@mailwoman/corpus 10.1.0 → 10.2.1

package/lib/es/adapters/catastro/adapter.ts

@@ -7,32 +7,32 @@
  * 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.
+ * lists 55 entries. 52 are the Dirección General del Catastro's own provinces. Three link the foral
+ * cadastres of Bizkaia, Gipuzkoa and Navarra. Those three 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
+ * The archive holds 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
+ * therefore selects the metadata record. That record holds 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 GML declares `<?xml version="1.0" encoding="ISO-8859-1"?>` and is not valid UTF-8. Ceuta's
+ * 11,861,975-byte member holds 26 bytes above 0x7F, one of which is the `Ñ` of a street name.
+ * `streamMarkupElements` decodes with a UTF-8 `TextDecoder`. That decoder answers those bytes with
+ * U+FFFD. The bytes therefore 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.
+ * markup stream hold 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.
@@ -52,10 +52,10 @@
  * | `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.
+ * So the designator is `pnp` with `plp` appended, and the street is `tv` followed by `nv`. That
+ * street matches 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. `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
@@ -114,13 +114,12 @@ export const ES_CATASTRO_COUNTRIES: readonly string[] = ["ES"]
  *
  * 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.
+ * `all rights reserved` template the feed still shows. No Creative Commons label or SPDX identifier
+ * covers that instrument, so the row records the instrument's own title.
  *
- * That decision reads `refused` for `redistribute-data`: clause 1 withholds distribution
+ * The register 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.
+ * Ingest, transformation and training are permitted, and this adapter performs those.
  */
 export const ES_CATASTRO_LICENSE =
 	"Licencia de acceso y uso de los servicios y conjuntos de datos INSPIRE de la Dirección General del Catastro"
@@ -130,7 +129,8 @@ export const ES_CATASTRO_LICENSE =
  *
  * The archive ships two members and the other is `A.ES.SDGC.AD.MD.<code>.xml`,
  * the ISO 19139 metadata record.
- * A selector matching `.xml` reads that record, which carries no `AD:Address`.
+ * A selector matching `.xml` reads that record.
+ * That record holds no `AD:Address`.
  */
 const GML_MEMBER = /\.gml$/iu
 
@@ -169,10 +169,12 @@ const MUNICIPALITY_LEVELS = new Set(["4", "4thOrder"])
 /**
  * The designator type the publisher writes on every address.
  *
- * Measured over Ceuta: 8,126 of 8,126 addresses carry exactly one designator
+ * Measured over Ceuta: 8,126 of 8,126 addresses hold exactly one designator,
  * and its `AD:type` is the text `1`.
- * The value is the house number, which `Consulta_DNPRC`'s `pnp` and `plp` fields
- * confirm for the two addresses tabulated in this module's header.
+ * The value is the house number.
+ *
+ * `Consulta_DNPRC`'s `pnp` and `plp` fields confirm that for the two addresses
+ * tabulated in this module's header.
  */
 const HOUSE_NUMBER_TYPE = "1"
 
@@ -263,22 +265,22 @@ export function createESCatastroAdapter(): CorpusAdapter {
 }
 
 /**
- * One address's row, or undefined when it carries too little to render.
+ * One address's row, or null when it holds too little to render.
  *
  * The reference index is complete before this is called, so a reference resolving to no
  * feature is a value the reader asked for and could not read, and it raises.
  */
-function composeRow(address: MarkupElement, referenced: ReadonlyMap<string, MarkupElement>): CanonicalRow | undefined {
-	const addressID = address.attributes["gml:id"]?.trim()
+function composeRow(address: MarkupElement, referenced: ReadonlyMap<string, MarkupElement>): CanonicalRow | null {
+	const addressID = address.attributes["gml:id"]?.trim() ?? null
 
 	if (voidDesignatorTypes(address).has(HOUSE_NUMBER_TYPE)) {
 		throw new VoidDesignatorError(ES_CATASTRO_ADAPTER_ID, addressID, HOUSE_NUMBER_TYPE)
 	}
 
