@mailwoman/corpus 10.2.2 → 10.4.0

package/lib/cz/adapters/cuzk/adapter.ts

@@ -15,7 +15,7 @@
  * to read. The `ad:component` hrefs look external, because ČÚZK writes an absolute WFS stored-query
  * URL on `services.cuzk.cz`. The `Id=` parameter of each one matches a `gml:id` in the file at hand.
  * Measured over `584282.xml`: 5,460 component references across 1,365 addresses, 0 of them reaching
- * outside the archive. Only `ad:parcel` and `ad:building` point at genuinely separate themes, and
+ * outside the archive. Only `ad:parcel` and `ad:building` point at separate themes, and
  * the adapter reads neither.
  *
  * ČÚZK states one condition of use and states it uniformly. Every one of the 6,259 `rights` elements

package/lib/cz/tools/fetch/cuzk.ts

@@ -123,11 +123,11 @@ export const CZ_CUZK_REQUEST_INTERVAL_MS = 250
 const ZIP_SIGNATURE = "PK"
 
 /**
- * How many archives are written between two manifest writes.
+ * The number of archives written between two manifest writes.
  *
  * The manifest is what makes an interrupted harvest resumable, and writing it
  * after every archive would rewrite a 6,258-entry document 6,258 times.
- * At this interval an interruption re-fetches at most this many archives that are already on disk.
+ * At this interval an interruption re-fetches at most this number of archives that are already on disk.
  */
 const MANIFEST_INTERVAL = 25
 
@@ -352,7 +352,7 @@ export interface CzCuzkHarvestManifest extends SourceCollectionManifest {
 	 */
 	projection: CzCuzkProjection
 	/**
-	 * How many municipalities the service document listed when this manifest was written.
+	 * The number of municipalities the service document listed when this manifest was written.
 	 *
 	 * The denominator for `files`, read from the feed on each run rather than fixed here.
 	 */
@@ -379,7 +379,7 @@ export interface HarvestCzCuzkOptions {
 	 */
 	municipalities?: readonly string[]
 	/**
-	 * Stop after this many municipalities, counting the ones already on disk.
+	 * Stop after this number of municipalities, counting the ones already on disk.
 	 *
 	 * A bounded run rather than all 6,258.
 	 */

package/lib/de/recipes/locale.ts

@@ -15,7 +15,7 @@
  * Two registers are absent from the OA tuples. A comma-free single line (`Neusser Str. 12 Nippes 50733
  * Köln`) segments as one unit at stage 2. With one segment, the placetype-pair prior never fires.
  * The model then reads `Nippes` as a second street. The one-value-per-tag projection deletes it.
- * `--comma-free-fraction` renders that many native-order rows with `" "` as the line separator. OA
+ * `--comma-free-fraction` renders that share of native-order rows with `" "` as the line separator. OA
  * rows also lack a district, so `--ortsteil-fraction` rows borrow a WOF Ortsteil of the tuple's
  * own locality as `dependent_locality`. The German spelling is recovered from the `names` table
  * ({@link ortsteilSurface}) because WOF's `spr.name` for DE neighborhoods is the ascii-folded label.

package/lib/dk/adapters/inspire/adapter.ts

@@ -93,7 +93,7 @@ export const DK_INSPIRE_COUNTRIES: readonly string[] = ["DK"]
 export const DK_INSPIRE_REQUIRED_TABLES: readonly string[] = ["address", "postaldescriptor", "thoroughfarename"]
 
