@mailwoman/corpus 9.2.0 → 9.3.0

package/out/src/shard-recipes/street-bare.js

@@ -1,86 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `street-bare` shard recipe — BARE-street rows (the v0.8.0 harness lever, 2026-06-05). The
- *   `functional.test.ts` cluster (bare street names — "10th Ave", "Main St", "1 Main Pl") was
- *   mislabeled `locality` because {@link synthesizeStreetRow} only ever emitted streets WITH a ",
- *   City, ST ZIP" tail. This recipe emits streets BARE (`--bare-prob`, default 0.6) — no tail, only
- *   `street_prefix`/`street`/`street_suffix` (+ optional `house_number` at `--hn-prob`, default
- *   0.85) — over the built-in {@link DEFAULT_US_BASES} pool (round-robin). Generate-mode, US-only,
- *   in-distribution (no German-collapse risk). Ported from scripts/build-street-bare-shard.mjs.
- *
- *   Byte-fidelity: the legacy script seeded its own mulberry32 from `--seed`
- *   (`mulberry32(opts.seed)`); this recipe re-creates the SAME generator
- *   (`makeMulberry32(opts.seed)`) and preserves the synthesis call order exactly, so `--seed N`
- *   reproduces the legacy run byte-for-byte.
- */
-import { DEFAULT_US_BASES } from "#synthesizers/intersection";
-import { synthesizeStreetRow } from "#synthesizers/street";
-import { alignRow } from "#utils";
-import { makeMulberry32, shardSourceID } from "./scaffold.js";
-/**
- * Shard recipe registered with the corpus builder — see the file header for the parse behaviour it exists to exercise,
- * and `description` below for the surface form it generates.
- */
-export const streetBareRecipe = {
-    name: "street-bare",
-    description: "Bare-street rows (US): DEFAULT_US_BASES → synthesizeStreetRow (bare) → aligned LabeledRow",
-    mode: "generate",
-    options: [
-        { flag: "--bare-prob <p>", description: "P(emit the street BARE — no city/region/postcode tail). Default 0.6" },
-        { flag: "--hn-prob <p>", description: "P(emit a house number). Default 0.85" },
-    ],
-    async run(opts, write) {
-        const random = makeMulberry32(opts.seed);
-        const count = opts.count ?? 3000;
-        const bareProb = opts.bareProb ?? 0.6;
-        const hnProb = opts.hnProb ?? 0.85;
-        const source = opts.sourceName ?? "synth-street-bare";
-        let emitted = 0;
-        let skipped = 0;
-        let guard = 0;
-        while (emitted < count && guard++ < count * 5) {
-            const base = DEFAULT_US_BASES[emitted % DEFAULT_US_BASES.length];
-            const synth = synthesizeStreetRow(base, {
-                random,
-                bareProb,
-                includeHouseNumberProb: hnProb,
-            });
-            if (!synth) {
-                skipped++;
-                continue;
-            }
-            const isBare = synth.components.region === undefined;
-            const canonical = {
-                raw: synth.raw,
-                components: synth.components,
-                country: base.country,
-                locale: synth.locale,
-                source,
-                source_id: shardSourceID(source, {
-                    street: synth.components.street,
-                    street_suffix: synth.components.street_suffix,
-                    house_number: synth.components.house_number,
-                    bare: String(isBare),
-                    n: String(emitted),
-                }),
-                corpus_version: "0.4.0",
-                license: "Synthetic — US street templates, public-domain street/city pools",
-            };
-            // Strict labeled-only check (matches the legacy builder): alignRow always returns a `row`
-            // (labeled OR quarantined), so the scaffold's `alignAndWrite` would write quarantined rows
-            // too — call alignRow directly and skip anything not "labeled".
