@mailwoman/corpus 10.0.0 → 10.1.0

package/lib/{recipes/index.ts → recipes.ts}

@@ -3,9 +3,11 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  *
- *   The synthetic-corpus SLICE RECIPE registry. Each recipe ({@link CorpusRecipe}) is what one legacy
- *   `build-*-slice.mjs` script used to be; the `mailwoman corpus slice <recipe>` command dispatches
- *   to them. A new slice is a new entry here, not a new script.
+ *   The recipe registry. A recipe derives rows from attested data; `recipes/sources.ts` records which
+ *   operation each one performs, and only the `invented-*` recipes compose a row with no real record
+ *   behind it. Each recipe ({@link CorpusRecipe}) is what one root-level
+ *   build script used to be. the `mailwoman corpus slice <recipe>` command dispatches to them. A new
+ *   recipe is a new entry here rather than a new script.
  */
 
 import { cnOrganizationalUnitsRecipe } from "#cn/recipes/organizational-units"
@@ -19,22 +21,22 @@ import { frOrderRecipe } from "#fr/recipes/order"
 import { localeRecipe } from "#international/recipes/locale"
 import { nlPostcodeRecipe } from "#nl/recipes/postcode"
 import { noFragmentRecipe } from "#no/recipes/fragment"
-import { noStreetRecipe } from "#no/recipes/street/index"
+import { noStreetRecipe } from "#no/recipes/street"
 import { noStreetLedRecipe } from "#no/recipes/street/led"
 import { anchorAbsorptionRecipe } from "#recipes/anchor-absorption"
 import { bareCountryRecipe } from "#recipes/bare/country"
-import { barePostcodeRecipe } from "#recipes/bare/postcode/index"
+import { barePostcodeRecipe } from "#recipes/bare/postcode"
 import { boundaryStressRecipe } from "#recipes/boundary-stress"
 import { countryBalancedRecipe } from "#recipes/country-balanced"
 import { houseVenueRecipe } from "#recipes/house-venue"
 import { intersectionRecipe } from "#recipes/intersection"
+import { poBoxRecipe } from "#recipes/po/box"
 import { poBoxCedexRecipe } from "#recipes/po/box/cedex"
-import { poBoxRecipe } from "#recipes/po/box/index"
 import type { CorpusRecipe } from "#recipes/scaffold"
 import { streetRecipe } from "#recipes/street"
 import { streetAffixRecipe, suffixBoundaryRecipe } from "#recipes/street/affix"
 import { streetBareRecipe } from "#recipes/street/bare"
-import { subVenueRecipe } from "#recipes/sub/venue/index"
+import { subVenueRecipe } from "#recipes/sub/venue"
 import { trailingRegionRecipe } from "#recipes/trailing-region"
 import { unitRecipe } from "#recipes/unit"
 import { sgRegisterRecipe } from "#sg/recipes/register"
@@ -87,18 +89,18 @@ const RECIPES: readonly CorpusRecipe[] = [
 /**
  * Recipe name → recipe.
  */
-export const SLICE_RECIPES: ReadonlyMap<string, CorpusRecipe> = new Map(RECIPES.map((r) => [r.name, r]))
+export const RECIPES_BY_NAME: ReadonlyMap<string, CorpusRecipe> = new Map(RECIPES.map((r) => [r.name, r]))
 
 /**
  * Look up a recipe by its `<recipe>` name.
  */
-export function getSliceRecipe(name: string): CorpusRecipe | undefined {
-	return SLICE_RECIPES.get(name)
+export function getRecipe(name: string): CorpusRecipe | undefined {
+	return RECIPES_BY_NAME.get(name)
 }
 
 /**
  * All recipes (for `--list` / help).
  */
-export function listSliceRecipes(): readonly CorpusRecipe[] {
+export function listRecipes(): readonly CorpusRecipe[] {
 	return RECIPES
 }

package/lib/registers.ts

@@ -0,0 +1,278 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Stable ids for the publications that corpus rows come from.
+ */
+
+/**
+ * The publications this repository reads, keyed by stable id.
+ *
+ * A row's `source` identifies the adapter or recipe that emitted it.
+ * A row's `register` identifies the publication behind its record.
+ * That publication determines the governing terms.
+ *
+ * Built corpora store these ids on every row, so an existing id must never change.
