-- +goose Up
-- teaching_signal_aggregates holds the folded reviewer craft counts that feed the curation pipeline (WS8).
-- Each row is one craft pattern, at one similarity band, for one decision outcome, on one engine model
-- version, over one observation window, with the observation and success counts that generalize the
-- reviewers' work. It carries NO PII, no candidate id, and no raw score, only the contents free counts from
-- the scrub.TeachingSignal vocabulary. tenant_id is set for the live in tenant rows the producer writes
-- (dp_applied false); the flag on cross tenant path later writes tenant_id NULL, dp_applied true rows after
-- the differential privacy step. Global reference data: migrator owned, app SELECT only, outside RLS. The
-- unique index keys on coalesce(tenant_id, the nil uuid) so the in tenant and cross tenant rows share one
-- key space, and it includes model_version so a forked engine version is visible as a distinct row rather
-- than silently overwriting a prior version's aggregate.
CREATE TABLE teaching_signal_aggregates (
    id            uuid PRIMARY KEY DEFAULT gen_random_uuid(),
    tenant_id     uuid,                 -- set for live in tenant rows, NULL for cross tenant post DP rows
    dp_applied    boolean NOT NULL DEFAULT false,
    pattern       text NOT NULL,
    band          text NOT NULL,
    outcome       text NOT NULL,
    model_version text NOT NULL,
    observations  int NOT NULL,
    successes     int NOT NULL,
    window_start  date NOT NULL,
    window_end    date NOT NULL,
    created_at    timestamptz NOT NULL DEFAULT now(),
    CONSTRAINT teaching_signal_aggregates_counts_valid CHECK (successes >= 0 AND observations >= successes)
);

-- One aggregate per tenant (or the nil uuid for the cross tenant cohort), pattern, band, outcome, model
-- version, and window. coalesce maps a NULL tenant_id onto a fixed nil uuid so the cross tenant rows have a
-- stable key; model_version in the key makes a forked version visible instead of silent.
CREATE UNIQUE INDEX teaching_signal_aggregates_key
    ON teaching_signal_aggregates (
        coalesce(tenant_id, '00000000-0000-0000-0000-000000000000'::uuid),
        pattern, band, outcome, model_version, window_start
    );

GRANT SELECT ON teaching_signal_aggregates TO missionmatch_app;

-- +goose Down
DROP TABLE IF EXISTS teaching_signal_aggregates;