fix: corrections on typing and attestation parameters

Author: jcabrero

nilai-api/gunicorn.conf.py

@@ -5,7 +5,7 @@
 bind = ["0.0.0.0:8080"]
 
 # Set the number of workers (2)
-workers = SETTINGS["gunicorn_workers"]
+workers = SETTINGS.gunicorn_workers
 
 # Set the number of threads per worker (16)
 threads = 1

nilai-api/src/nilai_api/attestation/__init__.py

@@ -12,7 +12,7 @@ async def get_attestation_report(
     """Get the attestation report for the given nonce"""
 
     try:
-        attestation_url = f"http://{SETTINGS['attestation_host']}:{SETTINGS['attestation_port']}/attestation/report"
+        attestation_url = f"http://{SETTINGS.attestation_host}:{SETTINGS.attestation_port}/attestation/report"
         async with httpx.AsyncClient() as client:
             response: httpx.Response = await client.get(attestation_url, params=nonce)
             report = AttestationReport(**response.json())
@@ -24,10 +24,10 @@ async def get_attestation_report(
 async def verify_attestation_report(attestation_report: AttestationReport) -> bool:
     """Verify the attestation report"""
     try:
-        attestation_url = f"http://{SETTINGS['attestation_host']}:{SETTINGS['attestation_port']}/attestation/verify"
+        attestation_url = f"http://{SETTINGS.attestation_host}:{SETTINGS.attestation_port}/attestation/verify"
         async with httpx.AsyncClient() as client:
-            response: httpx.Response = await client.post(
-                attestation_url, json=attestation_report.model_dump()
+            response: httpx.Response = await client.get(
+                attestation_url, params=attestation_report.model_dump()
             )
             return response.json()
     except Exception as e:

nilai-attestation/gunicorn.conf.py

@@ -2,10 +2,10 @@
 from nilai_common.config import SETTINGS
 
 # Bind to address and port
-bind = [f"0.0.0.0:{SETTINGS['attestation_port']}"]
+bind = [f"0.0.0.0:{SETTINGS.attestation_port}"]
 
 # Set the number of workers (2)
-workers = SETTINGS["gunicorn_workers"]
+workers = SETTINGS.gunicorn_workers
 
 # Set the number of threads per worker (16)
 threads = 1

nilai-attestation/src/nilai_attestation/attestation/__init__.py

@@ -24,6 +24,7 @@ def get_attestation_report(nonce: Nonce | None = None) -> AttestationReport:
     logger.info(f"Nonce: {attestation_nonce}")
 
     load_sev_library()
+
     return AttestationReport(
         nonce=attestation_nonce,
         verifying_key="",

nilai-attestation/src/nilai_attestation/attestation/nvtrust/nv_attester.py

@@ -46,7 +46,7 @@ def is_nvidia_gpu_available() -> bool:
         return False
 
 
-def nv_attest(nonce: Nonce) -> NVAttestationToken:
+def nv_attest(nonce: Nonce, name: str = "thisNode1") -> NVAttestationToken:
     """Generate an attestation token from local evidence.
 
     Args:
@@ -57,7 +57,7 @@ def nv_attest(nonce: Nonce) -> NVAttestationToken:
     """
     # Create and configure the attestation client.
     client = attestation.Attestation()
-    client.set_name("thisNode1")
+    client.set_name(name)
     client.set_nonce(nonce)
 
     logger.info("Checking if NVIDIA GPU is available")

nilai-attestation/src/nilai_attestation/attestation/nvtrust/nv_verifier.py

@@ -43,7 +43,9 @@
 }
 
 
-def verify_attestation(attestation_report: AttestationReport) -> bool:
+def verify_attestation(
+    attestation_report: AttestationReport, name: str = "thisNode1"
+) -> bool:
     """Verify an NVIDIA attestation token against a policy.
 
     Args:
@@ -57,7 +59,7 @@ def verify_attestation(attestation_report: AttestationReport) -> bool:
     # Create an attestation client instance for token verification.
     logger.info(f"Attestation report: {attestation_report}")
     client = attestation.Attestation()
-    client.set_name("thisNode1")
+    client.set_name(name)
     client.set_nonce(attestation_report.nonce)
     client.add_verifier(
         attestation.Devices.GPU, attestation.Environment.REMOTE, NRAS_URL, ""

nilai-attestation/src/nilai_attestation/routers/private.py

@@ -1,6 +1,6 @@
 # Fast API and serving
 import logging
-from fastapi import APIRouter
+from fastapi import APIRouter, Depends
 
 # Internal libraries
 from nilai_attestation.attestation import (
@@ -35,12 +35,14 @@ async def get_attestation(nonce: Nonce | None = None) -> AttestationReport:
     return get_attestation_report(nonce)
 
 
-@router.post("/attestation/verify", tags=["Attestation"])
-async def post_attestation(attestation_report: AttestationReport) -> bool:
+@router.get("/attestation/verify", tags=["Attestation"])
+async def get_attestation_verification(
+    attestation_report: AttestationReport = Depends(),
+) -> bool:
     """
-    Verify a cryptographic attestation report.
+    Verify a cryptographic attestation report passed as query parameters.
 