-            const aligned = alignRow(canonical);
-            if (aligned.kind !== "labeled" || !aligned.row) {
-                skipped++;
-                continue;
-            }
-            write(JSON.stringify({ ...aligned.row, synth_method: "street-bare", synth_base_id: null }) + "\n");
-            emitted++;
-        }
-        return { emitted, skipped };
-    },
-};
-//# sourceMappingURL=street-bare.js.map

package/out/src/shard-recipes/street-bare.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"street-bare.js","sourceRoot":"","sources":["../../../src/shard-recipes/street-bare.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;GAiBG;AAEH,OAAO,EAAE,gBAAgB,EAAE,MAAM,4BAA4B,CAAA;AAC7D,OAAO,EAAE,mBAAmB,EAAwB,MAAM,sBAAsB,CAAA;AAChF,OAAO,EAAE,QAAQ,EAAE,MAAM,QAAQ,CAAA;AAEjC,OAAO,EAAE,cAAc,EAAE,aAAa,EAA4C,MAAM,eAAe,CAAA;AAEvG;;;GAGG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAgB;IAC5C,IAAI,EAAE,aAAa;IACnB,WAAW,EAAE,2FAA2F;IACxG,IAAI,EAAE,UAAU;IAChB,OAAO,EAAE;QACR,EAAE,IAAI,EAAE,iBAAiB,EAAE,WAAW,EAAE,qEAAqE,EAAE;QAC/G,EAAE,IAAI,EAAE,eAAe,EAAE,WAAW,EAAE,sCAAsC,EAAE;KAC9E;IACD,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,MAAM,MAAM,GAAG,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACxC,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,IAAI,IAAI,CAAA;QAChC,MAAM,QAAQ,GAAG,IAAI,CAAC,QAAQ,IAAI,GAAG,CAAA;QACrC,MAAM,MAAM,GAAG,IAAI,CAAC,MAAM,IAAI,IAAI,CAAA;QAClC,MAAM,MAAM,GAAG,IAAI,CAAC,UAAU,IAAI,mBAAmB,CAAA;QAErD,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,KAAK,GAAG,CAAC,CAAA;QAEb,OAAO,OAAO,GAAG,KAAK,IAAI,KAAK,EAAE,GAAG,KAAK,GAAG,CAAC,EAAE,CAAC;YAC/C,MAAM,IAAI,GAAG,gBAAgB,CAAC,OAAO,GAAG,gBAAgB,CAAC,MAAM,CAAE,CAAA;YAEjE,MAAM,KAAK,GAAG,mBAAmB,CAAC,IAAuB,EAAE;gBAC1D,MAAM;gBACN,QAAQ;gBACR,sBAAsB,EAAE,MAAM;aAC9B,CAAC,CAAA;YAEF,IAAI,CAAC,KAAK,EAAE,CAAC;gBACZ,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,MAAM,MAAM,GAAG,KAAK,CAAC,UAAU,CAAC,MAAM,KAAK,SAAS,CAAA;YAEpD,MAAM,SAAS,GAAsB;gBACpC,GAAG,EAAE,KAAK,CAAC,GAAG;gBACd,UAAU,EAAE,KAAK,CAAC,UAAU;gBAC5B,OAAO,EAAE,IAAI,CAAC,OAAO;gBACrB,MAAM,EAAE,KAAK,CAAC,MAAM;gBACpB,MAAM;gBACN,SAAS,EAAE,aAAa,CAAC,MAAM,EAAE;oBAChC,MAAM,EAAE,KAAK,CAAC,UAAU,CAAC,MAAM;oBAC/B,aAAa,EAAE,KAAK,CAAC,UAAU,CAAC,aAAa;oBAC7C,YAAY,EAAE,KAAK,CAAC,UAAU,CAAC,YAAY;oBAC3C,IAAI,EAAE,MAAM,CAAC,MAAM,CAAC;oBACpB,CAAC,EAAE,MAAM,CAAC,OAAO,CAAC;iBAClB,CAAC;gBACF,cAAc,EAAE,OAAO;gBACvB,OAAO,EAAE,kEAAkE;aAC3E,CAAA;YAED,0FAA0F;YAC1F,2FAA2F;YAC3F,gEAAgE;YAChE,MAAM,OAAO,GAAG,QAAQ,CAAC,SAA2C,CAAC,CAAA;YAErE,IAAI,OAAO,CAAC,IAAI,KAAK,SAAS,IAAI,CAAC,OAAO,CAAC,GAAG,EAAE,CAAC;gBAChD,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,KAAK,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,GAAG,OAAO,CAAC,GAAG,EAAE,YAAY,EAAE,aAAa,EAAE,aAAa,EAAE,IAAI,EAAE,CAAC,GAAG,IAAI,CAAC,CAAA;YAElG,OAAO,EAAE,CAAA;QACV,CAAC;QAED,OAAO,EAAE,OAAO,EAAE,OAAO,EAAE,CAAA;IAC5B,CAAC;CACD,CAAA"}

package/out/src/shard-recipes/street.d.ts

@@ -1,15 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `street` shard recipe — synthetic street-decomposition rows for Stage 3 (US-only): tuples →
- *   {@link synthesizeStreetRow} → aligned LabeledRow. Ported from scripts/build-street-shard.mjs.
