# BPHEN LAB v3.2.3 Master Copy Checkpoint: 2026-10-28 (end of day)

Authorized by JU on 2026-10-28 after the DIS audit and fixes. This is a checkpoint snapshot of the live state.

## What is in this checkpoint
- Site worker (bphen-archive-supabase, lab.bphen.us): `versions/BPHEN_LAB_v3.2.3_checkpoint_20261028.js` (9,298 lines, live version 536c09c0-edd5-488e-9a43-2c10b0578ce9). SHA-256: 6c08c90b478884efa23ad00b9025ff8ab8cdb338e5cd2d2e7b8fed17e340b259
- DIS Macro tracker, image v14: `trackers/dis-macro-tracker-v14/`
- DIS Memorabilia tracker, image v8: `trackers/memorabilia-tracker-v8/`
- Front-end files as served by the Pages origin: unchanged (`frontend-v3.2.0/`).
- Restart tokens in the tracker worker.ts files are redacted. No keys are stored here.

## Changes since the v3.2.2 checkpoint (earlier on 2026-10-28)
1. Site worker: DIS Micro renamed to DIS Redress in the front end and back end; BPHEN JA INDEX renamed BPHEN PENS INDEX; source-type filter is a plain type name; retired report header and primary-objective code removed; copyright line pure white; paragraph spacing inside quoted blocks in the Analytical Eras reports; desk install starts sooner.
2. DIS Memorabilia tracker v8: formatter fixes for "1920s" and "(O.S.)", peerage tails, repeated surnames, junk catalog fragments, quote and bracket imbalance in descriptions; writes an `origin` tag on every new row.
3. DIS Macro tracker v14: cleaner fixes encoding, removes publisher footers and banners (including Truthout) and repeated teaser paragraphs anywhere in a transcript, handles headline tails such as "RFK Jr."; byline extractor strips glued dates, keeps initials, shortens long name lists, reads ProPublica bylines.
4. Memorabilia data: `origin` tag on all 206 rows (manual_upload 14, dis_collected 192). 32 rows corrected against the PDFs and catalog records, including Jarena Lee (Cincinnati 1839), Black Fugitives in Colonial South Carolina (Shane White, 1980) and Life and Times of Frederick Douglass (Boston 1892).
5. Macro data: transcripts cleaned (boilerplate, related-link lines, banners, cut-offs finished from the source pages), 35 missing bylines filled, 4 Guardian author lists shortened, 3 duplicates flagged hidden. Nothing was deleted.
6. Standing check added: `dis-checks/standing_check.sh` (read-only; Memorabilia audit, Macro audit, site text-engine test).

## Verification at checkpoint
- Disk worker equals the deployed worker 536c09c0; `node --check` passes.
- DIS Macro v14 /healthz ok (audit loop and scanner running). DIS Memorabilia v8 /healthz ok.
- Supabase archive_items: 1,717 rows (1,495 LIVE, 206 HISTORICAL, 16 DIS_REDRESS). 119 Macro rows flagged as duplicates, hidden. Titles, links and hashes unchanged by the audit fixes.
- Site text engine: 1,701 records, 0 errors, 0 empty. Live site checked signed in: Macro 1,376 visible, Memorabilia 206.

## Revert
The previous live worker for the site was version 8d884798-1571-4784-b370-eeea1ec20a63 (v3.2.2 checkpoint). Trackers: macro previous image v13 (settings saved in backups_AL6/macro_wrangler_v13.toml), memorabilia previous image v7 (backups_AL6/memo_wrangler_v7.toml). Pre-change data rows are in backups_AL6/.