@mailwoman/corpus 8.0.0 → 8.2.0

package/src/adapters/gnaf/adapter.ts

@@ -34,8 +34,14 @@ 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 GNAF_ADAPTER_ID = "gnaf"
-/** Open G-NAF is freely redistributable with attribution to Geoscape Australia (CC-BY-style). */
+/**
+ * Open G-NAF is freely redistributable with attribution to Geoscape Australia (CC-BY-style).
+ */
 export const GNAF_DEFAULT_LICENSE = "CC-BY-4.0"
 
 interface GNAFTuple {
@@ -102,8 +108,10 @@ export function createGNAFAdapter(): CorpusAdapter {
 
 				const orders = renderOrders(t)
 				const order = idx % orders.length
+
 				idx++
 				const raw = orders[order]!
+
 				const components: CanonicalRow["components"] = {
 					house_number: t.house_number,
 					street: t.street,
@@ -119,7 +127,8 @@ export function createGNAFAdapter(): CorpusAdapter {
 
 				const aligned = reconcileComponents(components, raw)
 
-				if (Object.keys(aligned).length === 0) continue
+				if (!Object.keys(aligned).length) continue
+
 				yield {
 					raw,
 					components: aligned,
@@ -130,10 +139,14 @@ export function createGNAFAdapter(): CorpusAdapter {
 					corpus_version: "",
 					license: GNAF_DEFAULT_LICENSE,
 				}
+
 				emitted++
 			}
 		},
 	}
 }
 
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
 export const gnafAdapter = createGNAFAdapter()

package/src/adapters/gnaf/assemble.ts

@@ -33,17 +33,25 @@ import { join } from "node:path"
 import { PSVSpliterator } from "spliterator"
 
 export interface GNAFAssembleOptions {
-	/** G-NAF `Standard` directory (holds the per-state `*_psv.psv` tables). */
+	/**
+	 * G-NAF `Standard` directory (holds the per-state `*_psv.psv` tables).
+	 */
 	standardDir: string
-	/** Target sample size (uniform reservoir → population-proportional across states). */
+	/**
+	 * Target sample size (uniform reservoir → population-proportional across states).
+	 */
 	sampleSize: number
-	/** Output JSONL path. */
+	/**
+	 * Output JSONL path.
+	 */
 	out: string
 	/**
 	 * Optional held-out eval JSONL (rows with a `components` field) — its (street,locality,postcode) are excluded.
 	 */
 	holdoutPath?: string
-	/** Progress sink (the CLI passes a setter). */
+	/**
+	 * Progress sink (the CLI passes a setter).
+	 */
 	onProgress?: (message: string) => void
 }
 
@@ -54,25 +62,32 @@ export interface GNAFAssembleResult {
 	byState: Record<string, number>
 }
 
-/** UPPERCASE → Title Case, preserving intra-word apostrophes/hyphens (O'Brien, Coff's Harbour). */
+/**
+ * UPPERCASE → Title Case, preserving intra-word apostrophes/hyphens (O'Brien, Coff's Harbour).
+ */
 export function titlecase(s: string): string {
 	return s
 		.toLowerCase()
-		.replace(/(^|[\s'\-/])([a-z])/g, (_m, p: string, c: string) => p + c.toUpperCase())
+		.replaceAll(/(^|[\s'\-/])([a-z])/g, (_m, p: string, c: string) => p + c.toUpperCase())
 		.trim()
 }
 
-/** Holdout/dedup key: a street within a locality+postcode (house-number-agnostic — conservative). */
+/**
+ * Holdout/dedup key: a street within a locality+postcode (house-number-agnostic — conservative).
+ */
 export function gnafHoldoutKey(street: string, locality: string, postcode: string): string {
 	return `${street}|${locality}|${postcode}`.toLowerCase()
 }
 
 type Row = Record<string, string | number | undefined>
+
 async function* psvObjects(path: string): AsyncIterable<Row> {
 	yield* PSVSpliterator.fromAsync(path, { mode: "object", header: true }) as AsyncIterable<Row>
 }
 
-/** Load a small lookup table fully into a Map keyed by `keyCol`. */
+/**
+ * Load a small lookup table fully into a Map keyed by `keyCol`.
+ */
 async function loadMap<V>(paths: string[], keyCol: string, pick: (r: Row) => V): Promise<Map<string, V>> {
 	const m = new Map<string, V>()
 