+ *
+ * An aggregator such as OpenAddresses or Overture is recorded as itself,
+ * because its per-record upstream and terms vary.
+ */
+export const SourceRegister = {
+	/**
+	 * OpenAddresses, an aggregator over national and municipal address registers.
+	 */
+	OpenAddresses: "openaddresses",
+	/**
+	 * Overture Maps addresses, an aggregator whose rows include a per-record source and license.
+	 */
+	Overture: "overture",
+	/**
+	 * Who's On First, the gazetteer of places and postal codes.
+	 */
+	WhosOnFirst: "whos-on-first",
+	/**
+	 * OpenStreetMap.
+	 */
+	OpenStreetMap: "openstreetmap",
+	/**
+	 * GeoNames, the place-name gazetteer.
+	 */
+	GeoNames: "geonames",
+	/**
+	 * GeoNames postal-code tables, published separately from the place gazetteer.
+	 */
+	GeoNamesPostal: "geonames-postal",
+	/**
+	 * France's Base Adresse Nationale.
+	 */
+	BaseAdresseNationale: "fr-ban",
+	/**
+	 * Kartverket's `Matrikkelen - Adresse`, the Norwegian cadastre's address table.
+	 *
+	 * The prefix is the publisher's own country, as `fr-ban`'s is, and the dataset
+	 * covers Svalbard beside the mainland.
+	 */
+	MatrikkelenAdresse: "no-matrikkelen-adresse",
+	/**
+	 * The INSPIRE Addresses theme of the Czech Office for Surveying, Mapping and Cadastre.
+	 *
+	 * ČÚZK publishes one zipped GML per municipality, 6,258 of them, through an ATOM service.
+	 */
+	CuzkInspireAddresses: "cz-cuzk-inspire-addresses",
+	/**
+	 * Suomen ympäristökeskus' Ryhti built-environment system, whose building address data
+	 * covers Åland's sixteen municipalities beside the Finnish mainland in one national file.
+	 *
+	 * The prefix is the publisher's own country, as `fr-ban`'s is.
+	 */
+	RyhtiBuildingAddress: "fi-ryhti-building-address",
+	/**
+	 * Klimadatastyrelsen's INSPIRE Addresses theme for Denmark, published as a GeoPackage.
+	 *
+	 * One address is a join across five tables rather than a row.
+	 */
+	DKInspireAddresses: "dk-inspire-addresses",
+	/**
+	 * Estonia's Aadressiandmete süsteem, served as the INSPIRE Addresses theme by Maa- ja Ruumiamet.
+	 *
+	 * The tail is the system the theme is served out of.
+	 */
+	EstoniaADS: "ee-ads-aadress",
+	/**
+	 * The municipal EMUiA address registers, reaching the INSPIRE Addresses theme through
+	 * the Państwowy Rejestr Granic and served by Główny Urząd Geodezji i Kartografii.
+	 */
+	PolandEMUiA: "pl-emuia-adresy",
+	/**
+	 * The INSPIRE Addresses theme of agentschap Digitaal Vlaanderen.
+	 *
+	 * Coverage is the Flemish Region rather than Belgium.
+	 * Brussels and Wallonia publish the theme through their own services,
+	 * so a Belgian build reads three registers.
+	 */
+	VlaanderenInspireAddresses: "be-vlaanderen-inspire-ad",
+	/**
+	 * The INSPIRE Addresses theme of the Service public de Wallonie, out of its ICAR register.
+	 *
+	 * Coverage is the Walloon Region.
+	 * Flanders and Brussels publish the theme through their own services,
+	 * so a Belgian build reads three registers.
+	 */
+	WallonieInspireAddresses: "be-wallonie-inspire-ad",
+	/**
+	 * The INSPIRE Addresses theme of the Brussels Regional Informatics Center, out of UrbIS.
+	 *
+	 * Coverage is the Brussels-Capital Region.
+	 * The address-source register holds no row for this publisher, so no license is elected
+	 * and `createIneligibilityReader` refuses every row the adapter emits.
+	 *
+	 * The id exists so the adapter can name its publication once a row is written.
+	 */
+	BrusselsInspireAddresses: "be-brussels-inspire-ad",
+	/**
+	 * The INSPIRE Addresses theme of Kadaster, the Dutch cadastre.
