@mailwoman/corpus 10.0.0 → 10.1.0

package/out/tools/sub/venue/harvest.js

@@ -1,168 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   @file The harvest side of the sub-venue lexicon: what one row of a harvestable source looks like,
- *   how a file of them is read back, and what a pass over them extracts — attested surfaces plus the
- *   measured shape of each designator's identifier half.
- *
- *   {@link SubVenueHarvestRow} is SOURCE-NEUTRAL, and verified so: an Overture Places row from the
- *   `airport_terminal` category is `{ designatorID, name }` and fits unchanged. It is declared here
- *   rather than imported so the builder does not take a dependency on `@mailwoman/osm` (which
- *   `@mailwoman/corpus` does not depend on) just to name a shape it reads from a file.
- *
- *   Two measurements shape {@link extractAttestedPhrases}, and both are the reason it is safe to run
- *   over raw OSM at all:
- *
- *   **A feature's `name` is usually the VENUE's name, not a sub-venue phrase.** On the Berlin extract
- *   (2,060 matched features) a `railway=platform` is named `Stendaler Straße` and an `amenity=university`
- *   is named `Hertie School`; across 250,116 named Great Britain features only 6,003 (2.40%) contain a
- *   designator token at all. So names are not harvested wholesale — a name contributes only when it
- *   CONTAINS a phrase already in the surface index, which is what makes `Terminal E (Untere Ebene)`
- *   evidence and `Otto Lilienthal Flughafen Berlin Tegel` not.
- *
- *   **A matched phrase belongs to the record the PHRASE names, not the record the ROW carries.**
- *   Attributing every hit to `row.designatorID` — the rule that matched the FEATURE — produced `west →
- *   platform`, `hall → platform` and `biggin → platform` on the GB extract, because a bus stop tagged
- *   `public_transport=platform` is named "Village Hall" or "West Kensington"; 108 of 133 OSM-derived
- *   surfaces named a different record than the one they pointed at. Attribution therefore runs through
- *   the index, and the row's own designator is kept as `context` — exactly the axis a confound board
- *   needs, since a `hall` seen on a platform is a confound and a `hall` seen on a terminal is evidence.
- */
-import { readLocalTextFile } from "@mailwoman/core/fs/readers";
-import { parseJSONStrict } from "@mailwoman/core/json";
-import { TextSpliterator } from "spliterator";
-import { nameContainsSurfaces } from "#tools/sub/venue/surfaces";
-/**
- * Classify an OSM `ref` into an {@link IdentifierShape} class.
- *
- * The classes are the ones Berlin's gates actually produced, plus the two aviation forms the corpus task names
- * (`Terminal 2F` is digit-letter, `Concourse B` is letter). `range` covers both separators OSM uses for a gate serving
- * more than one stand: `16-18` and `0/1`.
- */
-export function classifyIdentifier(ref) {
-    const value = ref.trim();
-    if (/^[0-9]+$/.test(value))
-        return "digit";
-    if (/^[A-Za-z]$/.test(value))
-        return "letter";
-    if (/^[A-Za-z]+[0-9]+$/.test(value))
-        return "letter-digit";
-    if (/^[0-9]+[A-Za-z]+$/.test(value))
-        return "digit-letter";
-    if (/^[0-9A-Za-z]+\s*[-/]\s*[0-9A-Za-z]+$/.test(value))
-        return "range";
-    return "other";
-}
-/**
- * How many real `ref` values each {@link IdentifierShape} keeps.
- *
- * Eight, not "all" and not one. The field exists so a slice author can see what a class actually CONTAINS — GB's
- * `other` class turned out to be semicolon multi-values (`1;2;3`, `13;14`), which one example would have hidden and
- * which the class name does not say. Eight fits a terminal line and covers the variety inside every class the GB
- * extract produced. The COUNT lives in `observations`; this is a sample, not a census.
- */
-const IDENTIFIER_EXAMPLES_PER_SHAPE = 8;
-/**
- * Harvest attested phrases and identifier shapes out of a source's rows.
- *
- * `index` gates the name harvest — a name contributes only when it CONTAINS a phrase already in the table. That filter
- * is the whole reason this function is safe to run over raw OSM: see this file's header for the Berlin measurement that
- * motivated it. The index also decides ATTRIBUTION: a hit is a surface of the record the PHRASE names, and the row's
- * own designator is recorded as `context`.
- *
- * Returns surfaces with real `observations` counts, so the lexicon can rank `terminal` above a phrase attested once.