@@ -89,7 +104,9 @@ async function loadMap<V>(paths: string[], keyCol: string, pick: (r: Row) => V):
 	return m
 }
 
-/** Build the held-out key set from an eval JSONL whose rows carry a `components` object. */
+/**
+ * Build the held-out key set from an eval JSONL whose rows carry a `components` object.
+ */
 async function loadHoldout(path: string): Promise<Set<string>> {
 	const keys = new Set<string>()
 	const text = await readFile(path, "utf8")
@@ -114,6 +131,7 @@ async function loadHoldout(path: string): Promise<Set<string>> {
 export async function assembleGNAF(opts: GNAFAssembleOptions): Promise<GNAFAssembleResult> {
 	const progress = opts.onProgress ?? (() => {})
 	const files = await readdir(opts.standardDir)
+
 	const pick = (re: RegExp, exclude?: RegExp) =>
 		files.filter((f) => re.test(f) && !(exclude && exclude.test(f))).map((f) => join(opts.standardDir, f))
 
@@ -129,16 +147,19 @@ export async function assembleGNAF(opts: GNAFAssembleOptions): Promise<GNAFAssem
 	}
 
 	progress(`loading STREET_LOCALITY (${streetPaths.length} files) + LOCALITY (${localityPaths.length})…`)
+
 	const streetMap = await loadMap(streetPaths, "STREET_LOCALITY_PID", (r) => ({
 		name: String(r.STREET_NAME ?? ""),
 		type: String(r.STREET_TYPE_CODE ?? ""),
 		suffix: String(r.STREET_SUFFIX_CODE ?? ""),
 	}))
+
 	const localityMap = await loadMap(localityPaths, "LOCALITY_PID", (r) => String(r.LOCALITY_NAME ?? ""))
 	progress(`streets=${streetMap.size.toLocaleString()} localities=${localityMap.size.toLocaleString()}`)
 
 	const reservoir: Array<{ house_number: string; street: string; locality: string; region: string; postcode: string }> =
 		[]
+
 	let seen = 0
 	let heldOut = 0
 
@@ -159,8 +180,10 @@ export async function assembleGNAF(opts: GNAFAssembleOptions): Promise<GNAFAssem
 
 			if (holdout.has(gnafHoldoutKey(street, locality, postcode))) {
 				heldOut++
+
 				continue
 			}
+
 			let house = numberFirst + (r.NUMBER_FIRST_SUFFIX ? String(r.NUMBER_FIRST_SUFFIX) : "")
 
 			if (r.NUMBER_LAST) {
@@ -170,7 +193,9 @@ export async function assembleGNAF(opts: GNAFAssembleOptions): Promise<GNAFAssem
 			if (r.FLAT_NUMBER) {
 				house = `${String(r.FLAT_NUMBER)}/${house}`
 			}
+
 			const tuple = { house_number: house, street, locality, region: state, postcode }
+
 			seen++
 
 			if (reservoir.length < opts.sampleSize) {
@@ -183,6 +208,7 @@ export async function assembleGNAF(opts: GNAFAssembleOptions): Promise<GNAFAssem
 				}
 			}
 		}
+
 		progress(`${state}: ${seen.toLocaleString()} valid joinable seen`)
 	}
 
@@ -193,7 +219,11 @@ export async function assembleGNAF(opts: GNAFAssembleOptions): Promise<GNAFAssem
 		out.write(JSON.stringify(t) + "\n")
 		byState[t.region] = (byState[t.region] ?? 0) + 1
 	}
-	await new Promise<void>((res) => out.end(res))
+
+	await new Promise<void>((res) => {
+		out.end(res)
+	})
+
 	progress(`wrote ${reservoir.length.toLocaleString()} tuples → ${opts.out}`)
 