-	let street: string | undefined
-	let postcode: string | undefined
-	let locality: string | undefined
-	let settlement: string | undefined
+	let street: string | null = null
+	let postcode: string | null = null
+	let locality: string | null = null
+	let settlement: string | null = null
 
 	for (const href of componentHrefs(address)) {
 		const key = componentJoinKey(href)

package/lib/es/adapters/gipuzkoa/adapter.ts

@@ -4,21 +4,18 @@
  * @author Teffen Ellis, et al.
  *
  * `es-gipuzkoa`: the INSPIRE Addresses theme the Diputación Foral de Gipuzkoa publishes, read from
- * the one zipped GML that carries the whole province.
+ * the one zipped GML that holds the whole province.
  *
- * The ATOM service at `https://b5m.gipuzkoa.eus/inspire/download/addresses.xml` is 4,469 bytes and
- * holds a single entry for the province rather than one per municipality. That entry links its
- * dataset through `rel="alternate"` and not `rel="enclosure"`, so a fetcher that looks only for an
- * enclosure finds no dataset here. The adapter is handed the archive and the feed is acquisition's
- * concern. The archive `ES.GFA.AD.zip` is 7,959,103 bytes and its one member `ES.GFA.AD.gml` is
- * 313,162,472 bytes, a ratio of 39, so the member is read as a chunk sequence and never buffered.
+ * The ATOM service at `https://b5m.gipuzkoa.eus/inspire/download/addresses.xml` holds a single entry
+ * for the province and links its dataset through `rel="alternate"` rather than `rel="enclosure"`, so
+ * a fetcher that looks only for an enclosure finds no dataset here. The adapter is handed the
+ * archive and the feed is acquisition's concern. The archive holds one 313 MB GML member. The
+ * adapter reads that member as a chunk sequence and never buffers it.
  *
- * The member is self-contained and its order decides the reader. Counted over all 313,162,472 bytes:
- * 68,744 `ad:Address`, 4,250 `ad:ThoroughfareName`, 178 `ad:PostalDescriptor` and 88
- * `ad:AdminUnitName`, which is every one of the 73,260 `gml:featureMember` elements. The 4,516
- * reference features are written last, beginning at byte 306,440,816. Holding an address until the
- * document ends would therefore hold all 68,744, so the adapter reads the member twice through
- * {@linkcode streamIndexedInspireRows} and indexes 4,516 features instead.
+ * The member is self-contained and its order decides the reader. Addresses number 68,744, and the
+ * 4,516 reference features they point at are written last. An address held until the document ends
+ * would keep all 68,744 pending, so the adapter reads the member twice through
+ * {@linkcode streamIndexedInspireRows} and indexes the 4,516 references instead.
  *
  * A reference is a WFS stored-query URL whose `ID` parameter equals the target's `gml:id`, written
  * against the internal host `http://b5mdev/` and with the id repeated as the URL's fragment:
@@ -28,31 +25,26 @@
  *
  * A postal-descriptor reference whose final segment is empty, `…&ID=ES.GFA.PD.059. #…`, states that
  * the publisher holds no postcode, and `referenceStatesNoValue` separates it from a broken join.
- * The Cadastre writes the same form, which is why the predicate has one home.
+ * The Cadastre writes the same form. The predicate therefore has one home.
  *
- * The locator vocabulary is one type. 68,743 of 68,744 addresses write exactly one designator, typed
- * `LocatorDesignatorTypeValue/addressIdentifierGeneral`; one address writes none. The municipality is
- * the only `ad:AdminUnitName` level published, `4thOrder`, and the province publishes 88 of them for
- * its 88 municipalities.
+ * The locator vocabulary is one type,
+ * `LocatorDesignatorTypeValue/addressIdentifierGeneral`. The municipality is the only
+ * `ad:AdminUnitName` level published, `4thOrder`.
  *
  * The publisher writes its street names in Basque, with the street type as a trailing word rather
  * than a leading abbreviation: `Samikolla ibilbidea`. The name is taken as written.
  *
- * `ad:LocatorName` is populated on 46,365 of 68,744 addresses and holds a building name as a Basque
- * and Spanish pair, `Gasolinera` with `Gasolindegia` and `Agirre, baserria` with `Agirre, caserío`.
- * The adapter writes none of them. The address-source register's personal-data review reads `absent`
- * on the ground that these name the building rather than its occupant, and it records that a Basque
- * *baserri* name is commonly also a surname. Keeping them out of the corpus keeps that reading from
- * resting on the stronger claim that the name is safe to publish as address text.
+ * `ad:LocatorName` holds a building name as a Basque and Spanish pair, `Gasolinera` with
+ * `Gasolindegia` and `Agirre, baserria` with `Agirre, caserío`. The adapter writes none of them: the
+ * address-source register's personal-data review reads `absent` because these identify the building
+ * rather than its occupant, and a Basque *baserri* name is commonly also a surname. The review
+ * therefore does not rest on the stronger claim that the name is safe to publish as address text.
  *
- * No publisher-side rendering of a whole address was found for this source. `ad:alternativeIdentifier`
- * is populated on all 68,744 addresses and holds a serial number such as `36798` rather than an
- * address. What the file does state twice is the join: the address `gml:id`
- * `ES.GFA.AD.056_1110_003` concatenates the municipality code `056`, the thoroughfare code `1110`
- * and the number `003`, and the references it carries are `AU_ADMINISTRATIVEUNIT_34162020056`,
- * `ES.GFA.TN.056.1110` and `ES.GFA.PD.056.20830`. The adapter's test checks the composed street and
- * number against that identifier, which is a check that the right features were joined rather than a
- * check of the address the publisher would print.
+ * The address `gml:id` `ES.GFA.AD.056_1110_003` concatenates the municipality code `056`, the
+ * thoroughfare code `1110` and the number `003`, and holds references
+ * `AU_ADMINISTRATIVEUNIT_34162020056`, `ES.GFA.TN.056.1110` and `ES.GFA.PD.056.20830`. The
+ * adapter's test checks the composed street and number against that identifier. That check confirms
+ * the right features were joined. It does not check the address the publisher would print.
  */
 