- */
-import { type ShardRecipe } from "./scaffold.ts";
-/**
- * Shard recipe registered with the corpus builder — see the file header for the parse behaviour it exists to exercise,
- * and `description` below for the surface form it generates.
- */
-export declare const streetRecipe: ShardRecipe;
-//# sourceMappingURL=street.d.ts.map

package/out/src/shard-recipes/street.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"street.d.ts","sourceRoot":"","sources":["../../../src/shard-recipes/street.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAKH,OAAO,EAAsC,KAAK,WAAW,EAAE,MAAM,eAAe,CAAA;AAEpF;;;GAGG;AACH,eAAO,MAAM,YAAY,EAAE,WA+D1B,CAAA"}

package/out/src/shard-recipes/street.js

@@ -1,69 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   `street` shard recipe — synthetic street-decomposition rows for Stage 3 (US-only): tuples →
- *   {@link synthesizeStreetRow} → aligned LabeledRow. Ported from scripts/build-street-shard.mjs.
- */
-import { stableSourceID } from "#adapters/utils";
-import { synthesizeStreetRow } from "#synthesizers/street";
-import { alignAndWrite, makeLcg, readTuples } from "./scaffold.js";
-/**
- * Shard recipe registered with the corpus builder — see the file header for the parse behaviour it exists to exercise,
- * and `description` below for the surface form it generates.
- */
-export const streetRecipe = {
-    name: "street",
-    description: "Street-decomposition rows (US): tuples → synthesizeStreetRow → aligned LabeledRow",
-    mode: "tuples",
-    options: [{ flag: "--house-number-prob <p>", description: "P(emit a house number). Default 0.85" }],
-    async run(opts, write) {
-        if (!opts.input)
-            throw new Error("street recipe requires --input <tuples.jsonl>");
-        const random = makeLcg(opts.seed);
-        const includeHouseNumberProb = opts.houseNumberProb ?? 0.85;
-        let read = 0;
-        let emitted = 0;
-        let skipped = 0;
-        for await (const tuple of readTuples(opts.input)) {
-            read++;
-            if (!tuple.locality || !tuple.region || !tuple.postcode || !tuple.country) {
-                skipped++;
-                continue;
-            }
-            if (tuple.country !== "US") {
-                skipped++;
-                continue;
-            }
-            for (let v = 0; v < opts.variants; v++) {
-                const synth = synthesizeStreetRow(tuple, { random, includeHouseNumberProb });
-                if (!synth)
-                    continue;
-                const ok = alignAndWrite(write, {
-                    raw: synth.raw,
-                    components: synth.components,
-                    country: tuple.country,
-                    locale: synth.locale,
-                    source: "synth-street",
-                    source_id: stableSourceID("synth-street", {
-                        locality: `${tuple.locality}#${v}`,
-                        region: tuple.region,
-                        postcode: tuple.postcode,
-                        country: tuple.country,
-                    }),
-                    corpus_version: "0.4.0",
-                    license: "Synthetic — public-domain street name + tuple combination",
-                }, "street-decomp");
-                if (ok) {
-                    emitted++;
-                }
-                else {
-                    skipped++;
-                }
-            }
-        }
-        return { read, emitted, skipped };
-    },
-};
-//# sourceMappingURL=street.js.map

