@mailwoman/corpus 10.0.0 → 10.1.0

package/out/env.d.ts

@@ -13,7 +13,9 @@ export declare const PublicCorpusEnvSchema: z.ZodObject<{
     MAILWOMAN_RESUME: z.ZodOptional<z.ZodString>;
 }, z.core.$strip>;
 /**
- * Credentials the corpus acquisition and golden-expansion tooling send. Never log their values.
+ * Credentials the corpus acquisition and golden-expansion tooling send.
+ *
+ * Never log their values.
  */
 export declare const PrivateCorpusEnvSchema: z.ZodObject<{
     OA_BATCH_TOKEN: z.ZodOptional<z.ZodString>;
@@ -36,7 +38,9 @@ export declare const $public: Record<never, never> & {
     MAILWOMAN_RESUME?: string | undefined;
 };
 /**
- * Live corpus credentials over core's. Never log their values.
+ * Live corpus credentials over core's.
+ *
+ * Never log their values.
  */
 export declare const $private: Record<never, never> & Record<string, never> & {
     OA_BATCH_TOKEN?: string | undefined;

package/out/env.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"env.d.ts","sourceRoot":"","sources":["../lib/env.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAGH,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAA;AAEvB;;GAEG;AACH,eAAO,MAAM,qBAAqB;;iBAKhC,CAAA;AAEF;;GAEG;AACH,eAAO,MAAM,sBAAsB;;;;iBAejC,CAAA;AAEF;;GAEG;AACH,eAAO,MAAM,OAAO;;;;;;;;;;;CAA6C,CAAA;AAEjE;;GAEG;AACH,eAAO,MAAM,QAAQ;;;;CAA+C,CAAA"}
+{"version":3,"file":"env.d.ts","sourceRoot":"","sources":["../lib/env.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAGH,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAA;AAEvB;;GAEG;AACH,eAAO,MAAM,qBAAqB;;iBAKhC,CAAA;AAEF;;;;GAIG;AACH,eAAO,MAAM,sBAAsB;;;;iBAejC,CAAA;AAEF;;GAEG;AACH,eAAO,MAAM,OAAO;;;;;;;;;;;CAA6C,CAAA;AAEjE;;;;GAIG;AACH,eAAO,MAAM,QAAQ;;;;CAA+C,CAAA"}

package/out/env.js

@@ -17,7 +17,9 @@ export const PublicCorpusEnvSchema = z.object({
     }),
 });
 /**
- * Credentials the corpus acquisition and golden-expansion tooling send. Never log their values.
+ * Credentials the corpus acquisition and golden-expansion tooling send.
+ *
+ * Never log their values.
  */
 export const PrivateCorpusEnvSchema = z.object({
     // OpenAddresses batch-download API token (`corpus/lib/tools/fetch/openaddresses.ts`).
@@ -40,7 +42,9 @@ export const PrivateCorpusEnvSchema = z.object({
  */
 export const $public = liveEnv(PublicCorpusEnvSchema, corePublic);
 /**
- * Live corpus credentials over core's. Never log their values.
+ * Live corpus credentials over core's.
+ *
+ * Never log their values.
  */
 export const $private = liveEnv(PrivateCorpusEnvSchema, corePrivate);
 //# sourceMappingURL=env.js.map

package/out/env.js.map

@@ -1 +1 @@
-{"version":3,"file":"env.js","sourceRoot":"","sources":["../lib/env.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAEH,OAAO,EAAE,QAAQ,IAAI,WAAW,EAAE,OAAO,IAAI,UAAU,EAAE,OAAO,EAAE,MAAM,qBAAqB,CAAA;AAC7F,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAA;AAEvB;;GAEG;AACH,MAAM,CAAC,MAAM,qBAAqB,GAAG,CAAC,CAAC,MAAM,CAAC;IAC7C,gBAAgB,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,IAAI,CAAC;QAC5C,KAAK,EAAE,cAAc;QACrB,WAAW,EAAE,qDAAqD;KAClE,CAAC;CACF,CAAC,CAAA;AAEF;;GAEG;AACH,MAAM,CAAC,MAAM,sBAAsB,GAAG,CAAC,CAAC,MAAM,CAAC;IAC9C,sFAAsF;IACtF,cAAc,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,IAAI,CAAC;QAC1C,KAAK,EAAE,2BAA2B;QAClC,WAAW,EAAE,oEAAoE;KACjF,CAAC;IACF,wDAAwD;IACxD,gBAAgB,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,IAAI,CAAC;QAC5C,KAAK,EAAE,kBAAkB;QACzB,WAAW,EAAE,wEAAwE;KACrF,CAAC;IACF,iBAAiB,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,IAAI,CAAC;QAC7C,KAAK,EAAE,mBAAmB;QAC1B,WAAW,EAAE,yEAAyE;KACtF,CAAC;CACF,CAAC,CAAA;AAEF;;GAEG;AACH,MAAM,CAAC,MAAM,OAAO,GAAG,OAAO,CAAC,qBAAqB,EAAE,UAAU,CAAC,CAAA;AAEjE;;GAEG;AACH,MAAM,CAAC,MAAM,QAAQ,GAAG,OAAO,CAAC,sBAAsB,EAAE,WAAW,CAAC,CAAA"}
+{"version":3,"file":"env.js","sourceRoot":"","sources":["../lib/env.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAEH,OAAO,EAAE,QAAQ,IAAI,WAAW,EAAE,OAAO,IAAI,UAAU,EAAE,OAAO,EAAE,MAAM,qBAAqB,CAAA;AAC7F,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAA;AAEvB;;GAEG;AACH,MAAM,CAAC,MAAM,qBAAqB,GAAG,CAAC,CAAC,MAAM,CAAC;IAC7C,gBAAgB,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,IAAI,CAAC;QAC5C,KAAK,EAAE,cAAc;QACrB,WAAW,EAAE,qDAAqD;KAClE,CAAC;CACF,CAAC,CAAA;AAEF;;;;GAIG;AACH,MAAM,CAAC,MAAM,sBAAsB,GAAG,CAAC,CAAC,MAAM,CAAC;IAC9C,sFAAsF;IACtF,cAAc,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,IAAI,CAAC;QAC1C,KAAK,EAAE,2BAA2B;QAClC,WAAW,EAAE,oEAAoE;KACjF,CAAC;IACF,wDAAwD;IACxD,gBAAgB,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,IAAI,CAAC;QAC5C,KAAK,EAAE,kBAAkB;QACzB,WAAW,EAAE,wEAAwE;KACrF,CAAC;IACF,iBAAiB,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC,QAAQ,EAAE,CAAC,IAAI,CAAC;QAC7C,KAAK,EAAE,mBAAmB;QAC1B,WAAW,EAAE,yEAAyE;KACtF,CAAC;CACF,CAAC,CAAA;AAEF;;GAEG;AACH,MAAM,CAAC,MAAM,OAAO,GAAG,OAAO,CAAC,qBAAqB,EAAE,UAAU,CAAC,CAAA;AAEjE;;;;GAIG;AACH,MAAM,CAAC,MAAM,QAAQ,GAAG,OAAO,CAAC,sBAAsB,EAAE,WAAW,CAAC,CAAA"}

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

@@ -0,0 +1,143 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * `es-bizkaia`: the INSPIRE Addresses theme the Diputación Foral de Bizkaia publishes, read from its
+ * per-municipality archives together with the component features its WFS serves.