 /**
- * How many rows Kysely buffers per chunk while streaming the join.
+ * The number of rows Kysely buffers per chunk while streaming the join.
  */
 const STREAM_CHUNK_ROWS = 2000
 

package/lib/ee/tools/fetch/ads.ts

@@ -76,7 +76,7 @@ export const EE_ADS_SORT_BY = "gml_id"
  * Features per page.
  *
  * The service's `CountDefault` is 1,000,000 and it honors `count=10000`, which is where the page
- * size is set: a 10,000-feature page is 31,198,909 bytes, and a larger page buys fewer requests
+ * size is set: a 10,000-feature page is 31,198,909 bytes, and a larger page needs fewer requests
  * at the cost of holding a larger body in memory and losing more work when one request fails.
  */
 export const EE_ADS_PAGE_SIZE = 10_000
@@ -135,7 +135,7 @@ export interface HarvestADSOptions {
 	 */
 	pageSize?: number
 	/**
-	 * Stop once the manifest holds this many pages, for a probe rather than a full harvest.
+	 * Stop once the manifest holds this number of pages, for a probe rather than a full harvest.
 	 */
 	maxPages?: number
 	signal?: AbortSignal

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

@@ -164,7 +164,7 @@ const MUNICIPALITY_LEVEL = "3rdOrder"
 const HOUSE_NUMBER_TYPE = "buildingIdentifier"
 
 /**
- * How many dot-separated fields a designator has.
+ * The number of dot-separated fields a designator has.
  *
  * Municipality, street, number: `48.001.00026.003` splits the municipality code across two.
  */

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

@@ -14,8 +14,8 @@
  * 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`,
+ * This publisher is read entirely from its references' titles. No other INSPIRE reader
+ * here reads its addresses that way. 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:

package/lib/es/cadastre.ts

@@ -72,8 +72,8 @@ export function spanishHouseNumber(
  * `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.
+ * A reader taking every unresolved reference as an absence would also swallow a broken join,
+ * so the two are separated here and every other unresolved reference raises.
  */
 export function referenceStatesNoValue(key: string): boolean {
 	return key.trim().endsWith(".")

package/lib/es/tools/fetch/bizkaia.ts

@@ -16,10 +16,10 @@
  *   archives therefore reads every address in the province without its street, its postcode or
  *   its municipality. This harvest acquires both services into one directory for that reason.
  *
- *   Each component type is read in one request, and that is the only way this service serves it. Its
+ *   Each component type is read in one request. This service serves a type no other way. Its
  *   capabilities document states `ImplementsResultPaging` as `FALSE` and `CountDefault` as `1000`, so
  *   a reader that paged a type would receive the first page again. The harvest therefore asks
- *   `RESULTTYPE=hits` for the type's count and then asks for that many features in one `GetFeature`,
+ *   `RESULTTYPE=hits` for the type's count and then asks for that number of features in one `GetFeature`,
  *   and {@linkcode fetchBizkaiaComponentDocument} raises where the document returns fewer than the
  *   count. A truncated component document is the failure that would otherwise reach the adapter as
  *   an unresolved reference per address.
@@ -121,11 +121,11 @@ export const ES_BIZKAIA_REQUEST_INTERVAL_MS = 250
 const BODY_TIMEOUT_MS = 600_000
 
 /**
- * How many archives are written between two manifest writes.
+ * The number of archives written between two manifest writes.
  *
  * The manifest is what makes an interrupted harvest resumable, and writing it
  * after every archive would rewrite the whole document once per municipality.
- * At this interval an interruption re-fetches at most this many archives already on disk.
+ * At this interval an interruption re-fetches at most this number of archives already on disk.
  */
 const MANIFEST_INTERVAL = 25
 
