Managed batches - Address PR bot comments from #22464

litellm/files/main.py

@@ -336,7 +336,7 @@ async def afile_retrieve(
 @client
 def file_retrieve(
     file_id: str,
-    custom_llm_provider: Literal["openai", "azure", "hosted_vllm", "manus"] = "openai",
+    custom_llm_provider: Literal["openai", "azure", "gemini", "vertex_ai", "hosted_vllm", "manus"] = "openai",
     extra_headers: Optional[Dict[str, str]] = None,
     extra_body: Optional[Dict[str, str]] = None,
     **kwargs,

litellm/llms/vertex_ai/batches/handler.py

@@ -115,9 +115,10 @@ class VertexAIBatchPrediction(VertexLLM):
                 data=json.dumps(vertex_batch_request),
             )
         except httpx.HTTPStatusError as e:
-            error_body = e.response.text if hasattr(e, 'response') else "N/A"
+            error_body = e.response.text
             litellm.verbose_logger.error(
-                f"Vertex AI batch create failed: status={e.response.status_code}, body={error_body[:1000]}"
+                "Vertex AI batch create failed: status=%s, body=%s",
+                e.response.status_code, error_body[:1000],
             )
             raise
         if response.status_code != 200:

litellm/llms/vertex_ai/files/transformation.py

@@ -408,10 +408,11 @@ class VertexAIFilesConfig(VertexBase, BaseFilesConfig):
         file_id = "deleted"
         if hasattr(raw_response, "request") and raw_response.request:
             url = str(raw_response.request.url)
-            if "/o/" in url:
+            if "/b/" in url and "/o/" in url:
                 import urllib.parse
+                bucket_part = url.split("/b/")[-1].split("/o/")[0]
                 encoded_name = url.split("/o/")[-1].split("?")[0]
-                file_id = f"gs://{urllib.parse.unquote(encoded_name)}"
+                file_id = f"gs://{bucket_part}/{urllib.parse.unquote(encoded_name)}"
         return FileDeleted(id=file_id, deleted=True, object="file")
 
     def transform_list_files_request(

tests/test_litellm/llms/vertex_ai/files/test_file_retrieve_provider_routing.py

@@ -0,0 +1,127 @@
+"""
+Tests for Fix 1: file_retrieve Literal type was missing 'vertex_ai' and 'gemini',
+causing a type mismatch when afile_retrieve delegated to the sync function.
+"""
+
+import pytest
+from unittest.mock import MagicMock, patch
+
+from litellm.files.main import file_retrieve
+
+
+class TestFileRetrieveProviderRouting:
+    """
+    Verify that file_retrieve accepts 'vertex_ai' and 'gemini' providers and
+    routes them through ProviderConfigManager / base_llm_http_handler.
+    """
+
+    def _make_mock_file_object(self):
+        mock = MagicMock()
+        mock.model_dump.return_value = {
+            "id": "gs://my-bucket/file.jsonl",
+            "object": "file",
+            "bytes": 1024,
+            "created_at": 0,
+            "filename": "file.jsonl",
+            "purpose": "batch",
+            "status": "processed",
+        }
+        return mock
+
+    def test_should_route_vertex_ai_through_provider_config(self):
+        """
+        Regression: file_retrieve Literal type was missing 'vertex_ai',
+        so passing custom_llm_provider='vertex_ai' would fail type-checking
+        and potentially cause a routing failure at runtime.
+        """
+        mock_file = self._make_mock_file_object()
+
+        with patch(
+            "litellm.files.main.base_llm_http_handler.retrieve_file",
+            return_value=mock_file,
+        ) as mock_retrieve:
+            result = file_retrieve(
+                file_id="gs://my-bucket/file.jsonl",
+                custom_llm_provider="vertex_ai",
+            )
+
+        mock_retrieve.assert_called_once()
+        assert result is not None
+
+    def test_should_route_gemini_through_provider_config(self):
+        """
+        Regression: file_retrieve Literal type was also missing 'gemini'.
+        """
+        mock_file = self._make_mock_file_object()
+
+        with patch(
+            "litellm.files.main.base_llm_http_handler.retrieve_file",
+            return_value=mock_file,
+        ) as mock_retrieve:
+            result = file_retrieve(
+                file_id="some-gemini-file-id",
+                custom_llm_provider="gemini",
+            )
+
+        mock_retrieve.assert_called_once()
+        assert result is not None
+
+    def test_should_pass_file_id_to_handler_for_vertex_ai(self):
+        """Verify the file_id is forwarded correctly to the underlying handler."""
+        mock_file = self._make_mock_file_object()
+        expected_file_id = "gs://my-bucket/path/to/file.jsonl"
+
+        with patch(
+            "litellm.files.main.base_llm_http_handler.retrieve_file",
+            return_value=mock_file,
+        ) as mock_retrieve:
+            file_retrieve(
+                file_id=expected_file_id,
+                custom_llm_provider="vertex_ai",
+            )
+
+        call_kwargs = mock_retrieve.call_args.kwargs
+        assert call_kwargs.get("file_id") == expected_file_id
+
+    def test_should_not_raise_bad_request_for_vertex_ai(self):
+        """
+        Before the fix, vertex_ai fell through to the else-branch which raised
+        BadRequestError. Verify it no longer does.
+        """
+        import litellm
+
+        mock_file = self._make_mock_file_object()
+
+        with patch(
+            "litellm.files.main.base_llm_http_handler.retrieve_file",
+            return_value=mock_file,
+        ):
+            try:
+                file_retrieve(
+                    file_id="gs://my-bucket/file.jsonl",
+                    custom_llm_provider="vertex_ai",
+                )
+            except litellm.exceptions.BadRequestError as e:
+                pytest.fail(
+                    f"file_retrieve raised BadRequestError for vertex_ai: {e}"
+                )
+
+    def test_should_not_raise_bad_request_for_gemini(self):
+        """Same as above but for 'gemini'."""
+        import litellm
+
+        mock_file = self._make_mock_file_object()
+
+        with patch(
+            "litellm.files.main.base_llm_http_handler.retrieve_file",
+            return_value=mock_file,
+        ):
+            try:
+                file_retrieve(
+                    file_id="some-file-id",
+                    custom_llm_provider="gemini",
+                )
+            except litellm.exceptions.BadRequestError as e:
+                pytest.fail(
+                    f"file_retrieve raised BadRequestError for gemini: {e}"
+                )