 	return { written: reservoir.length, seen, heldOut, byState }

package/src/adapters/index.ts

@@ -81,55 +81,68 @@ for (const adapter of BUILTIN_ADAPTERS) {
 
 export { BAN_ADAPTER_ID, banAdapter } from "./ban/adapter.ts"
 export { FCC_BDC_ADAPTER_ID, FCC_BDC_DEFAULT_LICENSE, fccBdcAdapter } from "./fcc-bdc/adapter.ts"
+
 export {
 	GEONAMES_POSTAL_ADAPTER_ID,
 	GEONAMES_POSTAL_DEFAULT_LICENSE,
 	geonamesPostalAdapter,
 } from "./geonames-postal/adapter.ts"
+
 export { GEONAMES_ADAPTER_ID, GEONAMES_DEFAULT_LICENSE, geonamesAdapter } from "./geonames/adapter.ts"
+
 export {
 	OPENADDRESSES_ADAPTER_ID,
 	OPENADDRESSES_DEFAULT_LICENSE,
 	openaddressesAdapter,
 } from "./openaddresses/adapter.ts"
+
 export {
 	STATE_HI_SCHOOLS_ADAPTER_ID,
 	STATE_HI_SCHOOLS_DEFAULT_LICENSE,
 	stateHiSchoolsAdapter,
 } from "./state-hi-schools/adapter.ts"
+
 export {
 	STATE_IA_CONTRACTORS_ADAPTER_ID,
 	STATE_IA_CONTRACTORS_DEFAULT_LICENSE,
 	stateIaContractorsAdapter,
 } from "./state-ia-contractors/adapter.ts"
+
 export {
 	STATE_NY_NOTARIES_ADAPTER_ID,
 	STATE_NY_NOTARIES_DEFAULT_LICENSE,
 	stateNyNotariesAdapter,
 } from "./state-ny-notaries/adapter.ts"
+
 export {
 	STATE_TX_NOTARIES_ADAPTER_ID,
 	STATE_TX_NOTARIES_DEFAULT_LICENSE,
 	stateTxNotariesAdapter,
 } from "./state-tx-notaries/adapter.ts"
+
 export { TIGER_ADAPTER_ID, TIGER_DEFAULT_LICENSE, tigerAdapter } from "./tiger/adapter.ts"
+
 export {
 	USGOV_HRSA_FQHC_ADAPTER_ID,
 	USGOV_HRSA_FQHC_DEFAULT_LICENSE,
 	usgovHrsaFqhcAdapter,
 } from "./usgov-hrsa-fqhc/adapter.ts"
+
 export {
 	USGOV_IMLS_PLS_ADAPTER_ID,
 	USGOV_IMLS_PLS_DEFAULT_LICENSE,
 	usgovImlsPlsAdapter,
 } from "./usgov-imls-pls/adapter.ts"
+
 export { USGOV_IRS_BMF_ADAPTER_ID, USGOV_IRS_BMF_DEFAULT_LICENSE, usgovIrsBmfAdapter } from "./usgov-irs-bmf/adapter.ts"
 export { USGOV_NAD_ADAPTER_ID, USGOV_NAD_DEFAULT_LICENSE, usgovNADAdapter } from "./usgov-nad/adapter.ts"
 export { USGOV_NPPES_ADAPTER_ID, USGOV_NPPES_DEFAULT_LICENSE, usgovNPPESAdapter } from "./usgov-nppes/adapter.ts"
+
 export {
 	USGOV_SAMHSA_ADAPTER_ID,
 	USGOV_SAMHSA_DEFAULT_LICENSE,
 	usgovSamhsaTreatmentLocatorAdapter,
 } from "./usgov-samhsa-treatment-locator/adapter.ts"
+
 export { WOF_ADMIN_ADAPTER_ID, wofAdminAdapter } from "./wof-admin-json/adapter.ts"
 export { WOF_POSTALCODE_ADAPTER_ID, wofPostalcodeAdapter } from "./wof-postalcode-json/adapter.ts"

package/src/adapters/openaddresses/adapter.ts

@@ -43,7 +43,15 @@ import { formatAddress, reconcileComponents } from "../../format.ts"
 import { SHARE_ALIKE_PATTERN } from "../../license.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 OPENADDRESSES_ADAPTER_ID = "openaddresses"
+/**
+ * License carried by this source (CC-BY-4.0), attached to each row so downstream consumers inherit the terms rather
+ * than having to look them up.
+ */
 export const OPENADDRESSES_DEFAULT_LICENSE = "CC-BY-4.0"
 
 /**
@@ -63,7 +71,9 @@ interface OaProperties {
 	license?: string
 }
 
-/** Return a lowercase-keyed view of a Feature's properties so case variants both work. */
+/**
+ * Return a lowercase-keyed view of a Feature's properties so case variants both work.
+ */
 function normalizeProperties(raw: unknown): OaProperties {
 	if (!raw || typeof raw !== "object") return {}
 	const out: Record<string, string> = {}
@@ -79,7 +89,9 @@ function normalizeProperties(raw: unknown): OaProperties {
 	return out as OaProperties
 }
 