@@ -306,7 +306,7 @@ export interface ESBizkaiaHarvestManifest extends SourceCollectionManifest {
 	 */
 	feed_updated: string | null
 	/**
-	 * How many municipalities the service document listed when this manifest was written.
+	 * The number of municipalities the service document listed when this manifest was written.
 	 *
 	 * The denominator for `files`, read from the feed on each run rather than fixed here.
 	 */
@@ -332,7 +332,7 @@ export interface HarvestESBizkaiaOptions {
 	 */
 	municipalities?: readonly string[]
 	/**
-	 * Stop after this many municipalities, counting the ones already on disk.
+	 * Stop after this number of municipalities, counting the ones already on disk.
 	 */
 	limit?: number
 	/**
@@ -390,7 +390,7 @@ async function archiveIsCurrent(
 /**
  * Reads one component feature type whole and writes it beside the archives.
  *
- * Two requests per type: `RESULTTYPE=hits` for the count, then one `GetFeature` for that many features.
+ * Two requests per type: `RESULTTYPE=hits` for the count, then one `GetFeature` for that number of features.
  * `COUNT` is asked one above the stated count, so a service whose hits response
  * undercounts its own matches answers with more features than it claimed
  * rather than with exactly the number the harvest asked for.

package/lib/es/tools/fetch/catastro.ts

@@ -105,11 +105,11 @@ export const ES_CATASTRO_REQUEST_INTERVAL_MS = 250
 const SLUG = ES_CATASTRO_ADAPTER_ID
 
 /**
- * How many archives are written between two manifest writes.
+ * The number of archives written between two manifest writes.
  *
  * The manifest is what makes an interrupted harvest resumable, and writing it after every
  * archive would rewrite a document with one entry per municipality once per municipality.
- * At this interval an interruption re-fetches at most this many archives that are already on disk.
+ * At this interval an interruption re-fetches at most this number of archives that are already on disk.
  */
 const MANIFEST_INTERVAL = 25
 
@@ -416,18 +416,18 @@ export interface ESCatastroArchiveEntry extends SourceManifest {
  */
 export interface ESCatastroHarvestManifest extends SourceCollectionManifest {
 	/**
-	 * How many territorial offices the service document listed, read from the feed on each run.
+	 * The number of territorial offices the service document listed, read from the feed on each run.
 	 */
 	provinces_listed: number
 	/**
-	 * How many of those offices this run read a province feed for.
+	 * The number of those offices this run read a province feed for.
 	 *
 	 * A bounded run reads fewer, so this is the denominator for `municipalities_listed`
 	 * rather than a count of the country.
 	 */
 	provinces_harvested: number
 	/**
-	 * How many municipalities the province feeds this run read listed between them.
+	 * The number of municipalities the province feeds this run read listed between them.
 	 */
 	municipalities_listed: number
 	files: ESCatastroArchiveEntry[]
@@ -453,7 +453,7 @@ export interface HarvestESCatastroOptions {
 	 */
 	municipalities?: readonly string[]
 	/**
-	 * Stop after this many municipalities, counting the ones already on disk.
+	 * Stop after this number of municipalities, counting the ones already on disk.
 	 */
 	limit?: number
 	/**

package/lib/es/tools/fetch/navarra.ts

@@ -91,10 +91,10 @@ export const ES_NAVARRA_REQUEST_INTERVAL_MS = 250
 const SLUG = ES_NAVARRA_ADAPTER_ID
 
 /**
- * How many archives are written between two manifest writes.
+ * The number of archives written between two manifest writes.
  *
  * The manifest is what makes an interrupted harvest resumable.
- * At this interval an interruption re-fetches at most this many archives that are already on disk.
+ * At this interval an interruption re-fetches at most this number of archives that are already on disk.
  */
 const MANIFEST_INTERVAL = 25
 
@@ -217,7 +217,7 @@ export interface ESNavarraArchiveEntry extends SourceManifest {
  */
 export interface ESNavarraHarvestManifest extends SourceCollectionManifest {
 	/**
-	 * How many partitions the service document listed when this manifest was written.
+	 * The number of partitions the service document listed when this manifest was written.
 	 *
 	 * The denominator for `files`, read from the feed on each run rather than fixed here.
 	 */
@@ -238,7 +238,7 @@ export interface HarvestESNavarraOptions {
 	 */
 	partitions?: readonly string[]
 	/**
-	 * Stop after this many partitions, counting the ones already on disk.
+	 * Stop after this number of partitions, counting the ones already on disk.
 	 */
 	limit?: number
 	/**