 import type { MarkupElement } from "@mailwoman/core/html/elements"
@@ -88,7 +80,8 @@ export const ES_GIPUZKOA_ADAPTER_ID = "es-gipuzkoa"
 /**
  * The jurisdiction this adapter emits.
  *
- * Coverage is the province of Gipuzkoa, which the feed states as the dataset's whole extent.
+ * Coverage is the province of Gipuzkoa.
+ * The feed states that as the dataset's whole extent.
  */
 export const ES_GIPUZKOA_COUNTRIES: readonly string[] = ["ES"]
 
@@ -97,15 +90,15 @@ export const ES_GIPUZKOA_COUNTRIES: readonly string[] = ["ES"]
  *
  * Gipuzkoa states four things, and the register elects the one addressing this dataset:
  * the `otherConstraints` of its ISO 19139 dataset record, `CC BY-SA 4.0`.
- * Its feed states attribution alone and its website states CC BY 4.0, and the register
- * records why the stricter reading is the one to err towards: the share-alike filter reads
- * this label, so recording CC BY 4.0 would hide an obligation the publisher may hold.
+ * Its feed states attribution only and its website states CC BY 4.0, and the register records
+ * why the stricter reading is the one to err towards: the share-alike filter reads this label,
+ * so recording CC BY 4.0 would hide an obligation the publisher may hold.
  *
  * The value is the SPDX identifier the register elected rather than the license's title,
  * because `licenseVerdict` resolves an identifier and reads a title as unrecognized.
  * Under `LicensePolicy.ShareAlikeFree` the title returns `{refusal: null, mentionsShareAlike: false}`
- * and `CC-BY-SA-4.0` returns `{refusal: "share-alike-carried"}`, so the title admits
- * these rows to a corpus assembled to hold no share-alike source.
+ * and `CC-BY-SA-4.0` returns a share-alike refusal, so the title admits these rows
+ * to a corpus assembled to hold no share-alike source.
  */
 export const ES_GIPUZKOA_LICENSE = "CC-BY-SA-4.0"
 
@@ -181,22 +174,22 @@ export function createESGipuzkoaAdapter(): CorpusAdapter {
 }
 
 /**
- * One address's row, or undefined when it carries too little to render.
+ * One address's row, or null when it holds too little to render.
  *
  * The reference index is complete before this is called, so a reference resolving to no
  * feature is a value the reader asked for and could not read, and it raises.
  */
-function composeRow(address: MarkupElement, referenced: ReadonlyMap<string, MarkupElement>): CanonicalRow | undefined {
-	const addressID = address.attributes["gml:id"]?.trim()
+function composeRow(address: MarkupElement, referenced: ReadonlyMap<string, MarkupElement>): CanonicalRow | null {
+	const addressID = address.attributes["gml:id"]?.trim() ?? null
 
 	if (voidDesignatorTypes(address).has(HOUSE_NUMBER_TYPE)) {
 		throw new VoidDesignatorError(ES_GIPUZKOA_ADAPTER_ID, addressID, HOUSE_NUMBER_TYPE)
 	}
 