-/** Parse a single ND-GeoJSON line; return null for blanks, comments, or non-Feature shapes. */
+/**
+ * Parse a single ND-GeoJSON line; return null for blanks, comments, or non-Feature shapes.
+ */
 function parseFeatureLine(line: string): OaProperties | null {
 	const trimmed = line.trim()
 
@@ -135,6 +147,7 @@ export function createOpenaddressesAdapter(opts: OpenaddressesAdapterOptions = {
 					"openaddresses adapter: --country is required (OpenAddresses files are country-partitioned but rows lack a country field)"
 				)
 			}
+
 			const country = adapterOpts.country
 
 			// TextSpliterator streams string lines (parseFeatureLine keeps tolerating blank/`#`/
@@ -172,6 +185,7 @@ export function createOpenaddressesAdapter(opts: OpenaddressesAdapterOptions = {
 
 					if (!allowShareAlike && SHARE_ALIKE_PATTERN.test(license)) {
 						shareAlikeBlocked++
+
 						continue
 					}
 
@@ -207,9 +221,10 @@ export function createOpenaddressesAdapter(opts: OpenaddressesAdapterOptions = {
 
 					const aligned = reconcileComponents(components, raw)
 
-					if (Object.keys(aligned).length === 0) continue
+					if (!Object.keys(aligned).length) continue
 
 					const sourceIDSeed = props.hash?.trim() || props.id?.trim()
+
 					const sourceID = sourceIDSeed
 						? `${OPENADDRESSES_ADAPTER_ID}-${sourceIDSeed}`
 						: stableSourceID(OPENADDRESSES_ADAPTER_ID, aligned)
@@ -223,6 +238,7 @@ export function createOpenaddressesAdapter(opts: OpenaddressesAdapterOptions = {
 						corpus_version: "",
 						license,
 					}
+
 					emitted++
 				}
 			} finally {
@@ -234,4 +250,7 @@ export function createOpenaddressesAdapter(opts: OpenaddressesAdapterOptions = {
 	}
 }
 
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
 export const openaddressesAdapter = createOpenaddressesAdapter()

package/src/adapters/overture/adapter.ts

@@ -36,10 +36,20 @@ import { stableSourceID } from "../../adapter.ts"
 import { formatAddress, 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 OVERTURE_ADAPTER_ID = "overture"
+/**
+ * License carried by this source (CDLA-Permissive-2.0), attached to each row so downstream consumers inherit the terms
+ * rather than having to look them up.
+ */
 export const OVERTURE_DEFAULT_LICENSE = "CDLA-Permissive-2.0"
 
-/** The flattened per-row shape emitted by `ingest-overture-addresses.ts --corpus-jsonl`. */
+/**
+ * The flattened per-row shape emitted by `ingest-overture-addresses.ts --corpus-jsonl`.
+ */
 interface OvertureCorpusRow {
 	street?: string
 	number?: string
@@ -74,6 +84,7 @@ export function createOvertureAdapter(): CorpusAdapter {
 					"overture adapter: --country is required (the Overture JSONL is per-country and rows omit a country field)"
 				)
 			}
+
 			const country = opts.country
 
 			// TextSpliterator streams string lines (parseLine keeps tolerating blank/`#`/malformed
@@ -109,6 +120,7 @@ export function createOvertureAdapter(): CorpusAdapter {
 				if (/^\d/.test(number)) {
 					components.house_number = number
 				}
+
 				components.street = street
 
 				if (unit) {
@@ -129,7 +141,7 @@ export function createOvertureAdapter(): CorpusAdapter {
 
 				const aligned = reconcileComponents(components, raw)
 
-				if (Object.keys(aligned).length === 0) continue
+				if (!Object.keys(aligned).length) continue
 
 				yield {
 					raw,
@@ -140,10 +152,14 @@ export function createOvertureAdapter(): CorpusAdapter {
 					corpus_version: "",
 					license: OVERTURE_DEFAULT_LICENSE,
 				}
+
 				emitted++
 			}
 		},
 	}
 }
 
