docs: finalize BuscarMisas network importer spec (all review issues resolved)

docs/superpowers/specs/2026-03-16-buscarmisas-network-importer-design.md

@@ -23,6 +23,26 @@ Add a single importer `scripts/import-buscarmisas-network.ts` that scrapes churc
 
 ---
 
+## Schema Migration (prerequisite)
+
+A new column must be added in **BethelGuide** (schema source of truth) before implementation:
+
+```prisma
+buscarmisasNetworkId  String?  @unique  @map("buscamissas_network_id")
+@@index([buscarmisasNetworkId])
+```
+
+After merging the migration in BethelGuide, copy the updated `schema.prisma` to ScraperControl and run `npx prisma generate`.
+
+The external ID format is `{domain-slug}/{church-slug}`, e.g.:
+`horariosmissa-com-br/paroquia-nossa-senhora-dos-remedios`
+
+where `domain-slug` replaces `.` with `-`, and `church-slug` is the final path segment of the church URL.
+
+`church-matcher.ts` and the `ExistingChurch` / `ChurchCandidate` interfaces must be updated to include `buscarmisasNetworkId` alongside the existing external ID fields, with a corresponding ID-match pass in `findDuplicateChurch()`.
+
+---
+
 ## Architecture
 
 ### Config Map
@@ -49,17 +69,19 @@ const NETWORK_SITES: Record<string, SiteConfig> = {
 # Single domain
 npx tsx scripts/import-buscarmisas-network.ts --domain horariosmissa.com.br
 
-# All domains sequentially
-npx tsx scripts/import-buscarmisas-network.ts --all
-
-# Resume after interruption
+# Single domain with resume (--resume-from only valid with --domain)
 npx tsx scripts/import-buscarmisas-network.ts --domain horariosmissa.com.br --resume-from 1200
 
