@mailwoman/corpus 8.6.0 → 9.1.0

package/src/shard-recipes/country-balanced.ts

@@ -24,15 +24,14 @@
  *   legacy script used.
  */
 
-import { spawnSync } from "node:child_process"
-
 import { COUNTRY_SURFACE_FORMS, CountryNames } from "@mailwoman/codex/country"
 import type { ComponentTag } from "@mailwoman/core/types"
+import { dataRootPath } from "@mailwoman/core/utils"
+import { stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import type { CanonicalRow } from "@mailwoman/corpus/types"
+import { alignRow } from "@mailwoman/corpus/utils"
 
-import { stableSourceID } from "../adapter.ts"
-import { alignRow } from "../align.ts"
-import type { CanonicalRow } from "../types.ts"
-import { makeMulberry32, type ShardRecipe } from "./scaffold.ts"
+import { makeMulberry32, readZippedCSVRecords, type ShardRecipe } from "./scaffold.ts"
 
 // v2: the country TOKEN is decoupled from the skeleton's locale and drawn from a BROAD pool — every
 // ISO canonical name + every curated surface form (endonyms/abbrevs). Surface forms are over-weighted
@@ -72,24 +71,72 @@ interface CountrySource {
  * countrywide extracts (FR/IT/NL) read region from the CSV when present.
  */
 const SOURCES: readonly CountrySource[] = [
-	{ zip: "/tmp/oa-cache/us__ia__statewide.zip", csv: "us/ia/statewide.csv", iso2: "US", region: "IA", order: "us" },
-	{ zip: "/tmp/oa-cache/us__il__cook.zip", csv: "us/il/cook.csv", iso2: "US", region: "IL", order: "us" },
-	{ zip: "/tmp/oa-cache/us__mt__statewide.zip", csv: "us/mt/statewide.csv", iso2: "US", region: "MT", order: "us" },
-	{ zip: "/tmp/oa-cache/us__sd__statewide.zip", csv: "us/sd/statewide.csv", iso2: "US", region: "SD", order: "us" },
-	{ zip: "/tmp/oa-cache/de__sn__statewide.zip", csv: "de/sn/statewide.csv", iso2: "DE", region: "", order: "eu" },
-	{ zip: "/tmp/oa-cache/fr__countrywide.zip", csv: "fr/countrywide.csv", iso2: "FR", region: "", order: "fr" },
+	{
+		zip: dataRootPath("oa-cache", "us__ia__statewide.zip"),
+		csv: "us/ia/statewide.csv",
+		iso2: "US",
+		region: "IA",
+		order: "us",
+	},
+	{ zip: dataRootPath("oa-cache", "us__il__cook.zip"), csv: "us/il/cook.csv", iso2: "US", region: "IL", order: "us" },
+	{
+		zip: dataRootPath("oa-cache", "us__mt__statewide.zip"),
+		csv: "us/mt/statewide.csv",
+		iso2: "US",
+		region: "MT",
+		order: "us",
+	},
+	{
+		zip: dataRootPath("oa-cache", "us__sd__statewide.zip"),
+		csv: "us/sd/statewide.csv",
+		iso2: "US",
+		region: "SD",
+		order: "us",
+	},
+	{
+		zip: dataRootPath("oa-cache", "de__sn__statewide.zip"),
+		csv: "de/sn/statewide.csv",
+		iso2: "DE",
+		region: "",
+		order: "eu",
+	},
+	{
+		zip: dataRootPath("oa-cache", "fr__countrywide.zip"),
+		csv: "fr/countrywide.csv",
+		iso2: "FR",
+		region: "",
+		order: "fr",
+	},
 	// ES uses the Spanish IGN schema, not the OA standard columns — skipped here (codex still recognizes
 	// "España"/"Spain"). A dedicated IGN adapter is a follow-up.
-	{ zip: "/tmp/oa-cache/it__countrywide.zip", csv: "it/countrywide.csv", iso2: "IT", region: "", order: "eu" },
-	{ zip: "/tmp/oa-cache/nl__countrywide.zip", csv: "nl/countrywide.csv", iso2: "NL", region: "", order: "eu" },
+	{
+		zip: dataRootPath("oa-cache", "it__countrywide.zip"),
+		csv: "it/countrywide.csv",
+		iso2: "IT",
+		region: "",
+		order: "eu",
+	},
+	{
+		zip: dataRootPath("oa-cache", "nl__countrywide.zip"),
+		csv: "nl/countrywide.csv",
+		iso2: "NL",
+		region: "",
+		order: "eu",
+	},
 ]
 