package/lib/fingerprints.ts

@@ -4,8 +4,8 @@
  * @author Teffen Ellis, et al.
  * @file A fixed-capacity set of 128-bit fingerprints in one flat `Uint32Array`.
  *
- *   The adapter runner deduplicates by holding every key it has seen. A V8 `Set` refuses its 16,777,217th
- *   entry outright, so the runner capped itself at 10,000,000 keys. Three adapters passed that cap in
+ *   The adapter runner deduplicates by holding every key it has seen. A V8 `Set` throws on its 16,777,217th
+ *   entry, so the runner capped itself at 10,000,000 keys. Three adapters passed that cap in
  *   `v0.7.0-de-holdout`. The duplicate census over that build measured what the cap costs: of the
  *   2,491,399 duplicate rows the three adapters wrote, 1,490,992 duplicate a key first seen after the cap.
  *   A membership test on a capped set cannot refuse those rows.
@@ -77,7 +77,7 @@ export class FingerprintSet {
 	}
 
 	/**
-	 * How many keys this table accepts before {@linkcode add} throws.
+	 * The number of keys this table accepts before {@linkcode add} throws.
 	 */
 	get limit(): number {
 		return this.#limit

package/lib/fr/adapters/finess/adapter.ts

@@ -32,8 +32,8 @@
  *
  * Every FINESS establishment is an organization's site: a hospital, a pharmacy, a laboratory, a
  * service office. Three categories house the people they serve in a private dwelling. In `462`
- * Lieux de vie et d'accueil the staff live with the people they receive. A `238` Centre d'accueil
- * familial spécialisé places patients with host families, as a `236` Service de placement familial
+ * Lieux de vie et d'accueil the staff live with the people they receive. A FINESS category `238`
+ * establishment, specialized family reception, places patients with host families, as a `236` Service de placement familial
  * does. An address in those three can be a family's home, so they are refused under
  * {@linkcode FinessRefusal.FamilyPlacement}, as is an establishment whose name states family
  * placement or a lieu de vie under another category. A row whose street type is TRE-R35's `CHEZ`,
@@ -56,7 +56,7 @@
  *
  * A digit in `voie` is refused under {@linkcode FinessRefusal.StreetDigitUnplaced} unless the digits
  * name the street, as a numbered route (`NATIONALE 2`) or a date (`DU 22 MAI`) does. The number has
- * its own field, so a digit left in the name is a box, a building or a door: `VALLÉE 3 BP 208`,
+ * its own field, so a digit left in the street name is a box, a building or a door: `VALLÉE 3 BP 208`,
  * `DIAKA BAT A4 PORTE 2`.
  *
  * ## The lieu-dit and box line, and the routing line
@@ -166,8 +166,9 @@ export const FR_FINESS_COUNTRIES: readonly string[] = [
 ]
 
 /**
- * The establishment categories whose address can be a private dwelling: lieux de vie et
- * d'accueil, centres d'accueil familial spécialisés and services de placement familial.
+ * The establishment categories whose address can be a private dwelling: FINESS
+ * categories `462` (lieux de vie et d'accueil), `238` (specialized family reception)
+ * and `236` (services de placement familial).
  */
 export const FINESS_FAMILY_PLACEMENT_CATEGORIES: ReadonlySet<string> = new Set(["236", "238", "462"])
 

package/lib/fr/adapters/finess/voie-type.ts

@@ -41,7 +41,7 @@ import { decomposeFrStreet } from "#fr/street-decompose"
  *
  * `VAL` keeps the code as its word.
  * TRE-R35 displays it as `Vallée, vallon`, two words the code does not choose between,
- * and `Val` is itself the written word of many French place names.
+ * and `Val` is itself the written word in French place names such as `Val-d'Isère`.
  */
 export const TRE_R35_VOIE_TYPES: Readonly<Record<string, string | null>> = {
 	ANSE: "anse",