+ *
+ * The ATOM service at `https://apli.bizkaia.eus/apps/Danok/INSPIRE/addresses.xml` lists 112
+ * municipality entries, one zipped GML each. Catastro's national feed links this service rather than
+ * serving its data, which is why the publisher is a register of its own.
+ *
+ * The archives carry addresses alone, and that is what shapes this reader. Abadiño's
+ * `ES.BFA.AD.001.zip` holds one member, `ES.BFA.AD.gml`, whose 1,360 `ad:Address` features write
+ * 4,080 `ad:component` references and not one `ad:ThoroughfareName`, `ad:PostalDescriptor` or
+ * `ad:AdminUnitName` feature. Every reference is a stored-query URL on
+ * `https://geo.bizkaia.eus/arcgisserverinspire/…` addressing an `adComponent.<n>` the ATOM feed never
+ * publishes. A reader of the archive alone therefore writes every address in the province without
+ * its street, its postcode or its municipality.
+ *
+ * Those features are on the publisher's WFS, which serves the whole province in three requests. The
+ * service answers `ad:ThoroughfareName` with `numberMatched="6921"`, `ad:PostalDescriptor` with
+ * `96` and `ad:AdminUnitName` with `113`, and a `GetFeature` with `COUNT=10000` returns each type
+ * whole: 6,921 thoroughfare names arrive in one 13,916,056-byte document. One request per type is
+ * also the only way to read them. The capabilities document does not advertise
+ * `ImplementsResultPaging` and its `CountDefault` is 1,000, so a reader that paged a type would
+ * receive the first page again. Acquisition saves those three documents and the adapter reads a
+ * directory holding them beside the municipality archives.
+ *
+ * The join is therefore across files, and the index is complete before the first address is read:
+ * the first pass reads every input for component features and the second reads every input for
+ * addresses. An unresolved reference is a value the reader asked for and could not read, and it
+ * raises.
+ *
+ * Two of the publisher's own spellings decide what a row carries.
+ *
+ * A thoroughfare name is a cadastral field rather than a street name:
+ * `BO\TRAÑA-MATIENA(1)` is the street-type abbreviation `BO`, a backslash, the name, and the
+ * street's own sequence code in parentheses. All 6,921 take that form. The row keeps the
+ * abbreviation, as the Dirección General del Catastro's own rendered line does with `CL SANTIAGO
+ * APOSTOL`, writes a space where the cadastre writes its separator, and drops the sequence code,
+ * which is an identifier rather than part of the name.
+ *
+ * The house number is the last of the four dot-separated fields of the designator, and the
+ * designator is a cadastral code: `48.001.00026.003` is municipality 48001, street 00026 and number
+ * 003. All 1,360 of Abadiño's designators have exactly four fields, its street field is five digits
+ * wide on 1,360 of 1,360, and its number field is fixed-width three. The row writes that field with
+ * its padding removed and every other character as written. Four shapes appear, counted over Abadiño:
+ *
+ * | shape | count | written |
+ * | --- | --- | --- |
+ * | `003` | 688 | `3` |
+ * | `008A` | 470 | `8A` |
+ * | `001BIS`, `004IZ`, `004DR` | 21 | `1BIS`, `4IZ`, `4DR` |
+ * | `033005`, `019A02` | 181 | `33005`, `19A02` |
+ *
+ * The first three are an ordinary Spanish number, a number with a letter, and a number with a
+ * *bis*, *izquierda* or *derecha* qualifier. The fourth is two fields concatenated and what its
+ * second field designates is not established: it is three digits wide like the number itself, and
+ * no document the publisher publishes says what it counts. Those 181 of 1,360 rows therefore carry the
+ * publisher's own concatenation with its padding removed, which transcribes the designator without
+ * interpreting it. Splitting the field on a guessed separator would state a number the cadastre does
+ * not hold.
+ *
+ * The municipality is `ad:AdminUnitName` at `3rdOrder`, which is the only level the service
+ * publishes: 113 of 113 of its administrative-unit features state it, and their names are padded
+ * with trailing spaces that `#inspire/address` trims. A reader keyed on the `4thOrder` that Czechia,
+ * Slovakia and Gipuzkoa use finds no municipality here.
+ *
+ * The publisher's own cross-rendering is `ad:LocatorName`, populated on 1,360 of 1,360 addresses:
+ * `48 001 CL\TRAÑAETXOSTE(00066) 002` for the address whose designator is `48.001.00066.002`. It
+ * restates the municipality code, the street type and name, the street code zero-padded to five, and
+ * the number field, and it is written inside the archive while the street itself comes from a
+ * separate WFS document. Putting the resolved street and the composed number beside it therefore
+ * checks that the right feature was joined, which is what the adapter's test does. The register's
+ * personal-data review reads these values as internal cadastral references rather than names, and
+ * the adapter writes none of them into a row.
