@mailwoman/corpus 10.0.0 → 10.2.0

package/out/registers.js

@@ -0,0 +1,275 @@
+/**
+ * @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.
+     *
+     * Its 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 hold addresses only, 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` The title states no 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 holds 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",
+};
+//# sourceMappingURL=registers.js.map

package/out/registers.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"registers.js","sourceRoot":"","sources":["../lib/registers.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAEH;;;;;;;;;;;GAWG;AACH,MAAM,CAAC,MAAM,cAAc,GAAG;IAC7B;;OAEG;IACH,aAAa,EAAE,eAAe;IAC9B;;OAEG;IACH,QAAQ,EAAE,UAAU;IACpB;;OAEG;IACH,WAAW,EAAE,eAAe;IAC5B;;OAEG;IACH,aAAa,EAAE,eAAe;IAC9B;;OAEG;IACH,QAAQ,EAAE,UAAU;IACpB;;OAEG;IACH,cAAc,EAAE,iBAAiB;IACjC;;OAEG;IACH,oBAAoB,EAAE,QAAQ;IAC9B;;;;;OAKG;IACH,kBAAkB,EAAE,wBAAwB;IAC5C;;;;OAIG;IACH,oBAAoB,EAAE,2BAA2B;IACjD;;;;;;;OAOG;IACH,oBAAoB,EAAE,2BAA2B;IACjD;;;;OAIG;IACH,kBAAkB,EAAE,sBAAsB;IAC1C;;;;OAIG;IACH,UAAU,EAAE,gBAAgB;IAC5B;;;OAGG;IACH,WAAW,EAAE,iBAAiB;IAC9B;;;;;;OAMG;IACH,0BAA0B,EAAE,0BAA0B;IACtD;;;;;;OAMG;IACH,wBAAwB,EAAE,wBAAwB;IAClD;;;;;;;;OAQG;IACH,wBAAwB,EAAE,wBAAwB;IAClD;;;;OAIG;IACH,wBAAwB,EAAE,wBAAwB;IAClD;;;OAGG;IACH,wBAAwB,EAAE,oBAAoB;IAC9C;;;;;;;;;OASG;IACH,wBAAwB,EAAE,wBAAwB;IAClD;;;;;;OAMG;IACH,uBAAuB,EAAE,uBAAuB;IAChD;;;;OAIG;IACH,wBAAwB,EAAE,wBAAwB;IAClD;;;;;OAKG;IACH,uBAAuB,EAAE,uBAAuB;IAChD;;;;;;OAMG;IACH,UAAU,EAAE,oBAAoB;IAChC;;;;;;OAMG;IACH,aAAa,EAAE,wBAAwB;IACvC;;OAEG;IACH,YAAY,EAAE,sBAAsB;IACpC;;;OAGG;IACH,MAAM,EAAE,WAAW;IACnB;;OAEG;IACH,iBAAiB,EAAE,uBAAuB;IAC1C;;;;;OAKG;IACH,kBAAkB,EAAE,wBAAwB;IAC5C;;;OAGG;IACH,KAAK,EAAE,UAAU;IACjB;;OAEG;IACH,eAAe,EAAE,uBAAuB;IACxC;;OAEG;IACH,qBAAqB,EAAE,yBAAyB;IAChD;;OAEG;IACH,IAAI,EAAE,SAAS;IACf;;OAEG;IACH,WAAW,EAAE,iBAAiB;IAC9B;;OAEG;IACH,uBAAuB,EAAE,YAAY;IACrC;;OAEG;IACH,gBAAgB,EAAE,YAAY;IAC9B;;OAEG;IACH,KAAK,EAAE,UAAU;IACjB;;OAEG;IACH,iBAAiB,EAAE,cAAc;IACjC;;OAEG;IACH,mBAAmB,EAAE,aAAa;IAClC;;OAEG;IACH,qBAAqB,EAAE,YAAY;IACnC;;OAEG;IACH,sBAAsB,EAAE,6BAA6B;IACrD;;OAEG;IACH,aAAa,EAAE,mBAAmB;IAClC;;OAEG;IACH,eAAe,EAAE,gBAAgB;IACjC;;OAEG;IACH,aAAa,EAAE,gBAAgB;IAC/B;;OAEG;IACH,eAAe,EAAE,gCAAgC;IACjD;;OAEG;IACH,KAAK,EAAE,iBAAiB;IACxB;;;;OAIG;IACH,cAAc,EAAE,oBAAoB;IACpC;;;;OAIG;IACH,aAAa,EAAE,0BAA0B;CAChC,CAAA"}