+	 *
+	 * One ATOM entry serves the whole country as a single gzipped GML.
+	 */
+	KadasterInspireAddresses: "nl-kadaster-inspire-ad",
+	/**
+	 * The INSPIRE Addresses theme the Ministerstvo vnútra Slovenskej republiky serves
+	 * out of the national address register, as a paged WFS 2.0.
+	 */
+	SlovakiaInspireAddresses: "sk-minv-inspire-ad",
+	/**
+	 * The INSPIRE Addresses theme of the Dirección General del Catastro.
+	 *
+	 * Coverage is Spain outside the foral cadastres: 52 of the 55 entries in the national
+	 * ATOM feed, one province feed each, listing one zipped GML per municipality.
+	 * The other three entries link the publications below.
+	 *
+	 * Each of those states its own terms.
+	 * Spain therefore needs four registers.
+	 */
+	CatastroInspireAddresses: "es-catastro-inspire-ad",
+	/**
+	 * The INSPIRE Addresses theme of the Diputación Foral de Bizkaia.
+	 *
+	 * Coverage is the province of Bizkaia, as 112 municipality archives.
+	 * The archives carry addresses alone, and the features they reference are served
+	 * by the publisher's WFS rather than by the ATOM feed.
+	 */
+	BizkaiaInspireAddresses: "es-bizkaia-inspire-ad",
+	/**
+	 * The INSPIRE Addresses theme of the Diputación Foral de Gipuzkoa.
+	 *
+	 * One ATOM entry serves the whole province as a single zipped GML.
+	 */
+	GipuzkoaInspireAddresses: "es-gipuzkoa-inspire-ad",
+	/**
+	 * The INSPIRE Addresses theme of the Gobierno de Navarra.
+	 *
+	 * The province is split across 272 zipped GML partitions, each titled `Address Navarra`
+	 * rather than named for a municipality.
+	 */
+	NavarraInspireAddresses: "es-navarra-inspire-ad",
+	/**
+	 * BD-Adresses, the georeferenced address file the Administration du cadastre et de
+	 * la topographie publishes for Luxembourg through `data.public.lu`.
+	 *
+	 * The publisher describes it as a subset of the national register of localities and streets,
+	 * enriched with coordinates, so the publication is narrower than that register.
+	 */
+	BDAdresses: "lu-act-bd-adresses",
+	/**
+	 * The OCDS release Autorità Nazionale Anticorruzione publishes for Italy's National Database
+	 * of Public Contracts, reached through the Open Contracting Partnership's Data Registry.
+	 *
+	 * Every address in the release belongs to a contracting authority or a paying office,
+	 * because the `supplier` role carries no address object.
+	 */
+	ANACContracts: "it-anac-ocds-contracts",
+	/**
+	 * Romania's school network, the Ministry of Education's list of every education unit and its address.
+	 */
+	ReteaScolara: "ro-edu-retea-scolara",
+	/**
+	 * FINESS, the health ministry's register of health and social establishments,
+	 * read from its establishment extract on data.gouv.fr.
+	 */
+	FINESS: "fr-finess",
+	/**
+	 * The education ministry's Annuaire de l'éducation, the directory of schools and education offices.
+	 */
+	AnnuaireEducation: "fr-annuaire-education",
+	/**
+	 * The CORDIS organization files for the participants in the EU's research framework programmes
+	 * (FP7, Horizon 2020, Horizon Europe), published by the Publications Office of the European Union.
+	 *
+	 * Each address is the one a participating legal entity registered with the participant register.
+	 */
+	CORDISParticipants: "eu-cordis-participants",
+	/**
+	 * Latvia's State Address Register (Valsts adrešu reģistrs), the open data Valsts zemes
+	 * dienests publishes as the data.gov.lv dataset `varis-atvertie-dati`.
+	 */
+	VARIS: "lv-varis",
+	/**
+	 * GLEIF's Level 1 LEI golden copy: the legal and headquarters addresses of every LEI holder.
+	 */
+	GLEIFGoldenCopy: "gleif-lei-golden-copy",
+	/**
+	 * HM Land Registry Price Paid Data, England and Wales.
+	 */
+	LandRegistryPricePaid: "gb-hm-land-registry-ppd",
+	/**
+	 * Australia's Geocoded National Address File.
