coverage

Freshness AND honesty per dataset. as_of = the newest period that has landed and has actually ELAPSED (bounded by today); complete_through = the newest COMPLETE period, which is what default windows serve; newest_period = the newest period KEY present, which can sit in the future when a family files forward-dated (FIP is keyed 31-December); landed_at = when ingest last SUCCEEDED for that source, from cvm_ingest_log (status ok with a finish time, so a later failed run never advances it); landed_git_sha = the git commit of THAT run — which code produced this data — NULL when the run recorded none (before migration 44, or run outside GitHub Actions), never borrowed from an older run. funds_<family> rows report each filing cadence separately. notes carries a caveat the dates cannot: the quotes row states where the cash tape starts (quote_history refuses a window before an instrument's first session); the funds_fidc row states the 2020-11 delinquency regime break (null for a fund with no tab VI row before, filed on every row after — never read that null as zero); funds_fip states why its newest_period runs ahead; fund_nav points at catalog().applicability and api.metric_coverage(); the fidc_tranches and fidc_aging rows state that those informe tabs come from CVM's yearly HIST archive through 2024-12 and the monthly informe from 2025-01; the fnet_documents row (the FNET register behind fund_documents and fund_restatements) is keyed on the DELIVERY day, with complete_through the day before as_of, and states that its history begins at first capture / backfill and that fund links come from a fortnightly sweep, so recent documents may have no cnpj yet; the company_events row (IPE filings, keyed on the delivery date, complete_through NULL as on financials) states that history starts in 2015 and that filings CVM published without a protocol number are not held; the macro_series and ptax rows carry their units and cadences (SELIC_META is published ahead, so its newest_period can sit in the future); the di_futures and reference_curves rows (B3 Price Report DI1 contracts from 2018, B3 reference curves from 2008) state where each history starts and that the long curve vertices are B3's extrapolation, not prices; and the five B3 lending / flow rows (short_interest, short_interest_by_sector, lending_trades, lending_participants, investor_flow) state the RATCHET — B3 keeps ~21 business days and publishes no archive, so their span starts at first capture and no backfill exists — along with the float_basis, brokerage-not-owner and first-difference traps that make those series easy to read wrongly. Their landed_at is split by ingest doc_type, so a COTAHIST run never reports as the lending group's freshness.

POST /rpc/coverage

coverage

cURL

curl --request POST \
  --url https://zcjbtpxuhdekpwcxmepn.supabase.co/rest/v1/rpc/coverage \
  --header 'Content-Type: application/json' \
  --header 'apikey: <api-key>' \
  --data '{}'

200

[
  {
    "dataset": "<string>",
    "as_of": "2023-12-25",
    "complete_through": "2023-12-25",
    "source": "<string>",
    "notes": "<string>",
    "newest_period": "2023-12-25",
    "landed_at": "2023-11-07T05:31:56Z",
    "landed_git_sha": "<string>"
  }
]

Authorizations

apikey · string · header · required

The publishable key (sb_publishable_…), on every request. It is NOT a JWT — sending it as Authorization: Bearer is rejected as an invalid JWT. It identifies the project, not you, and maps to the Postgres anon role.

Body

application/json

This function takes no arguments; send {}.

Response

200

application/json

The result set, oldest first. An unknown id is an empty array, not a 404 and never a guessed value.

dataset · string | null

as_of · string<date> | null

complete_through · string<date> | null

source · string | null

notes · string | null

newest_period · string<date> | null

landed_at · string<date-time> | null

landed_git_sha · string | null