@mailwoman/corpus 8.0.0 → 8.2.0

package/src/adapters/synth-po-box/adapter.ts

@@ -32,7 +32,15 @@ import {
 } from "../../synthesize-po-box.ts"
 import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
 
+/**
+ * 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 SYNTH_PO_BOX_ADAPTER_ID = "synth-po-box"
+/**
+ * License for the synthetic PO-box rows. The output is generated, but it inherits the terms of the real tuples it is
+ * derived from, so the attribution travels with it.
+ */
 export const SYNTH_PO_BOX_LICENSE = "Synthetic — derived from CC-BY / public-domain input tuples"
 
 export interface PoBoxInputRow extends PoBoxBaseTuple {
@@ -69,9 +77,9 @@ function makeRandom(seed: number): () => number {
 	let s = seed
 
 	return () => {
-		s = (s * 1664525 + 1013904223) % 4294967296
+		s = (s * 1_664_525 + 1_013_904_223) % 4_294_967_296
 
-		return s / 4294967296
+		return s / 4_294_967_296
 	}
 }
 
@@ -113,6 +121,7 @@ export function createSynthPoBoxAdapter(opts: SynthPoBoxAdapterOptions = {}): Co
 					input = JSON.parse(trimmed) as PoBoxInputRow
 				} catch {
 					skipped++
+
 					continue
 				}
 
@@ -123,6 +132,7 @@ export function createSynthPoBoxAdapter(opts: SynthPoBoxAdapterOptions = {}): Co
 
 				if (!input.locality || !input.postcode || !input.country || (!input.region && !regionOptional)) {
 					skipped++
+
 					continue
 				}
 
@@ -152,6 +162,7 @@ export function createSynthPoBoxAdapter(opts: SynthPoBoxAdapterOptions = {}): Co
 						corpus_version: "",
 						license: SYNTH_PO_BOX_LICENSE,
 					}
+
 					emitted++
 
 					if (options.limit !== undefined && emitted >= options.limit) break
@@ -171,12 +182,14 @@ export function createSynthPoBoxAdapter(opts: SynthPoBoxAdapterOptions = {}): Co
 					random() < militaryRatio
 				) {
 					const mil = synthesizeMilitaryPoBoxRow({ random })
+
 					const sourceID = stableSourceID(SYNTH_PO_BOX_ADAPTER_ID, {
 						po_box: `${mil.components.po_box}#mil${militarySeq++}`,
 						locality: mil.components.locality!,
 						region: mil.components.region!,
 						postcode: mil.components.postcode!,
 					})
+
 					yield {
 						raw: mil.raw,
 						components: mil.components,
@@ -187,6 +200,7 @@ export function createSynthPoBoxAdapter(opts: SynthPoBoxAdapterOptions = {}): Co
 						corpus_version: "",
 						license: SYNTH_PO_BOX_LICENSE,
 					}
+
 					emitted++
 				}
 			}
@@ -194,4 +208,7 @@ export function createSynthPoBoxAdapter(opts: SynthPoBoxAdapterOptions = {}): Co
 	}
 }
 
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
 export const synthPoBoxAdapter = createSynthPoBoxAdapter()

package/src/adapters/tiger/adapter.ts

@@ -41,7 +41,15 @@ import { formatAddress, reconcileComponents } from "../../format.ts"
 import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
 import { decomposeStreet } from "./street-decompose.ts"
 
+/**
+ * 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 TIGER_ADAPTER_ID = "tiger"
+/**
+ * License carried by this source (Public Domain), attached to each row so downstream consumers inherit the terms rather
+ * than having to look them up.
+ */
 export const TIGER_DEFAULT_LICENSE = "Public Domain"
 
 /**
@@ -125,7 +133,9 @@ function* streetVariants(row: TigerStreetRow): Iterable<{
 	}
 }
 
-/** Three locality-level variants, mirroring `wof-admin`'s fan-out. */
+/**
+ * Three locality-level variants, mirroring `wof-admin`'s fan-out.
+ */
 function* placeVariants(row: TigerPlaceRow): Iterable<{
 	components: CanonicalRow["components"]
 	variantKey: string
@@ -141,17 +151,21 @@ function* placeVariants(row: TigerPlaceRow): Iterable<{
 		components: { locality: name },
 		variantKey: "locality-only",
 	}
+
 	yield {
 		components: { locality: name, region: state.abbreviation },
 		variantKey: "with-region",
 	}
+
 	yield {
 		components: { locality: name, region: state.abbreviation, country: US_COUNTRY_DISPLAY },
 		variantKey: "with-region-country",
 	}
 }
 