package/out/src/shard-recipes/street.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"street.js","sourceRoot":"","sources":["../../../src/shard-recipes/street.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,OAAO,EAAE,cAAc,EAAE,MAAM,iBAAiB,CAAA;AAChD,OAAO,EAAE,mBAAmB,EAAwB,MAAM,sBAAsB,CAAA;AAEhF,OAAO,EAAE,aAAa,EAAE,OAAO,EAAE,UAAU,EAAoB,MAAM,eAAe,CAAA;AAEpF;;;GAGG;AACH,MAAM,CAAC,MAAM,YAAY,GAAgB;IACxC,IAAI,EAAE,QAAQ;IACd,WAAW,EAAE,mFAAmF;IAChG,IAAI,EAAE,QAAQ;IACd,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,yBAAyB,EAAE,WAAW,EAAE,sCAAsC,EAAE,CAAC;IACnG,KAAK,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK;QACpB,IAAI,CAAC,IAAI,CAAC,KAAK;YAAE,MAAM,IAAI,KAAK,CAAC,+CAA+C,CAAC,CAAA;QACjF,MAAM,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACjC,MAAM,sBAAsB,GAAG,IAAI,CAAC,eAAe,IAAI,IAAI,CAAA;QAC3D,IAAI,IAAI,GAAG,CAAC,CAAA;QACZ,IAAI,OAAO,GAAG,CAAC,CAAA;QACf,IAAI,OAAO,GAAG,CAAC,CAAA;QAEf,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,UAAU,CAAC,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;YAClD,IAAI,EAAE,CAAA;YAEN,IAAI,CAAC,KAAK,CAAC,QAAQ,IAAI,CAAC,KAAK,CAAC,MAAM,IAAI,CAAC,KAAK,CAAC,QAAQ,IAAI,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC;gBAC3E,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,IAAI,KAAK,CAAC,OAAO,KAAK,IAAI,EAAE,CAAC;gBAC5B,OAAO,EAAE,CAAA;gBAET,SAAQ;YACT,CAAC;YAED,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,QAAQ,EAAE,CAAC,EAAE,EAAE,CAAC;gBACxC,MAAM,KAAK,GAAG,mBAAmB,CAAC,KAAwB,EAAE,EAAE,MAAM,EAAE,sBAAsB,EAAE,CAAC,CAAA;gBAE/F,IAAI,CAAC,KAAK;oBAAE,SAAQ;gBAEpB,MAAM,EAAE,GAAG,aAAa,CACvB,KAAK,EACL;oBACC,GAAG,EAAE,KAAK,CAAC,GAAG;oBACd,UAAU,EAAE,KAAK,CAAC,UAAU;oBAC5B,OAAO,EAAE,KAAK,CAAC,OAAO;oBACtB,MAAM,EAAE,KAAK,CAAC,MAAM;oBACpB,MAAM,EAAE,cAAc;oBACtB,SAAS,EAAE,cAAc,CAAC,cAAc,EAAE;wBACzC,QAAQ,EAAE,GAAG,KAAK,CAAC,QAAQ,IAAI,CAAC,EAAE;wBAClC,MAAM,EAAE,KAAK,CAAC,MAAM;wBACpB,QAAQ,EAAE,KAAK,CAAC,QAAQ;wBACxB,OAAO,EAAE,KAAK,CAAC,OAAO;qBACtB,CAAC;oBACF,cAAc,EAAE,OAAO;oBACvB,OAAO,EAAE,2DAA2D;iBACpE,EACD,eAAe,CACf,CAAA;gBAED,IAAI,EAAE,EAAE,CAAC;oBACR,OAAO,EAAE,CAAA;gBACV,CAAC;qBAAM,CAAC;oBACP,OAAO,EAAE,CAAA;gBACV,CAAC;YACF,CAAC;QACF,CAAC;QAED,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,OAAO,EAAE,CAAA;IAClC,CAAC;CACD,CAAA"}

package/out/src/shard-recipes/sub-venue-sources.d.ts

@@ -1,228 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   The READ half of the `sub-venue` shard recipe (#35 step 4): which surfaces a locale may emit,
- *   what identifier follows them in that region, and which real names become the confound negatives.
- *   `sub-venue.ts` owns the WRITE half (line rendering + the emit loop) and the recipe registration;
- *   split because the two halves together run past the 750-line file cap, and this is the seam — one
- *   side reads disk and the ledger, the other side never touches either.
- *
- *   Every rule in this file is a rule about EVIDENCE, and each one has a measurement behind it:
- *
- *   - {@link promotedSurfacesFor} — a promotion names a designator, a phrase AND a locale, because the
- *       same token is a designator in one language and a disaster in another (`hall` is 0-of-3,273 in
- *       Great Britain and 35-of-40 in France).
- *   - {@link hasPromotedShape} — an `identifier-required` promotion is exercised only as
- *       `<phrase> <identifier>`. The de-DE `halle` board is the founding case: its 168-hit confound
- *       includes the CITY Halle (Saale), and only the identifier-bearing shape separates them.
- *   - {@link buildIdentifierModel} — the identifier distribution is measured per REGION, because it
- *       differs by country far more than the shared vocabulary suggests (GB gates 71% bare digit, ES
- *       35% ranges).
- *   - {@link isVenueSlotName} / {@link isSignIdentifier} — the filters that keep bus-stop codes, route
- *       descriptions and street names out of the slots they would mislabel. Both were written FROM
- *       smoke output, not predicted; the docstrings name the strings that produced them.