+ *
+ * No service either publisher runs was found to render a postal address for this province: the
+ * ArcGIS layer behind the WFS exposes INSPIRE fields alone. So the number's unexplained second field
+ * stays an open question rather than one this adapter resolved.
+ */
+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_BIZKAIA_ADAPTER_ID = "es-bizkaia";
+/**
+ * The jurisdiction this adapter emits.
+ */
+export declare const ES_BIZKAIA_COUNTRIES: readonly string[];
+/**
+ * The license the address-source register elected for this publisher.
+ *
+ * Bizkaia names no license instrument, so no SPDX identifier is written and this records the
+ * publisher's own words. Its INSPIRE metadata carries the controlled values `No limitations to
+ * public access` and `No conditions to access and use`, and its feed-level `rights` adds the
+ * condition that authorship and ownership of the BFA be stated as «©Bizkaiko Foru Aldundia». The
+ * register records two contradictions it did not resolve: all 112 municipality entries also read
+ * `Copyright (c) 2017", ES.BFA; all rights reserved`, and Catastro's national feed labels this
+ * publisher `CC BY 3.0 ES`, which no document Bizkaia publishes names.
+ */
+export declare const ES_BIZKAIA_LICENSE = "No conditions to access and use, with attribution to \u00AB\u00A9Bizkaiko Foru Aldundia\u00BB";
+/**
+ * The street name and type a thoroughfare feature states, without its cadastral decoration.
+ *
+ * `BO\TRAÑA-MATIENA(1)` answers `BO TRAÑA-MATIENA`.
+ * The backslash is the cadastre's separator between the street-type abbreviation
+ * and the name, and the trailing parenthesis holds the street's sequence code,
+ * which the address's own designator repeats as its third field.
+ *
+ * @returns The street, or undefined where the feature states no name.
+ */
+export declare function bizkaiaStreet(feature: MarkupElement): string | undefined;
+/**
+ * The street's sequence code a thoroughfare feature states, as the feature writes it.
+ *
+ * The code the address's designator joins on, which is what makes the publisher's own
+ * `ad:LocatorName` a check of the join rather than a copy of it.
+ *
+ * @returns The digits inside the trailing parentheses, or undefined where there are none.
+ */
+export declare function bizkaiaStreetCode(feature: MarkupElement): string | undefined;
+/**
+ * The house number the designator's fourth field states.
+ *
+ * The field is fixed-width three.
+ * The row writes it with that padding removed and every other character as written.
+ *
+ * This module's header tabulates the four shapes it takes, with their counts.
+ *
+ * @returns The number, or undefined where the designator is absent or does not carry four fields.
+ */
+export declare function bizkaiaHouseNumber(byType: ReadonlyMap<string, readonly string[]>): string | undefined;
+export declare function createESBizkaiaAdapter(): CorpusAdapter;
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
+export declare const esBizkaiaAdapter: CorpusAdapter;
+//# sourceMappingURL=adapter.d.ts.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.d.ts","sourceRoot":"","sources":["../../../../lib/es/adapters/bizkaia/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAiFG;AAGH,OAAO,EAAE,KAAK,aAAa,EAAwB,MAAM,+BAA+B,CAAA;AAqBxF,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;;;;;;;;;;GAUG;AACH,eAAO,MAAM,kBAAkB,kGAAmF,CAAA;AA2ClH;;;;;;;;;GASG;AACH,wBAAgB,aAAa,CAAC,OAAO,EAAE,aAAa,GAAG,MAAM,GAAG,SAAS,CAYxE;AAED;;;;;;;GAOG;AACH,wBAAgB,iBAAiB,CAAC,OAAO,EAAE,aAAa,GAAG,MAAM,GAAG,SAAS,CAE5E;AAED;;;;;;;;;GASG;AACH,wBAAgB,kBAAkB,CAAC,MAAM,EAAE,WAAW,CAAC,MAAM,EAAE,SAAS,MAAM,EAAE,CAAC,GAAG,MAAM,GAAG,SAAS,CAYrG;AAED,wBAAgB,sBAAsB,IAAI,aAAa,CA8DtD;AAyFD;;GAEG;AACH,eAAO,MAAM,gBAAgB,eAA2B,CAAA"}

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

@@ -0,0 +1,332 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * `es-bizkaia`: the INSPIRE Addresses theme the Diputación Foral de Bizkaia publishes, read from its
+ * per-municipality archives together with the component features its WFS serves.
+ *
+ * The ATOM service at `https://apli.bizkaia.eus/apps/Danok/INSPIRE/addresses.xml` lists 112
+ * municipality entries, one zipped GML each. Catastro's national feed links this service rather than
+ * serving its data, which is why the publisher is a register of its own.
+ *
+ * The archives carry addresses alone, and that is what shapes this reader. Abadiño's
+ * `ES.BFA.AD.001.zip` holds one member, `ES.BFA.AD.gml`, whose 1,360 `ad:Address` features write
+ * 4,080 `ad:component` references and not one `ad:ThoroughfareName`, `ad:PostalDescriptor` or
+ * `ad:AdminUnitName` feature. Every reference is a stored-query URL on
+ * `https://geo.bizkaia.eus/arcgisserverinspire/…` addressing an `adComponent.<n>` the ATOM feed never
+ * publishes. A reader of the archive alone therefore writes every address in the province without
+ * its street, its postcode or its municipality.
+ *
+ * Those features are on the publisher's WFS, which serves the whole province in three requests. The
+ * service answers `ad:ThoroughfareName` with `numberMatched="6921"`, `ad:PostalDescriptor` with
+ * `96` and `ad:AdminUnitName` with `113`, and a `GetFeature` with `COUNT=10000` returns each type
+ * whole: 6,921 thoroughfare names arrive in one 13,916,056-byte document. One request per type is
+ * also the only way to read them. The capabilities document does not advertise
+ * `ImplementsResultPaging` and its `CountDefault` is 1,000, so a reader that paged a type would
+ * receive the first page again. Acquisition saves those three documents and the adapter reads a
+ * directory holding them beside the municipality archives.