 /**
  * Held-out for --golden: Vermont (US holdout) + Berlin (DE holdout) — geographic split, never trained.
  */
 const EVAL_SOURCES: readonly CountrySource[] = [
-	{ zip: "/tmp/oa-cache/us__vt__statewide.zip", csv: "us/vt/statewide.csv", iso2: "US", region: "VT", order: "us" },
-	{ zip: "/tmp/oa-cache/de__berlin.zip", csv: "de/berlin.csv", iso2: "DE", region: "", order: "eu" },
+	{
+		zip: dataRootPath("oa-cache", "us__vt__statewide.zip"),
+		csv: "us/vt/statewide.csv",
+		iso2: "US",
+		region: "VT",
+		order: "us",
+	},
+	{ zip: dataRootPath("oa-cache", "de__berlin.zip"), csv: "de/berlin.csv", iso2: "DE", region: "", order: "eu" },
 ]
 
 /**
@@ -105,81 +152,20 @@ interface CountryTuple {
 	order: string
 }
 
-function splitCSV(line: string): string[] {
-	const out: string[] = []
-
-	let cur = "",
-		inQ = false
-
-	for (let i = 0; i < line.length; i++) {
-		const c = line[i]
-
-		if (inQ) {
-			if (c === '"') {
-				if (line[i + 1] === '"') {
-					cur += '"'
-
-					i++
-				} else {
-					inQ = false
-				}
-			} else {
-				cur += c
-			}
-		} else if (c === '"') {
-			inQ = true
-		} else if (c === ",") {
-			out.push(cur)
-			cur = ""
-		} else {
-			cur += c
-		}
-	}
-
-	out.push(cur)
-
-	return out
-}
-
-function readTuples(source: CountrySource, limit: number): CountryTuple[] {
-	// countrywide extracts (FR/IT/NL) are GB-scale — cap the bytes with `head` (read ~8 lines per wanted
-	// tuple to survive dedup/skips) so the toString stays under V8's string limit.
-	const maxLines = Math.max(limit * 8, 20_000) + 1
-
-	const r = spawnSync("bash", ["-c", `unzip -p "${source.zip}" "${source.csv}" | head -n ${maxLines}`], {
-		maxBuffer: 1024 * 1024 * 1024,
-		encoding: "buffer",
-	})
-
-	if (r.status !== 0) {
-		console.error(`  WARN: unzip failed for ${source.zip} (status ${r.status})`)
-
-		return []
-	}
-
-	const lines = r.stdout.toString("utf8").split(/\r?\n/)
-
-	if (lines.length < 2) return []
-	const header = splitCSV(lines[0]!).map((h) => h.trim().toLowerCase())
-	const idx = (n: string): number => header.indexOf(n)
-
-	const iNum = idx("number"),
-		iStreet = idx("street"),
-		iCity = idx("city"),
-		iRegion = idx("region"),
-		iPost = idx("postcode")
-
-	const get = (cells: string[], i: number): string => (i >= 0 && i < cells.length ? (cells[i] ?? "").trim() : "")
+/**
+ * The countrywide extracts (FR/IT/NL) are GB-scale, so this reads only as far as `limit` distinct tuples: the `break`
+ * closes the reader, which releases the archive without inflating the rest of the member.
+ */
+async function readTuples(source: CountrySource, limit: number): Promise<CountryTuple[]> {
 	const tuples: CountryTuple[] = []
 	const seen = new Set<string>()
 
-	for (let li = 1; li < lines.length && tuples.length < limit; li++) {
-		if (!lines[li]) continue
-		const cells = splitCSV(lines[li]!)
+	for await (const row of readZippedCSVRecords(source.zip, source.csv)) {
+		if (tuples.length >= limit) break
 