- */
-import type { LocaleBaseTuple } from "#synthesizers/german";
-import { type SubVenueLexiconTable, type SubVenuePromotion } from "#tools";
-/**
- * The committed lexicon, resolved through the package manifest.
- */
-export declare function defaultLexiconPath(): string;
-/**
- * Read and parse the lexicon. Strict: a corrupt lexicon is a build failure, not a fallback.
- */
-export declare function readSubVenueLexicon(path?: string): SubVenueLexiconTable;
-/**
- * Reject a name that is a route description, a junction, or a code rather than a name: embedded `/`, `;`, `,`, `:`,
- * parentheses, no letters, or a bare source code. Measured motivation, not taste — the GB extract's `platform` tier
- * contributes 7,549 `other`-shaped names like `kntgwdgj` and `SPEKE HALL ROAD/HILLFOOT AVE`.
- */
-export declare function isCleanName(name: string): boolean;
-/**
- * Is this name usable as a VENUE-slot string (positive venue or negative confound venue)?
- */
-export declare function isVenueSlotName(name: string): boolean;
-/**
- * A (designator, surface) pair usable in one locale, with the shape constraint the ledger attached to it.
- */
-export interface PromotedSurface {
-    designatorID: string;
-    phrase: string;
-    /**
-     * Rendering form — the phrase title-cased, which is how a sign writes it.
-     */
-    surface: string;
-    identifierRequired: boolean;
-    modifierEligible: boolean;
-}
-/**
- * Title-case a designator phrase for rendering (`flugsteig` → `Flugsteig`). Single-token by construction: every
- * promoted phrase in the ledger is one word.
- */
-export declare function titleCase(phrase: string): string;
-/**
- * Word-boundary containment, script-aware enough for the Latin legs this recipe runs.
- *
- * The boundary matters: without it `gate` matches Briggate and `wing` matches Wingate, which is how a harvest teaches
- * itself that a Yorkshire street is a sub-venue. (The lexicon's own harvest carries the same rule, and it drops the
- * boundary only for Han/Kana, which has no word boundaries and no leg here.)
- */
-export declare function containsPhrase(lowerName: string, phrase: string): boolean;
-/**
- * Is `value` what a sub-venue identifier looks like on a sign — an atom, or a range of two?
- *
- * Applied on top of {@link USABLE_IDENTIFIER_SHAPES}, which classifies but does not bound: `WSW3687` classifies as
- * `letter-digit` and is a station code.
- */
-export declare function isSignIdentifier(value: string): boolean;
-/**
- * Does a real extract name exercise `promoted` in one of the two shapes this shard teaches?
- *
- * Only `<phrase> <identifier>` and (English legs) `<modifier> <phrase>` qualify. That is stricter than "contains the
- * phrase", and the 2026-08-05 smoke is why: the loose test put `Glasgow Clyde College - Langside Campus`, `Terminal de
- * Ferry de Bilbao` and — worst — `Halle Wohnstadt Nord` into the attested pool, the last of which is a bare German
- * `Halle` wearing a name where the ledger requires an identifier. A whole venue's name is not a sub-venue string, and
- * an attested string that violates the promotion's own shape constraint is not attestation of it.
- *
- * The follower is checked with {@link classifyIdentifier} plus {@link isSignIdentifier} rather than "any following
- * word": `Wohnstadt` is a word, `8` is an identifier, and the de-DE board turns on exactly that difference.
- */
-export declare function hasPromotedShape(lowerName: string, promoted: PromotedSurface, modifiers?: readonly string[], english?: boolean): boolean;
-/**
- * {@link hasPromotedShape} plus the length cap that makes a string usable AS a sub-venue span.
- *
- * The two are separate because the difference decides which POOL a name lands in. `navette n2 vers terminal 2g` carries
- * the promoted shape and is five tokens: too long to be a sub-venue string, and disqualified from being a NEGATIVE
- * precisely because it does contain the shape. It belongs to neither pool, and only splitting the test says so.
- */
-export declare function matchesPromotedShape(lowerName: string, promoted: PromotedSurface, modifiers?: readonly string[], english?: boolean): boolean;
-/**
- * The surfaces a locale may emit as `unit`.
- *
- * Two inputs, and the difference between them is the advisory/binding split the ledger's docstring names:
- *
- * - The SHIPPED English vocabulary (`neural/venue-structure.ts`, re-declared in the lexicon as `shipped: true`) is a flat
- *   English list with no locale gate. It is promoted-by-shipping for the English legs, because the span proposer fires
- *   on it there today and the eval board's target cases are drawn from it.