+ *
+ * The join is therefore across files, and the index is complete before the first address is read:
+ * the first pass reads every input for component features and the second reads every input for
+ * addresses. An unresolved reference is a value the reader asked for and could not read, and it
+ * raises.
+ *
+ * Two of the publisher's own spellings decide what a row carries.
+ *
+ * A thoroughfare name is a cadastral field rather than a street name:
+ * `BO\TRAÑA-MATIENA(1)` is the street-type abbreviation `BO`, a backslash, the name, and the
+ * street's own sequence code in parentheses. All 6,921 take that form. The row keeps the
+ * abbreviation, as the Dirección General del Catastro's own rendered line does with `CL SANTIAGO
+ * APOSTOL`, writes a space where the cadastre writes its separator, and drops the sequence code,
+ * which is an identifier rather than part of the name.
+ *
+ * The house number is the last of the four dot-separated fields of the designator, and the
+ * designator is a cadastral code: `48.001.00026.003` is municipality 48001, street 00026 and number
+ * 003. All 1,360 of Abadiño's designators have exactly four fields, its street field is five digits
+ * wide on 1,360 of 1,360, and its number field is fixed-width three. The row writes that field with
+ * its padding removed and every other character as written. Four shapes appear, counted over Abadiño:
+ *
+ * | shape | count | written |
+ * | --- | --- | --- |
+ * | `003` | 688 | `3` |
+ * | `008A` | 470 | `8A` |
+ * | `001BIS`, `004IZ`, `004DR` | 21 | `1BIS`, `4IZ`, `4DR` |
+ * | `033005`, `019A02` | 181 | `33005`, `19A02` |
+ *
+ * The first three are an ordinary Spanish number, a number with a letter, and a number with a
+ * *bis*, *izquierda* or *derecha* qualifier. The fourth is two fields concatenated and what its
+ * second field designates is not established: it is three digits wide like the number itself, and
+ * no document the publisher publishes says what it counts. Those 181 of 1,360 rows therefore carry the
+ * publisher's own concatenation with its padding removed, which transcribes the designator without
+ * interpreting it. Splitting the field on a guessed separator would state a number the cadastre does
+ * not hold.
+ *
+ * The municipality is `ad:AdminUnitName` at `3rdOrder`, which is the only level the service
+ * publishes: 113 of 113 of its administrative-unit features state it, and their names are padded
+ * with trailing spaces that `#inspire/address` trims. A reader keyed on the `4thOrder` that Czechia,
+ * Slovakia and Gipuzkoa use finds no municipality here.
+ *
+ * The publisher's own cross-rendering is `ad:LocatorName`, populated on 1,360 of 1,360 addresses:
+ * `48 001 CL\TRAÑAETXOSTE(00066) 002` for the address whose designator is `48.001.00066.002`. It
+ * restates the municipality code, the street type and name, the street code zero-padded to five, and
+ * the number field, and it is written inside the archive while the street itself comes from a
+ * separate WFS document. Putting the resolved street and the composed number beside it therefore
+ * checks that the right feature was joined, which is what the adapter's test does. The register's
+ * personal-data review reads these values as internal cadastral references rather than names, and
+ * the adapter writes none of them into a row.
+ *
+ * No service either publisher runs was found to render a postal address for this province: the
+ * ArcGIS layer behind the WFS exposes INSPIRE fields alone. So the number's unexplained second field
+ * stays an open question rather than one this adapter resolved.
+ */
+import { openReadStream } from "@mailwoman/core/fs/streams";
+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 } from "#es/cadastre";
+import { adminUnitLevel, componentHrefs, componentJoinKey, designator, designatorsByType, inspireNameIs, placeName, postalDescriptorCode, thoroughfareName, voidDesignatorTypes, } from "#inspire/address";
+import { inspireGMLChunks } from "#inspire/archive";
+import { InspireArchiveError, UnresolvedComponentReferenceError, 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_BIZKAIA_ADAPTER_ID = "es-bizkaia";
+/**
+ * The jurisdiction this adapter emits.
+ */
+export const ES_BIZKAIA_COUNTRIES = ["ES"];
+/**
+ * The license the address-source register elected for this publisher.
+ *
+ * Bizkaia names no license instrument, so no SPDX identifier is written and this records the
+ * publisher's own words. Its INSPIRE metadata carries the controlled values `No limitations to
+ * public access` and `No conditions to access and use`, and its feed-level `rights` adds the
+ * condition that authorship and ownership of the BFA be stated as «©Bizkaiko Foru Aldundia». The
+ * register records two contradictions it did not resolve: all 112 municipality entries also read
+ * `Copyright (c) 2017", ES.BFA; all rights reserved`, and Catastro's national feed labels this
+ * publisher `CC BY 3.0 ES`, which no document Bizkaia publishes names.
+ */
+export const ES_BIZKAIA_LICENSE = "No conditions to access and use, with attribution to «©Bizkaiko Foru Aldundia»";
+/**
+ * The member of a municipality's archive that holds the GML.
+ *
+ * Each archive ships one member, named `ES.BFA.AD.gml` whichever municipality it holds,
+ * so the extension selects it and the archive's own name carries the municipality.
+ */
+const GML_MEMBER = /\.gml$/iu;
+/**
+ * The feature types the saved WFS documents carry, indexed before any address is read.
+ */
+const COMPONENT_ELEMENTS = [
+    "ad:ThoroughfareName",
+    "ad:PostalDescriptor",
+    "ad:AdminUnitName",
+    "ad:AddressAreaName",
+];
+/**
+ * The administrative level that is the municipality.
+ *
+ * The only level the service publishes, on 113 of 113 `ad:AdminUnitName` features.
+ */
+const MUNICIPALITY_LEVEL = "3rdOrder";
+/**
+ * The designator type the publisher writes.
+ *
+ * The href the publisher writes repeats the value,
+ * `…/LocatorDesignatorTypeValue/buildingIdentifier/buildingIdentifier`, and `codelistValue`
+ * reads the final segment, so the type name is the same as everyone else's.
+ */
+const HOUSE_NUMBER_TYPE = "buildingIdentifier";
+/**
+ * How many dot-separated fields a designator carries.
