {
"status": "ok",
"generation_id": generation_id,
"new_status": new_status
}
@router.get("/health")
async def webhook_health():
"""Health check endpoint for webhook"""
return {"status": "ok", "message": "Webhook endpoint is healthy"}