- * - {@link SUBVENUE_PROMOTIONS} adds the localized surfaces and SUBTRACTS the rejections. A rejection of a shipped
- *   designator cannot un-ship it (nothing here stops the proposer firing on "Red Wing"), but it absolutely stops this
- *   recipe generating a positive: en-US `wing` produces negatives instead.
- */
-export declare function promotedSurfacesFor(locale: string, lexicon: SubVenueLexiconTable, promotions?: readonly SubVenuePromotion[]): PromotedSurface[];
-/**
- * The phrases REJECTED in a locale — the negatives' vocabulary.
- */
-export declare function rejectedPhrasesFor(locale: string, promotions?: readonly SubVenuePromotion[]): string[];
-interface ShapeBucket {
-    shape: string;
-    observations: number;
-    examples: string[];
-}
-/**
- * A region's identifier distributions: per designator, plus the pooled fallback.
- */
-export interface IdentifierModel {
-    byDesignator: Map<string, ShapeBucket[]>;
-    pooled: ShapeBucket[];
-}
-/**
- * Build the per-region identifier model out of the lexicon's `identifierShapes`.
- */
-export declare function buildIdentifierModel(lexicon: SubVenueLexiconTable, region: string): IdentifierModel;
-/**
- * Draw one identifier for `designatorID` in this region.
- *
- * Own distribution when it has {@link MIN_OWN_SHAPE_OBSERVATIONS} usable observations, else the region's pooled one.
- * Shapes are weighted by observation count and an example is drawn uniformly inside the chosen shape — the lexicon
- * ships up to eight per shape, which is the resolution available.
- */
-export declare function sampleIdentifier(model: IdentifierModel, designatorID: string, random: () => number): string | null;
-/**
- * Per-leg pools read off disk once.
- */
-export interface LegPools {
-    context: LocaleBaseTuple[];
-    /**
-     * Real venue names for the venue slot (stations, airports, campuses; US: airports, terminals, hospitals, rail).
-     */
-    venues: string[];
-    /**
-     * Real sub-venue strings, already filtered to this locale's promoted surfaces and their shape constraint.
-     */
-    attested: string[];
-    /**
-     * Real names carrying a surface REJECTED in this locale, for the venue slot of a negative row.
-     */
-    rejectedVenues: string[];
-    /**
-     * Real names that contain a designator inside a longer proper name — "Lochaline Ferry Terminal", "Kingdom Hall". The
-     * whole string is `venue`; nothing in it is `unit`.
-     */
-    longerNames: string[];
-    /**
-     * Real names carrying a PROMOTED phrase in a shape the promotion does NOT cover — `Halle Rosengarten`, `PHOENIX
-     * Halle`, `Halle-Südstadt`. The other half of an `identifier-required` ruling, and the only thing that teaches the
-     * shape boundary rather than the word: de-DE has no `reject` row at all, so without this class its 168-hit confound
-     * (97 of them the CITY Halle) would go untaught while its 32-hit promotion got 11,000 rows.
-     */
-    unpromotedShapes: string[];
-}
-/**
- * The name pools a source contributes. `attested` and `unpromotedShapes` come only from an extract — poi.db carries no
- * `tier` and no localized names, so it cannot say which side of a shape boundary a name sits on.
- */
-export type NamePools = Pick<LegPools, "venues" | "attested" | "rejectedVenues" | "longerNames" | "unpromotedShapes">;
-/**
- * The pools a source that has nothing to say contributes — en-US has no OSM extract, and DE/ES/GB are outside poi.db's
- * four countries. Empty rather than absent so a leg's merge is unconditional.
- */
-export declare const EMPTY_NAME_POOLS: NamePools;
-/**
- * What every pool reader needs to know about the leg it is reading for.
- */
-export interface PoolQuery {
-    promoted: readonly PromotedSurface[];
-    rejectedPhrases: readonly string[];
-    designatorPhrases: readonly string[];
-    modifiers: readonly string[];
-    english: boolean;
-}
-/**
- * Read one OSM extract and split it into name pools.
- */
-export declare function readExtractPools(path: string, query: PoolQuery): NamePools;
-/**
- * Read the venue + confound pools for a country out of `poi.db`.
- *
- * Poi.db is FOUR COUNTRIES — US 11,521,612 / CA 794,418 / FR 721,352 / MX 644,316 — so this is reachable for en-US and
- * fr-FR and nothing else, and a zero here is evidence of absence in four countries rather than in the world.