package/out/ro/adapters/retea-scolara/adapter.d.ts

@@ -0,0 +1,114 @@
+/**
+ * @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 titled `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 holds 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 holds 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 records.
+ */
+import type { PathBuilderLike } from "path-ts";
+import { type XLSXCellValue } from "spliterator";
+import { type CanonicalRow, type CorpusAdapter } from "#types";
+/**
+ * Registry id for this adapter, stamped into every row it emits.
+ */
+export declare const RO_RETEA_SCOLARA_ADAPTER_ID = "ro-retea-scolara";
+/**
+ * The one jurisdiction this adapter emits.
+ */
+export declare const RO_RETEA_SCOLARA_COUNTRIES: readonly string[];
+/**
+ * The license data.gov.ro names for the dataset.
+ */
+export declare const RO_RETEA_SCOLARA_LICENSE = "CC-BY-4.0";
+/**
+ * The credit CC BY 4.0 obliges.
+ */
+export declare const RO_RETEA_SCOLARA_ATTRIBUTION = "Ministerul Educa\u021Biei, Re\u021Bea \u0219colar\u0103, via data.gov.ro";
+/**
+ * Why a unit did not become a row.
+ */
+export declare const ReteaScolaraRefusal: {
+    /**
+     * The unit states no locality.
+     */
+    readonly LocalityAbsent: "row:locality-absent";
+    /**
+     * The unit states neither a street nor a house number, so the locality by itself would be a place name.
+     */
+    readonly PremiseAbsent: "row:premise-absent";
+    /**
+     * The components do not render as a Romanian address line.
+     */
+    readonly Unrenderable: "row:unrenderable";
+    /**
+     * Another unit already produced this exact row.
+     */
+    readonly Duplicate: "row:duplicate";
+};
+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;
+};
+/**
+ * The house number a `Numar` cell states, or an empty string where it states none.
+ */
+export declare function readHouseNumber(value: string): string;
+/**
+ * Reads one unit into a row or the reason it is refused.
+ */
+export declare function readReteaScolaraUnit(unit: ReteaScolaraUnit): ReteaScolaraReading;
+/**
+ * The sheet's units keyed by its header, skipping the preamble above the row that holds `Strada`.
+ */
+export declare function readReteaScolaraUnits(inputPath: PathBuilderLike): AsyncGenerator<ReteaScolaraUnit>;
+export declare function createReteaScolaraAdapter(): CorpusAdapter;
+/**
+ * The registered `ro-retea-scolara` adapter instance, built by {@linkcode createReteaScolaraAdapter}.
+ */
+export declare const reteaScolaraAdapter: CorpusAdapter;
+//# sourceMappingURL=adapter.d.ts.map