-/** Build a TIGER adapter. Pure factory so multiple instances can be created in tests. */
+/**
+ * Build a TIGER adapter. Pure factory so multiple instances can be created in tests.
+ */
 export function createTigerAdapter(): CorpusAdapter {
 	return {
 		id: TIGER_ADAPTER_ID,
@@ -181,7 +195,7 @@ export function createTigerAdapter(): CorpusAdapter {
 						if (!raw) continue
 						const aligned = reconcileComponents(variant.components, raw)
 
-						if (Object.keys(aligned).length === 0) continue
+						if (!Object.keys(aligned).length) continue
 
 						yield {
 							raw,
@@ -193,6 +207,7 @@ export function createTigerAdapter(): CorpusAdapter {
 							corpus_version: "",
 							license: TIGER_DEFAULT_LICENSE,
 						}
+
 						emitted++
 					}
 				}
@@ -207,7 +222,7 @@ export function createTigerAdapter(): CorpusAdapter {
 						if (!raw) continue
 						const aligned = reconcileComponents(variant.components, raw)
 
-						if (Object.keys(aligned).length === 0) continue
+						if (!Object.keys(aligned).length) continue
 
 						yield {
 							raw,
@@ -219,6 +234,7 @@ export function createTigerAdapter(): CorpusAdapter {
 							corpus_version: "",
 							license: TIGER_DEFAULT_LICENSE,
 						}
+
 						emitted++
 					}
 				}
@@ -229,4 +245,7 @@ export function createTigerAdapter(): CorpusAdapter {
 	}
 }
 
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
 export const tigerAdapter = createTigerAdapter()

package/src/adapters/tiger/street-decompose.ts

@@ -54,6 +54,7 @@ function loadDictionary(filename: string): Set<string> {
 			// try next candidate
 		}
 	}
+
 	throw new Error(`Could not load libpostal dictionary: ${filename}`)
 }
 

package/src/adapters/usgov-hrsa-fqhc/adapter.ts

@@ -38,7 +38,15 @@ import { lookupStateAbbreviation } from "../../codex/us-fips-state.ts"
 import { reconcileComponents } from "../../format.ts"
 import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
 
+/**
+ * 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 USGOV_HRSA_FQHC_ADAPTER_ID = "usgov-hrsa-fqhc"
+/**
+ * License carried by this source (Public Domain), attached to each row so downstream consumers inherit the terms rather
+ * than having to look them up.
+ */
 export const USGOV_HRSA_FQHC_DEFAULT_LICENSE = "Public Domain"
 
 /**
@@ -52,7 +60,9 @@ interface HrsaSiteRow {
 	"Site City": string
 	"Site State Abbreviation": string
 	"Site Postal Code": string
-	/** Optional. Falls back to `stableSourceID` derived from components when missing. */
+	/**
+	 * Optional. Falls back to `stableSourceID` derived from components when missing.
+	 */
 	"Site ID"?: string
 }
 
@@ -113,6 +123,7 @@ export function createUsgovHrsaFqhcAdapter(): CorpusAdapter {
 			}
 
 			const stream = createReadStream(opts.inputPath, { encoding: "utf8" })
+
 			const parser = stream.pipe(
 				csvParse({
 					columns: true,
@@ -160,9 +171,10 @@ export function createUsgovHrsaFqhcAdapter(): CorpusAdapter {
 
 					const aligned = reconcileComponents(components, raw)
 
-					if (Object.keys(aligned).length === 0) continue
+					if (!Object.keys(aligned).length) continue
 
 					const siteID = (record["Site ID"] ?? "").trim()
+
 					const sourceID = siteID
 						? `${USGOV_HRSA_FQHC_ADAPTER_ID}-${siteID}`
 						: stableSourceID(USGOV_HRSA_FQHC_ADAPTER_ID, aligned)
@@ -177,6 +189,7 @@ export function createUsgovHrsaFqhcAdapter(): CorpusAdapter {
 						corpus_version: "",
 						license: USGOV_HRSA_FQHC_DEFAULT_LICENSE,
 					}
+
 					emitted++
 				}
 			} finally {
@@ -186,4 +199,7 @@ export function createUsgovHrsaFqhcAdapter(): CorpusAdapter {
 	}
 }
 
