Files
cobble/src/functions/backend/db/usage/bucket-usage-days.ts
T
tben d9aea8e9c8 Record what every model call spends, and show it
Nothing anywhere recorded what a model call cost: the harness used a
provider's usage numbers only to decide whether a turn had finished, then
dropped them. So a reply could not say what it took to make, one model
could not be compared with another on price, and there was no way to ask
what a day, a week or a month of the crew's work came to.

- `token_usages` records one row per call: user, bot, workspace, session,
  job and message, the harness, the provider and model that actually
  answered, the prompt and completion counts, and the cost when the
  provider reports one.
- `harness-usage-scope.ts` carries those ids in an AsyncLocalStorage
  scope that `run-bot-job.ts` enters once, so a run's turns, its session
  summary, its transcriptions, images and embeddings are all attributed
  without threading ids through every harness function. Nested runs merge
  rather than replace, and a null genuinely clears a field (a bot's own
  chat has no workspace).
- Work with no run behind it is charged to its owner instead:
  `run-with-fallback-scope.ts` reaches the workspace's creator else the
  bot's, and deliberately does not override a real run, so an ingest a
  bot asked for is not misattributed to the bot's owner.
- All six harnesses record through `withCandidates`'s `on_success`.
  Recording never throws: a model that answered must not lose its answer
  to a failed insert.
- A job's calls land before its message exists, so `complete-bot-job.ts`
  back-fills `message_id` once the message is inserted.
- The message's actions row shows what it cost, beside the timestamp and
  the play button, with the breakdown in its title.
- `/admin/usage` totals the install's spending for 7, 30 or 90 days: the
  total against the previous period, a day-by-day chart, a split per kind
  of call, and a table per model with the price per thousand tokens
  worked out from cost and tokens, which is what makes two models
  comparable. Crew managers only, since it is the whole install's work
  rather than one user's conversation. It sits with Routines and Activity
  in the user popover rather than under Settings, because it is a record
  of what happened, not something to configure.

Days are grouped in local time (`date(created_at / 1000, 'unixepoch',
'localtime')`), matching the local midnights the window builds and the
`YYYY-MM-DD` buckets it looks up. Grouping by the bare UTC day put a late
evening's calls under a name no bucket carried, so they vanished and the
chart's total silently disagreed with the totals card; a test pins it.

`canSeeRailItem` now decides who sees a popover or rail entry. The two
each filtered with their own inline rule and only one understood
`can_view`, so a gated page could be offered in the popover to everyone.
2026-10-23 15:41:41 +01:00

116 lines
3.3 KiB
TypeScript

import { DayMs, startOfDay } from "./grab-usage-window";
import { toSummary } from "./grab-usage-totals";
import type { TokenUsageSummary } from "@/src/harness/usage/types";
type Row = {
/**
* The day the call was made, `YYYY-MM-DD`, already in local time.
*/
local_day: string;
calls: number | null;
prompt_tokens: number | null;
completion_tokens: number | null;
cost: number | null;
};
type Params = {
rows: Row[];
start: number;
end: number;
};
export type UsageDay = {
/**
* The day's local midnight (epoch ms).
*/
at: number;
/**
* `YYYY-MM-DD`, in local time.
*/
label: string;
} & TokenUsageSummary;
/**
* The rows SQL grouped by local day, laid onto one bucket per day of the
* window, oldest first. Days with nothing in them are still there, at
* zero, so the bars keep their spacing, and every row lands in a bucket:
* a day outside the window would be a bug in the query's own bounds.
*/
export default function bucketUsageDays({ rows, start, end }: Params): UsageDay[] {
const by_day = new Map<string, TokenUsageSummary>();
for (const row of rows) {
/**
* Accumulated, not replaced: the query groups by day, so a repeated
* day means someone asked for two sets of rows, and the later one
* must add to the earlier rather than hide it.
*/
const label = String(row.local_day);
const summary = toSummary({ row });
by_day.set(label, mergeSummaries({ into: by_day.get(label), summary }));
}
const days: UsageDay[] = [];
const last = startOfDay({ at: end });
for (let at = startOfDay({ at: start }); at <= last; at += DayMs) {
const label = localDate({ at });
days.push({
at,
label,
...(by_day.get(label) || emptySummary()),
});
}
return days;
}
/**
* The day a local midnight falls on, in local time.
*
* Never `toISOString`: an epoch millisecond read in UTC is a different
* calendar day for anyone east or west of Greenwich, and a call counted
* under one day would then look for a bucket named after another and
* disappear from the chart (which is exactly what it did before).
*/
export function localDate({ at }: { at: number }) {
const date = new Date(at);
return [
date.getFullYear(),
String(date.getMonth() + 1).padStart(2, "0"),
String(date.getDate()).padStart(2, "0"),
].join("-");
}
function emptySummary(): TokenUsageSummary {
return { prompt_tokens: 0, completion_tokens: 0, total_tokens: 0, calls: 0 };
}
/**
* One summary added onto another. A day that reported no cost still costs
* nothing; only a real cost is carried, so an unpriced day doesn't turn a
* priced one into zero.
*/
function mergeSummaries({
into,
summary,
}: {
into?: TokenUsageSummary;
summary: TokenUsageSummary;
}): TokenUsageSummary {
const prompt_tokens = (into?.prompt_tokens || 0) + summary.prompt_tokens;
const completion_tokens = (into?.completion_tokens || 0) + summary.completion_tokens;
const cost = (into?.cost || 0) + (summary.cost || 0);
return {
prompt_tokens,
completion_tokens,
total_tokens: prompt_tokens + completion_tokens,
...(cost > 0 ? { cost } : {}),
calls: (into?.calls || 0) + summary.calls,
};
}