- */
-export declare function readPOIPools(dbPath: string, country: string, query: PoolQuery): NamePools;
-/**
- * Merge two sources' name pools.
- */
-export declare function mergeNamePools(a: NamePools, b: NamePools): NamePools;
-/**
- * Load the address skeletons every leg renders onto: GB / US / FR from the house-venue v3 tuples (the same 176,519 real
- * rows the `synth-house-venue` shard is built from, so the two shards' address halves are drawn from one pool), DE and
- * ES streamed out of OpenAddresses.
- */
-export declare function loadContextTuples(tuplesPath: string, seed: number): Promise<Map<string, LocaleBaseTuple[]>>;
-/**
- * The street-side confound classes, mined from the leg's OWN address tuples.
- *
- * Real streets, not invented ones. The 176,519-row context pool carries 195 GB `hall` streets, 114 GB `gate` streets,
- * 134 distinct GB `-gate` single tokens and a two-figure `<modifier> <designator>` population in both GB and US — small
- * absolute numbers, but every one of them a street somebody lives on, which is the property an invented list cannot
- * have.
- */
-export interface StreetNegatives {
-    designator: LocaleBaseTuple[];
-    modifierDesignator: LocaleBaseTuple[];
-    gateSuffix: LocaleBaseTuple[];
-}
-export declare function buildStreetNegatives(context: readonly LocaleBaseTuple[], designatorPhrases: readonly string[], modifiers: readonly string[], country: string): StreetNegatives;
-export {};
-//# sourceMappingURL=sub-venue-sources.d.ts.map

package/out/src/shard-recipes/sub-venue-sources.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"sub-venue-sources.d.ts","sourceRoot":"","sources":["../../../src/shard-recipes/sub-venue-sources.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;GAyBG;AASH,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,sBAAsB,CAAA;AAC3D,OAAO,EAGN,KAAK,oBAAoB,EAEzB,KAAK,iBAAiB,EACtB,MAAM,QAAQ,CAAA;AAOf;;GAEG;AACH,wBAAgB,kBAAkB,IAAI,MAAM,CAE3C;AAED;;GAEG;AACH,wBAAgB,mBAAmB,CAAC,IAAI,GAAE,MAA6B,GAAG,oBAAoB,CAE7F;AAyBD;;;;GAIG;AACH,wBAAgB,WAAW,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CAYjD;AA6FD;;GAEG;AACH,wBAAgB,eAAe,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CAUrD;AAMD;;GAEG;AACH,MAAM,WAAW,eAAe;IAC/B,YAAY,EAAE,MAAM,CAAA;IACpB,MAAM,EAAE,MAAM,CAAA;IACd;;OAEG;IACH,OAAO,EAAE,MAAM,CAAA;IACf,kBAAkB,EAAE,OAAO,CAAA;IAC3B,gBAAgB,EAAE,OAAO,CAAA;CACzB;AAED;;;GAGG;AACH,wBAAgB,SAAS,CAAC,MAAM,EAAE,MAAM,GAAG,MAAM,CAEhD;AAED;;;;;;GAMG;AACH,wBAAgB,cAAc,CAAC,SAAS,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