1. Removed the 45-day index-walk cap: it discarded the older part of a long
outage while still advancing source_max_date, permanently losing filings.
The walk now covers every unprocessed date (a large gap is one-time cost).
2. Discrepancy detection meets the immutability contract: it compares ALL source
snapshot fields (not five), read-only during stage/validate, reports the
differing accessions + fields in validation_json, and promote emits a warning
system event (in-transaction) — never mutating the stored row.
3. Malformed companyfacts (missing facts/units structure) are recorded separately
and FAIL validation, instead of silently degrading to skipped rows that the
50% backfill coverage floor could still pass.
Also corrected the stale "sum share classes" / DEI-only wording in the snapshot
model docstring and the A3 design doc to describe the us-gaap fallback.
Tests: +4 regressions (>45-day gap loses nothing, newly-added issuer backfills
without filing, malformed payload fails, shares discrepancy detected + evented).
23 passed, 1 skipped.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A1 carry-forward. The SEC cover-page share count
(dei:EntityCommonStockSharesOutstanding) is reported "as of" its own date, which
can differ from the fiscal period_end — store that date so market cap uses the
right point-in-time count. Migration 026 edited in place (never run with data).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
First reviewable slice of workstream A: schema only, no importers, no data.
- data_import_runs: lean batch-import audit (source/revision/status,
row_counts_json + validation_json as Text-holding-JSON per repo convention).
- fundamental_snapshots: CIK-keyed, one immutable row per accession; stores
per-period raw facts (duration = cumulative YTD/FY, balance-sheet =
period-end) plus period_start/period_end/fiscal_year/fiscal_period so
discrete quarters, Q4, TTM and YoY are derived at read time.
- earnings_events: Dolt-sourced calendar + surprise history, unique
(ticker_id, announce_date).
- tickers: nullable cik/sic/sic_description — the ticker<->issuer join point.
fundamental_data is left untouched (cutover gated separately at A5). Models
registered in app/models/__init__.py; Ticker gains an earnings_events
relationship. Verified: create_all builds the tables, mappers configure, and
migration 026 renders valid Postgres DDL up and down.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
dennisthiessenandClaude Opus 4.8