+ *
+ * Municipality, street, number: `48.001.00026.003` splits the municipality code across two.
+ */
+const DESIGNATOR_FIELDS = 4;
+/**
+ * The street name and type a thoroughfare feature states, without its cadastral decoration.
+ *
+ * `BO\TRAÑA-MATIENA(1)` answers `BO TRAÑA-MATIENA`.
+ * The backslash is the cadastre's separator between the street-type abbreviation
+ * and the name, and the trailing parenthesis holds the street's sequence code,
+ * which the address's own designator repeats as its third field.
+ *
+ * @returns The street, or undefined where the feature states no name.
+ */
+export function bizkaiaStreet(feature) {
+    const text = thoroughfareName(feature);
+    if (!text)
+        return undefined;
+    const street = text
+        .replace(/\((\d+)\)\s*$/u, "")
+        .replaceAll("\\", " ")
+        .replaceAll(/\s+/gu, " ")
+        .trim();
+    return street || undefined;
+}
+/**
+ * The street's sequence code a thoroughfare feature states, as the feature writes it.
+ *
+ * The code the address's designator joins on, which is what makes the publisher's own
+ * `ad:LocatorName` a check of the join rather than a copy of it.
+ *
+ * @returns The digits inside the trailing parentheses, or undefined where there are none.
+ */
+export function bizkaiaStreetCode(feature) {
+    return /\((\d+)\)\s*$/u.exec(thoroughfareName(feature) ?? "")?.[1];
+}
+/**
+ * The house number the designator's fourth field states.
+ *
+ * The field is fixed-width three.
+ * The row writes it with that padding removed and every other character as written.
+ *
+ * This module's header tabulates the four shapes it takes, with their counts.
+ *
+ * @returns The number, or undefined where the designator is absent or does not carry four fields.
+ */
+export function bizkaiaHouseNumber(byType) {
+    const code = designator(byType, HOUSE_NUMBER_TYPE);
+    if (!code)
+        return undefined;
+    const fields = code.split(".");
+    if (fields.length !== DESIGNATOR_FIELDS)
+        return undefined;
+    const number = fields[DESIGNATOR_FIELDS - 1].replace(/^0+(?=.)/u, "");
+    return number || undefined;
+}
+export function createESBizkaiaAdapter() {
+    return {
+        id: ES_BIZKAIA_ADAPTER_ID,
+        defaultLicense: ES_BIZKAIA_LICENSE,
+        addressRole: AddressRole.Premise,
+        register: SourceRegister.BizkaiaInspireAddresses,
+        surface: SurfaceOrigin.Rendered,
+        description: "INSPIRE Addresses (Diputación Foral de Bizkaia): house-number-level addresses for the province, from 112 municipality archives joined against saved WFS component pages.",
+        async *rows(opts) {
+            if (opts.country && !ES_BIZKAIA_COUNTRIES.includes(opts.country)) {
+                throw new UnsupportedCountryError(ES_BIZKAIA_ADAPTER_ID, ES_BIZKAIA_COUNTRIES, opts.country);
+            }
+            const inputs = await inputPaths(opts.inputPath);
+            const referenced = new Map();
+            for (const path of inputs) {
+                if (opts.signal?.aborted)
+                    return;
+                for await (const feature of streamMarkupElements(memberChunks(path), COMPONENT_ELEMENTS, { xml: true })) {
+                    const id = feature.attributes["gml:id"];
+                    if (id) {
+                        referenced.set(id, feature);
+                    }
+                }
+            }
+            // Every address references a street, a postcode and a municipality, so an empty
+            // index turns each row into a row with no place rather than into fewer rows.
+            if (!referenced.size) {
+                throw new InspireArchiveError(ES_BIZKAIA_ADAPTER_ID, `${PathBuilder.from(opts.inputPath).toString()} holds no component feature. The municipality archives carry addresses alone, and the features they reference come from the publisher's WFS.`);
+            }
+            let emitted = 0;
+            for (const path of inputs) {
+                if (opts.signal?.aborted)
+                    break;
+                if (opts.limit !== undefined && emitted >= opts.limit)
+                    break;
+                for await (const address of streamMarkupElements(memberChunks(path), "ad:Address", { xml: true })) {
+                    if (opts.signal?.aborted)
+                        break;
+                    if (opts.limit !== undefined && emitted >= opts.limit)
+                        break;
+                    const row = composeRow(address, referenced);
+                    if (!row)
+                        continue;
+                    yield row;
+                    emitted++;
+                }
+            }
+        },
+    };
+}
+/**
+ * One address's row, or undefined when it carries too little to render.
+ */
+function composeRow(address, referenced) {
+    const addressID = address.attributes["gml:id"]?.trim();
+    if (voidDesignatorTypes(address).has(HOUSE_NUMBER_TYPE)) {
+        throw new VoidDesignatorError(ES_BIZKAIA_ADAPTER_ID, addressID, HOUSE_NUMBER_TYPE);
+    }
+    let street;
+    let postcode;
+    let locality;
+    let settlement;
+    for (const href of componentHrefs(address)) {
+        const key = componentJoinKey(href);
+        const feature = key ? referenced.get(key) : undefined;
+        if (!feature)
+            throw new UnresolvedComponentReferenceError(ES_BIZKAIA_ADAPTER_ID, addressID, href);
+        const name = feature.name;
+        if (inspireNameIs(name, "ad:ThoroughfareName")) {
+            street = bizkaiaStreet(feature);
+        }
+        else if (inspireNameIs(name, "ad:PostalDescriptor")) {
+            postcode = postalDescriptorCode(feature);
+        }
+        else if (inspireNameIs(name, "ad:AddressAreaName")) {
+            settlement = placeName(feature);
+        }
+        else if (inspireNameIs(name, "ad:AdminUnitName") && adminUnitLevel(feature) === MUNICIPALITY_LEVEL) {
+            locality = placeName(feature);
+        }
+    }
+    // Bizkaia's number is read by its own reader rather than `spanishHouseNumber`, because its
+    // designator packs a fixed-width number field that the other three cadastres do not write.