package/out/ro/adapters/retea-scolara/adapter.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../lib/ro/adapters/retea-scolara/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA6CG;AAGH,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,SAAS,CAAA;AAC9C,OAAO,EAAmB,KAAK,aAAa,EAAE,MAAM,aAAa,CAAA;AAKjE,OAAO,EAGN,KAAK,YAAY,EACjB,KAAK,aAAa,EAGlB,MAAM,QAAQ,CAAA;AAEf;;GAEG;AACH,eAAO,MAAM,2BAA2B,qBAAqB,CAAA;AAE7D;;GAEG;AACH,eAAO,MAAM,0BAA0B,EAAE,SAAS,MAAM,EAAW,CAAA;AAEnE;;GAEG;AACH,eAAO,MAAM,wBAAwB,cAAc,CAAA;AAEnD;;GAEG;AACH,eAAO,MAAM,4BAA4B,6EAAyD,CAAA;AAElG;;GAEG;AACH,eAAO,MAAM,mBAAmB;IAC/B;;OAEG;aACH,cAAc,EAAE,qBAAqB;IACrC;;OAEG;aACH,aAAa,EAAE,oBAAoB;IACnC;;OAEG;aACH,YAAY,EAAE,kBAAkB;IAChC;;OAEG;aACH,SAAS,EAAE,eAAe;CACjB,CAAA;AAEV,MAAM,MAAM,mBAAmB,GAAG,CAAC,OAAO,mBAAmB,CAAC,CAAC,MAAM,OAAO,mBAAmB,CAAC,CAAA;AAEhG;;GAEG;AACH,MAAM,MAAM,gBAAgB,GAAG,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,aAAa,GAAG,SAAS,CAAC,CAAC,CAAA;AAElF,MAAM,MAAM,mBAAmB,GAAG;IAAE,QAAQ,CAAC,QAAQ,EAAE,YAAY,CAAA;CAAE,GAAG;IAAE,QAAQ,CAAC,OAAO,EAAE,mBAAmB,CAAA;CAAE,CAAA;AAcjH;;GAEG;AACH,wBAAgB,eAAe,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAIrD;AAED;;GAEG;AACH,wBAAgB,oBAAoB,CAAC,IAAI,EAAE,gBAAgB,GAAG,mBAAmB,CAiDhF;AAED;;GAEG;AACH,wBAAuB,qBAAqB,CAAC,SAAS,EAAE,eAAe,GAAG,cAAc,CAAC,gBAAgB,CAAC,CAqBzG;AAED,wBAAgB,yBAAyB,IAAI,aAAa,CA2CzD;AAED;;GAEG;AACH,eAAO,MAAM,mBAAmB,eAA8B,CAAA"}

package/out/ro/adapters/retea-scolara/adapter.js

@@ -0,0 +1,209 @@
+/**
+ * @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 titled `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 holds 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 holds 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 records.
+ */
+import { formatAddressRow } from "@mailwoman/codex/address/format";
+import { XLSXSpliterator } from "spliterator";
+import { UnsupportedCountryError } from "#adapters/errors";
+import { stableSourceID } from "#adapters/source-id";
+import { SourceRegister } from "#registers";
+import { AddressRole, 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 = ["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 by itself 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",
+};
+const NO_NUMBER = 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, header) {
+    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) {
+    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) {
+    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 = {};
+    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) {
+    let header = null;
+    for await (const row of XLSXSpliterator.fromAsync(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() {
+    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) {
+            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();
+            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();
+//# sourceMappingURL=adapter.js.map

