@mailwoman/corpus 10.0.0 → 10.2.0

package/out/es/adapters/navarra/adapter.d.ts

@@ -0,0 +1,106 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * `es-navarra`: the INSPIRE Addresses theme the Gobierno de Navarra publishes, read from its zipped
+ * GML partitions.
+ *
+ * The ATOM service at
+ * `https://filescartografia.navarra.es/2_CARTOGRAFIA_TEMATICA/2_7_CATASTRO/2_7_3_INSPIRE_ATOM/2_7_3_3_AD/Addresses_ServiceATOM_Navarra.xml`
+ * is 671,131 bytes and splits the province across 272 entries. Every entry is titled `Address
+ * Navarra`, and the feed states no place for a partition. The three measured each held one
+ * municipality: partition 1 Abáigar, partition 100 Eulate and partition 250 Bera. The feed states
+ * no such mapping and this adapter claims none. It reads one archive or a directory of them. A
+ * caller harvests all 272 by pointing it at the directory.
+ *
+ * This publisher is read entirely from its references' titles, and that is what separates it from
+ * every other INSPIRE reader here. Its own file publishes no `AD:ThoroughfareName`,
+ * `AD:PostalDescriptor` or `AD:AdminUnitName` feature at all: partition 1 holds 29 distinct element
+ * names across 115 addresses and none of them is a component feature. Each address instead holds
+ * six `AD:component` references whose `xlink:title` states the value:
+ *
+ * | `xlink:href` | `xlink:title` | read as |
+ * | --- | --- | --- |
+ * | `…&ID=AD_ADMINUNITNAME_PAI_34000000000#…` | `España` | skipped |
+ * | `…&ID=AD_ADMINUNITNAME_COM_34150000000#…` | `Comunidad Foral de Navarra` | skipped |
+ * | `…&ID=AD_ADMINUNITNAME_MUN_34153131001#…` | `Abáigar` | the locality |
+ * | `http://inspire.ec.europa.eu/codelist/AdministrativeHierarchyLevel/5thOrder` | `Abáigar` | the settlement |
+ * | `ThoroughfareName` | `CALLE CALLEJA` | the street |
+ * | `PostalDescriptor` | `31280` | the postcode |
+ *
+ * The counts are exact and regular: 690 references over 115 addresses in partition 1, 1,728 over 288
+ * in partition 100 and 6,324 over 1,054 in partition 250. Each address holds six references, one of
+ * each shape above.
+ *
+ * The two place titles are not the same value. In Abáigar and Eulate the `5thOrder` title repeats the
+ * municipality, and in Bera it states a *concejo* the municipality contains: `Suspela` and `Dornaku`
+ * beside `Bera`. The fifth order is below the municipal fourth, so the title is written as a
+ * dependent locality where it differs from the municipality and left out where it repeats it.
+ *
+ * `Comunidad Foral de Navarra` is the autonomous community's own name rather than the province name a
+ * Spanish address writes, and the publisher's own address layer has no region column, so no
+ * `region` is written.
+ *
+ * The publisher's own rendering is in a different service, and it agrees. `IDENA` publishes
+ * `IDENA:DIRECC_Txt_Direcciones`, 181,098 features, each with `MUNICIPIO`, `ENTIDAD`, `CODPOSTAL`,
+ * `VIA` and `PORTAL`. Partition 1's `AD.Address.1` composes `CALLE CALLEJA, 4, 31280 Abáigar`, and
+ * that layer's `DIRECC_Txt_Direcciones.1` reads `MUNICIPIO` `Abáigar`, `ENTIDAD` `Abáigar`,
+ * `CODPOSTAL` `31280`, `VIA` `CALLE CALLEJA`, `PORTAL` `4`. The adapter's integration test checks a
+ * sample of rows against that layer, the publisher-side oracle for this source.
+ *
+ * Two values state an absence rather than a value. A `PostalDescriptor` title of `0` is the publisher
+ * holding no postcode, measured in Bera. A designator of `S/N` is *sin número*, and `#es/cadastre`
+ * reads it for all four Spanish cadastres. A designator may also hold a space-separated letter,
+ * `18 A` and `34 A` in Eulate. The adapter keeps that letter as the publisher wrote it.
+ */
+import { type MarkupElement } from "@mailwoman/core/html/elements";
+import { type CorpusAdapter } from "#types";
+/**
+ * Registry id for this adapter.
+ *
+ * Stamped into every row it emits, so a corpus record can be traced back to the dataset it came from.
+ */
+export declare const ES_NAVARRA_ADAPTER_ID = "es-navarra";
+/**
+ * The jurisdiction this adapter emits.
+ */
+export declare const ES_NAVARRA_COUNTRIES: readonly string[];
+/**
+ * The license the address-source register elected for this publisher.
+ *
+ * Navarra states one license and states it in its own feed: the feed-level `rights` of
+ * `Addresses_ServiceATOM_Navarra.xml` reads `This layer is published under the terms of the license
+ * Creative Commons Attribution 4.0 International (CC BY 4.0)`, and all 272 of its entries repeat it.
+ * The version is stated, so the register records `CC-BY-4.0` as the SPDX identifier.
+ *
+ * The row states that identifier rather than the license's title.
+ * `licenseVerdict` resolves an identifier and reads a title as unrecognized.
+ *
+ * A row whose license resolves to no expression has unknown obligations rather than none.
+ * The build reports that separately from a refusal.
+ */
+export declare const ES_NAVARRA_LICENSE = "CC-BY-4.0";
+/**
+ * What one address's references state, read from their titles.
+ */
+interface ReferencedTitles {
+    street: string | null;
+    postcode: string | null;
+    locality: string | null;
+    settlement: string | null;
+}
+/**
+ * The street, postcode and place names an address's own references state.