-    - **attestation_report**: Attestation report to verify
+    - **attestation_report**: Attestation report to verify (fields passed as query parameters)
     - **Returns**: True if the attestation report is valid, False otherwise
     """
     return verify_attestation_report(attestation_report)

nilai-models/src/nilai_models/daemon.py

@@ -21,7 +21,7 @@ async def get_metadata(num_retries=30):
     while True:
         url = None
         try:
-            url = f"http://{SETTINGS['host']}:{SETTINGS['port']}/v1/models"
+            url = f"http://{SETTINGS.host}:{SETTINGS.port}/v1/models"
             # Request model metadata from localhost:8000/v1/models
             async with httpx.AsyncClient() as client:
                 response = await client.get(url)
@@ -37,7 +37,7 @@ async def get_metadata(num_retries=30):
                     license="Apache 2.0",  # Usage license
                     source=f"https://huggingface.co/{model_name}",  # Model source
                     supported_features=["chat_completion"],  # Capabilities
-                    tool_support=SETTINGS["tool_support"],  # Tool support
+                    tool_support=SETTINGS.tool_support,  # Tool support
                 )
 
         except Exception as e:
@@ -81,12 +81,12 @@ async def main():
     try:
         # Initialize discovery service
         discovery_service = ModelServiceDiscovery(
-            host=SETTINGS["etcd_host"], port=SETTINGS["etcd_port"]
+            host=SETTINGS.etcd_host, port=SETTINGS.etcd_port
         )
 
         metadata = await get_metadata()
         model_endpoint = ModelEndpoint(
-            url=f"http://{SETTINGS['host']}:{SETTINGS['port']}", metadata=metadata
+            url=f"http://{SETTINGS.host}:{SETTINGS.port}", metadata=metadata
         )
 
         # Setup signal handlers

packages/nilai-common/src/nilai_common/config.py

@@ -1,24 +1,25 @@
 import os
+from pydantic import BaseModel
 
-# from dotenv import load_dotenv
 
-# load_dotenv()  # Only needed locally if using a .env file
+class HostSettings(BaseModel):
+    host: str = "localhost"
+    port: int = 8000
+    etcd_host: str = "localhost"
+    etcd_port: int = 2379
+    tool_support: bool = False
+    gunicorn_workers: int = 10
+    attestation_host: str = "localhost"
+    attestation_port: int = 8081
 
-SETTINGS = {
-    "host": os.getenv("SVC_HOST", "localhost"),
-    "port": os.getenv("SVC_PORT", 8000),
-    "etcd_host": os.getenv("ETCD_HOST", "localhost"),
-    "etcd_port": os.getenv("ETCD_PORT", 2379),
-    "tool_support": os.getenv("TOOL_SUPPORT", False),
-    "gunicorn_workers": os.getenv("NILAI_GUNICORN_WORKERS", 10),
-    "attestation_host": os.getenv("ATTESTATION_HOST", "localhost"),
-    "attestation_port": os.getenv("ATTESTATION_PORT", 8080),
-}
-# if environment == "docker":
-#     config = "docker_settings.py"
-# else:
-#     config = "local_settings.py"
 
-# # Import the appropriate config dynamically
-# from importlib import import_module
-# settings = import_module(config)
+SETTINGS: HostSettings = HostSettings(
+    host=str(os.getenv("SVC_HOST", "localhost")),
+    port=int(os.getenv("SVC_PORT", 8000)),
+    etcd_host=str(os.getenv("ETCD_HOST", "localhost")),
+    etcd_port=int(os.getenv("ETCD_PORT", 2379)),
+    tool_support=bool(os.getenv("TOOL_SUPPORT", False)),
+    gunicorn_workers=int(os.getenv("NILAI_GUNICORN_WORKERS", 10)),
+    attestation_host=str(os.getenv("ATTESTATION_HOST", "localhost")),
+    attestation_port=int(os.getenv("ATTESTATION_PORT", 8081)),
+)

tests/e2e/test_http.py

@@ -25,6 +25,7 @@ def client():
             "Content-Type": "application/json",
             "Authorization": f"Bearer {AUTH_TOKEN}",
         },
+        timeout=None,
     )