-		const street = get(cells, iStreet),
-			locality = get(cells, iCity),
-			house_number = get(cells, iNum)
+		const street = row.street ?? "",
+			locality = row.city ?? "",
+			house_number = row.number ?? ""
 
 		if (!street || !locality || !house_number) continue
 		const key = `${house_number}|${street}|${locality}`.toLowerCase()
@@ -191,8 +177,8 @@ function readTuples(source: CountrySource, limit: number): CountryTuple[] {
 			house_number,
 			street,
 			locality,
-			region: get(cells, iRegion) || source.region,
-			postcode: get(cells, iPost),
+			region: row.region || source.region,
+			postcode: row.postcode ?? "",
 			iso2: source.iso2,
 			order: source.order,
 		})
@@ -450,7 +436,7 @@ export const countryBalancedRecipe: ShardRecipe = {
 		const pool: CountryTuple[] = []
 
 		for (const s of sources) {
-			const t = readTuples(s, perSource)
+			const t = await readTuples(s, perSource)
 
 			console.error(`  ${s.csv} (${s.iso2}): ${t.length} tuples`)
 
@@ -460,7 +446,7 @@ export const countryBalancedRecipe: ShardRecipe = {
 		}
 
 		if (!pool.length) {
-			throw new Error("No tuples — are the cached OA zips present in /tmp/oa-cache?")
+			throw new Error(`No tuples — are the cached OA zips present in ${dataRootPath("oa-cache")}?`)
 		}
 
 		let emitted = 0

package/src/shard-recipes/fr-admin-split.ts

@@ -32,11 +32,11 @@
 
 import { departementForCodePostal } from "@mailwoman/codex/fr"
 import type { ComponentTag } from "@mailwoman/core/types"
+import { stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import type { CanonicalRow } from "@mailwoman/corpus/types"
+import { alignRow } from "@mailwoman/corpus/utils"
 import { CSVSpliterator, Delimiters } from "spliterator"
 
-import { stableSourceID } from "../adapter.ts"
-import { alignRow } from "../align.ts"
-import type { CanonicalRow } from "../types.ts"
 import { makeMulberry32, type ShardRecipe } from "./scaffold.ts"
 
 /* oxlint-disable sister-software/no-unnamed-threshold -- the bare decimals below are weighted-sampler

package/src/shard-recipes/fr-fragment.ts

@@ -68,7 +68,7 @@
  *   blow-up). Disable that mask for any run including this shard.
  */
 
-import { readFileSync } from "node:fs"
+import { TextSpliterator } from "spliterator"
 
 import { decomposeFrStreet } from "../adapters/ban/street-decompose.ts"
 import { alignAndWrite, makeMulberry32, readTuples, type ShardRecipe, shardSourceID } from "./scaffold.ts"
@@ -187,12 +187,15 @@ export const frFragmentRecipe: ShardRecipe = {
 			)
 		}
 
-		const excluded = new Set(
-			readFileSync(excludePath, "utf8")
-				.split("\n")
-				.map((line) => line.trim())
-				.filter((line) => line && !line.startsWith("#"))
-		)
+		const excluded = new Set<string>()
+
+		for await (const line of TextSpliterator.fromAsync(excludePath)) {
+			const trimmed = line.trim()
+
+			if (trimmed && !trimmed.startsWith("#")) {
+				excluded.add(trimmed)
+			}
+		}
 
 		if (!excluded.size) throw new Error(`fr-fragment: --exclude-surfaces "${excludePath}" listed no surfaces`)
 

package/src/shard-recipes/fr-lieudit.ts

@@ -31,11 +31,11 @@ import { extractBANAddrPoints } from "@mailwoman/ban/sdk"
 import { COUNTRY_SURFACE_FORMS } from "@mailwoman/codex/country"
 import type { ComponentTag } from "@mailwoman/core/types"
 import { dataRootPath } from "@mailwoman/core/utils"
+import { stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import type { CanonicalRow } from "@mailwoman/corpus/types"
+import { alignRow } from "@mailwoman/corpus/utils"
 
-import { stableSourceID } from "../adapter.ts"
 import { decomposeFrStreet } from "../adapters/ban/street-decompose.ts"
-import { alignRow } from "../align.ts"
-import type { CanonicalRow } from "../types.ts"
 import { makeMulberry32, type ShardRecipe } from "./scaffold.ts"
 
 /**

package/src/shard-recipes/fr-order.ts

@@ -30,14 +30,13 @@
  *   inlined the variant logic and never called it).
  */
 
-import { spawnSync } from "node:child_process"
-
 import type { ComponentTag } from "@mailwoman/core/types"
+import { dataRootPath } from "@mailwoman/core/utils"
+import { stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import type { CanonicalRow } from "@mailwoman/corpus/types"
+import { alignRow } from "@mailwoman/corpus/utils"
 