+ *
+ * Each reference is placed by its href rather than by its position, because the six are written in one
+ * order in the three partitions measured and that order is the publisher's rather than the schema's.
+ */
+export declare function navarraReferencedTitles(address: MarkupElement): ReferencedTitles;
+export declare function createESNavarraAdapter(): CorpusAdapter;
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
+export declare const esNavarraAdapter: CorpusAdapter;
+export {};
+//# sourceMappingURL=adapter.d.ts.map

package/out/es/adapters/navarra/adapter.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../lib/es/adapters/navarra/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAuDG;AAEH,OAAO,EAAE,KAAK,aAAa,EAAwB,MAAM,+BAA+B,CAAA;AAgBxF,OAAO,EAAuD,KAAK,aAAa,EAAiB,MAAM,QAAQ,CAAA;AAE/G;;;;GAIG;AACH,eAAO,MAAM,qBAAqB,eAAe,CAAA;AAEjD;;GAEG;AACH,eAAO,MAAM,oBAAoB,EAAE,SAAS,MAAM,EAAW,CAAA;AAE7D;;;;;;;;;;;;;GAaG;AACH,eAAO,MAAM,kBAAkB,cAAc,CAAA;AAoD7C;;GAEG;AACH,UAAU,gBAAgB;IACzB,MAAM,EAAE,MAAM,GAAG,IAAI,CAAA;IACrB,QAAQ,EAAE,MAAM,GAAG,IAAI,CAAA;IACvB,QAAQ,EAAE,MAAM,GAAG,IAAI,CAAA;IACvB,UAAU,EAAE,MAAM,GAAG,IAAI,CAAA;CACzB;AAED;;;;;GAKG;AACH,wBAAgB,uBAAuB,CAAC,OAAO,EAAE,aAAa,GAAG,gBAAgB,CAoBhF;AAED,wBAAgB,sBAAsB,IAAI,aAAa,CA2CtD;AAoDD;;GAEG;AACH,eAAO,MAAM,gBAAgB,eAA2B,CAAA"}

package/out/es/adapters/navarra/adapter.js

@@ -0,0 +1,242 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * `es-navarra`: the INSPIRE Addresses theme the Gobierno de Navarra publishes, read from its zipped
+ * GML partitions.
+ *
+ * The ATOM service at
+ * `https://filescartografia.navarra.es/2_CARTOGRAFIA_TEMATICA/2_7_CATASTRO/2_7_3_INSPIRE_ATOM/2_7_3_3_AD/Addresses_ServiceATOM_Navarra.xml`
+ * is 671,131 bytes and splits the province across 272 entries. Every entry is titled `Address
+ * Navarra`, and the feed states no place for a partition. The three measured each held one
+ * municipality: partition 1 Abáigar, partition 100 Eulate and partition 250 Bera. The feed states
+ * no such mapping and this adapter claims none. It reads one archive or a directory of them. A
+ * caller harvests all 272 by pointing it at the directory.
+ *
+ * This publisher is read entirely from its references' titles, and that is what separates it from
+ * every other INSPIRE reader here. Its own file publishes no `AD:ThoroughfareName`,
+ * `AD:PostalDescriptor` or `AD:AdminUnitName` feature at all: partition 1 holds 29 distinct element
+ * names across 115 addresses and none of them is a component feature. Each address instead holds
+ * six `AD:component` references whose `xlink:title` states the value:
+ *
+ * | `xlink:href` | `xlink:title` | read as |
+ * | --- | --- | --- |
+ * | `…&ID=AD_ADMINUNITNAME_PAI_34000000000#…` | `España` | skipped |
+ * | `…&ID=AD_ADMINUNITNAME_COM_34150000000#…` | `Comunidad Foral de Navarra` | skipped |
+ * | `…&ID=AD_ADMINUNITNAME_MUN_34153131001#…` | `Abáigar` | the locality |
+ * | `http://inspire.ec.europa.eu/codelist/AdministrativeHierarchyLevel/5thOrder` | `Abáigar` | the settlement |
+ * | `ThoroughfareName` | `CALLE CALLEJA` | the street |
+ * | `PostalDescriptor` | `31280` | the postcode |
+ *
+ * The counts are exact and regular: 690 references over 115 addresses in partition 1, 1,728 over 288
+ * in partition 100 and 6,324 over 1,054 in partition 250. Each address holds six references, one of
+ * each shape above.
+ *
+ * The two place titles are not the same value. In Abáigar and Eulate the `5thOrder` title repeats the
+ * municipality, and in Bera it states a *concejo* the municipality contains: `Suspela` and `Dornaku`
+ * beside `Bera`. The fifth order is below the municipal fourth, so the title is written as a
+ * dependent locality where it differs from the municipality and left out where it repeats it.
+ *
+ * `Comunidad Foral de Navarra` is the autonomous community's own name rather than the province name a
+ * Spanish address writes, and the publisher's own address layer has no region column, so no
+ * `region` is written.
+ *
+ * The publisher's own rendering is in a different service, and it agrees. `IDENA` publishes
+ * `IDENA:DIRECC_Txt_Direcciones`, 181,098 features, each with `MUNICIPIO`, `ENTIDAD`, `CODPOSTAL`,
+ * `VIA` and `PORTAL`. Partition 1's `AD.Address.1` composes `CALLE CALLEJA, 4, 31280 Abáigar`, and
+ * that layer's `DIRECC_Txt_Direcciones.1` reads `MUNICIPIO` `Abáigar`, `ENTIDAD` `Abáigar`,
+ * `CODPOSTAL` `31280`, `VIA` `CALLE CALLEJA`, `PORTAL` `4`. The adapter's integration test checks a
+ * sample of rows against that layer, the publisher-side oracle for this source.