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
 export const overtureAdapter = createOvertureAdapter()

package/src/adapters/state-hi-schools/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 STATE_HI_SCHOOLS_ADAPTER_ID = "state-hi-schools"
+/**
+ * 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 STATE_HI_SCHOOLS_DEFAULT_LICENSE = "Public Domain"
 
 const HOUSE_NUMBER_PREFIX = /^(\d+(?:-\d+)?[A-Za-z]?)\s+(.+)$/
@@ -88,6 +96,7 @@ export function createStateHiSchoolsAdapter(): CorpusAdapter {
 			}
 
 			const stream = createReadStream(opts.inputPath, { encoding: "utf8" })
+
 			const parser = stream.pipe(
 				csvParse({
 					columns: true,
@@ -132,6 +141,7 @@ export function createStateHiSchoolsAdapter(): CorpusAdapter {
 					}
 
 					const streetPart = [split.house_number, split.street].filter(Boolean).join(" ").trim()
+
 					const raw = [
 						name,
 						streetPart,
@@ -145,6 +155,7 @@ export function createStateHiSchoolsAdapter(): CorpusAdapter {
 					if (Object.keys(aligned).length <= 2) continue
 
 					const code = (record.code ?? "").toString().trim()
+
 					const sourceID = code
 						? `${STATE_HI_SCHOOLS_ADAPTER_ID}-${code}`
 						: stableSourceID(STATE_HI_SCHOOLS_ADAPTER_ID, aligned)
@@ -159,6 +170,7 @@ export function createStateHiSchoolsAdapter(): CorpusAdapter {
 						corpus_version: "",
 						license: STATE_HI_SCHOOLS_DEFAULT_LICENSE,
 					}
+
 					emitted++
 				}
 			} finally {
@@ -168,4 +180,7 @@ export function createStateHiSchoolsAdapter(): CorpusAdapter {
 	}
 }
 
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
 export const stateHiSchoolsAdapter = createStateHiSchoolsAdapter()

package/src/adapters/state-ia-contractors/adapter.ts

@@ -25,7 +25,15 @@ 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 STATE_IA_CONTRACTORS_ADAPTER_ID = "state-ia-contractors"
+/**
+ * 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 STATE_IA_CONTRACTORS_DEFAULT_LICENSE = "Public Domain"
 
 const HOUSE_NUMBER_PREFIX = /^(\d+(?:-\d+)?[A-Za-z]?)\s+(.+)$/
@@ -66,6 +74,7 @@ export function createStateIaContractorsAdapter(): CorpusAdapter {
 			}
 
 			const stream = createReadStream(opts.inputPath, { encoding: "utf8" })
+
 			const parser = stream.pipe(
 				csvParse({
 					columns: true,
@@ -113,6 +122,7 @@ export function createStateIaContractorsAdapter(): CorpusAdapter {
 					}
 
 					const streetPart = [split.house_number, split.street].filter(Boolean).join(" ").trim()
+
 					const raw = [venue, streetPart, [city, [stateAbbr, zip].filter(Boolean).join(" ")].filter(Boolean).join(", ")]
 						.filter(Boolean)
 						.join(", ")
@@ -122,6 +132,7 @@ export function createStateIaContractorsAdapter(): CorpusAdapter {
 					if (Object.keys(aligned).length <= 2) continue
 
 					const regNum = (record["Registration #"] ?? "").trim()
+
 					const sourceID = regNum
 						? `${STATE_IA_CONTRACTORS_ADAPTER_ID}-${regNum}`
 						: stableSourceID(STATE_IA_CONTRACTORS_ADAPTER_ID, aligned)
@@ -136,6 +147,7 @@ export function createStateIaContractorsAdapter(): CorpusAdapter {
 						corpus_version: "",
 						license: STATE_IA_CONTRACTORS_DEFAULT_LICENSE,
 					}
+
 					emitted++
 				}
 			} finally {
@@ -145,4 +157,7 @@ export function createStateIaContractorsAdapter(): CorpusAdapter {
 	}
 }
 