-import { stableSourceID } from "../adapter.ts"
-import { alignRow } from "../align.ts"
-import type { CanonicalRow } from "../types.ts"
-import { makeMulberry32, type ShardRecipe } from "./scaffold.ts"
+import { makeMulberry32, readZippedCSVRecords, type ShardRecipe } from "./scaffold.ts"
 
 /* oxlint-disable sister-software/no-unnamed-threshold -- the bare decimals below are weighted-sampler
    cutoffs, not thresholds: `const r = random()` followed by a cascade of `r < 0.4` branches IS the
@@ -45,7 +44,7 @@ import { makeMulberry32, type ShardRecipe } from "./scaffold.ts"
    would hide the distribution behind a wall of identifiers. Genuine thresholds in these files are
    extracted as named constants above. */
 
-const SOURCE = { zip: "/tmp/oa-cache/fr__countrywide.zip", csv: "fr/countrywide.csv" }
+const SOURCE = { zip: dataRootPath("oa-cache", "fr__countrywide.zip"), csv: "fr/countrywide.csv" }
 
 /**
  * Ordinal suffixes used in French house numbers (BAN corpus), to cover the "8 bis" sub-mode.
@@ -71,84 +70,20 @@ interface FrTuple {
 }
 
 /**
- * Minimal RFC-4180-ish splitter (handles quoted fields with doubled-quote escaping).
- */
-function splitCSV(line: string): string[] {
-	const out: string[] = []
-	let cur = ""
-	let inQ = false
-
-	for (let i = 0; i < line.length; i++) {
-		const c = line[i]
-
-		if (inQ) {
-			if (c === '"') {
-				if (line[i + 1] === '"') {
-					cur += '"'
-
-					i++
-				} else {
-					inQ = false
-				}
-			} else {
-				cur += c
-			}
-		} else if (c === '"') {
-			inQ = true
-		} else if (c === ",") {
-			out.push(cur)
-			cur = ""
-		} else {
-			cur += c
-		}
-	}
-
-	out.push(cur)
-
-	return out
-}
-
-/**
- * Stream FR tuples out of the cached OA zip. The countrywide extract is GB-scale; cap with `head` to stay under V8's
- * string limit. Only keeps rows with a house_number (the shard's core signal) and a postcode (required for
- * reversed-order rendering to be meaningful).
+ * Stream FR tuples out of the cached OA zip. The countrywide extract is GB-scale, so this reads only as far as `limit`
+ * distinct tuples — the `break` closes the reader and releases the archive. Only keeps rows with a house_number (the
+ * shard's core signal) and a postcode (required for reversed-order rendering to be meaningful).
  */
-function readTuples(limit: number): FrTuple[] {
-	const maxLines = Math.max(limit * 8, 40_000) + 1
-
-	const r = spawnSync("bash", ["-c", `unzip -p "${SOURCE.zip}" "${SOURCE.csv}" | head -n ${maxLines}`], {
-		maxBuffer: 1024 * 1024 * 1024,
-		encoding: "buffer",
-	})
-
-	if (r.status !== 0) {
-		console.error(`  WARN: unzip failed for ${SOURCE.zip} (status ${r.status})`)
-
-		return []
-	}
-
-	const lines = r.stdout.toString("utf8").split(/\r?\n/)
-
-	if (lines.length < 2) return []
-	const header = splitCSV(lines[0]!).map((h) => h.trim().toLowerCase())
-	const idx = (name: string): number => header.indexOf(name)
-
-	const iNum = idx("number"),
-		iStreet = idx("street"),
-		iCity = idx("city"),
-		iPost = idx("postcode")
-
-	const get = (cells: string[], i: number): string => (i >= 0 && i < cells.length ? (cells[i] ?? "").trim() : "")
+async function readTuples(limit: number): Promise<FrTuple[]> {
 	const tuples: FrTuple[] = []
 	const seen = new Set<string>()
 
-	for (let li = 1; li < lines.length && tuples.length < limit; li++) {
-		if (!lines[li]) continue
-		const cells = splitCSV(lines[li]!)
-		const street = get(cells, iStreet)
-		const locality = get(cells, iCity)
-		const house_number = get(cells, iNum)
-		const postcode = get(cells, iPost)
+	for await (const row of readZippedCSVRecords(SOURCE.zip, SOURCE.csv)) {
+		if (tuples.length >= limit) break
+		const street = row.street ?? ""
+		const locality = row.city ?? ""
+		const house_number = row.number ?? ""
+		const postcode = row.postcode ?? ""
 