+ *
+ * Two values state an absence rather than a value. A `PostalDescriptor` title of `0` is the publisher
+ * holding no postcode, measured in Bera. A designator of `S/N` is *sin número*, and `#es/cadastre`
+ * reads it for all four Spanish cadastres. A designator may also hold a space-separated letter,
+ * `18 A` and `34 A` in Eulate. The adapter keeps that letter as the publisher wrote it.
+ */
+import { streamMarkupElements } from "@mailwoman/core/html/elements";
+import { PathBuilder } from "path-ts";
+import { Globerator } from "spliterator/node/fs";
+import { UnsupportedCountryError } from "#adapters/errors";
+import { cadastralRow, spanishHouseNumber } from "#es/cadastre";
+import { codelistValue, componentJoinKey, componentLinks, designatorsByType, voidDesignatorTypes, } from "#inspire/address";
+import { inspireGMLChunks } from "#inspire/archive";
+import { VoidDesignatorError } from "#inspire/errors";
+import { SourceRegister } from "#registers";
+import { AddressRole, SurfaceOrigin } from "#types";
+/**
+ * Registry id for this adapter.
+ *
+ * Stamped into every row it emits, so a corpus record can be traced back to the dataset it came from.
+ */
+export const ES_NAVARRA_ADAPTER_ID = "es-navarra";
+/**
+ * The jurisdiction this adapter emits.
+ */
+export const ES_NAVARRA_COUNTRIES = ["ES"];
+/**
+ * The license the address-source register elected for this publisher.
+ *
+ * Navarra states one license and states it in its own feed: the feed-level `rights` of
+ * `Addresses_ServiceATOM_Navarra.xml` reads `This layer is published under the terms of the license
+ * Creative Commons Attribution 4.0 International (CC BY 4.0)`, and all 272 of its entries repeat it.
+ * The version is stated, so the register records `CC-BY-4.0` as the SPDX identifier.
+ *
+ * The row states that identifier rather than the license's title.
+ * `licenseVerdict` resolves an identifier and reads a title as unrecognized.
+ *
+ * A row whose license resolves to no expression has unknown obligations rather than none.
+ * The build reports that separately from a refusal.
+ */
+export const ES_NAVARRA_LICENSE = "CC-BY-4.0";
+/**
+ * The member of a partition's archive that holds the GML.
+ *
+ * Each archive ships one member, `AD_Navarra_<n>.gml`, so the extension selects
+ * it without the adapter knowing the partition number.
+ */
+const GML_MEMBER = /\.gml$/iu;
+/**
+ * The element each address is written as.
+ *
+ * The publisher binds the INSPIRE namespaces to upper-case prefixes.
+ */
+const ADDRESS_ELEMENT = "AD:Address";
+/**
+ * The designator type the publisher writes, on every address of the three partitions measured.
+ */
+const HOUSE_NUMBER_TYPE = "addressIdentifierGeneral";
+/**
+ * The reference href that states a street name in its title.
+ */
+const THOROUGHFARE_HREF = "thoroughfarename";
+/**
+ * The reference href that states a postcode in its title.
+ */
+const POSTAL_DESCRIPTOR_HREF = "postaldescriptor";
+/**
+ * The stored-query id prefix of the administrative-unit reference that is the municipality.
+ *
+ * The other two are `AD_ADMINUNITNAME_PAI_`, the country, and `AD_ADMINUNITNAME_COM_`,
+ * the autonomous community, and neither belongs to a written address.
+ */
+const MUNICIPALITY_ID_PREFIX = "AD_ADMINUNITNAME_MUN_";
+/**
+ * The administrative level whose title is a settlement within the municipality.
+ */
+const SETTLEMENT_LEVEL = "5thOrder";
+/**
+ * The postcode title that states that the publisher holds none.
+ *
+ * Bera writes `0` beside its `31780`.
+ */
+const NO_POSTCODE_TITLE = "0";
+/**
+ * The street, postcode and place names an address's own references state.
+ *
+ * Each reference is placed by its href rather than by its position, because the six are written in one
+ * order in the three partitions measured and that order is the publisher's rather than the schema's.
+ */
+export function navarraReferencedTitles(address) {
+    const titles = { street: null, postcode: null, locality: null, settlement: null };
+    for (const { href, title } of componentLinks(address)) {
+        if (!title)
+            continue;
+        const lowered = href.toLowerCase();
+        if (lowered === THOROUGHFARE_HREF) {
+            titles.street = title;
+        }
+        else if (lowered === POSTAL_DESCRIPTOR_HREF) {
+            titles.postcode = title === NO_POSTCODE_TITLE ? null : title;
+        }
+        else if (codelistValue(href) === SETTLEMENT_LEVEL) {
+            titles.settlement = title;
+        }
+        else if (componentJoinKey(href)?.startsWith(MUNICIPALITY_ID_PREFIX)) {
+            titles.locality = title;
+        }
+    }
+    return titles;
+}
+export function createESNavarraAdapter() {
+    return {
+        id: ES_NAVARRA_ADAPTER_ID,
+        defaultLicense: ES_NAVARRA_LICENSE,
+        addressRole: AddressRole.Premise,
+        register: SourceRegister.NavarraInspireAddresses,
+        surface: SurfaceOrigin.Rendered,
+        description: "INSPIRE Addresses (Gobierno de Navarra): house-number-level addresses for Navarra, as 272 zipped GML partitions.",
+        async *rows(opts) {
+            if (opts.country && !ES_NAVARRA_COUNTRIES.includes(opts.country)) {
+                throw new UnsupportedCountryError(ES_NAVARRA_ADAPTER_ID, ES_NAVARRA_COUNTRIES, opts.country);
+            }
+            let emitted = 0;
+            for (const archivePath of await archivePaths(opts.inputPath)) {
+                if (opts.signal?.aborted)
+                    break;
+                if (opts.limit !== undefined && emitted >= opts.limit)
+                    break;
+                // The markup stream is read directly rather than through `#inspire/stream`,
+                // because this publisher writes no component feature for an address to reference.