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
 export const stateIaContractorsAdapter = createStateIaContractorsAdapter()

package/src/adapters/state-ny-notaries/adapter.ts

@@ -23,7 +23,15 @@ 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 STATE_NY_NOTARIES_ADAPTER_ID = "state-ny-notaries"
+/**
+ * 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 STATE_NY_NOTARIES_DEFAULT_LICENSE = "Public Domain"
 
 const HOUSE_NUMBER_PREFIX = /^(\d+(?:-\d+)?[A-Za-z]?)\s+(.+)$/
@@ -75,6 +83,7 @@ export function createStateNyNotariesAdapter(): CorpusAdapter {
 			}
 
 			const stream = createReadStream(opts.inputPath, { encoding: "utf8" })
+
 			const parser = stream.pipe(
 				csvParse({
 					columns: true,
@@ -134,6 +143,7 @@ export function createStateNyNotariesAdapter(): CorpusAdapter {
 					}
 
 					const streetPart = [split.house_number, split.street].filter(Boolean).join(" ").trim()
+
 					const raw = [venue, streetPart, [city, [stateAbbr, zip].filter(Boolean).join(" ")].filter(Boolean).join(", ")]
 						.filter(Boolean)
 						.join(", ")
@@ -143,6 +153,7 @@ export function createStateNyNotariesAdapter(): CorpusAdapter {
 					if (Object.keys(aligned).length <= 2) continue
 
 					const commNum = (record["Commission Number (UID)"] ?? "").trim()
+
 					const sourceID = commNum
 						? `${STATE_NY_NOTARIES_ADAPTER_ID}-${commNum}`
 						: stableSourceID(STATE_NY_NOTARIES_ADAPTER_ID, aligned)
@@ -157,6 +168,7 @@ export function createStateNyNotariesAdapter(): CorpusAdapter {
 						corpus_version: "",
 						license: STATE_NY_NOTARIES_DEFAULT_LICENSE,
 					}
+
 					emitted++
 				}
 			} finally {
@@ -166,4 +178,7 @@ export function createStateNyNotariesAdapter(): CorpusAdapter {
 	}
 }
 
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
 export const stateNyNotariesAdapter = createStateNyNotariesAdapter()

package/src/adapters/state-tx-notaries/adapter.ts

@@ -24,12 +24,22 @@ 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 STATE_TX_NOTARIES_ADAPTER_ID = "state-tx-notaries"
+/**
+ * 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 STATE_TX_NOTARIES_DEFAULT_LICENSE = "Public Domain"
 
 const HOUSE_NUMBER_PREFIX = /^(\d+(?:-\d+)?[A-Za-z]?)\s+(.+)$/
 
-/** Match trailing "CITY, ST ZIP" or "CITY, ST" at the end of an address line. */
+/**
+ * Match trailing "CITY, ST ZIP" or "CITY, ST" at the end of an address line.
+ */
 const CITY_STATE_ZIP_SUFFIX = /[,]?\s*([^,]+),\s*([A-Z]{2})\s*(\d{5}(?:-\d{4})?)?\s*$/i
 
 interface TxNotaryRow {
@@ -63,6 +73,7 @@ export function createStateTxNotariesAdapter(): CorpusAdapter {
 			}
 
 			const stream = createReadStream(opts.inputPath, { encoding: "utf8" })
+
 			const parser = stream.pipe(
 				csvParse({
 					columns: true,
@@ -90,7 +101,7 @@ export function createStateTxNotariesAdapter(): CorpusAdapter {
 
 					// Parse embedded city/state/zip from the trailing portion of the address.
 					// Addresses look like: "1215 MCMILLAN DR\nCEDAR HILL, TX 75104"
-					const addrSingleLine = rawAddress.replace(/\n/g, ", ")
+					const addrSingleLine = rawAddress.replaceAll("\n", ", ")
 					const cszMatch = CITY_STATE_ZIP_SUFFIX.exec(addrSingleLine)
 
 					if (!cszMatch) continue
@@ -126,6 +137,7 @@ export function createStateTxNotariesAdapter(): CorpusAdapter {
 					}
 
 					const streetPart = [split.house_number, split.street].filter(Boolean).join(" ").trim()
+
 					const raw = [venue, streetPart, [city, [stateAbbr, zip].filter(Boolean).join(" ")].filter(Boolean).join(", ")]
 						.filter(Boolean)
 						.join(", ")
@@ -148,6 +160,7 @@ export function createStateTxNotariesAdapter(): CorpusAdapter {
 						corpus_version: "",
 						license: STATE_TX_NOTARIES_DEFAULT_LICENSE,
 					}
+
 					emitted++
 				}
 			} finally {
@@ -157,4 +170,7 @@ export function createStateTxNotariesAdapter(): CorpusAdapter {
 	}
 }
 
+/**
+ * The configured adapter instance registered with the corpus builder.
+ */
 export const stateTxNotariesAdapter = createStateTxNotariesAdapter()