@mailwoman/corpus 8.1.0 → 8.3.0

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()

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()