feat(e2e): Add files external provider tests

.github/workflows/ci.yaml

@@ -338,6 +338,8 @@ jobs:
         env:
           _TYPER_FORCE_DISABLE_TERMINAL: "1"
           E2E_SERVICES_LOG: ${{ runner.temp }}/services.log
+          NGC_API_KEY: ${{ secrets.NGC_API_KEY }}
+          HF_TOKEN: ${{ secrets.HF_TOKEN }}
       - name: Dump server logs
         if: always()
         run: |

e2e/files/test_storage_backends.py

@@ -0,0 +1,225 @@
+"""E2E tests for external storage backends (NGC, Hugging Face).
+
+These tests verify that the files service can create filesets backed by
+external storage providers and read files from them via the SDK.
+
+NGC tests require ``NGC_API_KEY`` in the environment and are skipped
+otherwise.  Hugging Face tests use a small public repo; when ``HF_TOKEN``
+is set the request is authenticated (avoids rate-limits in CI).
+"""
+
+import os
+import tempfile
+import uuid
+from collections.abc import Iterator
+from pathlib import Path
+
+import pytest
+from nemo_platform import NeMoPlatform
+from nemo_platform.types.files import HuggingfaceStorageConfigParam, NGCStorageConfigParam
+
+# ---------------------------------------------------------------------------
+# NGC configuration
+# ---------------------------------------------------------------------------
+NGC_API_KEY_ENV = "NGC_API_KEY"
+
+NGC_ORG = "nvidian"
+NGC_TEAM = "nemo-llm"
+NGC_TARGET = "nemo-platform-quickstart"
+NGC_TARGET_TYPE = "resource"
+
+# ---------------------------------------------------------------------------
+# Hugging Face configuration — small public model
+# ---------------------------------------------------------------------------
+HF_TOKEN_ENV = "HF_TOKEN"
+
+HF_REPO_ID = "hf-internal-testing/tiny-random-bert"
+HF_REPO_TYPE = "model"
+
+# ---------------------------------------------------------------------------
+# Fixtures
+# ---------------------------------------------------------------------------
+
+
+@pytest.fixture
+def ngc_api_key() -> str:
+    """Return the NGC API key from the environment."""
+    key = os.environ.get(NGC_API_KEY_ENV)
+    assert key, f"{NGC_API_KEY_ENV} must be set"
+    return key
+
+
+@pytest.fixture
+def ngc_secret(sdk: NeMoPlatform, workspace: str, ngc_api_key: str) -> Iterator[str]:
+    """Create a secret containing the NGC API key, cleaned up after test."""
+    secret_name = f"e2e-ngc-key-{uuid.uuid4().hex[:8]}"
+    sdk.secrets.create(workspace=workspace, name=secret_name, value=ngc_api_key)
+    yield secret_name
+    try:
+        sdk.secrets.delete(workspace=workspace, name=secret_name)
+    except Exception:
+        pass
+
+
+@pytest.fixture
+def ngc_fileset(sdk: NeMoPlatform, workspace: str, ngc_secret: str) -> Iterator[str]:
+    """Create an NGC-backed fileset, cleaned up after test."""
+    fileset_name = f"e2e-ngc-fs-{uuid.uuid4().hex[:8]}"
+    sdk.files.filesets.create(
+        workspace=workspace,
+        name=fileset_name,
+        description="E2E test NGC-backed fileset",
+        storage=NGCStorageConfigParam(
+            api_key_secret=ngc_secret,
+            org=NGC_ORG,
+            team=NGC_TEAM,
+            target=NGC_TARGET,
+            target_type=NGC_TARGET_TYPE,
+        ),
+    )
+    yield fileset_name
+    try:
+        sdk.files.filesets.delete(fileset_name, workspace=workspace)
+    except Exception:
+        pass
+
+
+@pytest.fixture
+def hf_secret(sdk: NeMoPlatform, workspace: str) -> Iterator[str | None]:
+    """Create a secret for the HF token if available, otherwise yield None."""
+    token = os.environ.get(HF_TOKEN_ENV)
+    if not token:
+        yield None
+        return
+
+    secret_name = f"e2e-hf-tok-{uuid.uuid4().hex[:8]}"
+    sdk.secrets.create(workspace=workspace, name=secret_name, value=token)
+    yield secret_name
+    try:
+        sdk.secrets.delete(workspace=workspace, name=secret_name)
+    except Exception:
+        pass
+
+
+@pytest.fixture
+def hf_fileset(sdk: NeMoPlatform, workspace: str, hf_secret: str | None) -> Iterator[str]:
+    """Create a Hugging Face-backed fileset, cleaned up after test."""
+    fileset_name = f"e2e-hf-fs-{uuid.uuid4().hex[:8]}"
+
+    storage = HuggingfaceStorageConfigParam(
+        repo_id=HF_REPO_ID,
+        repo_type=HF_REPO_TYPE,
+    )
+    if hf_secret is not None:
+        storage["token_secret"] = hf_secret
+
+    sdk.files.filesets.create(