+                // The index would stay empty and no address would ever be deferred.
+                for await (const address of streamMarkupElements(inspireGMLChunks(archivePath, GML_MEMBER), ADDRESS_ELEMENT, {
+                    xml: true,
+                })) {
+                    if (opts.signal?.aborted)
+                        break;
+                    if (opts.limit !== undefined && emitted >= opts.limit)
+                        break;
+                    const row = composeRow(address);
+                    if (!row)
+                        continue;
+                    yield row;
+                    emitted++;
+                }
+            }
+        },
+    };
+}
+/**
+ * One address's row, or null when it holds too little to render.
+ */
+function composeRow(address) {
+    const addressID = address.attributes["gml:id"]?.trim() ?? null;
+    if (voidDesignatorTypes(address).has(HOUSE_NUMBER_TYPE)) {
+        throw new VoidDesignatorError(ES_NAVARRA_ADAPTER_ID, addressID, HOUSE_NUMBER_TYPE);
+    }
+    const { street, postcode, locality, settlement } = navarraReferencedTitles(address);
+    // Navarra's settlement is a fifth-order title, a *concejo* within the municipality.
+    // `cadastralRow` writes it as a dependent locality only where it differs from the municipality.
+    return cadastralRow({
+        street,
+        house: spanishHouseNumber(designatorsByType(address), HOUSE_NUMBER_TYPE),
+        postcode,
+        locality,
+        settlement,
+        addressID,
+    }, { adapterID: ES_NAVARRA_ADAPTER_ID, license: ES_NAVARRA_LICENSE });
+}
+/**
+ * Every archive under `inputPath`, which is one `.zip`, one unpacked `.gml`, or a directory holding either.
+ *
+ * A directory is the ordinary case, because the publisher ships 272 partitions
+ * and a caller harvesting the province points at the directory it fetched them into.
+ * A directory holding neither raises rather than yielding zero rows, so a wrong path
+ * reports itself instead of reading as an empty publisher.
+ */
+async function archivePaths(inputPath) {
+    const path = PathBuilder.from(inputPath);
+    const basename = path.basename().toLowerCase();
+    if (basename.endsWith(".zip") || basename.endsWith(".gml"))
+        return [path];
+    const names = await Globerator.from("*.{zip,gml}", { cwd: path.toString(), absolute: false }).toSorted();
+    if (!names.length) {
+        throw new Error(`${ES_NAVARRA_ADAPTER_ID} adapter: ${path.toString()} holds no .zip archive or .gml member`);
+    }
+    return names.map((name) => path(name));
+}
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
+export const esNavarraAdapter = createESNavarraAdapter();
+//# sourceMappingURL=adapter.js.map

package/out/es/adapters/navarra/adapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../lib/es/adapters/navarra/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAuDG;AAEH,OAAO,EAAsB,oBAAoB,EAAE,MAAM,+BAA+B,CAAA;AACxF,OAAO,EAAE,WAAW,EAAwB,MAAM,SAAS,CAAA;AAC3D,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAA;AAEhD,OAAO,EAAE,uBAAuB,EAAE,MAAM,kBAAkB,CAAA;AAC1D,OAAO,EAAE,YAAY,EAAE,kBAAkB,EAAE,MAAM,cAAc,CAAA;AAC/D,OAAO,EACN,aAAa,EACb,gBAAgB,EAChB,cAAc,EACd,iBAAiB,EACjB,mBAAmB,GACnB,MAAM,kBAAkB,CAAA;AACzB,OAAO,EAAE,gBAAgB,EAAE,MAAM,kBAAkB,CAAA;AACnD,OAAO,EAAE,mBAAmB,EAAE,MAAM,iBAAiB,CAAA;AACrD,OAAO,EAAE,cAAc,EAAE,MAAM,YAAY,CAAA;AAC3C,OAAO,EAAE,WAAW,EAA8D,aAAa,EAAE,MAAM,QAAQ,CAAA;AAE/G;;;;GAIG;AACH,MAAM,CAAC,MAAM,qBAAqB,GAAG,YAAY,CAAA;AAEjD;;GAEG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAAsB,CAAC,IAAI,CAAC,CAAA;AAE7D;;;;;;;;;;;;;GAaG;AACH,MAAM,CAAC,MAAM,kBAAkB,GAAG,WAAW,CAAA;AAE7C;;;;;GAKG;AACH,MAAM,UAAU,GAAG,UAAU,CAAA;AAE7B;;;;GAIG;AACH,MAAM,eAAe,GAAG,YAAY,CAAA;AAEpC;;GAEG;AACH,MAAM,iBAAiB,GAAG,0BAA0B,CAAA;AAEpD;;GAEG;AACH,MAAM,iBAAiB,GAAG,kBAAkB,CAAA;AAE5C;;GAEG;AACH,MAAM,sBAAsB,GAAG,kBAAkB,CAAA;AAEjD;;;;;GAKG;AACH,MAAM,sBAAsB,GAAG,uBAAuB,CAAA;AAEtD;;GAEG;AACH,MAAM,gBAAgB,GAAG,UAAU,CAAA;AAEnC;;;;GAIG;AACH,MAAM,iBAAiB,GAAG,GAAG,CAAA;AAY7B;;;;;GAKG;AACH,MAAM,UAAU,uBAAuB,CAAC,OAAsB;IAC7D,MAAM,MAAM,GAAqB,EAAE,MAAM,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,UAAU,EAAE,IAAI,EAAE,CAAA;IAEnG,KAAK,MAAM,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,cAAc,CAAC,OAAO,CAAC,EAAE,CAAC;QACvD,IAAI,CAAC,KAAK;YAAE,SAAQ;QAEpB,MAAM,OAAO,GAAG,IAAI,CAAC,WAAW,EAAE,CAAA;QAElC,IAAI,OAAO,KAAK,iBAAiB,EAAE,CAAC;YACnC,MAAM,CAAC,MAAM,GAAG,KAAK,CAAA;QACtB,CAAC;aAAM,IAAI,OAAO,KAAK,sBAAsB,EAAE,CAAC;YAC/C,MAAM,CAAC,QAAQ,GAAG,KAAK,KAAK,iBAAiB,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,CAAA;QAC7D,CAAC;aAAM,IAAI,aAAa,CAAC,IAAI,CAAC,KAAK,gBAAgB,EAAE,CAAC;YACrD,MAAM,CAAC,UAAU,GAAG,KAAK,CAAA;QAC1B,CAAC;aAAM,IAAI,gBAAgB,CAAC,IAAI,CAAC,EAAE,UAAU,CAAC,sBAAsB,CAAC,EAAE,CAAC;YACvE,MAAM,CAAC,QAAQ,GAAG,KAAK,CAAA;QACxB,CAAC;IACF,CAAC;IAED,OAAO,MAAM,CAAA;AACd,CAAC;AAED,MAAM,UAAU,sBAAsB;IACrC,OAAO;QACN,EAAE,EAAE,qBAAqB;QACzB,cAAc,EAAE,kBAAkB;QAClC,WAAW,EAAE,WAAW,CAAC,OAAO;QAChC,QAAQ,EAAE,cAAc,CAAC,uBAAuB;QAChD,OAAO,EAAE,aAAa,CAAC,QAAQ;QAC/B,WAAW,EACV,kHAAkH;QAEnH,KAAK,CAAC,CAAC,IAAI,CAAC,IAAoB;YAC/B,IAAI,IAAI,CAAC,OAAO,IAAI,CAAC,oBAAoB,CAAC,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC;gBAClE,MAAM,IAAI,uBAAuB,CAAC,qBAAqB,EAAE,oBAAoB,EAAE,IAAI,CAAC,OAAO,CAAC,CAAA;YAC7F,CAAC;YAED,IAAI,OAAO,GAAG,CAAC,CAAA;YAEf,KAAK,MAAM,WAAW,IAAI,MAAM,YAAY,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,CAAC;gBAC9D,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,4EAA4E;gBAC5E,kFAAkF;gBAClF,oEAAoE;gBACpE,IAAI,KAAK,EAAE,MAAM,OAAO,IAAI,oBAAoB,CAAC,gBAAgB,CAAC,WAAW,EAAE,UAAU,CAAC,EAAE,eAAe,EAAE;oBAC5G,GAAG,EAAE,IAAI;iBACT,CAAC,EAAE,CAAC;oBACJ,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;wBAAE,MAAK;oBAE/B,IAAI,IAAI,CAAC,KAAK,KAAK,SAAS,IAAI,OAAO,IAAI,IAAI,CAAC,KAAK;wBAAE,MAAK;oBAE5D,MAAM,GAAG,GAAG,UAAU,CAAC,OAAO,CAAC,CAAA;oBAE/B,IAAI,CAAC,GAAG;wBAAE,SAAQ;oBAElB,MAAM,GAAG,CAAA;oBAET,OAAO,EAAE,CAAA;gBACV,CAAC;YACF,CAAC;QACF,CAAC;KACD,CAAA;AACF,CAAC;AAED;;GAEG;AACH,SAAS,UAAU,CAAC,OAAsB;IACzC,MAAM,SAAS,GAAG,OAAO,CAAC,UAAU,CAAC,QAAQ,CAAC,EAAE,IAAI,EAAE,IAAI,IAAI,CAAA;IAE9D,IAAI,mBAAmB,CAAC,OAAO,CAAC,CAAC,GAAG,CAAC,iBAAiB,CAAC,EAAE,CAAC;QACzD,MAAM,IAAI,mBAAmB,CAAC,qBAAqB,EAAE,SAAS,EAAE,iBAAiB,CAAC,CAAA;IACnF,CAAC;IAED,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,QAAQ,EAAE,UAAU,EAAE,GAAG,uBAAuB,CAAC,OAAO,CAAC,CAAA;IAEnF,oFAAoF;IACpF,gGAAgG;IAChG,OAAO,YAAY,CAClB;QACC,MAAM;QACN,KAAK,EAAE,kBAAkB,CAAC,iBAAiB,CAAC,OAAO,CAAC,EAAE,iBAAiB,CAAC;QACxE,QAAQ;QACR,QAAQ;QACR,UAAU;QACV,SAAS;KACT,EACD,EAAE,SAAS,EAAE,qBAAqB,EAAE,OAAO,EAAE,kBAAkB,EAAE,CACjE,CAAA;AACF,CAAC;AAED;;;;;;;GAOG;AACH,KAAK,UAAU,YAAY,CAAC,SAA0B;IACrD,MAAM,IAAI,GAAG,WAAW,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;IACxC,MAAM,QAAQ,GAAG,IAAI,CAAC,QAAQ,EAAE,CAAC,WAAW,EAAE,CAAA;IAE9C,IAAI,QAAQ,CAAC,QAAQ,CAAC,MAAM,CAAC,IAAI,QAAQ,CAAC,QAAQ,CAAC,MAAM,CAAC;QAAE,OAAO,CAAC,IAAI,CAAC,CAAA;IAEzE,MAAM,KAAK,GAAG,MAAM,UAAU,CAAC,IAAI,CAAC,aAAa,EAAE,EAAE,GAAG,EAAE,IAAI,CAAC,QAAQ,EAAE,EAAE,QAAQ,EAAE,KAAK,EAAE,CAAC,CAAC,QAAQ,EAAE,CAAA;IAExG,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC;QACnB,MAAM,IAAI,KAAK,CAAC,GAAG,qBAAqB,aAAa,IAAI,CAAC,QAAQ,EAAE,uCAAuC,CAAC,CAAA;IAC7G,CAAC;IAED,OAAO,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAA;AACvC,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAG,sBAAsB,EAAE,CAAA"}

package/out/es/cadastre.d.ts

@@ -0,0 +1,110 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * The two ways a Spanish cadastre states that it holds no value. All four cadastres share them.