- */
-export function extractAttestedPhrases(rows, index, options = {}) {
-    const source = options.source ?? "osm";
-    const region = options.region ?? "";
-    /**
-     * `phrase\0lang\0source` → { count, context }.
-     */
-    const surfaceCounts = new Map();
-    /**
-     * `designatorID\0shape` → { count, examples }.
-     */
-    const shapes = new Map();
-    const note = (phrase, lang, sourceTag, context) => {
-        const key = `${phrase}\0${lang}\0${sourceTag}`;
-        const entry = surfaceCounts.get(key) ?? { count: 0, context: new Map() };
-        entry.count++;
-        entry.context.set(context, (entry.context.get(context) ?? 0) + 1);
-        surfaceCounts.set(key, entry);
-    };
-    for (const row of rows) {
-        if (row.name) {
-            for (const hit of nameContainsSurfaces(row.name, index)) {
-                // `und` — the default `name` tag carries no language. Overture's `name` is the same: a
-                // primary name in whatever language the place uses, untagged.
-                note(hit, "und", `${source}:name`, row.designatorID);
-            }
-        }
-        for (const [lang, localized] of Object.entries(row.localizedNames ?? {})) {
-            for (const hit of nameContainsSurfaces(localized, index)) {
-                note(hit, lang, `${source}:name:${lang}`, row.designatorID);
-            }
-        }
-        if (row.ref) {
-            const shape = classifyIdentifier(row.ref);
-            const key = `${row.designatorID}\0${shape}`;
-            const entry = shapes.get(key) ?? { count: 0, examples: new Set() };
-            entry.count++;
-            if (entry.examples.size < IDENTIFIER_EXAMPLES_PER_SHAPE) {
-                entry.examples.add(row.ref.trim());
-            }
-            shapes.set(key, entry);
-        }
-    }
-    const surfaces = [...surfaceCounts].map(([key, entry]) => {
-        const [phrase, lang, sourceTag] = key.split("\0");
-        const record = index.get(phrase);
-        return {
-            phrase,
-            recordID: record.recordID,
-            recordKind: record.recordKind,
-            lang,
-            region,
-            source: sourceTag,
-            curated: false,
-            observations: entry.count,
-            context: Object.fromEntries([...entry.context].toSorted((a, b) => b[1] - a[1] || a[0].localeCompare(b[0]))),
-        };
-    });
-    const identifierShapes = [...shapes].map(([key, entry]) => {
-        const [designatorID, shape] = key.split("\0");
-        return {
-            designatorID,
-            region,
-            shape,
-            observations: entry.count,
-            examples: [...entry.examples].toSorted((a, b) => a.localeCompare(b)),
-        };
-    });
-    return { surfaces, identifierShapes };
-}
-/**
- * Read a JSONL file of {@link SubVenueHarvestRow}s. Blank lines and unparseable rows are skipped rather than fatal — an
- * extract is a build output, and one malformed line should not cost the whole lexicon.
- */
-export async function readSubVenueJSONL(path) {
-    const out = [];
-    // `TextSpliterator` rather than `split("\n")` — a whole-country extract runs to 250,000 lines
-    // (52 MB for Great Britain), and materializing every segment before reading the first is exactly
-    // what the repo lint rule exists to prevent.
