spend: ONE all-time figure, computed once, for every surface that shows it
ci/woodpecker/push/woodpecker Pipeline was canceled

The panel's card said $941.37 and the member dashboard's said $900.51 for the
same claim. The panel used its own per-team query; the dashboard summed the
chart series IN THE BROWSER, which disagrees with the server whenever a row is
null, capped or stale.

coop_spend_totals() is now the definition: one GROUP BY over SpendLogs joined to
the team table, so
  spend_all_time      = every row (members, operators, agent teams, sandboxes) —
                        the same universe litellm_spend_by_day() sums, so the
                        cards and the money chart agree by construction
  own_all_time_spend  = the part that is the co-op's own (non-member teams, plus
                        rows with no team)
and member_all_time_spend = the rest. Exposed through /admin/overview (the
panel's boot payload), the money payload, and /broker/metrics (the dashboard's).
_coop_spend_totals_safe() returns {} rather than a zero: a failed computation
must never read as "the co-op has spent nothing".

tests/test_coop_spend_totals.py pins the arithmetic (total == member + own ==
every row) and the no-fake-zero rule.
This commit is contained in:
inference-admin committed 2026-11-13 21:44:13 +00:00
1 parent e15a01c338
commit 2a77b0c988
2 files changed
+180 -1

No files matched your search

+101
View File
@@ -0,0 +1,101 @@
#!/usr/bin/env python3
"""Regression test: ONE definition of the co-op's all-time spend.
WHY THIS EXISTS
On 2026-11-13 the panel's card said $941.37 and the member dashboard's said
$900.51 — the same claim, two numbers. The panel used the portal's per-team
query; the dashboard summed the chart series IN THE BROWSER, which disagrees
with the server whenever a row is null, capped or stale. coop_spend_totals()
is now the single source, exposed to the panel's boot payload, the money
payload and the broker's metrics.
The split has to be the same split on every surface, and a failure to compute
it must NEVER look like "$0 spent": the safe wrapper returns {} so callers
omit the key instead of printing a zero that reads as fact.
"""
import pathlib
import sys
ROOT = pathlib.Path(__file__).resolve().parent.parent
sys.path.insert(0, str(ROOT))
import app.main as m # noqa: E402
FAILS = []
def check(cond, msg):
print((" ok " if cond else " FAIL ") + msg)
if not cond:
FAILS.append(msg)
class FakeCursor:
def __init__(self, rows):
self._rows = rows
def execute(self, *a, **k):
pass
def fetchall(self):
return self._rows
def __enter__(self):
return self
def __exit__(self, *a):
return False
class FakeConn:
def __init__(self, rows):
self._rows = rows
def cursor(self):
return FakeCursor(self._rows)
def close(self):
pass
ROWS = [
("member:alice@example.org", 10.005),
("member:bob@example.org", 5.005),
("agent:infrastructure-bot", 1.5),
("sandbox:throwaway", 0.02),
("", 1.07), # spend rows with no team at all
]
real_ready, real_connect = m.litellm_db_ready, m._pg_connect
try:
m.litellm_db_ready = lambda: True
m._pg_connect = lambda: FakeConn(ROWS)
t = m.coop_spend_totals()
print("1. the split adds up and covers every row")
check(t["member_all_time_spend"] == 15.01,
f"member side is the member: teams ($10.005 + $5.005 -> 15.01, got {t['member_all_time_spend']})")
check(t["own_all_time_spend"] == 2.59,
f"own side is everything else: agents, sandboxes, unattributed (got {t['own_all_time_spend']})")
check(t["spend_all_time"] == round(t["member_all_time_spend"] + t["own_all_time_spend"], 2),
"total == member + own, so no dollar is counted twice or dropped")
check(t["spend_all_time"] == round(sum(r[1] for r in ROWS), 2),
f"total is every SpendLogs row ({round(sum(r[1] for r in ROWS), 2)}), the same universe the money chart sums")
print("2. a failure to compute must not look like $0")
def boom():
raise RuntimeError("LiteLLM DB credentials not configured")
m._pg_connect = boom
check(m._coop_spend_totals_safe() == {},
"the safe wrapper returns {} so callers omit the key instead of showing a fake zero")
m.litellm_db_ready = lambda: False
check(m._coop_spend_totals_safe() == {},
"and it degrades the same way when the DB is simply not configured")
finally:
m.litellm_db_ready, m._pg_connect = real_ready, real_connect
print()
if FAILS:
print(f"coop spend totals: {len(FAILS)} FAILURE(S)")
sys.exit(1)
print("coop spend totals: OK — one definition, one number, no fake zeros")