-	let street: string | undefined
-	let postcode: string | undefined
-	let locality: string | undefined
-	let settlement: string | undefined
+	let street: string | null = null
+	let postcode: string | null = null
+	let locality: string | null = null
+	let settlement: string | null = null
 
 	for (const href of componentHrefs(address)) {
 		const key = componentJoinKey(href)

package/lib/es/adapters/navarra/adapter.ts

@@ -9,15 +9,15 @@
  * 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`, so a partition cannot be selected by place from the feed alone. The three measured each
- * turned out to hold one municipality, partition 1 Abáigar, partition 100 Eulate and partition 250
- * Bera, but the feed states no such mapping and this adapter claims none. It reads one archive or a
- * directory of them, which is how a caller harvests all 272.
+ * 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 carries
+ * 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 |
@@ -30,7 +30,7 @@
  * | `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, which is six per address in each, one of
+ * 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
@@ -39,7 +39,7 @@
  * 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 carries no region column, so no
+ * 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
@@ -47,12 +47,12 @@
  * `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, which is this source's publisher-side oracle.
+ * 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, and a designator of `S/N` is *sin número*, which
- * `#es/cadastre` reads for all four Spanish cadastres. A designator may also carry a space-separated
- * letter, `18 A` and `34 A` in Eulate, which is kept as the publisher wrote it.
+ * 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, streamMarkupElements } from "@mailwoman/core/html/elements"
@@ -93,10 +93,11 @@ export const ES_NAVARRA_COUNTRIES: readonly string[] = ["ES"]
  * 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 carries that identifier rather than the license's title.
- * `licenseVerdict` resolves an identifier and reads a title as unrecognized,
- * and a row whose license resolves to no expression has unknown obligations
- * rather than none, which the build reports separately from a refusal.
+ * 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"
 
@@ -146,7 +147,7 @@ const SETTLEMENT_LEVEL = "5thOrder"
 /**
  * The postcode title that states that the publisher holds none.
  *
- * Measured in Bera, which writes `0` beside its `31780`.
+ * Bera writes `0` beside its `31780`.
  */
 const NO_POSTCODE_TITLE = "0"
 
@@ -154,10 +155,10 @@ const NO_POSTCODE_TITLE = "0"
  * What one address's references state, read from their titles.
  */
 interface ReferencedTitles {
-	street?: string
-	postcode?: string
-	locality?: string
-	settlement?: string
+	street: string | null
+	postcode: string | null
+	locality: string | null
+	settlement: string | null
 }
 
 /**
@@ -167,7 +168,7 @@ interface ReferencedTitles {
  * order in the three partitions measured and that order is the publisher's rather than the schema's.
  */
 export function navarraReferencedTitles(address: MarkupElement): ReferencedTitles {
-	const titles: ReferencedTitles = {}
+	const titles: ReferencedTitles = { street: null, postcode: null, locality: null, settlement: null }
 
 	for (const { href, title } of componentLinks(address)) {
 		if (!title) continue
@@ -177,7 +178,7 @@ export function navarraReferencedTitles(address: MarkupElement): ReferencedTitle
 		if (lowered === THOROUGHFARE_HREF) {
 			titles.street = title
 		} else if (lowered === POSTAL_DESCRIPTOR_HREF) {
-			titles.postcode = title === NO_POSTCODE_TITLE ? undefined : title
+			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)) {
@@ -234,10 +235,10 @@ export function createESNavarraAdapter(): CorpusAdapter {
 }
 
 /**
- * One address's row, or undefined when it carries too little to render.
+ * One address's row, or null when it holds too little to render.
  */
-function composeRow(address: MarkupElement): CanonicalRow | undefined {
-	const addressID = address.attributes["gml:id"]?.trim()
+function composeRow(address: MarkupElement): CanonicalRow | null {
+	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)
@@ -245,8 +246,8 @@ function composeRow(address: MarkupElement): CanonicalRow | undefined {
 
 	const { street, postcode, locality, settlement } = navarraReferencedTitles(address)
 