+	 */
+	GNAF: "au-gnaf",
+	/**
+	 * The United States Census Bureau's TIGER/Line road and address-range files.
+	 */
+	CensusTIGER: "us-census-tiger",
+	/**
+	 * The National Address Database, published by the US Department of Transportation.
+	 */
+	NationalAddressDatabase: "us-dot-nad",
+	/**
+	 * The FCC Broadband Data Collection location fabric.
+	 */
+	FCCBroadbandData: "us-fcc-bdc",
+	/**
+	 * The National Plan and Provider Enumeration System.
+	 */
+	NPPES: "us-nppes",
+	/**
+	 * HRSA's Federally Qualified Health Center file.
+	 */
+	HRSAHealthCenters: "us-hrsa-fqhc",
+	/**
+	 * The IMLS Public Libraries Survey.
+	 */
+	IMLSPublicLibraries: "us-imls-pls",
+	/**
+	 * The IRS Business Master File of tax-exempt organizations.
+	 */
+	IRSBusinessMasterFile: "us-irs-bmf",
+	/**
+	 * SAMHSA's behavioral health treatment services locator.
+	 */
+	SAMHSATreatmentLocator: "us-samhsa-treatment-locator",
+	/**
+	 * Hawaii Department of Education's school directory.
+	 */
+	HawaiiSchools: "us-hi-doe-schools",
+	/**
+	 * New York State's notary public register.
+	 */
+	NewYorkNotaries: "us-ny-notaries",
+	/**
+	 * Texas's notary public register.
+	 */
+	TexasNotaries: "us-tx-notaries",
+	/**
+	 * Iowa's Active Construction Contractor Registrations.
+	 */
+	IowaContractors: "us-ia-contractor-registrations",
+	/**
+	 * This repository's curated codex tables, such as country names, address layouts and postcode formats.
+	 */
+	Codex: "mailwoman-codex",
+	/**
+	 * Facts that a person read from a publisher and recorded here by hand.
+	 *
+	 * The reviewed file records the publisher.
+	 */
+	ReviewedByHand: "mailwoman-reviewed",
+	/**
+	 * A tuple extract under `$MAILWOMAN_DATA_ROOT/corpus/tuples/` whose upstream publication is unknown.
+	 *
+	 * A recipe whose tuples record their upstream passes that register instead.
+	 */
+	DerivedTuples: "mailwoman-derived-tuples",
+} as const
+
+/**
+ * One of the {@link SourceRegister} ids.
+ */
+export type SourceRegister = (typeof SourceRegister)[keyof typeof SourceRegister]

package/lib/ro/adapters/retea-scolara/adapter.ts

@@ -0,0 +1,264 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * `ro-retea-scolara`: Romania's school network (`Rețea școlară`), the Ministry of Education's list of every
+ * education unit for a school year, read for each unit's postal address.
+ *
+ * Input is the publisher's own workbook, one sheet named `Export`, as data.gov.ro serves it for the
+ * dataset `69392a50-5750-4a54-90c1-d461de44da6d`. Three rows precede the header. The adapter finds the
+ * header by its `Strada` cell rather than by position, because the preamble is a generation timestamp
+ * whose length the publisher does not fix.
+ *
+ * ## Which fields make the address
+ *
+ * The sheet splits the address into `Strada`, `Numar`, `Cod postal` and `Localitate unitate`, so the
+ * adapter splits no street line. Measured over the 2025–2026 edition's 18,022 units:
+ *
+ * - `Strada` is the street name as the school writes it, usually without the street-type word:
+ *   `REPUBLICII`, `PRINCIPALA`, `Piața 1848`. 198 values begin with `STR.` and 247 with `CALEA`.
+ *   The value is copied unchanged, because a street type the publisher did not write is not evidence.
+ * - `Numar` holds the house number. `FN` (`fără număr`, "without number") and `-` state that the premise
+ *   has none, on 601 and 180 units. Both are read as absent. A range such as `12-14` is a house number.
+ * - `Cod postal` holds a six-digit Romanian postcode on 17,484 units. A value of any other length is
+ *   left out of the row rather than refusing the row, because the address is complete without it.
+ * - `Judet PJ` holds the county as a two-letter code. The row carries no region, because libaddressinput's
+ *   Romanian layout places the county between postcode and locality, and this reader has no source for
+ *   the county name in the form an address prints it.