+ *
+ * Spain publishes the INSPIRE Addresses theme through four cadastres: the Dirección General del
+ * Catastro for 52 provinces, and the Diputación Foral de Bizkaia, the Diputación Foral de Gipuzkoa
+ * and the Gobierno de Navarra for their own territories. Each has its own adapter, because each
+ * states its own terms and writes its own file layout. They agree on these two conventions, and
+ * reading either of them as data rather than as an absence puts a token where a corpus row expects a
+ * number or a postcode.
+ */
+import type { CanonicalRow } from "#types";
+/**
+ * Whether a designator value states that the address has no number.
+ *
+ * The whole value has to be the token.
+ * A value that holds more, such as Gipuzkoa's `S/N-A`, `S/N-C` and `S/N-5`,
+ * states something further and is kept as the publisher wrote it: what the trailing part
+ * designates is not established, dropping it would merge several addresses of one building,
+ * and rewriting it would state a number the publisher did not.
+ */
+export declare function isSinNumero(value: string): boolean;
+/**
+ * The house number a Spanish cadastre states, or null where it states that there is none.
+ *
+ * @param types The publisher's own designator types, in the caller's order of precedence.
+ * The Dirección General del Catastro writes the type as the element text `1`, while Navarra
+ * and Gipuzkoa write the `LocatorDesignatorTypeValue/addressIdentifierGeneral` codelist URI.
+ */
+export declare function spanishHouseNumber(byType: ReadonlyMap<string, readonly string[]>, ...types: readonly string[]): string | null;
+/**
+ * Whether a component reference's key states that the publisher holds no value for that component.
+ *
+ * A Spanish cadastre's feature identifier is a dotted path whose last segment is the value:
+ * the Cadastre's postal descriptors in Ceuta are `ES.SDGC.PD.55.101.51001` through
+ * `…51005`, and Gipuzkoa's are `ES.GFA.PD.056.20830`.
+ * Where the cadastre holds no postcode it writes the same identifier with that segment empty,
+ * `ES.SDGC.PD.55.101.` and `ES.GFA.PD.059.`, against no feature of that id.
+ *
+ * Those truncated references are the only unresolvable ones measured in either file.
+ * A reader taking every unresolved reference as an absence would also swallow a genuinely
+ * broken join, so the two are separated here and every other unresolved reference raises.
+ */
+export declare function referenceStatesNoValue(key: string): boolean;
+/**
+ * What a cadastre's reader has read from one address, before it becomes a row.
+ *
+ * Each of the four readers resolves these from a different file layout, and `cadastralRow` turns any
+ * of them into the same row, so the layout stays in the adapter and the row shape has one home.
+ */
+export interface CadastralAddress {
+    /**
+     * The thoroughfare name the address references, absent where the publisher writes none.
+     */
+    street: string | null;
+    /**
+     * The house number this cadastre's own designator type states.
+     */
+    house: string | null;
+    /**
+     * The postal descriptor's code.
+     */
+    postcode: string | null;
+    /**
+     * The municipality-level administrative unit's name.
+     */
+    locality: string | null;
+    /**
+     * The address-area name.
+     * It is a settlement within the municipality.
+     */
+    settlement: string | null;
+    /**
+     * The publisher's own identifier for the address.
+     * It keys `source_id` where it exists.
+     */
+    addressID: string | null;
+}
+/**
+ * What distinguishes one cadastre's rows from another's.
+ *
+ * The license differs per publisher, so it is passed rather than derived: Gipuzkoa's register
+ * entry elects `CC-BY-SA-4.0` where the other three elect no share-alike obligation.
+ */
+export interface CadastralSource {
+    adapterID: string;
+    license: string;
+}
+/**
+ * One canonical row from what a Spanish cadastre states about one address.
+ *
+ * All four cadastres publish the INSPIRE Addresses theme, so they agree on how an
+ * address becomes a row even though each writes a different file layout.
+ * The municipality supplies the locality.
+ *
+ * The address area supplies a dependent locality only where the two differ.
+ * The line renders in Spain's order.
+ *
+ * Each adapter would otherwise repeat those 42 lines four times.
+ *
+ * Returns `null` in two cases.
+ * The address holds neither a municipality nor an address area, so the row would have no locality.
+ *
+ * Otherwise `formatAddressRow` returns `null` for the components it was given.