+# All domains sequentially (no --resume-from; use --domain for resuming individual runs)
+npx tsx scripts/import-buscarmisas-network.ts --all
+
 # Dry run (no DB writes)
 npx tsx scripts/import-buscarmisas-network.ts --domain horariosmissa.com.br --dry-run
 ```
 
-Source slug stored in DB: domain with dots replaced by dashes, e.g. `horariosmissa-com-br`.
+**Validation:** If `--domain` is provided but not present in `NETWORK_SITES`, exit immediately with a clear error message listing valid domains. `--resume-from` combined with `--all` is also an error — exit with usage message.
+
+Source slug stored in DB `source` field: `buscarmisas-network` (same value for all domains — the `buscarmisasNetworkId` distinguishes per-church).
 
 ---
 
@@ -69,9 +91,9 @@ Source slug stored in DB: domain with dots replaced by dashes, e.g. `horariosmis
 
 - Fetch `https://{domain}/sitemap_index.xml`
 - Extract child sitemap URLs
-- For `sitemapType: 'page'`: collect all `page-sitemap*.xml` URLs
-- For `sitemapType: 'post'`: collect `post-sitemap.xml`
-- Fetch each child sitemap, filter to 3-segment church URLs (path depth = `/{region}/{city}/{church-slug}/`)
+- For `sitemapType: 'page'`: collect all `page-sitemap*.xml` URLs (ignore `post-sitemap*.xml` and `page-sitemap.xml` city-only entries for Chile)
+- For `sitemapType: 'post'`: collect `post-sitemap.xml` only
+- Fetch each child sitemap, filter to 3-segment church URLs (path segments = `/{region}/{city}/{church-slug}/`)
 - Collect deduplicated list of church page URLs
 
 ### 2. Church Page Parsing
@@ -83,49 +105,76 @@ For each church URL, fetch the HTML and extract:
 | Name | Table cell after `<strong>Nome</strong>` (PT) or `<strong>Nombre</strong>` (ES) |
 | Address | Table cell after `<strong>Endereço</strong>` (PT) or `<strong>Dirección</strong>` (ES) |
 | Phone | `href="tel:..."` anchor |
-| Latitude/Longitude | Google Maps iframe `src` — `center={lat}%2C{lng}` parameter |
+| Latitude/Longitude | Google Maps iframe `src` — `center={lat}%2C{lng}` parameter (confirmed present on all 5 network sites; same API key `AIzaSyCNTEOso0tZG6YMSJFoaJEY5Th1stEWrJI` used across the network) |
 | Country | From `SiteConfig.country` |
-| State/Region | 1st path segment of church URL |
-| City | 2nd path segment of church URL |
+| State/Region | 1st path segment of church URL (URL-decoded) |
+| City | 2nd path segment of church URL (URL-decoded) |
 | Mass schedule | Mon–Sun table rows: day name → time string (skip `-` entries) |
+| External ID | `{domain-slug}/{church-slug}` |
+
+If `center=` is absent from the iframe src, skip the church with a warning log. Do not fall back to the `q=` parameter (it contains a search query, not coordinates).
 
 ### 3. Schedule Parsing
 
-- Day names resolved via existing `src/scrapers/i18n/day-names.ts` patterns
-- Portuguese: `Segunda-feira`, `Terça-feira`, `Quarta-feira`, `Quinta-feira`, `Sexta-feira`, `Sábado`, `Domingo`
-- Spanish: `Lunes`, `Martes`, `Miércoles`, `Jueves`, `Viernes`, `Sábado`, `Domingo`
-- Times are comma-separated within a single cell (e.g. `10:00, 18:00`) — split and create one schedule entry per time
+- Use `getDayNamesForCountry(config.country)` from `src/scrapers/i18n/day-names.ts` to get the day-name map keyed by country code (`'BR'`, `'MX'`, etc.)
+- Build patterns with `buildDayPatterns(dayNames)` and match against the table's day-name cell text
+- Times are comma-separated within a single cell (e.g. `10:00, 18:00`) — split on `,` and create one schedule entry per time
 - `-` entries indicate no mass that day — skip
 
 ### 4. Upsert to Database
 
-- Use `findDuplicateChurch` (existing church-matcher) to check for duplicates before insert
-- Upsert church with `source = {domain-slug}`
+- Load all existing `buscarmisasNetworkId` values from DB into a `Set` at startup — skip already-imported churches (same pattern as `import-discovermass.ts`)
+- Use `findDuplicateChurch()` for new churches to detect cross-source duplicates
+- Upsert church with `source = 'buscarmisas-network'` and `buscarmisasNetworkId = externalId`
 - Upsert mass schedules linked to church
-- Track progress: log every 100 churches, support `--resume-from {index}`
+- Log progress every 100 churches; support `--resume-from {index}` (single-domain mode only)
 
 ### 5. Rate Limiting
 
-- 2-second delay between church page fetches (same as `import-discovermass.ts`)
+- 2-second delay between church page fetches
 - 5-second delay between domains when running `--all`
-- Respect HTTP 429 / 503 with exponential backoff (up to 3 retries)
+- On HTTP 429 or 503: exponential backoff, up to 3 retries, then skip with warning
 
 ---
 
 ## Error Handling
 
-- Skip churches where lat/lng cannot be extracted (log warning, continue)
-- Skip churches where name is empty
-- On fetch error: log and continue to next URL (don't abort the run)
+- Skip churches where `center=` lat/lng is absent (log warning, continue)
+- Skip churches where name is empty after parsing
+- On fetch error: log and continue to next URL
 - On DB error: log and continue
 
 ---
 
 ## Integration
 
-- Add `import:buscarmisas-network` script to `package.json`
-- Add to scheduler pipeline alongside other importers
-- No new dependencies required
+### package.json
+
+```json
+"import:buscarmisas-network": "tsx scripts/import-buscarmisas-network.ts"
+```
+
+### Scheduler (`scripts/scheduler.ts`)
+
+Add one `PipelinePhase` per domain (5 total) so each country can be scheduled and monitored independently. Each phase's `type` string must match exactly between `PIPELINE_GROUPS` and the `case` label in `getJobCommand()` — mismatch silently throws "Unknown job type". The `type` field in the DB `BackgroundJob` model is a plain `String`, consistent with existing values like `'discovermass-import'`.
+
+All 5 phases and their corresponding `case` blocks:
+
+| Phase `type` | `--domain` |
+|---|---|
+| `buscarmisas-network-BR` | `horariosmissa.com.br` |
+| `buscarmisas-network-MX` | `buscarmisas.com.mx` |
+| `buscarmisas-network-AR` | `horariosmisa.com.ar` |
+| `buscarmisas-network-CO` | `buscarmisas.co` |
+| `buscarmisas-network-CL` | `horariomisa.cl` |
+
+```ts
+case 'buscarmisas-network-BR':
+  return `npx tsx scripts/import-buscarmisas-network.ts --domain horariosmissa.com.br`;
+case 'buscarmisas-network-MX':
+  return `npx tsx scripts/import-buscarmisas-network.ts --domain buscarmisas.com.mx`;
+// ... etc for AR, CO, CL
+```
 
 ---