+ *
+ * `Denumire lunga unitate`, the unit's full name, is the venue.
+ *
+ * ## Organizations only
+ *
+ * Every unit is a school, kindergarten, children's club, county inspectorate or similar public body, and
+ * the sheet carries no natural person's address. The venue is the institution's name.
+ *
+ * ## Identity and license
+ *
+ * The row id is content-addressed over the aligned components. A school and its kindergarten often share
+ * one address under two names, so they are two rows.
+ *
+ * data.gov.ro publishes the dataset under `CC-BY-4.0`, the license its CKAN record names for the
+ * publisher `Ministerul Educației`. CC BY 4.0 requires credit, a link to the license and a statement that
+ * changes were made, so every row records the license and {@linkcode RO_RETEA_SCOLARA_ATTRIBUTION} is the
+ * credit the model card carries.
+ */
+
+import { formatAddressRow } from "@mailwoman/codex/address/format"
+import type { PathBuilderLike } from "path-ts"
+import { XLSXSpliterator, type XLSXCellValue } from "spliterator"
+
+import { UnsupportedCountryError } from "#adapters/errors"
+import { stableSourceID } from "#adapters/source-id"
+import { SourceRegister } from "#registers"
+import {
+	AddressRole,
+	type AdapterOptions,
+	type CanonicalRow,
+	type CorpusAdapter,
+	countDropped,
+	SurfaceOrigin,
+} from "#types"
+
+/**
+ * Registry id for this adapter, stamped into every row it emits.
+ */
+export const RO_RETEA_SCOLARA_ADAPTER_ID = "ro-retea-scolara"
+
+/**
+ * The one jurisdiction this adapter emits.
+ */
+export const RO_RETEA_SCOLARA_COUNTRIES: readonly string[] = ["RO"]
+
+/**
+ * The license data.gov.ro names for the dataset.
+ */
+export const RO_RETEA_SCOLARA_LICENSE = "CC-BY-4.0"
+
+/**
+ * The credit CC BY 4.0 obliges.
+ */
+export const RO_RETEA_SCOLARA_ATTRIBUTION = "Ministerul Educației, Rețea școlară, via data.gov.ro"
+
+/**
+ * Why a unit did not become a row.
+ */
+export const ReteaScolaraRefusal = {
+	/**
+	 * The unit states no locality.
+	 */
+	LocalityAbsent: "row:locality-absent",
+	/**
+	 * The unit states neither a street nor a house number, so the locality alone would be a place name.
+	 */
+	PremiseAbsent: "row:premise-absent",
+	/**
+	 * The components do not render as a Romanian address line.
+	 */
+	Unrenderable: "row:unrenderable",
+	/**
+	 * Another unit already produced this exact row.
+	 */
+	Duplicate: "row:duplicate",
+} as const
+
+export type ReteaScolaraRefusal = (typeof ReteaScolaraRefusal)[keyof typeof ReteaScolaraRefusal]
+
+/**
+ * One unit's fields, keyed by the sheet's header.
+ */
+export type ReteaScolaraUnit = Readonly<Record<string, XLSXCellValue | undefined>>
+
+export type ReteaScolaraReading = { readonly admitted: CanonicalRow } | { readonly refused: ReteaScolaraRefusal }
+
+const NO_NUMBER: ReadonlySet<string> = new Set(["FN", "F.N.", "F/N", "-", "0", "SN", "S/N"])
+
+const NUMBER_MARKER = /^(?:NR|NUM|NUMAR)\.?\s*/iu
+
+const POSTCODE = /^\d{6}$/u
+
+function cell(unit: ReteaScolaraUnit, header: string): string {
+	return String(unit[header] ?? "")
+		.replaceAll(/\s+/gu, " ")
+		.trim()
+}
+
+/**
+ * The house number a `Numar` cell states, or an empty string where it states none.
+ */
+export function readHouseNumber(value: string): string {
+	const stripped = value.replace(NUMBER_MARKER, "").trim()
+
+	return NO_NUMBER.has(stripped.toUpperCase()) ? "" : stripped
+}
+
+/**
+ * Reads one unit into a row or the reason it is refused.
