apunkt/SurfSense
1
0
Fork 0
mirror of https://github.com/MODSetter/SurfSense.git synced 2026-05-25 19:15:18 +02:00
Code Issues Projects Releases Packages Wiki Activity

perf(tokens): add per-call latency to capture log

Browse source
This commit is contained in:
CREDO23 2026-05-19 21:30:25 +02:00
parent b3b66e4c48
commit 581bbfb5c1
1 changed files with 12 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

13
surfsense_backend/app/services/token_tracking_service.py
View file

@ -357,9 +357,19 @@ class TokenTrackingCallback(CustomLogger):
cost_micros=cost_micros,
call_kind=call_kind,
)
# Per-LLM-call wall-clock latency (LiteLLM passes datetime objects).
call_latency_s: float | None = None
try:
if start_time is not None and end_time is not None:
delta = end_time - start_time
call_latency_s = getattr(delta, "total_seconds", lambda: float(delta))()
except Exception:
call_latency_s = None
logger.info(
"[TokenTracking] Captured: model=%s kind=%s prompt=%d completion=%d total=%d "
"cost=$%.6f (%d micros) (accumulator now has %d calls)",
"cost=$%.6f (%d micros) (accumulator now has %d calls)%s",
model,
call_kind,
prompt_tokens,
@ -368,6 +378,7 @@ class TokenTrackingCallback(CustomLogger):
cost_usd,
cost_micros,
len(acc.calls),
f" latency={call_latency_s:.3f}s" if call_latency_s is not None else "",
)