 		// Require all four fields: HN is the signal; postcode drives reversed-order variants.
 		if (!street || !locality || !house_number || !postcode) continue
@@ -211,12 +146,12 @@ export const frOrderRecipe: ShardRecipe = {
 
 		// Over-read from the CSV so the dedup + filter pass can fill `count` rows.
 		const poolLimit = Math.max(count * 8, 40_000)
-		const pool = readTuples(poolLimit)
+		const pool = await readTuples(poolLimit)
 
 		console.error(`  ${SOURCE.csv}: ${pool.length} unique tuples (capped read)`)
 
 		if (!pool.length) {
-			throw new Error("No FR tuples found — is /tmp/oa-cache/fr__countrywide.zip present?")
+			throw new Error(`No FR tuples found — is ${SOURCE.zip} present?`)
 		}
 
 		let emitted = 0

package/src/shard-recipes/german.ts

@@ -15,12 +15,12 @@
  *   docs/articles/evals/resolver-geo/2026-06-06-anchor-pilot.md (the order-artifact correction).
  */
 
-import { spawnSync } from "node:child_process"
+import { dataRootPath } from "@mailwoman/core/utils"
+import { stableSourceID } from "@mailwoman/corpus/adapters/utils"
+import { synthesizeGermanRow, type LocaleBaseTuple } from "@mailwoman/corpus/synthesizers/german"
+import { alignRow } from "@mailwoman/corpus/utils"
 
-import { stableSourceID } from "../adapter.ts"
-import { alignRow } from "../align.ts"
-import { synthesizeGermanRow, type LocaleBaseTuple } from "../synthesize-german.ts"
-import { makeMulberry32, type ShardRecipe } from "./scaffold.ts"
+import { makeMulberry32, readZippedCSVRecords, type ShardRecipe } from "./scaffold.ts"
 
 /**
  * A German OA source (cached zip) + the Bundesland the file covers (OA's REGION column is empty for DE).
@@ -37,89 +37,26 @@ interface GermanSource {
  * Berlin (a city-state, region==locality); sn/statewide → Sachsen.
  */
 const SOURCES: GermanSource[] = [
-	{ zip: "/tmp/oa-cache/de__berlin.zip", csv: "de/berlin.csv", region: "Berlin" },
-	{ zip: "/tmp/oa-cache/de__sn__statewide.zip", csv: "de/sn/statewide.csv", region: "Sachsen" },
+	{ zip: dataRootPath("oa-cache", "de__berlin.zip"), csv: "de/berlin.csv", region: "Berlin" },
+	{ zip: dataRootPath("oa-cache", "de__sn__statewide.zip"), csv: "de/sn/statewide.csv", region: "Sachsen" },
 ]
 
 /**
- * Minimal RFC-4180-ish splitter (handles quoted fields).
+ * Stream real German tuples out of a cached OA zip.
  */
-function splitCSV(line: string): string[] {
-	const out: string[] = []
-	let cur = ""
-	let inQ = false
-
-	for (let i = 0; i < line.length; i++) {
-		const c = line[i]!
-
-		if (inQ) {
-			if (c === '"') {
-				if (line[i + 1] === '"') {
-					cur += '"'
-
-					i++
-				} else {
-					inQ = false
-				}
-			} else {
-				cur += c
-			}
-		} else if (c === '"') {
-			inQ = true
-		} else if (c === ",") {
-			out.push(cur)
-			cur = ""
-		} else {
-			cur += c
-		}
-	}
-
-	out.push(cur)
-
-	return out
-}
-
-/**
- * Stream real German tuples out of a cached OA zip (buffered `unzip -p`).
- */
-function readGermanTuples(source: GermanSource): LocaleBaseTuple[] {
-	const r = spawnSync("unzip", ["-p", source.zip, source.csv], { maxBuffer: 1024 * 1024 * 1024, encoding: "buffer" })
-
-	if (r.status !== 0) {
-		console.error(`  WARN: unzip failed for ${source.zip} (status ${r.status})`)
-
-		return []
-	}
-
-	const lines = r.stdout.toString("utf8").split(/\r?\n/)
-
-	if (lines.length < 2) return []
-	const header = splitCSV(lines[0]!).map((h) => h.trim().toLowerCase())
-	const idx = (name: string): number => header.indexOf(name)
-
-	const iNum = idx("number"),
-		iStreet = idx("street"),
-		iCity = idx("city"),
-		iRegion = idx("region"),
-		iPost = idx("postcode")
-
-	const get = (cells: string[], i: number): string => (i >= 0 && i < cells.length ? (cells[i] ?? "").trim() : "")
+async function readGermanTuples(source: GermanSource): Promise<LocaleBaseTuple[]> {
 	const tuples: LocaleBaseTuple[] = []
 	const seen = new Set<string>()
 
