From 5692db812389efd118a14fb403b1da1e278c56ee Mon Sep 17 00:00:00 2001
From: Sameer Kankute <sameer@berri.ai>
Date: Tue, 17 Mar 2026 15:33:11 +0530
Subject: [PATCH] fix(file_search): address latest greptile feedback

Strip internal logging ids from emulated sub-calls, dedupe included search_results by file_id, clean unused imports, and add unit coverage for dedupe behavior.

Made-with: Cursor
---
 .../responses/file_search/emulated_handler.py | 12 +++++++----
 litellm/responses/main.py                     |  3 ++-
 .../llms/test_file_search_responses.py        | 21 +++++++++++++++++++
 3 files changed, 31 insertions(+), 5 deletions(-)

diff --git a/litellm/responses/file_search/emulated_handler.py b/litellm/responses/file_search/emulated_handler.py
index 85a8f789988..5d2c23fdfa3 100644
--- a/litellm/responses/file_search/emulated_handler.py
+++ b/litellm/responses/file_search/emulated_handler.py
@@ -14,9 +14,7 @@ Flow:
 import json
 import time
 import uuid
-from typing import Any, Dict, Iterable, List, Literal, Optional, Tuple, Union, cast
-
-import httpx
+from typing import Any, Dict, Iterable, List, Optional, Tuple, cast
 
 from litellm._logging import verbose_logger
 from litellm.types.llms.openai import ResponsesAPIResponse
@@ -220,7 +218,13 @@ def _build_search_results_for_include(
     file_search_call.search_results (mirrors OpenAI's include= format).
     """
     formatted: List[Dict[str, Any]] = []
+    seen_file_ids: set = set()
     for result in results:
+        file_id = _get_field(result, "file_id") or ""
+        if file_id and file_id in seen_file_ids:
+            continue
+        if file_id:
+            seen_file_ids.add(file_id)
         content_items = _get_field(result, "content") or []
         text_chunks = [
             c.get("text", "") if isinstance(c, dict) else getattr(c, "text", "")
@@ -229,7 +233,7 @@ def _build_search_results_for_include(
         text = " ".join(t for t in text_chunks if t)
         formatted.append(
             {
-                "file_id": _get_field(result, "file_id") or "",
+                "file_id": file_id,
                 "filename": _get_field(result, "filename") or "",
                 "score": _get_field(result, "score"),
                 "text": text,
diff --git a/litellm/responses/main.py b/litellm/responses/main.py
index 761be5a2e14..4404e6b366a 100644
--- a/litellm/responses/main.py
+++ b/litellm/responses/main.py
@@ -732,6 +732,7 @@ def responses(
                 aresponses_with_emulated_file_search,
             )
 
+            _internal_skip = {"litellm_logging_obj", "litellm_call_id", "aresponses"}
             emulated_kwargs = {
                 "include": include,
                 "instructions": instructions,
@@ -759,7 +760,7 @@ def responses(
                 "extra_body": extra_body,
                 "timeout": timeout,
                 "custom_llm_provider": custom_llm_provider,
-                **kwargs,
+                **{k: v for k, v in kwargs.items() if k not in _internal_skip},
             }
             if _is_async:
                 return aresponses_with_emulated_file_search(
diff --git a/tests/test_litellm/llms/test_file_search_responses.py b/tests/test_litellm/llms/test_file_search_responses.py
index f1eb2650542..63599781c17 100644
--- a/tests/test_litellm/llms/test_file_search_responses.py
+++ b/tests/test_litellm/llms/test_file_search_responses.py
@@ -659,6 +659,27 @@ class TestEmulatedFileSearchHandler:
         annotations = _build_file_citation_annotations([r1, r2], "text")
         assert len(annotations) == 1
 
+    def test_H14_include_search_results_dedupes_by_file_id(self):
+        from litellm.responses.file_search.emulated_handler import (
+            _build_search_results_for_include,
+        )
+
+        r1, r2 = MagicMock(), MagicMock()
+        r1.file_id = "file-abc"
+        r1.filename = "doc.pdf"
+        r1.score = 0.9
+        r1.attributes = {}
+        r1.content = [{"type": "text", "text": "first hit"}]
+        r2.file_id = "file-abc"  # same file appears for a second query
+        r2.filename = "doc.pdf"
+        r2.score = 0.85
+        r2.attributes = {}
+        r2.content = [{"type": "text", "text": "second hit"}]
+
+        search_results = _build_search_results_for_include([r1, r2])
+        assert len(search_results) == 1
+        assert search_results[0]["file_id"] == "file-abc"
+
     # --- End-to-end (mocked) ---
 
     @pytest.mark.asyncio