+        workspace=workspace,
+        name=fileset_name,
+        description="E2E test HF-backed fileset",
+        storage=storage,
+    )
+    yield fileset_name
+    try:
+        sdk.files.filesets.delete(fileset_name, workspace=workspace)
+    except Exception:
+        pass
+
+
+# ===================================================================
+# NGC tests
+# ===================================================================
+
+
+@pytest.mark.skipif(not os.environ.get(NGC_API_KEY_ENV), reason=f"{NGC_API_KEY_ENV} not set")
+class TestNGCFileset:
+    """Tests for NGC-backed filesets."""
+
+    def test_list_files(self, sdk: NeMoPlatform, workspace: str, ngc_fileset: str):
+        """Listing an NGC-backed fileset returns files with paths and sizes."""
+        files = sdk.files.list(fileset=ngc_fileset, workspace=workspace)
+        assert len(files.data) > 0, "NGC fileset should contain at least one file"
+
+        for f in files.data:
+            assert f.path, "Each file should have a path"
+            assert f.size > 0, "Each file should have a non-zero size"
+
+    def test_download_file(self, sdk: NeMoPlatform, workspace: str, ngc_fileset: str):
+        """Downloading the smallest file from an NGC fileset succeeds and size matches."""
+        files = sdk.files.list(fileset=ngc_fileset, workspace=workspace)
+        assert len(files.data) > 0
+
+        target = min(files.data, key=lambda f: f.size)
+
+        with tempfile.TemporaryDirectory() as tmpdir:
+            local_path = Path(tmpdir) / target.path.replace("/", "_")
+            sdk.files.download(
+                fileset=ngc_fileset,
+                workspace=workspace,
+                remote_path=target.path,
+                local_path=str(local_path),
+            )
+            assert local_path.exists()
+            assert local_path.stat().st_size == target.size
+
+    def test_cache_status(self, sdk: NeMoPlatform, workspace: str, ngc_fileset: str):
+        """NGC-backed files report a cacheable status."""
+        files = sdk.files.list(
+            fileset=ngc_fileset,
+            workspace=workspace,
+            include_cache_status=True,
+        )
+        assert len(files.data) > 0
+
+        for f in files.data:
+            assert f.cache_status is not None
+            assert f.cache_status != "not_cacheable"
+
+
+# ===================================================================
+# Hugging Face tests
+# ===================================================================
+
+
+class TestHuggingFaceFileset:
+    """Tests for Hugging Face-backed filesets."""
+
+    def test_list_files(self, sdk: NeMoPlatform, workspace: str, hf_fileset: str):
+        """Listing an HF-backed fileset returns files with paths and sizes."""
+        files = sdk.files.list(fileset=hf_fileset, workspace=workspace)
+        assert len(files.data) > 0, "HF fileset should contain at least one file"
+
+        for f in files.data:
+            assert f.path, "Each file should have a path"
+            assert f.size > 0, "Each file should have a non-zero size"
+
+    def test_download_file(self, sdk: NeMoPlatform, workspace: str, hf_fileset: str):
+        """Downloading the smallest file from an HF fileset succeeds and size matches."""
+        files = sdk.files.list(fileset=hf_fileset, workspace=workspace)
+        assert len(files.data) > 0
+
+        target = min(files.data, key=lambda f: f.size)
+
+        with tempfile.TemporaryDirectory() as tmpdir:
+            local_path = Path(tmpdir) / target.path.replace("/", "_")
+            sdk.files.download(
+                fileset=hf_fileset,
+                workspace=workspace,
+                remote_path=target.path,
+                local_path=str(local_path),
+            )
+            assert local_path.exists()
+            assert local_path.stat().st_size == target.size
+
+    def test_cache_status(self, sdk: NeMoPlatform, workspace: str, hf_fileset: str):
+        """HF-backed files report a cacheable status."""
+        files = sdk.files.list(
+            fileset=hf_fileset,
+            workspace=workspace,
+            include_cache_status=True,
+        )
+        assert len(files.data) > 0
+
+        for f in files.data:
+            assert f.cache_status is not None
+            assert f.cache_status != "not_cacheable"