package/lib/fr/tools/fetch/annuaire-education.ts

@@ -15,7 +15,7 @@
  *
  * Two checks keep a short transfer from reading as a small directory. The records endpoint is asked
  * for the `total_count` of the same filter before the export, and the written file must hold exactly
- * that many lines. A mismatch removes the file and fails the fetch. A re-run compares the catalog
+ * that number of lines. A mismatch removes the file and fails the fetch. A re-run compares the catalog
  * record's `modified` against the manifest and makes no transfer when they agree and the file is on
  * disk at the recorded length.
  */
@@ -75,7 +75,7 @@ export function annuaireExportURL(): string {
 }
 
 /**
- * The records URL that answers how many records the overseas filter selects.
+ * The records URL that answers with the count of records the overseas filter selects.
  */
 export function annuaireCountURL(): string {
 	return `${ANNUAIRE_EDUCATION_API_URL}/records?where=${encodeURIComponent(annuaireOverseasFilter())}&limit=0`
@@ -135,7 +135,7 @@ export async function readAnnuaireCatalog(
 }
 
 /**
- * Read how many records the overseas filter selects.
+ * Read the count of records the overseas filter selects.
  *
  * @throws When the endpoint answers no `total_count`.
  */

package/lib/gb/constituent-country.ts

@@ -0,0 +1,52 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   The constituent country of a GB postcode, read from the ONS National Statistics UPRN Lookup.
+ *
+ *   A GB row labels its constituent country `region`, because WOF records England, Scotland, Wales and
+ *   Northern Ireland as `macroregion` places under the United Kingdom.
+ *   The postcode-area table in the codex assigns a whole area to one country. That assignment is wrong on the border:
+ *   every `SY` postcode maps to Wales, and `SY1` to `SY5` lie in Shrewsbury, England.
+ *   NSUL records the country of each UPRN, so the postcode's country is a count of its UPRNs.
+ */
+
+import { GB_COUNTRIES } from "@mailwoman/codex/gb/country"
+import { NSULLookup } from "@mailwoman/resolver-wof-sqlite/nsul/lookup"
+import { nsulDatabasePath } from "@mailwoman/resolver-wof-sqlite/paths"
+import type { PathBuilderLike } from "path-ts"
+
+/**
+ * What {@link openConstituentCountryReader} answered for one postcode.
+ */
+export interface ConstituentCountryAnswer {
+	/**
+	 * The English name of the country holding most of the postcode's UPRNs, such as `Wales`.
+	 *
+	 * Null when the register assigns no UPRN to the postcode.
+	 */
+	name: string | null
+	/**
+	 * Whether the postcode's UPRNs lie in more than one constituent country.
+	 */
+	straddles: boolean
+}
+
+/**
+ * Opens `nsul.db` and returns a reader from a GB postcode to its constituent country.
+ *
+ * The reader holds the database open for the life of the process, as a corpus build reads millions of rows.
+ */
+export function openConstituentCountryReader(
+	databasePath: PathBuilderLike = nsulDatabasePath("nsul.db")
+): (postcode: string) => ConstituentCountryAnswer {
+	const lookup = new NSULLookup({ databasePath })
+
+	return (postcode) => {
+		const countries = lookup.countriesForPostcode(postcode)
+		const top = countries[0]
+
+		return { name: top ? GB_COUNTRIES[top.country].name : null, straddles: countries.length > 1 }
+	}
+}