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
 export const usgovHrsaFqhcAdapter = createUsgovHrsaFqhcAdapter()

package/src/adapters/usgov-imls-pls/adapter.ts

@@ -27,7 +27,15 @@ import { lookupStateAbbreviation } from "../../codex/us-fips-state.ts"
 import { reconcileComponents } from "../../format.ts"
 import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
 
+/**
+ * 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 USGOV_IMLS_PLS_ADAPTER_ID = "usgov-imls-pls"
+/**
+ * License carried by this source (Public Domain), attached to each row so downstream consumers inherit the terms rather
+ * than having to look them up.
+ */
 export const USGOV_IMLS_PLS_DEFAULT_LICENSE = "Public Domain"
 
 const HOUSE_NUMBER_PREFIX = /^(\d+(?:-\d+)?[A-Za-z]?)\s+(.+)$/
@@ -65,6 +73,7 @@ export function createUsgovImlsPlsAdapter(): CorpusAdapter {
 			}
 
 			const stream = createReadStream(opts.inputPath, { encoding: "utf8" })
+
 			const parser = stream.pipe(
 				csvParse({
 					columns: true,
@@ -113,6 +122,7 @@ export function createUsgovImlsPlsAdapter(): CorpusAdapter {
 					}
 
 					const streetPart = [split.house_number, split.street].filter(Boolean).join(" ").trim()
+
 					const raw = [
 						libName,
 						streetPart,
@@ -126,6 +136,7 @@ export function createUsgovImlsPlsAdapter(): CorpusAdapter {
 					if (Object.keys(aligned).length <= 2) continue
 
 					const fscsKey = (record.FSCSKEY ?? "").trim()
+
 					const sourceID = fscsKey
 						? `${USGOV_IMLS_PLS_ADAPTER_ID}-${fscsKey}`
 						: stableSourceID(USGOV_IMLS_PLS_ADAPTER_ID, aligned)
@@ -140,6 +151,7 @@ export function createUsgovImlsPlsAdapter(): CorpusAdapter {
 						corpus_version: "",
 						license: USGOV_IMLS_PLS_DEFAULT_LICENSE,
 					}
+
 					emitted++
 				}
 			} finally {
@@ -149,4 +161,7 @@ export function createUsgovImlsPlsAdapter(): CorpusAdapter {
 	}
 }
 
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
 export const usgovImlsPlsAdapter = createUsgovImlsPlsAdapter()

package/src/adapters/usgov-irs-bmf/adapter.ts

@@ -27,11 +27,21 @@ import { stableSourceID } from "../../adapter.ts"
 import { reconcileComponents } from "../../format.ts"
 import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
 
+/**
+ * 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 USGOV_IRS_BMF_ADAPTER_ID = "usgov-irs-bmf"
+/**
+ * License carried by this source (Public Domain), attached to each row so downstream consumers inherit the terms rather
+ * than having to look them up.
+ */
 export const USGOV_IRS_BMF_DEFAULT_LICENSE = "Public Domain"
 
 const HOUSE_NUMBER_PREFIX = /^(\d+(?:-\d+)?[A-Za-z]?)\s+(.+)$/
-// PO box in its many written forms: "PO BOX 12", "P.O. BOX 12", "P O BOX 12", "POB 12", "BOX 12".
+/**
+ * PO box in its many written forms: "PO BOX 12", "P.O. BOX 12", "P O BOX 12", "POB 12", "BOX 12".
+ */
 const PO_BOX = /^\s*(?:P\.?\s?O\.?\s*BOX|POB|BOX)\s+\w/i
 
 interface IrsBmfRow {
@@ -43,7 +53,9 @@ interface IrsBmfRow {
 	ZIP: string
 }
 
-/** Classify the street line into a `po_box` or a `{house_number?, street}` split. */
+/**
+ * Classify the street line into a `po_box` or a `{house_number?, street}` split.
+ */
 function splitStreetLine(street: string): { po_box: string } | { house_number?: string; street: string } | null {
 	const trimmed = street.trim()
 
