-- organizations.sheet_url: each university's own published Google Sheet. -- -- Until now the Sheet source was a single KNOWLEDGE_JSON_URL environment -- variable — one Sheet, feeding one collection. That works for one university -- and cannot work for several: a second university's rows would land in the -- first university's collection and answer the wrong students. -- -- Storing the URL here puts it beside the rest of that organization's tenant -- configuration (name, email_domain, plan_tier), so the sync can walk the -- organizations table and give each university its own Sheet feeding its own -- collection. -- -- The URL is the Apps Script "published JSON" endpoint for that university's -- Sheet, not the Sheet's edit link — see sheet_integration/apps_script/Code.gs. -- -- Run once against the live Aurora cluster (CloudShell + psql) before -- deploying the code that reads it. Safe to re-run (IF NOT EXISTS). ALTER TABLE organizations ADD COLUMN IF NOT EXISTS sheet_url text; -- Nullable on purpose: an organization without a Sheet is a normal, supported -- state. It simply has no Sheet-sourced content — its knowledge base can still -- be filled by document upload. The sync skips organizations with no URL -- rather than treating the absence as an error. -- Partial index: the sync loop only ever asks for organizations that actually -- have a Sheet, so the index only needs to cover those rows. CREATE INDEX IF NOT EXISTS idx_organizations_sheet_url ON organizations (id) WHERE sheet_url IS NOT NULL AND is_active;