completed local setup with compose

tests/e2e/test_backend_journey.py

@@ -0,0 +1,76 @@
+import asyncio
+
+import httpx
+import pytest
+
+from libs.events import EventTopics, NATSEventBus
+from libs.schemas.events import DocumentExtractedEventData
+
+# Configuration
+INGESTION_URL = "http://localhost:8000"
+NATS_URL = "nats://localhost:4222"
+TENANT_ID = "tenant_e2e_test"
+
+
+@pytest.mark.e2e
+@pytest.mark.asyncio
+async def test_backend_journey():
+    """
+    E2E test for the full backend journey: Ingest -> OCR -> Extract.
+    """
+    # 1. Initialize NATS bus
+    bus = NATSEventBus(
+        servers=[NATS_URL],
+        stream_name="TAX_AGENT_EVENTS",
+        consumer_group="e2e-test-consumer",
+    )
+    await bus.start()
+
+    # Future to capture the final event
+    extraction_future = asyncio.Future()
+
+    async def extraction_handler(topic, payload):
+        if payload.tenant_id == TENANT_ID:
+            extraction_future.set_result(payload)
+
+    # Subscribe to the final event in the chain
+    await bus.subscribe(EventTopics.DOC_EXTRACTED, extraction_handler)
+
+    try:
+        # 2. Upload a document
+        async with httpx.AsyncClient() as client:
+            # Create a dummy PDF file
+            files = {"file": ("test.pdf", b"%PDF-1.4 mock content", "application/pdf")}
+            response = await client.post(
+                f"{INGESTION_URL}/upload",
+                files=files,
+                data={"kind": "invoice", "source": "e2e_test"},
+                headers={"X-Tenant-ID": TENANT_ID, "X-User-ID": "e2e_tester"},
+            )
+            assert response.status_code == 200, f"Upload failed: {response.text}"
+            upload_data = response.json()
+            doc_id = upload_data["doc_id"]
+            print(f"Uploaded document: {doc_id}")
+
+        # 3. Wait for extraction event (with timeout)
+        try:
+            # Give it enough time for the whole chain to process
+            payload = await asyncio.wait_for(extraction_future, timeout=30.0)
+
+            # 4. Verify payload
+            data = payload.data
+            assert data["doc_id"] == doc_id
+            assert data["tenant_id"] == TENANT_ID
+            assert "extraction_results" in data
+
+            # Validate against schema
+            event_data = DocumentExtractedEventData(**data)
+            assert event_data.doc_id == doc_id
+
+            print("E2E Journey completed successfully!")
+
+        except TimeoutError:
+            pytest.fail("Timed out waiting for extraction event")
+
+    finally:
+        await bus.stop()