@@ -82,6 +94,7 @@ export function createUsgovIrsBmfAdapter(): CorpusAdapter {
 			}
 
 			const stream = createReadStream(opts.inputPath, { encoding: "utf8" })
+
 			const parser = stream.pipe(
 				csvParse({ columns: true, skip_empty_lines: true, relax_quotes: true, relax_column_count: true, trim: true })
 			)
@@ -143,6 +156,7 @@ export function createUsgovIrsBmfAdapter(): CorpusAdapter {
 						corpus_version: "",
 						license: USGOV_IRS_BMF_DEFAULT_LICENSE,
 					}
+
 					emitted++
 				}
 			} finally {
@@ -152,4 +166,7 @@ export function createUsgovIrsBmfAdapter(): CorpusAdapter {
 	}
 }
 
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
 export const usgovIrsBmfAdapter = createUsgovIrsBmfAdapter()

package/src/adapters/usgov-nad/adapter.ts

@@ -38,7 +38,15 @@ import { TextSpliterator } from "spliterator"
 import { reconcileComponents } from "../../format.ts"
 import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
 
+/**
+ * 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 USGOV_NAD_ADAPTER_ID = "usgov-nad"
+/**
+ * License carried by this source (Public Domain), attached to each row so downstream consumers inherit the terms rather
+ * than having to look them up.
+ */
 export const USGOV_NAD_DEFAULT_LICENSE = "Public Domain"
 
 interface NADRecord {
@@ -192,6 +200,7 @@ function decomposeNADStreet(r: NADRecord): DecomposedNADStreet | undefined {
 
 		return { prefix, street: name, suffix, full }
 	}
+
 	const full = (r.StNam_Full ?? "").toString().trim()
 
 	if (full) return { full, street: full }
@@ -243,7 +252,7 @@ export function createUsgovNADAdapter(): CorpusAdapter {
 			// Single-file inputs (e.g. a bulk-extracted CSV) are not currently supported — the
 			// featureserver shard pattern is the primary distribution.
 			const entries = await readdir(opts.inputPath)
-			const shards = entries.filter((n) => n.endsWith(".ndjson")).sort()
+			const shards = entries.filter((n) => n.endsWith(".ndjson")).toSorted()
 
 			let emitted = 0
 			outer: for (const shard of shards) {
@@ -328,6 +337,7 @@ export function createUsgovNADAdapter(): CorpusAdapter {
 						corpus_version: "",
 						license: USGOV_NAD_DEFAULT_LICENSE,
 					}
+
 					emitted++
 				}
 			}
@@ -335,4 +345,7 @@ export function createUsgovNADAdapter(): CorpusAdapter {
 	}
 }
 
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
 export const usgovNADAdapter = createUsgovNADAdapter()

package/src/adapters/usgov-nppes/adapter.ts

@@ -30,7 +30,15 @@ import { lookupStateAbbreviation } from "../../codex/us-fips-state.ts"
 import { reconcileComponents } from "../../format.ts"
 import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
 
+/**
+ * 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 USGOV_NPPES_ADAPTER_ID = "usgov-nppes"
+/**
+ * License carried by this source (Public Domain), attached to each row so downstream consumers inherit the terms rather
+ * than having to look them up.
+ */
 export const USGOV_NPPES_DEFAULT_LICENSE = "Public Domain"
 
 const HOUSE_NUMBER_PREFIX = /^(\d+(?:-\d+)?[A-Za-z]?)\s+(.+)$/
@@ -86,6 +94,7 @@ export function createUsgovNPPESAdapter(): CorpusAdapter {
 			}
 
 			const stream = createReadStream(opts.inputPath, { encoding: "utf8" })
+
 			const parser = stream.pipe(
 				csvParse({
 					columns: true,
@@ -157,6 +166,7 @@ export function createUsgovNPPESAdapter(): CorpusAdapter {
 						corpus_version: "",
 						license: USGOV_NPPES_DEFAULT_LICENSE,
 					}
+
 					emitted++
 				}
 			} finally {
@@ -166,4 +176,7 @@ export function createUsgovNPPESAdapter(): CorpusAdapter {
 	}
 }
 