+    return cadastralRow({
+        street,
+        house: bizkaiaHouseNumber(designatorsByType(address)),
+        postcode,
+        locality,
+        settlement,
+        addressID,
+    }, { adapterID: ES_BIZKAIA_ADAPTER_ID, license: ES_BIZKAIA_LICENSE });
+}
+/**
+ * One input's bytes, whether it arrives zipped or unpacked.
+ */
+function memberChunks(path) {
+    const resolved = PathBuilder.from(path);
+    if (resolved.basename().toLowerCase().endsWith(".zip"))
+        return inspireGMLChunks(resolved, GML_MEMBER);
+    return openReadStream(resolved);
+}
+/**
+ * Every input under `inputPath`, which is one file or a directory holding several.
+ *
+ * The two passes read the same list, so an input holding addresses and an input holding
+ * component features need not be told apart: a municipality archive contributes addresses
+ * to the second pass and no feature to the first, and a saved WFS page the reverse.
+ * Deciding by extension would be wrong on a hand-assembled input that holds both,
+ * and the publisher's own naming distinguishes neither.
+ *
+ * A path holding no input raises rather than yielding zero rows, so a wrong path
+ * reports itself instead of reading as an empty publisher.
+ */
+async function inputPaths(inputPath) {
+    const path = PathBuilder.from(inputPath);
+    const basename = path.basename().toLowerCase();
+    if (basename.endsWith(".zip") || basename.endsWith(".gml") || basename.endsWith(".xml"))
+        return [path];
+    const names = await Globerator.from("*.{zip,gml,xml}", { cwd: path.toString(), absolute: false }).toSorted();
+    if (!names.length) {
+        throw new InspireArchiveError(ES_BIZKAIA_ADAPTER_ID, `${path.toString()} holds no .zip archive or saved WFS page`);
+    }
+    return names.map((name) => path(name));
+}
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
+export const esBizkaiaAdapter = createESBizkaiaAdapter();
+//# sourceMappingURL=adapter.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"adapter.js","sourceRoot":"","sources":["../../../../lib/es/adapters/bizkaia/adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAiFG;AAEH,OAAO,EAAE,cAAc,EAAE,MAAM,4BAA4B,CAAA;AAC3D,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,MAAM,cAAc,CAAA;AAC3C,OAAO,EACN,cAAc,EACd,cAAc,EACd,gBAAgB,EAChB,UAAU,EACV,iBAAiB,EACjB,aAAa,EACb,SAAS,EACT,oBAAoB,EACpB,gBAAgB,EAChB,mBAAmB,GACnB,MAAM,kBAAkB,CAAA;AACzB,OAAO,EAAE,gBAAgB,EAAE,MAAM,kBAAkB,CAAA;AACnD,OAAO,EAAE,mBAAmB,EAAE,iCAAiC,EAAE,mBAAmB,EAAE,MAAM,iBAAiB,CAAA;AAC7G,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;;;;;;;;;;GAUG;AACH,MAAM,CAAC,MAAM,kBAAkB,GAAG,gFAAgF,CAAA;AAElH;;;;;GAKG;AACH,MAAM,UAAU,GAAG,UAAU,CAAA;AAE7B;;GAEG;AACH,MAAM,kBAAkB,GAAG;IAC1B,qBAAqB;IACrB,qBAAqB;IACrB,kBAAkB;IAClB,oBAAoB;CACX,CAAA;AAEV;;;;GAIG;AACH,MAAM,kBAAkB,GAAG,UAAU,CAAA;AAErC;;;;;;GAMG;AACH,MAAM,iBAAiB,GAAG,oBAAoB,CAAA;AAE9C;;;;GAIG;AACH,MAAM,iBAAiB,GAAG,CAAC,CAAA;AAE3B;;;;;;;;;GASG;AACH,MAAM,UAAU,aAAa,CAAC,OAAsB;IACnD,MAAM,IAAI,GAAG,gBAAgB,CAAC,OAAO,CAAC,CAAA;IAEtC,IAAI,CAAC,IAAI;QAAE,OAAO,SAAS,CAAA;IAE3B,MAAM,MAAM,GAAG,IAAI;SACjB,OAAO,CAAC,gBAAgB,EAAE,EAAE,CAAC;SAC7B,UAAU,CAAC,IAAI,EAAE,GAAG,CAAC;SACrB,UAAU,CAAC,OAAO,EAAE,GAAG,CAAC;SACxB,IAAI,EAAE,CAAA;IAER,OAAO,MAAM,IAAI,SAAS,CAAA;AAC3B,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,iBAAiB,CAAC,OAAsB;IACvD,OAAO,gBAAgB,CAAC,IAAI,CAAC,gBAAgB,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;AACnE,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,UAAU,kBAAkB,CAAC,MAA8C;IAChF,MAAM,IAAI,GAAG,UAAU,CAAC,MAAM,EAAE,iBAAiB,CAAC,CAAA;IAElD,IAAI,CAAC,IAAI;QAAE,OAAO,SAAS,CAAA;IAE3B,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAA;IAE9B,IAAI,MAAM,CAAC,MAAM,KAAK,iBAAiB;QAAE,OAAO,SAAS,CAAA;IAEzD,MAAM,MAAM,GAAG,MAAM,CAAC,iBAAiB,GAAG,CAAC,CAAE,CAAC,OAAO,CAAC,WAAW,EAAE,EAAE,CAAC,CAAA;IAEtE,OAAO,MAAM,IAAI,SAAS,CAAA;AAC3B,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,0KAA0K;QAE3K,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,MAAM,MAAM,GAAG,MAAM,UAAU,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;YAC/C,MAAM,UAAU,GAAG,IAAI,GAAG,EAAyB,CAAA;YAEnD,KAAK,MAAM,IAAI,IAAI,MAAM,EAAE,CAAC;gBAC3B,IAAI,IAAI,CAAC,MAAM,EAAE,OAAO;oBAAE,OAAM;gBAEhC,IAAI,KAAK,EAAE,MAAM,OAAO,IAAI,oBAAoB,CAAC,YAAY,CAAC,IAAI,CAAC,EAAE,kBAAkB,EAAE,EAAE,GAAG,EAAE,IAAI,EAAE,CAAC,EAAE,CAAC;oBACzG,MAAM,EAAE,GAAG,OAAO,CAAC,UAAU,CAAC,QAAQ,CAAC,CAAA;oBAEvC,IAAI,EAAE,EAAE,CAAC;wBACR,UAAU,CAAC,GAAG,CAAC,EAAE,EAAE,OAAO,CAAC,CAAA;oBAC5B,CAAC;gBACF,CAAC;YACF,CAAC;YAED,gFAAgF;YAChF,6EAA6E;YAC7E,IAAI,CAAC,UAAU,CAAC,IAAI,EAAE,CAAC;gBACtB,MAAM,IAAI,mBAAmB,CAC5B,qBAAqB,EACrB,GAAG,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,QAAQ,EAAE,8IAA8I,CAC5L,CAAA;YACF,CAAC;YAED,IAAI,OAAO,GAAG,CAAC,CAAA;YAEf,KAAK,MAAM,IAAI,IAAI,MAAM,EAAE,CAAC;gBAC3B,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,IAAI,KAAK,EAAE,MAAM,OAAO,IAAI,oBAAoB,CAAC,YAAY,CAAC,IAAI,CAAC,EAAE,YAAY,EAAE,EAAE,GAAG,EAAE,IAAI,EAAE,CAAC,EAAE,CAAC;oBACnG,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,EAAE,UAAU,CAAC,CAAA;oBAE3C,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,EAAE,UAA8C;IACzF,MAAM,SAAS,GAAG,OAAO,CAAC,UAAU,CAAC,QAAQ,CAAC,EAAE,IAAI,EAAE,CAAA;IAEtD,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,IAAI,MAA0B,CAAA;IAC9B,IAAI,QAA4B,CAAA;IAChC,IAAI,QAA4B,CAAA;IAChC,IAAI,UAA8B,CAAA;IAElC,KAAK,MAAM,IAAI,IAAI,cAAc,CAAC,OAAO,CAAC,EAAE,CAAC;QAC5C,MAAM,GAAG,GAAG,gBAAgB,CAAC,IAAI,CAAC,CAAA;QAClC,MAAM,OAAO,GAAG,GAAG,CAAC,CAAC,CAAC,UAAU,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,SAAS,CAAA;QAErD,IAAI,CAAC,OAAO;YAAE,MAAM,IAAI,iCAAiC,CAAC,qBAAqB,EAAE,SAAS,EAAE,IAAI,CAAC,CAAA;QAEjG,MAAM,IAAI,GAAG,OAAO,CAAC,IAAI,CAAA;QAEzB,IAAI,aAAa,CAAC,IAAI,EAAE,qBAAqB,CAAC,EAAE,CAAC;YAChD,MAAM,GAAG,aAAa,CAAC,OAAO,CAAC,CAAA;QAChC,CAAC;aAAM,IAAI,aAAa,CAAC,IAAI,EAAE,qBAAqB,CAAC,EAAE,CAAC;YACvD,QAAQ,GAAG,oBAAoB,CAAC,OAAO,CAAC,CAAA;QACzC,CAAC;aAAM,IAAI,aAAa,CAAC,IAAI,EAAE,oBAAoB,CAAC,EAAE,CAAC;YACtD,UAAU,GAAG,SAAS,CAAC,OAAO,CAAC,CAAA;QAChC,CAAC;aAAM,IAAI,aAAa,CAAC,IAAI,EAAE,kBAAkB,CAAC,IAAI,cAAc,CAAC,OAAO,CAAC,KAAK,kBAAkB,EAAE,CAAC;YACtG,QAAQ,GAAG,SAAS,CAAC,OAAO,CAAC,CAAA;QAC9B,CAAC;IACF,CAAC;IAED,2FAA2F;IAC3F,2FAA2F;IAC3F,OAAO,YAAY,CAClB;QACC,MAAM;QACN,KAAK,EAAE,kBAAkB,CAAC,iBAAiB,CAAC,OAAO,CAAC,CAAC;QACrD,QAAQ;QACR,QAAQ;QACR,UAAU;QACV,SAAS;KACT,EACD,EAAE,SAAS,EAAE,qBAAqB,EAAE,OAAO,EAAE,kBAAkB,EAAE,CACjE,CAAA;AACF,CAAC;AAED;;GAEG;AACH,SAAS,YAAY,CAAC,IAAqB;IAC1C,MAAM,QAAQ,GAAG,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;IAEvC,IAAI,QAAQ,CAAC,QAAQ,EAAE,CAAC,WAAW,EAAE,CAAC,QAAQ,CAAC,MAAM,CAAC;QAAE,OAAO,gBAAgB,CAAC,QAAQ,EAAE,UAAU,CAAC,CAAA;IAErG,OAAO,cAAc,CAAC,QAAQ,CAAC,CAAA;AAChC,CAAC;AAED;;;;;;;;;;;GAWG;AACH,KAAK,UAAU,UAAU,CAAC,SAA0B;IACnD,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,IAAI,QAAQ,CAAC,QAAQ,CAAC,MAAM,CAAC;QAAE,OAAO,CAAC,IAAI,CAAC,CAAA;IAEtG,MAAM,KAAK,GAAG,MAAM,UAAU,CAAC,IAAI,CAAC,iBAAiB,EAAE,EAAE,GAAG,EAAE,IAAI,CAAC,QAAQ,EAAE,EAAE,QAAQ,EAAE,KAAK,EAAE,CAAC,CAAC,QAAQ,EAAE,CAAA;IAE5G,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC;QACnB,MAAM,IAAI,mBAAmB,CAAC,qBAAqB,EAAE,GAAG,IAAI,CAAC,QAAQ,EAAE,0CAA0C,CAAC,CAAA;IACnH,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/adapters/catastro/adapter.d.ts

@@ -0,0 +1,104 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * `es-catastro`: the INSPIRE Addresses theme the Dirección General del Catastro publishes, read
+ * from its per-municipality zipped GML.
