refactor(passthrough): address multipart review feedback

Build form_data_dict in one pass with groupby instead of rescanning form_items per field name, and assert on the files list directly in the boundary regression test so repeated field names are not collapsed by dict().

Co-authored-by: Cursor <cursoragent@cursor.com>

litellm/proxy/pass_through_endpoints/pass_through_endpoints.py

@@ -6,6 +6,7 @@ import posixpath
 import traceback
 from base64 import b64encode
 from datetime import datetime
+from itertools import groupby
 from typing import Any, Dict, List, Mapping, Optional, Tuple, Union, cast
 from urllib.parse import urlencode, urlparse
 
@@ -499,21 +500,23 @@ class HttpPassThroughEndpointHelpers(BasePassthroughUtils):
             if isinstance(field_value, (StarletteUploadFile, UploadFile))
         ]
 
-        field_names = tuple(
-            dict.fromkeys(
-                field_name
-                for field_name, field_value in form_items
-                if not isinstance(field_value, (StarletteUploadFile, UploadFile))
-            )
+        non_file_items = tuple(
+            (field_name, field_value)
+            for field_name, field_value in form_items
+            if not isinstance(field_value, (StarletteUploadFile, UploadFile))
         )
+        field_order = {
+            field_name: index
+            for index, field_name in enumerate(
+                dict.fromkeys(field_name for field_name, _ in non_file_items)
+            )
+        }
         form_data_dict = {
-            name: [
-                field_value
-                for field_name, field_value in form_items
-                if field_name == name
-                and not isinstance(field_value, (StarletteUploadFile, UploadFile))
-            ]
-            for name in field_names
+            field_name: [value for _, value in group]
+            for field_name, group in groupby(
+                sorted(non_file_items, key=lambda item: field_order[item[0]]),
+                key=lambda item: item[0],
+            )
         }
 
         # Remove content-type header - httpx will set it correctly with the new boundary

tests/test_litellm/proxy/pass_through_endpoints/test_pass_through_endpoints.py

@@ -3276,8 +3276,10 @@ async def test_multipart_passthrough_preserves_boundary():
     async def mock_httpx_request(method, url, **kwargs):
         # Verify that files parameter is passed (not json)
         assert "files" in kwargs, "Files should be passed for multipart requests"
-        files = dict(kwargs["files"])
-        assert "file" in files, "File field should be in files"
+        file_parts = [
+            value for name, value in kwargs["files"] if name == "file"
+        ]
+        assert len(file_parts) == 1, "File field should be in files"
 
         # Verify content-type is NOT in headers (httpx will set it with correct boundary)
         headers = kwargs.get("headers", {})
@@ -3285,7 +3287,7 @@ async def test_multipart_passthrough_preserves_boundary():
             "content-type" not in headers
         ), "content-type should be removed for multipart"
 
-        filename, content, content_type = files["file"]
+        filename, content, content_type = file_parts[0]
         assert filename == "test.txt"
         assert content == b"test file content"
         assert content_type == "text/plain"