-	for (let li = 1; li < lines.length; li++) {
-		const lineStr = lines[li]
-
-		if (!lineStr) continue
-		const cells = splitCSV(lineStr)
-		const street = get(cells, iStreet)
-		const locality = get(cells, iCity)
+	for await (const row of readZippedCSVRecords(source.zip, source.csv)) {
+		const street = row.street ?? ""
+		const locality = row.city ?? ""
 
 		if (!street || !locality) continue
-		const house_number = get(cells, iNum)
-		const postcode = get(cells, iPost)
+		const house_number = row.number ?? ""
+		const postcode = row.postcode ?? ""
 		// OA's REGION column is empty for DE — fall back to the source's Bundesland (set per file).
-		const region = get(cells, iRegion) || source.region || ""
+		const region = row.region || source.region || ""
 		const key = `${house_number}|${street}|${locality}|${postcode}`.toLowerCase()
 
 		if (seen.has(key)) continue
@@ -155,7 +92,7 @@ export const germanRecipe: ShardRecipe = {
 		const pool: LocaleBaseTuple[] = []
 
 		for (const s of SOURCES) {
-			const t = readGermanTuples(s)
+			const t = await readGermanTuples(s)
 
 			console.error(`  ${s.csv}: ${t.length} unique tuples`)
 
@@ -165,7 +102,7 @@ export const germanRecipe: ShardRecipe = {
 		}
 
 		if (!pool.length) {
-			throw new Error("No German tuples found — are the cached zips present in /tmp/oa-cache?")
+			throw new Error(`No German tuples found — are the cached zips present in ${dataRootPath("oa-cache")}?`)
 		}
 
 		let emitted = 0

package/src/shard-recipes/house-venue.ts

@@ -10,7 +10,8 @@
  *   scripts/build-house-venue-shard.mjs.
  */
 
-import { synthesizeHouseVenueRow, type HouseVenueBaseTuple } from "../synthesize-house-venue.ts"
+import { synthesizeHouseVenueRow, type HouseVenueBaseTuple } from "@mailwoman/corpus/synthesizers/house-venue"
+
 import { alignAndWrite, makeLcg, readTuples, shardSourceID, type ShardRecipe } from "./scaffold.ts"
 
 const LICENSE = "Synthetic — derived from CC-BY / public-domain input tuples"

package/src/shard-recipes/index.ts

@@ -29,9 +29,10 @@ import { poBoxCedexRecipe } from "./po-box-cedex.ts"
 import { poBoxRecipe } from "./po-box.ts"
 import type { ShardRecipe } from "./scaffold.ts"
 import { siBareVillageRecipe } from "./si-bare-village.ts"
-import { streetAffixRecipe } from "./street-affix.ts"
+import { streetAffixRecipe, suffixBoundaryRecipe } from "./street-affix.ts"
 import { streetBareRecipe } from "./street-bare.ts"
 import { streetRecipe } from "./street.ts"
+import { subVenueRecipe } from "./sub-venue.ts"
 import { unitRecipe } from "./unit.ts"
 
 export * from "./scaffold.ts"
@@ -43,6 +44,7 @@ const RECIPES: readonly ShardRecipe[] = [
 	streetRecipe,
 	streetBareRecipe,
 	streetAffixRecipe,
+	suffixBoundaryRecipe,
 	noStreetRecipe,
 	houseVenueRecipe,
 	poBoxRecipe,
@@ -64,6 +66,7 @@ const RECIPES: readonly ShardRecipe[] = [
 	countryBalancedRecipe,
 	boundaryStressRecipe,
 	anchorAbsorptionRecipe,
+	subVenueRecipe,
 ]
 
 /**