package/out/ro/adapters/retea-scolara/adapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../lib/ro/adapters/retea-scolara/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA6CG;AAEH,OAAO,EAAE,gBAAgB,EAAE,MAAM,iCAAiC,CAAA;AAElE,OAAO,EAAE,eAAe,EAAsB,MAAM,aAAa,CAAA;AAEjE,OAAO,EAAE,uBAAuB,EAAE,MAAM,kBAAkB,CAAA;AAC1D,OAAO,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAA;AACpD,OAAO,EAAE,cAAc,EAAE,MAAM,YAAY,CAAA;AAC3C,OAAO,EACN,WAAW,EAIX,YAAY,EACZ,aAAa,GACb,MAAM,QAAQ,CAAA;AAEf;;GAEG;AACH,MAAM,CAAC,MAAM,2BAA2B,GAAG,kBAAkB,CAAA;AAE7D;;GAEG;AACH,MAAM,CAAC,MAAM,0BAA0B,GAAsB,CAAC,IAAI,CAAC,CAAA;AAEnE;;GAEG;AACH,MAAM,CAAC,MAAM,wBAAwB,GAAG,WAAW,CAAA;AAEnD;;GAEG;AACH,MAAM,CAAC,MAAM,4BAA4B,GAAG,sDAAsD,CAAA;AAElG;;GAEG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAAG;IAClC;;OAEG;IACH,cAAc,EAAE,qBAAqB;IACrC;;OAEG;IACH,aAAa,EAAE,oBAAoB;IACnC;;OAEG;IACH,YAAY,EAAE,kBAAkB;IAChC;;OAEG;IACH,SAAS,EAAE,eAAe;CACjB,CAAA;AAWV,MAAM,SAAS,GAAwB,IAAI,GAAG,CAAC,CAAC,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,GAAG,EAAE,GAAG,EAAE,IAAI,EAAE,KAAK,CAAC,CAAC,CAAA;AAE5F,MAAM,aAAa,GAAG,2BAA2B,CAAA;AAEjD,MAAM,QAAQ,GAAG,UAAU,CAAA;AAE3B,SAAS,IAAI,CAAC,IAAsB,EAAE,MAAc;IACnD,OAAO,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC;SAC/B,UAAU,CAAC,OAAO,EAAE,GAAG,CAAC;SACxB,IAAI,EAAE,CAAA;AACT,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,eAAe,CAAC,KAAa;IAC5C,MAAM,QAAQ,GAAG,KAAK,CAAC,OAAO,CAAC,aAAa,EAAE,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;IAExD,OAAO,SAAS,CAAC,GAAG,CAAC,QAAQ,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,QAAQ,CAAA;AAC7D,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,oBAAoB,CAAC,IAAsB;IAC1D,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,EAAE,oBAAoB,CAAC,CAAA;IAEjD,IAAI,CAAC,QAAQ;QAAE,OAAO,EAAE,OAAO,EAAE,mBAAmB,CAAC,cAAc,EAAE,CAAA;IAErE,MAAM,MAAM,GAAG,IAAI,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAA;IACnC,MAAM,WAAW,GAAG,eAAe,CAAC,IAAI,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC,CAAA;IAExD,IAAI,CAAC,MAAM,IAAI,CAAC,WAAW;QAAE,OAAO,EAAE,OAAO,EAAE,mBAAmB,CAAC,aAAa,EAAE,CAAA;IAElF,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,EAAE,YAAY,CAAC,CAAA;IACzC,MAAM,KAAK,GAAG,IAAI,CAAC,IAAI,EAAE,wBAAwB,CAAC,IAAI,IAAI,CAAC,IAAI,EAAE,yBAAyB,CAAC,CAAA;IAE3F,MAAM,UAAU,GAA+B,EAAE,CAAA;IAEjD,IAAI,KAAK,EAAE,CAAC;QACX,UAAU,CAAC,KAAK,GAAG,KAAK,CAAA;IACzB,CAAC;IAED,IAAI,MAAM,EAAE,CAAC;QACZ,UAAU,CAAC,MAAM,GAAG,MAAM,CAAA;IAC3B,CAAC;IAED,IAAI,WAAW,EAAE,CAAC;QACjB,UAAU,CAAC,YAAY,GAAG,WAAW,CAAA;IACtC,CAAC;IAED,IAAI,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,EAAE,CAAC;QAC7B,UAAU,CAAC,QAAQ,GAAG,QAAQ,CAAA;IAC/B,CAAC;IAED,UAAU,CAAC,QAAQ,GAAG,QAAQ,CAAA;IAE9B,MAAM,QAAQ,GAAG,gBAAgB,CAAC,UAAU,EAAE,IAAI,EAAE,EAAE,UAAU,EAAE,IAAI,EAAE,CAAC,CAAA;IAEzE,IAAI,CAAC,QAAQ;QAAE,OAAO,EAAE,OAAO,EAAE,mBAAmB,CAAC,YAAY,EAAE,CAAA;IAEnE,OAAO;QACN,QAAQ,EAAE;YACT,GAAG,EAAE,QAAQ,CAAC,GAAG;YACjB,UAAU,EAAE,QAAQ,CAAC,UAAU;YAC/B,OAAO,EAAE,IAAI;YACb,MAAM,EAAE,OAAO;YACf,MAAM,EAAE,2BAA2B;YACnC,SAAS,EAAE,cAAc,CAAC,2BAA2B,EAAE,QAAQ,CAAC,UAAU,CAAC;YAC3E,cAAc,EAAE,EAAE;YAClB,OAAO,EAAE,wBAAwB;SACjC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,KAAK,SAAS,CAAC,CAAC,qBAAqB,CAAC,SAA0B;IACtE,IAAI,MAAM,GAAoB,IAAI,CAAA;IAElC,IAAI,KAAK,EAAE,MAAM,GAAG,IAAI,eAAe,CAAC,SAAS,CAAkB,SAAS,CAAC,QAAQ,EAAE,EAAE;QACxF,IAAI,EAAE,OAAO;QACb,MAAM,EAAE,KAAK;KACb,CAAC,EAAE,CAAC;QACJ,IAAI,CAAC,MAAM,EAAE,CAAC;YACb,IAAI,GAAG,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,KAAK,QAAQ,CAAC,EAAE,CAAC;gBAC7C,MAAM,GAAG,GAAG,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,MAAM,CAAC,KAAK,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC,CAAA;YACxD,CAAC;YAED,SAAQ;QACT,CAAC;QAED,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,KAAK,IAAI,IAAI,KAAK,KAAK,EAAE,CAAC;YAAE,SAAQ;QAElE,MAAM,MAAM,CAAC,WAAW,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC,CAAC,IAAI,EAAE,GAAG,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAA;IAC1E,CAAC;IAED,IAAI,CAAC,MAAM;QAAE,MAAM,IAAI,KAAK,CAAC,GAAG,2BAA2B,2CAA2C,SAAS,EAAE,CAAC,CAAA;AACnH,CAAC;AAED,MAAM,UAAU,yBAAyB;IACxC,OAAO;QACN,EAAE,EAAE,2BAA2B;QAC/B,cAAc,EAAE,wBAAwB;QACxC,WAAW,EAAE,WAAW,CAAC,QAAQ;QACjC,QAAQ,EAAE,cAAc,CAAC,YAAY;QACrC,OAAO,EAAE,aAAa,CAAC,QAAQ;QAC/B,WAAW,EAAE,uFAAuF;QAEpG,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,IAAI,IAAI,CAAC,OAAO,IAAI,CAAC,0BAA0B,CAAC,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC;gBACxE,MAAM,IAAI,uBAAuB,CAAC,2BAA2B,EAAE,0BAA0B,EAAE,IAAI,CAAC,OAAO,CAAC,CAAA;YACzG,CAAC;YAED,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAA;YAC9B,IAAI,OAAO,GAAG,CAAC,CAAA;YAEf,IAAI,KAAK,EAAE,MAAM,IAAI,IAAI,qBAAqB,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,CAAC;gBAChE,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;oBAAE,MAAK;gBAE/B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;oBAAE,MAAK;gBAE5D,MAAM,OAAO,GAAG,oBAAoB,CAAC,IAAI,CAAC,CAAA;gBAE1C,IAAI,CAAC,CAAC,UAAU,IAAI,OAAO,CAAC,EAAE,CAAC;oBAC9B,YAAY,CAAC,IAAI,EAAE,OAAO,CAAC,OAAO,CAAC,CAAA;oBAEnC,SAAQ;gBACT,CAAC;gBAED,IAAI,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,QAAQ,CAAC,SAAS,CAAC,EAAE,CAAC;oBAC1C,YAAY,CAAC,IAAI,EAAE,mBAAmB,CAAC,SAAS,CAAC,CAAA;oBAEjD,SAAQ;gBACT,CAAC;gBAED,IAAI,CAAC,GAAG,CAAC,OAAO,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAA;gBACpC,MAAM,OAAO,CAAC,QAAQ,CAAA;gBAEtB,OAAO,EAAE,CAAA;YACV,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAAG,yBAAyB,EAAE,CAAA"}