From de0db82320d3ec274b8b60f1d32a981891d968c4 Mon Sep 17 00:00:00 2001
From: RoyVivat <royvivat@gmail.com>
Date: Wed, 18 Mar 2026 16:23:27 -0700
Subject: [PATCH] Updated get_logging_id in utils.py to strip s3://bucket from
 id. Added tests

---
 litellm/utils.py                              | 10 +++--
 tests/test_litellm/integrations/test_s3_v2.py | 37 +++++++++++++++++++
 2 files changed, 44 insertions(+), 3 deletions(-)

diff --git a/litellm/utils.py b/litellm/utils.py
index 81d749ab821..ef67ea74894 100644
--- a/litellm/utils.py
+++ b/litellm/utils.py
@@ -7401,9 +7401,13 @@ def print_args_passed_to_litellm(original_function, args, kwargs):
 
 def get_logging_id(start_time, response_obj):
     try:
-        response_id = (
-            "time-" + start_time.strftime("%H-%M-%S-%f") + "_" + response_obj.get("id")
-        )
+        raw_id = response_obj.get("id") or ""
+        # Batch API IDs may be S3 URIs (e.g. "s3://bucket/key"). Strip the scheme and bucket, keeping only the object key to use as the log ID.
+        if "://" in raw_id:
+            safe_id = raw_id.split("://", 1)[1].split("/", 1)[-1].replace("/", "-")
+        else:
+            safe_id = raw_id
+        response_id = "time-" + start_time.strftime("%H-%M-%S-%f") + "_" + safe_id
         return response_id
     except Exception:
         return None
diff --git a/tests/test_litellm/integrations/test_s3_v2.py b/tests/test_litellm/integrations/test_s3_v2.py
index b53c05fa241..7dc25043fd5 100644
--- a/tests/test_litellm/integrations/test_s3_v2.py
+++ b/tests/test_litellm/integrations/test_s3_v2.py
@@ -6,6 +6,7 @@ import pytest
 
 from litellm.integrations.s3_v2 import S3Logger
 from litellm.types.utils import StandardLoggingPayload
+from litellm.utils import get_logging_id
 
 
 class TestS3V2UnitTests:
@@ -661,6 +662,42 @@ async def test_strip_base64_recursive_redaction():
 
 
 
+# --------------------------------------------------------------
+# Tests for get_logging_id S3 URI sanitization (issue #23904)
+# --------------------------------------------------------------
+def test_get_logging_id_strips_s3_uri_scheme_and_bucket():
+    """Bedrock batch file IDs are S3 URIs — they must not appear verbatim in log paths."""
+    start_time = datetime(2026, 3, 9, 17, 40, 11, 901585)
+    response_obj = {
+        "id": "s3://bucket-int/litellm-bedrock-files-us.anthropic.claude-sonnet-4-5-20250929-v1-0-29ea93-452e-8a2f.jsonl"
+    }
+    result = get_logging_id(start_time, response_obj)
+    assert "://" not in result, "S3 URI scheme must be stripped from log ID"
+    assert result.startswith("time-17-40-11-901585_")
+    # Bucket name should be gone; object key should be present
+    assert "bucket-int" not in result
+    assert "litellm-bedrock-files-us.anthropic.claude-sonnet-4-5-20250929-v1-0-29ea93-452e-8a2f.jsonl" in result
+
+
+def test_get_logging_id_normal_id_unchanged():
+    """Normal completion IDs must pass through without modification."""
+    start_time = datetime(2026, 3, 9, 17, 40, 11, 901585)
+    response_obj = {"id": "chatcmpl-abc123"}
+    result = get_logging_id(start_time, response_obj)
+    assert result == "time-17-40-11-901585_chatcmpl-abc123"
+
+
+def test_get_logging_id_s3_uri_safe_for_url_path():
+    """The sanitized log ID must be embeddable in an S3 object key URL without creating a malformed path."""
+    start_time = datetime(2026, 3, 9, 17, 40, 11, 901585)
+    response_obj = {"id": "s3://my-bucket/subdir/file.jsonl"}
+    result = get_logging_id(start_time, response_obj)
+    # Simulate how s3_v2 embeds this in a URL
+    url = f"https://my-log-bucket.s3.us-east-1.amazonaws.com/LiteLLMAPPLogs/2026-03-09/{result}.json"
+    assert "s3://" not in url
+    assert url.count("://") == 1, "URL must contain exactly one scheme separator"
+
+
 # --------------------------------------------------------------
 # Shared fixture that silences asyncio.create_task during tests
 # --------------------------------------------------------------