,GAAG,OAAO,CAazE;AA4BD;;;;;GAKG;AACH,wBAAgB,gBAAgB,CAAC,KAAK,EAAE,MAAM,GAAG,OAAO,CAMvD;AAED;;;;;;;;;;;GAWG;AACH,wBAAgB,gBAAgB,CAC/B,SAAS,EAAE,MAAM,EACjB,QAAQ,EAAE,eAAe,EACzB,SAAS,GAAE,SAAS,MAAM,EAAO,EACjC,OAAO,UAAQ,GACb,OAAO,CAgBT;AAED;;;;;;GAMG;AACH,wBAAgB,oBAAoB,CACnC,SAAS,EAAE,MAAM,EACjB,QAAQ,EAAE,eAAe,EACzB,SAAS,GAAE,SAAS,MAAM,EAAO,EACjC,OAAO,UAAQ,GACb,OAAO,CAIT;AAED;;;;;;;;;;;GAWG;AACH,wBAAgB,mBAAmB,CAClC,MAAM,EAAE,MAAM,EACd,OAAO,EAAE,oBAAoB,EAC7B,UAAU,GAAE,SAAS,iBAAiB,EAAwB,GAC5D,eAAe,EAAE,CA6CnB;AAED;;GAEG;AACH,wBAAgB,kBAAkB,CACjC,MAAM,EAAE,MAAM,EACd,UAAU,GAAE,SAAS,iBAAiB,EAAwB,GAC5D,MAAM,EAAE,CAEV;AAMD,UAAU,WAAW;IACpB,KAAK,EAAE,MAAM,CAAA;IACb,YAAY,EAAE,MAAM,CAAA;IACpB,QAAQ,EAAE,MAAM,EAAE,CAAA;CAClB;AAED;;GAEG;AACH,MAAM,WAAW,eAAe;IAC/B,YAAY,EAAE,GAAG,CAAC,MAAM,EAAE,WAAW,EAAE,CAAC,CAAA;IACxC,MAAM,EAAE,WAAW,EAAE,CAAA;CACrB;AAoBD;;GAEG;AACH,wBAAgB,oBAAoB,CAAC,OAAO,EAAE,oBAAoB,EAAE,MAAM,EAAE,MAAM,GAAG,eAAe,CA0BnG;AAED;;;;;;GAMG;AACH,wBAAgB,gBAAgB,CAAC,KAAK,EAAE,eAAe,EAAE,YAAY,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,MAAM,GAAG,MAAM,GAAG,IAAI,CAkBlH;AAMD;;GAEG;AACH,MAAM,WAAW,QAAQ;IACxB,OAAO,EAAE,eAAe,EAAE,CAAA;IAC1B;;OAEG;IACH,MAAM,EAAE,MAAM,EAAE,CAAA;IAChB;;OAEG;IACH,QAAQ,EAAE,MAAM,EAAE,CAAA;IAClB;;OAEG;IACH,cAAc,EAAE,MAAM,EAAE,CAAA;IACxB;;;OAGG;IACH,WAAW,EAAE,MAAM,EAAE,CAAA;IACrB;;;;;OAKG;IACH,gBAAgB,EAAE,MAAM,EAAE,CAAA;CAC1B;AAED;;;GAGG;AACH,MAAM,MAAM,SAAS,GAAG,IAAI,CAAC,QAAQ,EAAE,QAAQ,GAAG,UAAU,GAAG,gBAAgB,GAAG,aAAa,GAAG,kBAAkB,CAAC,CAAA;AAErH;;;GAGG;AACH,eAAO,MAAM,gBAAgB,EAAE,SAM9B,CAAA;AAED;;GAEG;AACH,MAAM,WAAW,SAAS;IACzB,QAAQ,EAAE,SAAS,eAAe,EAAE,CAAA;IACpC,eAAe,EAAE,SAAS,MAAM,EAAE,CAAA;IAClC,iBAAiB,EAAE,SAAS,MAAM,EAAE,CAAA;IACpC,SAAS,EAAE,SAAS,MAAM,EAAE,CAAA;IAC5B,OAAO,EAAE,OAAO,CAAA;CAChB;AAUD;;GAEG;AACH,wBAAgB,gBAAgB,CAAC,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,SAAS,GAAG,SAAS,CAiD1E;AA8BD;;;;;GAKG;AACH,wBAAgB,YAAY,CAAC,MAAM,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,EAAE,KAAK,EAAE,SAAS,GAAG,SAAS,CA2DzF;AAED;;GAEG;AACH,wBAAgB,cAAc,CAAC,CAAC,EAAE,SAAS,EAAE,CAAC,EAAE,SAAS,GAAG,SAAS,CAQpE;AAuBD;;;;GAIG;AACH,wBAAsB,iBAAiB,CAAC,UAAU,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,OAAO,CAAC,GAAG,CAAC,MAAM,EAAE,eAAe,EAAE,CAAC,CAAC,CA2CjH;AAED;;;;;;;GAOG;AACH,MAAM,WAAW,eAAe;IAC/B,UAAU,EAAE,eAAe,EAAE,CAAA;IAC7B,kBAAkB,EAAE,eAAe,EAAE,CAAA;IACrC,UAAU,EAAE,eAAe,EAAE,CAAA;CAC7B;AAQD,wBAAgB,oBAAoB,CACnC,OAAO,EAAE,SAAS,eAAe,EAAE,EACnC,iBAAiB,EAAE,SAAS,MAAM,EAAE,EACpC,SAAS,EAAE,SAAS,MAAM,EAAE,EAC5B,OAAO,EAAE,MAAM,GACb,eAAe,CAkCjB"}