+ */
+export declare function cadastralRow(address: CadastralAddress, source: CadastralSource): CanonicalRow | null;
+//# sourceMappingURL=cadastre.d.ts.map

package/out/es/cadastre.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"cadastre.d.ts","sourceRoot":"","sources":["../../lib/es/cadastre.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAMH,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,QAAQ,CAAA;AAc1C;;;;;;;;GAQG;AACH,wBAAgB,WAAW,CAAC,KAAK,EAAE,MAAM,GAAG,OAAO,CAElD;AAED;;;;;;GAMG;AACH,wBAAgB,kBAAkB,CACjC,MAAM,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,MAAM,EAAE,CAAC,EAC9C,GAAG,KAAK,EAAE,SAAS,MAAM,EAAE,GACzB,MAAM,GAAG,IAAI,CAMf;AAED;;;;;;;;;;;;GAYG;AACH,wBAAgB,sBAAsB,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO,CAE3D;AAED;;;;;GAKG;AACH,MAAM,WAAW,gBAAgB;IAChC;;OAEG;IACH,MAAM,EAAE,MAAM,GAAG,IAAI,CAAA;IACrB;;OAEG;IACH,KAAK,EAAE,MAAM,GAAG,IAAI,CAAA;IACpB;;OAEG;IACH,QAAQ,EAAE,MAAM,GAAG,IAAI,CAAA;IACvB;;OAEG;IACH,QAAQ,EAAE,MAAM,GAAG,IAAI,CAAA;IACvB;;;OAGG;IACH,UAAU,EAAE,MAAM,GAAG,IAAI,CAAA;IACzB;;;OAGG;IACH,SAAS,EAAE,MAAM,GAAG,IAAI,CAAA;CACxB;AAED;;;;;GAKG;AACH,MAAM,WAAW,eAAe;IAC/B,SAAS,EAAE,MAAM,CAAA;IACjB,OAAO,EAAE,MAAM,CAAA;CACf;AAED;;;;;;;;;;;;;;;;GAgBG;AACH,wBAAgB,YAAY,CAAC,OAAO,EAAE,gBAAgB,EAAE,MAAM,EAAE,eAAe,GAAG,YAAY,GAAG,IAAI,CAiDpG"}

package/out/es/cadastre.js

@@ -0,0 +1,128 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * The two ways a Spanish cadastre states that it holds no value. All four cadastres share them.
+ *
+ * Spain publishes the INSPIRE Addresses theme through four cadastres: the Dirección General del
+ * Catastro for 52 provinces, and the Diputación Foral de Bizkaia, the Diputación Foral de Gipuzkoa
+ * and the Gobierno de Navarra for their own territories. Each has its own adapter, because each
+ * states its own terms and writes its own file layout. They agree on these two conventions, and
+ * reading either of them as data rather than as an absence puts a token where a corpus row expects a
+ * number or a postcode.
+ */
+import { formatAddressRow } from "@mailwoman/codex/address/format";
+import { stableSourceID } from "#adapters/source-id";
+import { designator } from "#inspire/address";
+/**
+ * The *sin número* tokens the Spanish cadastres write, lower-cased for comparison.
+ *
+ * `S-N` is the Dirección General del Catastro's spelling, on 856 of Ceuta's 8,126 addresses.
+ * `S/N` is the Gobierno de Navarra's and the Diputación Foral de Gipuzkoa's.
+ *
+ * All four readers accept both.
+ * Neither is a number under any jurisdiction's rules, and a reader scoped to one
+ * spelling would write the other spelling into a row.
+ */
+const SIN_NUMERO = new Set(["s-n", "s/n"]);
+/**
+ * Whether a designator value states that the address has no number.
+ *
+ * The whole value has to be the token.
+ * A value that holds more, such as Gipuzkoa's `S/N-A`, `S/N-C` and `S/N-5`,
+ * states something further and is kept as the publisher wrote it: what the trailing part
+ * designates is not established, dropping it would merge several addresses of one building,
+ * and rewriting it would state a number the publisher did not.
+ */
+export function isSinNumero(value) {
+    return SIN_NUMERO.has(value.trim().toLowerCase());
+}
+/**
+ * The house number a Spanish cadastre states, or null where it states that there is none.
+ *
+ * @param types The publisher's own designator types, in the caller's order of precedence.
+ * The Dirección General del Catastro writes the type as the element text `1`, while Navarra
+ * and Gipuzkoa write the `LocatorDesignatorTypeValue/addressIdentifierGeneral` codelist URI.
+ */
+export function spanishHouseNumber(byType, ...types) {
+    const value = designator(byType, ...types);
+    if (!value || isSinNumero(value))
+        return null;
+    return value;
+}
+/**
+ * Whether a component reference's key states that the publisher holds no value for that component.
+ *
+ * A Spanish cadastre's feature identifier is a dotted path whose last segment is the value:
+ * the Cadastre's postal descriptors in Ceuta are `ES.SDGC.PD.55.101.51001` through
+ * `…51005`, and Gipuzkoa's are `ES.GFA.PD.056.20830`.
+ * Where the cadastre holds no postcode it writes the same identifier with that segment empty,
+ * `ES.SDGC.PD.55.101.` and `ES.GFA.PD.059.`, against no feature of that id.
+ *
+ * Those truncated references are the only unresolvable ones measured in either file.
+ * A reader taking every unresolved reference as an absence would also swallow a genuinely
+ * broken join, so the two are separated here and every other unresolved reference raises.
+ */
+export function referenceStatesNoValue(key) {
+    return key.trim().endsWith(".");
+}
+/**
+ * One canonical row from what a Spanish cadastre states about one address.
+ *
+ * All four cadastres publish the INSPIRE Addresses theme, so they agree on how an
+ * address becomes a row even though each writes a different file layout.
+ * The municipality supplies the locality.
+ *
+ * The address area supplies a dependent locality only where the two differ.
+ * The line renders in Spain's order.
+ *
+ * Each adapter would otherwise repeat those 42 lines four times.
+ *
+ * Returns `null` in two cases.
+ * The address holds neither a municipality nor an address area, so the row would have no locality.
+ *
+ * Otherwise `formatAddressRow` returns `null` for the components it was given.
+ */
+export function cadastralRow(address, source) {
+    const place = address.locality ?? address.settlement;
+    if (!place)
+        return null;
+    const components = {};
+    if (address.house) {
+        components.house_number = address.house;
+    }
+    if (address.street) {
+        components.street = address.street;
+    }
+    if (address.postcode) {
+        components.postcode = address.postcode;
+    }
+    components.locality = place;
+    // The address area is a settlement within the municipality where the two names differ,
+    // and repeats the municipality where they do not.