tests/test_litellm/llms/vertex_ai/files/test_vertex_ai_files_transformation.py

@@ -167,7 +167,7 @@ class TestTransformDeleteFile:
         assert isinstance(result, FileDeleted)
         assert result.deleted is True
         assert result.object == "file"
-        assert "litellm-vertex-files/publishers/google/models/gemini-2.0-flash-001/abc" in result.id
+        assert result.id == "gs://my-bucket/litellm-vertex-files/publishers/google/models/gemini-2.0-flash-001/abc"
 
     def test_should_fallback_to_deleted_id_when_no_request(self, config):
         raw_response = MagicMock(spec=httpx.Response)
@@ -182,3 +182,49 @@ class TestTransformDeleteFile:
         assert isinstance(result, FileDeleted)
         assert result.id == "deleted"
         assert result.deleted is True
+
+    def test_should_include_bucket_name_in_reconstructed_delete_id(self, config):
+        """
+        Regression: the old code split on /o/ only, dropping the bucket from
+        the reconstructed gs:// URI. e.g. gs://path/to/file instead of
+        gs://my-bucket/path/to/file.
+        """
+        raw_response = MagicMock(spec=httpx.Response)
+        mock_request = MagicMock()
+        encoded_object = urllib.parse.quote("path/to/file.jsonl", safe="")
+        mock_request.url = (
+            f"https://storage.googleapis.com/storage/v1/b/my-bucket/o/{encoded_object}"
+        )
+        raw_response.request = mock_request
+
+        result = config.transform_delete_file_response(
+            raw_response=raw_response,
+            logging_obj=MagicMock(),
+            litellm_params={},
+        )
+
+        assert result.id == "gs://my-bucket/path/to/file.jsonl"
+
+    def test_should_include_bucket_in_nested_object_path(self, config):
+        """Verify bucket extraction works with deeply nested GCS object paths."""
+        raw_response = MagicMock(spec=httpx.Response)
+        mock_request = MagicMock()
+        encoded_object = urllib.parse.quote(
+            "litellm-vertex-files/publishers/google/models/gemini-2.0-flash-001/abc-123",
+            safe="",
+        )
+        mock_request.url = (
+            f"https://storage.googleapis.com/storage/v1/b/prod-bucket/o/{encoded_object}"
+        )
+        raw_response.request = mock_request
+
+        result = config.transform_delete_file_response(
+            raw_response=raw_response,
+            logging_obj=MagicMock(),
+            litellm_params={},
+        )
+
+        assert result.id == (
+            "gs://prod-bucket/litellm-vertex-files/publishers/google/"
+            "models/gemini-2.0-flash-001/abc-123"
+        )