package/lib/inspire/address.ts

@@ -123,7 +123,7 @@ export function codelistValue(uri: string | null | undefined): string | null {
 /**
  * Whether a publisher marked this element void.
  *
- * INSPIRE writes `xsi:nil="true"` and usually a `nilReason`.
+ * INSPIRE writes `xsi:nil="true"`, in most cases with a `nilReason`.
  * Brussels writes the nil marker and no reason on all 232,003 of its void `ad:validFrom`
  * elements, and Czechia writes four `gn:pronunciation` elements nil with no reason.
  *
@@ -355,7 +355,7 @@ const REFERENCE_ID_PARAMETERS = new Set(["id", "featureid", "resourceid"])
  * `http://vocab.belgif.be/auth/refnis1995/1000#id` addresses the same term as that URI without it.
  *
  * Whether a key joins is not a property of its spelling, so this reads a key
- * and the caller decides joinability against the features the publisher actually served.
+ * and the caller decides joinability against the features the publisher served.
  * A check based on the URL's shape once reported 15 of Flanders' 20 references as belonging
  * to another register, when every one of them addressed a feature of the same service.
  *
@@ -392,7 +392,7 @@ export function componentJoinKey(href: string): string | null {
 /**
  * The spelling a `gn:GeographicalName` holds, reached from a feature that has one.
  *
- * Every name-bearing INSPIRE feature nests the text at
+ * Every INSPIRE feature that has a name nests the text at
  * `…/gn:GeographicalName/gn:spelling/gn:SpellingOfName/gn:text`, under a per-feature
  * prefix: an `ad:ThoroughfareName` adds `ad:name/ad:ThoroughfareNameValue/ad:name`,
  * while an `ad:AdminUnitName` and an `ad:AddressAreaName` add only `ad:name`.

package/lib/international/recipes/locale.ts

@@ -19,6 +19,7 @@ import type { PathBuilderLike } from "path-ts"
 import { CSVSpliterator } from "spliterator"
 
 import { stableSourceID } from "#adapters/source-id"
+import { openConstituentCountryReader } from "#gb/constituent-country"
 import type { CorpusRecipe } from "#recipes/scaffold"
 import { defaultRecipeSource } from "#recipes/sources"
 import { SourceRegister } from "#registers"
@@ -47,6 +48,12 @@ export interface LocalePart {
 	 * The street combines `tipo_vial` with `nombre_via`.
 	 */
 	cnigRaw?: boolean
+	/**
+	 * Opens a reader that answers the row's region from its postcode, replacing the region column.
+	 *
+	 * A GB part answers the constituent country this way, and its county column is not read.
+	 */
+	regionOfPostcode?: () => (postcode: string) => string | null
 }
 
 /**
@@ -158,7 +165,18 @@ const COUNTRY_SOURCES: Record<string, LocaleCountrySource> = {
 		source: defaultRecipeSource("synth-gb"),
 		register: SourceRegister.LandRegistryPricePaid,
 		corpusVersion: "0.9.9",
-		parts: [{ path: dataRootPath("ppd", "2026-07-22", "gb-tuples.csv"), districtAsLocality: true }],
+		// The `REGION` column holds a county, so the constituent country is read from NSUL by postcode instead.
+		parts: [
+			{
+				path: dataRootPath("ppd", "2026-07-22", "gb-tuples.csv"),
+				districtAsLocality: true,
+				regionOfPostcode: () => {
+					const countryOf = openConstituentCountryReader()
+
+					return (postcode) => countryOf(postcode).name
+				},
+			},
+		],
 	},
 }
 
@@ -224,6 +242,8 @@ export async function readTuples(part: LocalePart, rng: () => number): Promise<L
 	let header: string[] | null = null
 	let seen = 0
 	let dropped = 0
+	const regionOf = part.regionOfPostcode?.()
+	let regionUnknown = 0
 
 	try {
 		// The first row holds the column names.
@@ -304,12 +324,24 @@ export async function readTuples(part: LocalePart, rng: () => number): Promise<L
 				continue
 			}
 