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
 export const usgovNPPESAdapter = createUsgovNPPESAdapter()

package/src/adapters/usgov-samhsa-treatment-locator/adapter.ts

@@ -37,7 +37,15 @@ import { lookupStateAbbreviation } from "../../codex/us-fips-state.ts"
 import { reconcileComponents } from "../../format.ts"
 import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
 
+/**
+ * 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 USGOV_SAMHSA_ADAPTER_ID = "usgov-samhsa-treatment-locator"
+/**
+ * License carried by this source (Public Domain), attached to each row so downstream consumers inherit the terms rather
+ * than having to look them up.
+ */
 export const USGOV_SAMHSA_DEFAULT_LICENSE = "Public Domain"
 
 /**
@@ -52,7 +60,9 @@ interface SamhsaSiteRow {
 	city: string
 	state: string
 	zip: string
-	/** Optional. Falls back to `stableSourceID` derived from components when missing. */
+	/**
+	 * Optional. Falls back to `stableSourceID` derived from components when missing.
+	 */
 	frid?: string
 }
 
@@ -109,7 +119,9 @@ function composeVenue(name1: string, name2: string | undefined): string {
 	return `${n1} - ${n2}`
 }
 
-/** Same envelope-style format as HRSA: venue prefix, street body, city/state/zip suffix. */
+/**
+ * Same envelope-style format as HRSA: venue prefix, street body, city/state/zip suffix.
+ */
 function composeRaw(
 	venue: string,
 	house: string | undefined,
@@ -137,6 +149,7 @@ export function createUsgovSamhsaTreatmentLocatorAdapter(): CorpusAdapter {
 			}
 
 			const stream = createReadStream(opts.inputPath, { encoding: "utf8" })
+
 			const parser = stream.pipe(
 				csvParse({
 					columns: true,
@@ -182,9 +195,10 @@ export function createUsgovSamhsaTreatmentLocatorAdapter(): CorpusAdapter {
 
 					const aligned = reconcileComponents(components, raw)
 
-					if (Object.keys(aligned).length === 0) continue
+					if (!Object.keys(aligned).length) continue
 
 					const frID = (record.frid ?? "").trim()
+
 					const sourceID = frID
 						? `${USGOV_SAMHSA_ADAPTER_ID}-${frID}`
 						: stableSourceID(USGOV_SAMHSA_ADAPTER_ID, aligned)
@@ -199,6 +213,7 @@ export function createUsgovSamhsaTreatmentLocatorAdapter(): CorpusAdapter {
 						corpus_version: "",
 						license: USGOV_SAMHSA_DEFAULT_LICENSE,
 					}
+
 					emitted++
 				}
 			} finally {
@@ -208,4 +223,7 @@ export function createUsgovSamhsaTreatmentLocatorAdapter(): CorpusAdapter {
 	}
 }
 
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
 export const usgovSamhsaTreatmentLocatorAdapter = createUsgovSamhsaTreatmentLocatorAdapter()

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

@@ -33,6 +33,15 @@ import { DatabaseSync } from "node:sqlite"
 
 import type { AdapterOptions, CanonicalRow, CorpusAdapter } from "../../types.ts"
 
+/**
+ * Ancestry hops walked upward from a WOF record before giving up — deeper than any real JP admin chain.
+ */
+const MAX_ANCESTRY_DEPTH = 6
+
+/**
+ * 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 WOF_ADMIN_JP_ADAPTER_ID = "wof-admin-jp"
 
 interface PlaceRow {
@@ -49,13 +58,15 @@ interface NameRow {
 	language: string
 }
 
-/** Walk parent chain up to 6 levels. */
+/**
+ * Walk parent chain up to 6 levels.
+ */
 function chainOf(db: DatabaseSync, startID: number, _jpnNames: Map<number, string>): PlaceRow[] {
 	const stmt = db.prepare(`SELECT id, name, placetype, parent_id, country FROM spr WHERE id = ?`)
 	const out: PlaceRow[] = []
 	let id = startID
 
