Update README to include usage examples and troubleshooting tips for the FastAPI service integration with Paperless-ngx and llama.cpp.

tests/test_llama_payload.py

@@ -0,0 +1,35 @@
+from __future__ import annotations
+
+import base64
+import json
+
+import respx
+from httpx import Response
+
+from notebook_tools.llama_client import LlamaClient
+
+
+@respx.mock
+async def test_llama_client_parses_openai_style_response() -> None:
+    # Arrange: mock llama endpoint
+    route = respx.post("http://llama.local/v1/chat/completions").mock(
+        return_value=Response(200, json={"choices": [{"message": {"content": "Hello\nWorld"}}]})
+    )
+
+    client = LlamaClient(base_url="http://llama.local", model="m")
+
+    # Act
+    out = await client.ocr_jpeg(jpeg_bytes=b"\xff\xd8\xff\xe0fakejpeg")
+
+    # Assert
+    assert out == "Hello\nWorld"
+    assert route.called
+
+    # Optional: sanity-check that we really sent a base64 data URL
+    sent = json.loads(route.calls[0].request.content.decode("utf-8"))
+    url = sent["messages"][0]["content"][1]["image_url"]["url"]
+    assert url.startswith("data:image/jpeg;base64,")
+    b64 = url.split(",", 1)[1]
+    # If this fails, our payload construction changed.
+    base64.b64decode(b64)
+