+			const postcode = get(cells, cols.post)
+
+			let region = get(cells, cols.region) || part.region || ""
+
+			if (regionOf) {
+				region = (postcode && regionOf(postcode)) || ""
+
+				if (!region) {
+					regionUnknown++
+				}
+			}
+
 			const tuple: LocaleBaseTuple = {
 				house_number: get(cells, cols.num),
 				street,
 				locality,
-				region: get(cells, cols.region) || part.region || "",
-				postcode: get(cells, cols.post),
+				region,
+				postcode,
 				...(dependent_locality ? { dependent_locality } : {}),
 			}
 
@@ -333,6 +365,10 @@ export async function readTuples(part: LocalePart, rng: () => number): Promise<L
 
 	console.error(`  ${part.path ?? part.csv}: ${reservoir.length} sampled of ${seen} rows (${dropped} city-noise drops)`)
 
+	if (regionOf) {
+		console.error(`  ${part.path ?? part.csv}: ${regionUnknown} rows have no region for their postcode`)
+	}
+
 	return reservoir
 }
 
@@ -425,10 +461,8 @@ export const localeRecipe: CorpusRecipe = {
 			throw new Error(`--country-fraction must be in [0, 1], got ${countryFraction}`)
 		}
 
-		// `COUNTRY_SOURCES` and `RECIPE_SOURCES` both use the retired spelling as their key.
-		// This lookup decides the vocabulary this recipe writes in one place
-		// rather than in twelve table entries.
-		const source = opts.sourceName ?? defaultRecipeSource(countrySource.source)
+		// Each `COUNTRY_SOURCES` entry already holds the current spelling from `defaultRecipeSource`.
+		const source = opts.sourceName ?? countrySource.source
 		const count = opts.count ?? 4000
 		// When unset, each part keeps its own setting.
 		// True also selects the ES pedanía parts.

package/lib/it/adapters/anac/adapter.ts

@@ -29,7 +29,7 @@
  * public bodies, and the one role whose parties can be sole traders publishes no address.
  * {@linkcode readANACParty} still applies a name check and refuses a party whose name takes the
  * `Surname, Firstname` form, because a later edition could add one. On the 2025 edition the check
- * refuses no party: 0 of the 18,740 address-bearing buyer and payer parties take that form, while 56
+ * refuses no party: 0 of the 18,740 buyer and payer parties that have an address take that form, while 56
  * hold a comma inside an organization name — `PARCO NAZIONALE DEL CILENTO, VALLO DI DIANO E
  * ALBURNI` and `SETTORE PATRIMONIO, AMBIENTE E LAVORI PUBBLICI` among them. The check therefore
  * requires one word on each side of the comma rather than any comma.
@@ -304,8 +304,9 @@ const KILOMETRE_POINT = /[\s,]+km\.?[\s,]*\d+(?:[.,]\d+)?$/iu
 /**
  * A name of the form `Surname, Firstname`: one comma, one word on each side, letters only.
  *
- * An organization name in this release holds a comma often and holds a word on each side of it rarely:
- * `PARCO NAZIONALE DEL CILENTO, VALLO DI DIANO E ALBURNI` has eight words after the comma.
+ * An organization name in this release can hold a comma, but a name with a single word on
+ * each side of the comma is rare: `PARCO NAZIONALE DEL CILENTO, VALLO DI DIANO E ALBURNI`
+ * has eight words after the comma.
  * The single word on each side is what the inverted personal form has
  * and what a department or authority name does not.
  */

package/lib/jp/adapters/wof-admin/adapter.ts

@@ -13,8 +13,8 @@
  *   1. **Reversed ordering** — region → county → locality → block → house_number "東京都世田谷区南町1-2-3" not
  *        "1-2-3 Minamicho, Setagaya, Tokyo"
  *   2. **No street names** — most JP addresses use a block/sub-block grid system. The "neighborhood"
- *        placetype (丁目) is the closest analog to a street but is actually a grid cell.
- *   3. **Postcode-first convention** — addresses are often prefixed with `〒NNN-nnnn`.
+ *        placetype (丁目) is the closest analog to a street but is a grid cell.
+ *   3. **Postcode-first convention** — addresses can be prefixed with `〒NNN-nnnn`.
  *
  *   Schema mapping to ComponentTags (subset of Stage 3 + JP-specific Phase 6 tags):
  *