+    // The repetition would state a dependent locality the publisher does not hold.
+    if (address.settlement && address.locality && address.settlement !== address.locality) {
+        components.dependent_locality = address.settlement;
+    }
+    const rendered = formatAddressRow(components, "ES", { singleLine: true });
+    if (!rendered)
+        return null;
+    const { raw, components: aligned } = rendered;
+    return {
+        raw,
+        components: aligned,
+        country: "ES",
+        locale: "es-ES",
+        // The row omits `register`.
+        // `runAdapter` stamps the adapter's own declaration onto it.
+        // A per-row `register` would repeat what the adapter already states once.
+        source: source.adapterID,
+        source_id: address.addressID
+            ? `${source.adapterID}-${address.addressID}`
+            : stableSourceID(source.adapterID, aligned),
+        corpus_version: "",
+        license: source.license,
+    };
+}
+//# sourceMappingURL=cadastre.js.map

package/out/es/cadastre.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"cadastre.js","sourceRoot":"","sources":["../../lib/es/cadastre.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAEH,OAAO,EAAE,gBAAgB,EAAE,MAAM,iCAAiC,CAAA;AAElE,OAAO,EAAE,cAAc,EAAE,MAAM,qBAAqB,CAAA;AACpD,OAAO,EAAE,UAAU,EAAE,MAAM,kBAAkB,CAAA;AAG7C;;;;;;;;;GASG;AACH,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC,CAAC,KAAK,EAAE,KAAK,CAAC,CAAC,CAAA;AAE1C;;;;;;;;GAQG;AACH,MAAM,UAAU,WAAW,CAAC,KAAa;IACxC,OAAO,UAAU,CAAC,GAAG,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,WAAW,EAAE,CAAC,CAAA;AAClD,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,kBAAkB,CACjC,MAA8C,EAC9C,GAAG,KAAwB;IAE3B,MAAM,KAAK,GAAG,UAAU,CAAC,MAAM,EAAE,GAAG,KAAK,CAAC,CAAA;IAE1C,IAAI,CAAC,KAAK,IAAI,WAAW,CAAC,KAAK,CAAC;QAAE,OAAO,IAAI,CAAA;IAE7C,OAAO,KAAK,CAAA;AACb,CAAC;AAED;;;;;;;;;;;;GAYG;AACH,MAAM,UAAU,sBAAsB,CAAC,GAAW;IACjD,OAAO,GAAG,CAAC,IAAI,EAAE,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAA;AAChC,CAAC;AAgDD;;;;;;;;;;;;;;;;GAgBG;AACH,MAAM,UAAU,YAAY,CAAC,OAAyB,EAAE,MAAuB;IAC9E,MAAM,KAAK,GAAG,OAAO,CAAC,QAAQ,IAAI,OAAO,CAAC,UAAU,CAAA;IAEpD,IAAI,CAAC,KAAK;QAAE,OAAO,IAAI,CAAA;IAEvB,MAAM,UAAU,GAA+B,EAAE,CAAA;IAEjD,IAAI,OAAO,CAAC,KAAK,EAAE,CAAC;QACnB,UAAU,CAAC,YAAY,GAAG,OAAO,CAAC,KAAK,CAAA;IACxC,CAAC;IAED,IAAI,OAAO,CAAC,MAAM,EAAE,CAAC;QACpB,UAAU,CAAC,MAAM,GAAG,OAAO,CAAC,MAAM,CAAA;IACnC,CAAC;IAED,IAAI,OAAO,CAAC,QAAQ,EAAE,CAAC;QACtB,UAAU,CAAC,QAAQ,GAAG,OAAO,CAAC,QAAQ,CAAA;IACvC,CAAC;IAED,UAAU,CAAC,QAAQ,GAAG,KAAK,CAAA;IAE3B,uFAAuF;IACvF,kDAAkD;IAClD,+EAA+E;IAC/E,IAAI,OAAO,CAAC,UAAU,IAAI,OAAO,CAAC,QAAQ,IAAI,OAAO,CAAC,UAAU,KAAK,OAAO,CAAC,QAAQ,EAAE,CAAC;QACvF,UAAU,CAAC,kBAAkB,GAAG,OAAO,CAAC,UAAU,CAAA;IACnD,CAAC;IAED,MAAM,QAAQ,GAAG,gBAAgB,CAAC,UAAU,EAAE,IAAI,EAAE,EAAE,UAAU,EAAE,IAAI,EAAE,CAAC,CAAA;IAEzE,IAAI,CAAC,QAAQ;QAAE,OAAO,IAAI,CAAA;IAE1B,MAAM,EAAE,GAAG,EAAE,UAAU,EAAE,OAAO,EAAE,GAAG,QAAQ,CAAA;IAE7C,OAAO;QACN,GAAG;QACH,UAAU,EAAE,OAAO;QACnB,OAAO,EAAE,IAAI;QACb,MAAM,EAAE,OAAO;QACf,4BAA4B;QAC5B,6DAA6D;QAC7D,0EAA0E;QAC1E,MAAM,EAAE,MAAM,CAAC,SAAS;QACxB,SAAS,EAAE,OAAO,CAAC,SAAS;YAC3B,CAAC,CAAC,GAAG,MAAM,CAAC,SAAS,IAAI,OAAO,CAAC,SAAS,EAAE;YAC5C,CAAC,CAAC,cAAc,CAAC,MAAM,CAAC,SAAS,EAAE,OAAO,CAAC;QAC5C,cAAc,EAAE,EAAE;QAClB,OAAO,EAAE,MAAM,CAAC,OAAO;KACvB,CAAA;AACF,CAAC"}