-	// Navarra's settlement is a fifth-order title, a *concejo* within the municipality, which
-	// `cadastralRow` writes as a dependent locality only where it differs from the municipality.
+	// 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,

package/lib/es/cadastre.ts

@@ -3,7 +3,7 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  *
- * The two ways a Spanish cadastre states that it holds no value, which all four of them share.
+ * 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
@@ -25,10 +25,9 @@ import type { CanonicalRow } from "#types"
  * `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.
  *
- * Both are accepted by each of the four readers.
- * Neither is a number under any jurisdiction's rules, and the files measured are a few municipalities
- * out of the 52 provinces, 112 municipalities and 272 partitions those services publish,
- * so a reader scoped to the one spelling its own sample carried would write the other into a row.
+ * 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"])
 
@@ -36,7 +35,7 @@ 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 carries more, such as Gipuzkoa's `S/N-A`, `S/N-C` and `S/N-5`,
+ * 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.
@@ -46,7 +45,7 @@ export function isSinNumero(value: string): boolean {
 }
 
 /**
- * The house number a Spanish cadastre states, or undefined where it states that there is none.
+ * 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
@@ -55,10 +54,10 @@ export function isSinNumero(value: string): boolean {
 export function spanishHouseNumber(
 	byType: ReadonlyMap<string, readonly string[]>,
 	...types: readonly string[]
-): string | undefined {
+): string | null {
 	const value = designator(byType, ...types)
 
-	if (!value || isSinNumero(value)) return undefined
+	if (!value || isSinNumero(value)) return null
 
 	return value
 }
@@ -72,9 +71,7 @@ export function spanishHouseNumber(
  * 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: 70 of Ceuta's 8,126 addresses carry one, while its thoroughfare
- * and administrative-unit references resolve 8,126 of 8,126 each.
+ * 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.
  */
@@ -92,27 +89,29 @@ export interface CadastralAddress {
 	/**
 	 * The thoroughfare name the address references, absent where the publisher writes none.
 	 */
-	street: string | undefined
+	street: string | null
 	/**
 	 * The house number this cadastre's own designator type states.
 	 */
-	house: string | undefined
+	house: string | null
 	/**
 	 * The postal descriptor's code.
 	 */
-	postcode: string | undefined
+	postcode: string | null
 	/**
 	 * The municipality-level administrative unit's name.
 	 */
-	locality: string | undefined
+	locality: string | null
 	/**
-	 * The address-area name, which is a settlement within the municipality.
+	 * The address-area name.
+	 * It is a settlement within the municipality.
 	 */
-	settlement: string | undefined
+	settlement: string | null
 	/**
-	 * The publisher's own identifier for the address, which keys `source_id` where it exists.
+	 * The publisher's own identifier for the address.
+	 * It keys `source_id` where it exists.
 	 */
-	addressID: string | undefined
+	addressID: string | null
 }
 
 /**
@@ -136,16 +135,17 @@ export interface CadastralSource {
  * The address area supplies a dependent locality only where the two differ.
  * The line renders in Spain's order.
  *
- * Composing that in each adapter repeated 42 lines four times.
+ * Each adapter would otherwise repeat those 42 lines four times.
  *
- * Returns `undefined` in two cases: the address carries neither a municipality
- * nor an address area, which would leave the row without a locality, or `formatAddressRow`
- * returns `undefined` for the components it was given.
+ * 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: CadastralAddress, source: CadastralSource): CanonicalRow | undefined {
+export function cadastralRow(address: CadastralAddress, source: CadastralSource): CanonicalRow | null {
 	const place = address.locality ?? address.settlement
 
-	if (!place) return undefined
+	if (!place) return null
 
 	const components: CanonicalRow["components"] = {}
 
@@ -165,14 +165,14 @@ export function cadastralRow(address: CadastralAddress, source: CadastralSource)
 
 	// The address area is a settlement within the municipality where the two names differ,
 	// and repeats the municipality where they do not.
-	// Writing the repetition would state a dependent locality the publisher does not hold.
+	// 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 undefined
+	if (!rendered) return null
 
 	const { raw, components: aligned } = rendered
 
@@ -181,9 +181,9 @@ export function cadastralRow(address: CadastralAddress, source: CadastralSource)
 		components: aligned,
 		country: "ES",
 		locale: "es-ES",
-		// `register` is left to `runAdapter`, which stamps the adapter's own
-		// declaration onto a row that omits it.
-		// Setting it here would state per row what the adapter already states once.
+		// 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}`