package/lib/kr/adapters/juso/label-rows.ts

@@ -152,7 +152,7 @@ function joinNumber(main: string, sub: string): string {
 
 export interface ReadJusoOptions {
 	/**
-	 * Stop after this many rows per 시도, for a fixture or a smoke run.
+	 * Stop after this number of rows per 시도, for a fixture or a smoke run.
 	 * Omit for the whole register.
 	 */
 	maxRowsPerRegion?: number

package/lib/kr/adapters/localdata/key-index.ts

@@ -142,7 +142,8 @@ export function aliasKeyIndex(index: KeyIndex): void {
 }
 
 /**
- * How many tokens from `start` form one of the region's 시군구 — 2 for `수원시 장안구`, 1 for `종로구`, 0 for none.
+ * The number of tokens from `start` that form one of the region's 시군구 —
+ * 2 for `수원시 장안구`, 1 for `종로구`, 0 for none.
  *
  * Two-token first, because `수원시` by itself is also listed and taking it would
  * leave `장안구` to be read as a road.

package/lib/lu/adapters/bd-adresses/adapter.ts

@@ -33,7 +33,7 @@
  * Measured over the 179,622 rows of the 2026-09-28 edition with a generic-word probe rather than a
  * language classifier: 133,449 values open with a French street generic such as `Rue`, `Avenue` or
  * `Montée`, 18,629 close with a Luxembourgish one such as `-strooss`, `-wee` or `-gaass`, and the
- * remaining 27,544 match neither probe and include plainly Luxembourgish forms such as
+ * remaining 27,544 match neither probe and include Luxembourgish forms such as
  * `Am Stengber`, `An Hiesel` and `Um Reebou`.
  * No column marks a row's language, so `locale` reads `und-LU`.
  * `und` is BCP-47's undetermined language subtag, and the repository already stores it in

package/lib/nl/adapters/kadaster/adapter.ts

@@ -360,8 +360,8 @@ function documentChunks(documentPath: PathBuilderLike): AsyncIterable<Uint8Array
 /**
  * Every document under `inputPath`, where `inputPath` is one GML file or a directory holding them.
  *
- * A directory is read because the download is one national file today and the
- * service document could list more than one entry tomorrow.
+ * A directory is read because the download is one national file now and a later
+ * service document could list more than one entry.
  * A directory holding no document raises rather than yielding zero rows, so a wrong
  * path reports itself instead of reading as an empty publisher.
  */

package/lib/nl/tools/fetch/kadaster.ts

@@ -54,7 +54,7 @@ export const NL_KADASTER_FEED_URL = "https://service.pdok.nl/kadaster/ad/atom/ad
  * The adapter's `inputPath` states this name.
  *
  * The publisher's own name for the file.
- * The adapter reads a `.gml.gz` by inflating it, so the suffix is load-bearing.
+ * The adapter reads a `.gml.gz` by inflating it, so the suffix is required.
  */
 export const NL_KADASTER_DOWNLOAD_FILENAME = "addresses.gml.gz"
 

package/lib/parquet/shuffle.ts

@@ -10,8 +10,8 @@ import { shuffleWith } from "@mailwoman/core/random"
 /**
  * Rows held in memory while shuffling, when the caller names no window.
  *
- * Five 50,000-row groups span more than one output row-group and reach `wof-admin`'s
- * whole 13-country set, while a window below `ROW_GROUP_SIZE` buys no wider mix.
+ * Five 50,000-row groups span more than one output row-group and reach `wof-admin`'s whole
+ * 13-country set, while a window below `ROW_GROUP_SIZE` draws from no more than one row-group.
  */
 export const DEFAULT_SHUFFLE_WINDOW = 250_000