# ASF other registers — handoff

State at 2026-05-11:
- `asf.entitati`: 849 entities (61 asigurator + 788 broker) — only the
  `/scr/ra` insurance registry is ingested.
- ASF has additional registries (private pensions, capital markets,
  secondary intermediaries, software providers, lecturers, etc.) at
  separate pages — NOT exposed via the same `/scr/ra/cautare` JSON endpoint.

## Why deferred

Each register appears to use a different access pattern:
- `/scr/ra` (used by current scraper) — only insurance + brokers.
- Pension funds (Pilonul II/III) — no `/scr/` endpoint visible. Likely PDF
  or static HTML on `asfromania.ro/ro/a/2365/...`.
- Capital markets entities — likely a different `/scr/...` path needs to
  be discovered via browser-network-tab inspection.

Confirmation needed via interactive exploration (curl with realistic
Referer + Cookie, or browser dev-tools). Cannot be done blindly from
high-level webpages.

## Registries discovered (from `/ro/a/1544/registre-entitati-autorizate`)

### Insurance (Asigurări)
- ✅ `/scr/ra/cautare` — currently scraped (asigurator + broker).
- ❓ `/ro/a/2082/registrul-asigurătorilor-și-intermediarilor-din-see` —
  EEA insurers and intermediaries (likely overlap with main register).
- ❓ `/app.php/ro/a/1704/intermediari-secundari` — secondary intermediaries
  (post-2019).
- ❓ `/ro/a/1997/intermediari-secundari---persoane-fizice` (pre-2019).
- ❓ `/ro/a/1998/intermediari-secundari---persoane-juridice` (pre-2019).
- ❓ `/ro/a/1999/specialisti-constatare-daune` — damage assessors.
- ❓ `/ro/a/2068/registrul-furnizorilor-de-programe-(activi)` — software
  providers.
- ❓ `/ro/a/2067/registrul-lectorilor` — authorized lecturers.

### Capital Markets (Piață de capital)
- ❓ `/app.php/ro/a/1705/registrul-instrumentelor-si-investitiilor-financiare`

### Private Pensions (Pensii private)
- ❓ `/ro/a/2365/registrul-entitatilor-din-piata-pensiilor-private` — Pilonul
  II + III administrators (SAFI), pension funds, fund managers.

## Recommended approach (~4-6h)

1. **Discovery phase (1h)**: open each `?` URL in browser, inspect Network
   tab for actual data endpoints. Note: most are likely Drupal/Symfony
   pages serving an embedded JSON or rendering an HTML table. Some may
   only offer PDF download (need OCR/parsing).
2. **Per-register scraper (1-2h each)**:
   - If it's a JSON endpoint similar to `/scr/ra/cautare`, clone the
     scrape-asf.ts pattern with a new `register_type` value
     (e.g., `pensie_administrator`, `intermediar_secundar`).
   - If it's an HTML table, parse with cheerio.
   - If it's a PDF, use pdftotext like CNAS.
3. **Schema**: `asf.entitati.register_type` is already a text column —
   add new enum-like values without DDL.
4. **Volume estimate**:
   - Pension funds: ~10 administrators (SAFI/SIF), ~20 funds.
   - Capital markets: ~50-200 entities.
   - Secondary intermediaries: ~3,000-10,000 individuals + firms.
   - Lecturers: ~50.
   - **Total ~3,500-10,300 new entities** if all done.

## Defer reason

Multi-day discovery + per-register scraper development. The 2-3h
single-candidate budget cannot accommodate even one full register
implementation without first doing the discovery for all of them.

Recommended next sub-agent: pick **secondary intermediaries** (largest
volume → 3-10k entities) as the first target, since the data shape
should mirror existing broker entries.