+ *
+ * The ATOM service at `https://www.catastro.hacienda.gob.es/INSPIRE/Addresses/ES.SDGC.AD.atom.xml`
+ * lists 55 entries, and 52 of them are the Dirección General del Catastro's own provinces. The other
+ * three link the foral cadastres of Bizkaia, Gipuzkoa and Navarra, which state their own terms and
+ * have an adapter each. Every DGC entry links a province feed listing one zipped GML per
+ * municipality, so a caller harvests municipality by municipality and this adapter reads one archive
+ * or a directory of them.
+ *
+ * Three properties of the file decide the reader, and each of the three would silently empty a row
+ * if it were read the way the other INSPIRE publishers are read.
+ *
+ * The archive carries two members. `A.ES.SDGC.AD.<code>.gml` is the data and
+ * `A.ES.SDGC.AD.MD.<code>.xml` is its ISO 19139 metadata record. A member selector matching `.xml`
+ * therefore selects the metadata record. That record carries no `AD:Address` at all, so
+ * {@linkcode GML_MEMBER} matches the extension that holds the features.
+ *
+ * The GML declares `<?xml version="1.0" encoding="ISO-8859-1"?>` and is not valid UTF-8: Ceuta's
+ * 11,861,975-byte member carries 26 bytes above 0x7F, one of which is the `Ñ` of a street name.
+ * `streamMarkupElements` decodes with a UTF-8 `TextDecoder`, which answers those bytes with U+FFFD,
+ * so the bytes pass through `decodeByteStream` first. {@linkcode assertLatin1Declaration} reads the
+ * declaration rather than trusting this note, because a publisher that re-encodes its output would
+ * otherwise have every accented name mojibaked under a reader that still claimed to be right.
+ *
+ * The file binds the INSPIRE namespaces to upper-case prefixes, `AD:Address` and `GN:text`, and
+ * states a designator type and an administrative level as element text rather than as a codelist
+ * URI. `#inspire/address` matches an element by its local name and accepts either form of a coded
+ * value, so the readers there serve this publisher. Only the element names this module passes to the
+ * markup stream carry the upper-case prefix.
+ *
+ * Each address references exactly three features by local `#` fragment, measured as 8,126 of 8,126
+ * addresses in Ceuta carrying one `ES.SDGC.PD`, one `ES.SDGC.TN` and one `ES.SDGC.AU` reference.
+ * Those features are written after the addresses — Ceuta's 511 of them sit in the last 530,000 bytes
+ * of the member — so the adapter reads the member twice through
+ * {@linkcode streamIndexedInspireRows} rather than holding every address until the document ends.
+ *
+ * The number is one designator, typed `1`. The publisher's own rendering fixes what that type means
+ * and what the value `S-N` means. `Consulta_DNPRC` on the OVC web service answers a cadastral
+ * reference with the address the Cadastre holds for it, as `<ldt>`, plus the fields it is built
+ * from: `tv` the street-type abbreviation, `nv` the street name, `pnp` the number, `plp` its letter,
+ * `dp` the postcode and `nm` the municipality. The cadastral reference is the tail of the address's
+ * own `gml:id`, so every row this adapter writes can be put beside the publisher's own line:
+ *
+ * | `gml:id` tail | designator | `<ldt>` |
+ * | --- | --- | --- |
+ * | `9744803TE8794S` | `13` | `CL SANTIAGO APOSTOL 13 51002 CEUTA (CEUTA)` |
+ * | `9745701TE8794S` | `13D` | `CL SANTIAGO APOSTOL 13(D) 51002 CEUTA (CEUTA)` |
+ *
+ * So the designator is `pnp` with `plp` appended, and the street is `tv` followed by `nv`, which is
+ * the `AD:ThoroughfareName` text verbatim once its leading space is trimmed. The abbreviation stays
+ * in the street component because the publisher writes it in its own rendered line. Expanding `CL`
+ * to `Calle` would state something the Cadastre did not.
+ *
+ * `S-N` is *sin número*, the publisher stating that the address has no number, and it is 856 of
+ * Ceuta's 8,126 designators. A row carrying it writes no `house_number`, because the string is a
+ * statement about the number rather than a number.
+ *
+ * One reference form states an absence the same way. The postal-descriptor reference's final segment
+ * is the postcode, and 70 of Ceuta's 8,126 addresses write `#ES.SDGC.PD.55.101.` with that segment
+ * empty, against five `AD:PostalDescriptor` features the file does publish. Those 70 are the only
+ * unresolvable references in the member: the thoroughfare and administrative-unit references resolve
+ * 8,126 of 8,126 each. A reference whose final segment is empty is therefore read as the publisher
+ * holding no value for that component, and every other unresolved reference raises.
+ */
+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_CATASTRO_ADAPTER_ID = "es-catastro";
+/**
+ * The jurisdiction this adapter emits.
+ *
+ * The 52 province feeds cover peninsular Spain, the Balearics, the Canaries, Ceuta and Melilla.
+ * The three Basque provinces and Navarra keep their own cadastres and their own adapters.
+ */
+export declare const ES_CATASTRO_COUNTRIES: readonly string[];
+/**
+ * The license the address-source register elected for this publisher.
+ *
+ * The register elects the titled instrument, `Licencia de acceso y uso de los servicios y conjuntos
+ * de datos INSPIRE de la Dirección General del Catastro` at version 1.0, rather than the older
+ * `all rights reserved` template the feed still carries. It names no license instrument a
+ * Creative Commons label or an SPDX identifier would cover, so the row records the instrument's
+ * own title.
+ *
+ * That decision reads `refused` for `redistribute-data`: clause 1 withholds distribution
+ * of the information as supplied, and requires transformation before any public use.
+ * Ingest, transformation and training are permitted, which is what this adapter performs.
+ */
+export declare const ES_CATASTRO_LICENSE = "Licencia de acceso y uso de los servicios y conjuntos de datos INSPIRE de la Direcci\u00F3n General del Catastro";
+export declare function createESCatastroAdapter(): CorpusAdapter;
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
+export declare const esCatastroAdapter: CorpusAdapter;
+//# sourceMappingURL=adapter.d.ts.map