tiennm99/litellm
1
0
Fork 0
mirror of https://github.com/tiennm99/litellm.git synced 2026-08-23 18:25:22 +00:00
Code Issues Packages Projects Releases Wiki Activity

Revert "Allow configuration to on what threshold to try truncating request content in db"

Browse Source 
This reverts commit 771052d67d.
This commit is contained in:
WilsonSunBritten
2025-08-28 16:12:42 -06:00
parent 771052d67d
commit 026480358b
2 changed files with 3 additions and 114 deletions
Download Patch File Download Diff File Expand all files Collapse all files
litellm/proxy/spend_tracking/spend_tracking_utils.py
+3 -17
View File
@@ -53,7 +53,7 @@ def _get_spend_logs_metadata(
guardrail_information: Optional[StandardLoggingGuardrailInformation] = None,
usage_object: Optional[dict] = None,
model_map_information: Optional[StandardLoggingModelInformation] = None,
cold_storage_object_key: Optional[str] = None,
cold_storage_object_key: Optional[str] = None
) -> SpendLogsMetadata:
if metadata is None:
return SpendLogsMetadata(
@@ -101,7 +101,7 @@ def _get_spend_logs_metadata(
clean_metadata["usage_object"] = usage_object
clean_metadata["model_map_information"] = model_map_information
clean_metadata["cold_storage_object_key"] = cold_storage_object_key
return clean_metadata
@@ -484,7 +484,6 @@ def _sanitize_request_body_for_spend_logs_payload(
Truncates strings longer than 1000 characters and handles nested dictionaries.
"""
from litellm.constants import LITELLM_TRUNCATED_PAYLOAD_FIELD
MAX_STRING_LENGTH = 1000
if visited is None:
@@ -523,21 +522,8 @@ def _get_proxy_server_request_for_spend_logs_payload(
)
if _proxy_server_request is not None:
_request_body = _proxy_server_request.get("body", {}) or {}
_request_body = _sanitize_request_body_for_spend_logs_payload(_request_body)
_request_body_json_str = json.dumps(_request_body, default=str)
# Check if request body size exceeds truncation threshold
max_body_size_before_trunc = litellm_params.get(
"max_request_size_before_trunc"
)
if (
max_body_size_before_trunc is None
or len(_request_body_json_str) > max_body_size_before_trunc
):
_request_body = _sanitize_request_body_for_spend_logs_payload(
_request_body
)
_request_body_json_str = json.dumps(_request_body, default=str)
return _request_body_json_str
return "{}"
tests/test_spend_logs.py
-97
View File
@@ -336,100 +336,3 @@ async def test_spend_report_endpoint():
print(f"Total Spend: {total_spend}")
print("Metadata: ", metadata)
print()
# Helper function to generate test request bodies
def _generate_test_request_bodies():
large_content = "x" * 2000 # 2000 characters
small_content = "x" * 50 # 50 characters
large_request_body = {
"model": "gpt-3.5-turbo",
"messages": [{"role": "user", "content": large_content}]
}
small_request_body = {
"model": "gpt-3.5-turbo",
"messages": [{"role": "user", "content": small_content}]
}
return large_request_body, small_request_body, large_content, small_content
def _setup_spend_logs_settings():
"""Helper to mock the settings for storing prompts in spend logs"""
import litellm.proxy.proxy_server
litellm.proxy.proxy_server.general_settings = {"store_prompts_in_spend_logs": True}
def test_large_request_no_truncation_threshold():
"""Test that large request with no threshold gets truncated"""
from litellm.proxy.spend_tracking.spend_tracking_utils import _get_proxy_server_request_for_spend_logs_payload
large_request_body, _, _, _ = _generate_test_request_bodies()
_setup_spend_logs_settings()
litellm_params = {
"proxy_server_request": {"body": large_request_body}
}
result = _get_proxy_server_request_for_spend_logs_payload({}, litellm_params)
# Should be truncated (contains truncation indicator)
assert "LITELLM_TRUNCATED_PAYLOAD_FIELD" in result
def test_large_request_high_truncation_threshold():
"""Test that large request with high threshold does not get truncated"""
from litellm.proxy.spend_tracking.spend_tracking_utils import _get_proxy_server_request_for_spend_logs_payload
large_request_body, _, large_content, _ = _generate_test_request_bodies()
_setup_spend_logs_settings()
litellm_params = {
"proxy_server_request": {"body": large_request_body},
"max_request_size_before_trunc": 10000 # Higher than request size
}
result = _get_proxy_server_request_for_spend_logs_payload({}, litellm_params)
# Should not be truncated (full content preserved)
assert "LITELLM_TRUNCATED_PAYLOAD_FIELD" not in result
assert large_content in result
def test_large_request_low_truncation_threshold():
"""Test that large request with low threshold gets truncated"""
from litellm.proxy.spend_tracking.spend_tracking_utils import _get_proxy_server_request_for_spend_logs_payload
large_request_body, _, _, _ = _generate_test_request_bodies()
_setup_spend_logs_settings()
litellm_params = {
"proxy_server_request": {"body": large_request_body},
"max_request_size_before_trunc": 100 # Lower than request size
}
result = _get_proxy_server_request_for_spend_logs_payload({}, litellm_params)
# Should be truncated
assert "LITELLM_TRUNCATED_PAYLOAD_FIELD" in result
def test_small_request_low_truncation_threshold():
"""Test that small request with low threshold does not get truncated"""
from litellm.proxy.spend_tracking.spend_tracking_utils import _get_proxy_server_request_for_spend_logs_payload
_, small_request_body, _, small_content = _generate_test_request_bodies()
_setup_spend_logs_settings()
litellm_params = {
"proxy_server_request": {"body": small_request_body},
"max_request_size_before_trunc": 100 # Higher than small request size
}
result = _get_proxy_server_request_for_spend_logs_payload({}, litellm_params)
# Should not be truncated (content is small enough)
assert "LITELLM_TRUNCATED_PAYLOAD_FIELD" not in result
assert small_content in result