+ */
+export function readReteaScolaraUnit(unit: ReteaScolaraUnit): ReteaScolaraReading {
+	const locality = cell(unit, "Localitate unitate")
+
+	if (!locality) return { refused: ReteaScolaraRefusal.LocalityAbsent }
+
+	const street = cell(unit, "Strada")
+	const houseNumber = readHouseNumber(cell(unit, "Numar"))
+
+	if (!street && !houseNumber) return { refused: ReteaScolaraRefusal.PremiseAbsent }
+
+	const postcode = cell(unit, "Cod postal")
+	const venue = cell(unit, "Denumire lunga unitate") || cell(unit, "Denumire scurta unitate")
+
+	const components: CanonicalRow["components"] = {}
+
+	if (venue) {
+		components.venue = venue
+	}
+
+	if (street) {
+		components.street = street
+	}
+
+	if (houseNumber) {
+		components.house_number = houseNumber
+	}
+
+	if (POSTCODE.test(postcode)) {
+		components.postcode = postcode
+	}
+
+	components.locality = locality
+
+	const rendered = formatAddressRow(components, "RO", { singleLine: true })
+
+	if (!rendered) return { refused: ReteaScolaraRefusal.Unrenderable }
+
+	return {
+		admitted: {
+			raw: rendered.raw,
+			components: rendered.components,
+			country: "RO",
+			locale: "ro-RO",
+			source: RO_RETEA_SCOLARA_ADAPTER_ID,
+			source_id: stableSourceID(RO_RETEA_SCOLARA_ADAPTER_ID, rendered.components),
+			corpus_version: "",
+			license: RO_RETEA_SCOLARA_LICENSE,
+		},
+	}
+}
+
+/**
+ * The sheet's units keyed by its header, skipping the preamble above the row that holds `Strada`.
+ */
+export async function* readReteaScolaraUnits(inputPath: PathBuilderLike): AsyncGenerator<ReteaScolaraUnit> {
+	let header: string[] | null = null
+
+	for await (const row of XLSXSpliterator.fromAsync<XLSXCellValue[]>(inputPath.toString(), {
+		mode: "array",
+		header: false,
+	})) {
+		if (!header) {
+			if (row.some((value) => value === "Strada")) {
+				header = row.map((value) => String(value ?? "").trim())
+			}
+
+			continue
+		}
+
+		if (!row.some((value) => value !== null && value !== "")) continue
+
+		yield Object.fromEntries(header.map((name, index) => [name, row[index]]))
+	}
+
+	if (!header) throw new Error(`${RO_RETEA_SCOLARA_ADAPTER_ID}: no header row with a "Strada" cell in ${inputPath}`)
+}
+
+export function createReteaScolaraAdapter(): CorpusAdapter {
+	return {
+		id: RO_RETEA_SCOLARA_ADAPTER_ID,
+		defaultLicense: RO_RETEA_SCOLARA_LICENSE,
+		addressRole: AddressRole.Facility,
+		register: SourceRegister.ReteaScolara,
+		surface: SurfaceOrigin.Rendered,
+		description: "Romania's school network (Rețea școlară): the postal address of every education unit.",
+
+		async *rows(opts: AdapterOptions): AsyncIterable<CanonicalRow> {
+			if (opts.country && !RO_RETEA_SCOLARA_COUNTRIES.includes(opts.country)) {
+				throw new UnsupportedCountryError(RO_RETEA_SCOLARA_ADAPTER_ID, RO_RETEA_SCOLARA_COUNTRIES, opts.country)
+			}
+
+			const seen = new Set<string>()
+			let emitted = 0
+
+			for await (const unit of readReteaScolaraUnits(opts.inputPath)) {
+				if (opts.signal?.aborted) break
+
+				if (opts.limit !== undefined && emitted >= opts.limit) break
+
+				const reading = readReteaScolaraUnit(unit)
+
+				if (!("admitted" in reading)) {
+					countDropped(opts, reading.refused)
+
+					continue
+				}
+
+				if (seen.has(reading.admitted.source_id)) {
+					countDropped(opts, ReteaScolaraRefusal.Duplicate)
+
+					continue
+				}
+
+				seen.add(reading.admitted.source_id)
+				yield reading.admitted
+
+				emitted++
+			}
+		},
+	}
+}
+
+/**
+ * The registered `ro-retea-scolara` adapter instance, built by {@linkcode createReteaScolaraAdapter}.
+ */
+export const reteaScolaraAdapter = createReteaScolaraAdapter()