-	for (let i = 0; i < 6 && id > 0; i++) {
+	for (let i = 0; i < MAX_ANCESTRY_DEPTH && id > 0; i++) {
 		const row = stmt.get(id) as PlaceRow | undefined
 
 		if (!row) break
@@ -66,7 +77,9 @@ function chainOf(db: DatabaseSync, startID: number, _jpnNames: Map<number, strin
 	return out
 }
 
-/** Pick the best display name for a place: prefer Japanese variant, fall back to English. */
+/**
+ * Pick the best display name for a place: prefer Japanese variant, fall back to English.
+ */
 function pickName(row: PlaceRow, jpnNames: Map<number, string>): string {
 	return jpnNames.get(row.id) ?? row.name
 }
@@ -164,6 +177,7 @@ export function createWOFAdminJpAdapter(): CorpusAdapter {
 						corpus_version: "",
 						license: "CC-BY-4.0",
 					}
+
 					emitted++
 				}
 			} finally {
@@ -173,4 +187,7 @@ export function createWOFAdminJpAdapter(): CorpusAdapter {
 	}
 }
 
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
 export const wofAdminJpAdapter = createWOFAdminJpAdapter()

package/src/adapters/wof-admin-json/adapter.ts

@@ -61,13 +61,17 @@ const COUNTRY_DISPLAY_NAME: Record<string, string> = {
 	FR: "France",
 }
 
-/** BCP-47 locale defaulting for the corpus row's `locale` field. Defaulted by country. */
+/**
+ * BCP-47 locale defaulting for the corpus row's `locale` field. Defaulted by country.
+ */
 const LOCALE_BY_COUNTRY: Record<string, string> = {
 	US: "en-US",
 	FR: "fr-FR",
 }
 
-/** Map a WOF placetype to a Mailwoman `ComponentTag`, or `undefined` to skip. */
+/**
+ * Map a WOF placetype to a Mailwoman `ComponentTag`, or `undefined` to skip.
+ */
 function placetypeToTag(placetype: WhosOnFirstPlacetype | string): ComponentTag | undefined {
 	switch (placetype) {
 		case "country":
@@ -93,10 +97,14 @@ function placetypeToTag(placetype: WhosOnFirstPlacetype | string): ComponentTag
 }
 
 interface VariantSpec {
-	/** Hierarchy-variant id appended to `source_id`. */
+	/**
+	 * Hierarchy-variant id appended to `source_id`.
+	 */
 	suffix: string
 
-	/** Component tag → display string the adapter will hand to the runner. */
+	/**
+	 * Component tag → display string the adapter will hand to the runner.
+	 */
 	components: Partial<Record<ComponentTag, string>>
 }
 
@@ -202,6 +210,10 @@ export function nameSlotsFor(rec: WOFRecord): Array<{ key: string; value: string
 	return slots
 }
 
+/**
+ * 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 WOF_ADMIN_ADAPTER_ID = "wof-admin"
 
 /**
@@ -236,7 +248,7 @@ export function createWOFAdminAdapter(): CorpusAdapter {
 			const ancestry = buildAncestryIndex(byID)
 
 			// Pass 2: emit rows in sorted-id order for deterministic JSONL.
-			const ids = [...byID.keys()].sort((a, b) => a - b)
+			const ids = [...byID.keys()].toSorted((a, b) => a - b)
 			let emitted = 0
 
 			for (const id of ids) {
@@ -256,7 +268,7 @@ export function createWOFAdminAdapter(): CorpusAdapter {
 						if (!raw) continue
 						const aligned = reconcileComponents(variant.components, raw)
 
-						if (Object.keys(aligned).length === 0) continue
+						if (!Object.keys(aligned).length) continue
 
 						yield {
 							raw,
@@ -268,6 +280,7 @@ export function createWOFAdminAdapter(): CorpusAdapter {
 							corpus_version: "",
 							license: "CC0-1.0",
 						}
+
 						emitted++
 					}
 				}
@@ -276,5 +289,7 @@ export function createWOFAdminAdapter(): CorpusAdapter {
 	}
 }
 
-/** Single shared instance, suitable for `defaultAdapterRegistry`. */
+/**
+ * Single shared instance, suitable for `defaultAdapterRegistry`.
+ */
 export const wofAdminAdapter = createWOFAdminAdapter()