-    for (const line of TextSpliterator.from(await readLocalTextFile(path))) {
-        const trimmed = line.trim();
-        if (!trimmed)
-            continue;
-        try {
-            out.push(parseJSONStrict(trimmed));
-        }
-        catch {
-            continue;
-        }
-    }
-    return out;
-}
-//# sourceMappingURL=harvest.js.map

package/out/tools/sub/venue/harvest.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"harvest.js","sourceRoot":"","sources":["../../../../lib/tools/sub/venue/harvest.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA+BG;AAEH,OAAO,EAAE,iBAAiB,EAAE,MAAM,4BAA4B,CAAA;AAC9D,OAAO,EAAE,eAAe,EAAE,MAAM,sBAAsB,CAAA;AACtD,OAAO,EAAE,eAAe,EAAE,MAAM,aAAa,CAAA;AAE7C,OAAO,EAAE,oBAAoB,EAAqB,MAAM,2BAA2B,CAAA;AA2BnF;;;;;;GAMG;AACH,MAAM,UAAU,kBAAkB,CAAC,GAAW;IAC7C,MAAM,KAAK,GAAG,GAAG,CAAC,IAAI,EAAE,CAAA;IAExB,IAAI,UAAU,CAAC,IAAI,CAAC,KAAK,CAAC;QAAE,OAAO,OAAO,CAAA;IAE1C,IAAI,YAAY,CAAC,IAAI,CAAC,KAAK,CAAC;QAAE,OAAO,QAAQ,CAAA;IAE7C,IAAI,mBAAmB,CAAC,IAAI,CAAC,KAAK,CAAC;QAAE,OAAO,cAAc,CAAA;IAE1D,IAAI,mBAAmB,CAAC,IAAI,CAAC,KAAK,CAAC;QAAE,OAAO,cAAc,CAAA;IAE1D,IAAI,sCAAsC,CAAC,IAAI,CAAC,KAAK,CAAC;QAAE,OAAO,OAAO,CAAA;IAEtE,OAAO,OAAO,CAAA;AACf,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,6BAA6B,GAAG,CAAC,CAAA;AAkBvC;;;;;;;;;GASG;AACH,MAAM,UAAU,sBAAsB,CACrC,IAAkC,EAClC,KAAmB,EACnB,OAAO,GAAmB,EAAE;IAE5B,MAAM,MAAM,GAAG,OAAO,CAAC,MAAM,IAAI,KAAK,CAAA;IACtC,MAAM,MAAM,GAAG,OAAO,CAAC,MAAM,IAAI,EAAE,CAAA;IACnC;;OAEG;IACH,MAAM,aAAa,GAAG,IAAI,GAAG,EAA2D,CAAA;IACxF;;OAEG;IACH,MAAM,MAAM,GAAG,IAAI,GAAG,EAAoD,CAAA;IAE1E,MAAM,IAAI,GAAG,CAAC,MAAc,EAAE,IAAY,EAAE,SAAiB,EAAE,OAAe,EAAQ,EAAE;QACvF,MAAM,GAAG,GAAG,GAAG,MAAM,KAAK,IAAI,KAAK,SAAS,EAAE,CAAA;QAC9C,MAAM,KAAK,GAAG,aAAa,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK,EAAE,CAAC,EAAE,OAAO,EAAE,IAAI,GAAG,EAAkB,EAAE,CAAA;QAExF,KAAK,CAAC,KAAK,EAAE,CAAA;QACb,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,OAAO,EAAE,CAAC,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAA;QACjE,aAAa,CAAC,GAAG,CAAC,GAAG,EAAE,KAAK,CAAC,CAAA;IAC9B,CAAC,CAAA;IAED,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QACxB,IAAI,GAAG,CAAC,IAAI,EAAE,CAAC;YACd,KAAK,MAAM,GAAG,IAAI,oBAAoB,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK,CAAC,EAAE,CAAC;gBACzD,uFAAuF;gBACvF,8DAA8D;gBAC9D,IAAI,CAAC,GAAG,EAAE,KAAK,EAAE,GAAG,MAAM,OAAO,EAAE,GAAG,CAAC,YAAY,CAAC,CAAA;YACrD,CAAC;QACF,CAAC;QAED,KAAK,MAAM,CAAC,IAAI,EAAE,SAAS,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,GAAG,CAAC,cAAc,IAAI,EAAE,CAAC,EAAE,CAAC;YAC1E,KAAK,MAAM,GAAG,IAAI,oBAAoB,CAAC,SAAS,EAAE,KAAK,CAAC,EAAE,CAAC;gBAC1D,IAAI,CAAC,GAAG,EAAE,IAAI,EAAE,GAAG,MAAM,SAAS,IAAI,EAAE,EAAE,GAAG,CAAC,YAAY,CAAC,CAAA;YAC5D,CAAC;QACF,CAAC;QAED,IAAI,GAAG,CAAC,GAAG,EAAE,CAAC;YACb,MAAM,KAAK,GAAG,kBAAkB,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;YACzC,MAAM,GAAG,GAAG,GAAG,GAAG,CAAC,YAAY,KAAK,KAAK,EAAE,CAAA;YAC3C,MAAM,KAAK,GAAG,MAAM,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,EAAE,KAAK,EAAE,CAAC,EAAE,QAAQ,EAAE,IAAI,GAAG,EAAU,EAAE,CAAA;YAE1E,KAAK,CAAC,KAAK,EAAE,CAAA;YAEb,IAAI,KAAK,CAAC,QAAQ,CAAC,IAAI,GAAG,6BAA6B,EAAE,CAAC;gBACzD,KAAK,CAAC,QAAQ,CAAC,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAA;YACnC,CAAC;YAED,MAAM,CAAC,GAAG,CAAC,GAAG,EAAE,KAAK,CAAC,CAAA;QACvB,CAAC;IACF,CAAC;IAED,MAAM,QAAQ,GAAsB,CAAC,GAAG,aAAa,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,EAAE,KAAK,CAAC,EAAE,EAAE;QAC3E,MAAM,CAAC,MAAM,EAAE,IAAI,EAAE,SAAS,CAAC,GAAG,GAAG,CAAC,KAAK,CAAC,IAAI,CAA6B,CAAA;QAC7E,MAAM,MAAM,GAAG,KAAK,CAAC,GAAG,CAAC,MAAM,CAAE,CAAA;QAEjC,OAAO;YACN,MAAM;YACN,QAAQ,EAAE,MAAM,CAAC,QAAQ;YACzB,UAAU,EAAE,MAAM,CAAC,UAAU;YAC7B,IAAI;YACJ,MAAM;YACN,MAAM,EAAE,SAAS;YACjB,OAAO,EAAE,KAAK;YACd,YAAY,EAAE,KAAK,CAAC,KAAK;YACzB,OAAO,EAAE,MAAM,CAAC,WAAW,CAAC,CAAC,GAAG,KAAK,CAAC,OAAO,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;SAC3G,CAAA;IACF,CAAC,CAAC,CAAA;IAEF,MAAM,gBAAgB,GAAsB,CAAC,GAAG,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,GAAG,EAAE,KAAK,CAAC,EAAE,EAAE;QAC5E,MAAM,CAAC,YAAY,EAAE,KAAK,CAAC,GAAG,GAAG,CAAC,KAAK,CAAC,IAAI,CAAqB,CAAA;QAEjE,OAAO;YACN,YAAY;YACZ,MAAM;YACN,KAAK;YACL,YAAY,EAAE,KAAK,CAAC,KAAK;YACzB,QAAQ,EAAE,CAAC,GAAG,KAAK,CAAC,QAAQ,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC;SACpE,CAAA;IACF,CAAC,CAAC,CAAA;IAEF,OAAO,EAAE,QAAQ,EAAE,gBAAgB,EAAE,CAAA;AACtC,CAAC;AAED;;;GAGG;AACH,MAAM,CAAC,KAAK,UAAU,iBAAiB,CAAC,IAAY;IACnD,MAAM,GAAG,GAAyB,EAAE,CAAA;IAEpC,8FAA8F;IAC9F,iGAAiG;IACjG,6CAA6C;IAC7C,KAAK,MAAM,IAAI,IAAI,eAAe,CAAC,IAAI,CAAC,MAAM,iBAAiB,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC;QACxE,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,EAAE,CAAA;QAE3B,IAAI,CAAC,OAAO;YAAE,SAAQ;QAEtB,IAAI,CAAC;YACJ,GAAG,CAAC,IAAI,CAAC,eAAe,CAAqB,OAAO,CAAC,CAAC,CAAA;QACvD,CAAC;QAAC,MAAM,CAAC;YACR,SAAQ;QACT,CAAC;IACF,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC"}

package/out/tools/sub/venue/promotions.d.ts

@@ -1,94 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   The sub-venue curation ledger (#35 wave 2) — every decision taken about whether a designator
- *   surface may be used for parsing IN A GIVEN LOCALE, with the census that backs it.
- *
- *   This file is HAND-AUTHORED. `sub-venue-lexicon.ts` is a pure function of its sources; nothing in it
- *   can decide that Spanish `terminal` is safe and British `hall` is not, because that judgement is
- *   about the confounds a token has in a language, and a confound is a fact about the world rather than
- *   about the data. What the builder does is APPLY these decisions — {@link SUBVENUE_PROMOTIONS} is the
- *   only thing that ever sets `curated: true` on a machine-derived surface.
- *
- *   ── Why every decision is per-LOCALE ─────────────────────────────────────────────────────────────
- *   Wave 1 measured `hall` at 3,274 hits in the Great Britain extract and stopped there, which reads as
- *   a verdict on the word. It is not. Re-censused per region 2026-08-05, the same token is a British
- *   disaster and a German designator, and no single global flag can express that. The rule this file
- *   enforces: a promotion names a designator, a phrase, AND a locale, and it is earned by a count taken
- *   in that locale's own data.
- *
- *   ── What "real" and "confound" mean in the numbers below ─────────────────────────────────────────
- *   Each census counts the surface's occurrences in one region's extract and splits them two ways:
- *
- *   - **real** — the phrase sits in a name that is genuinely venue-interior structure, judged by the
- *     rule designator of the feature carrying it (`terminal`, `gate`, `campus`) and by the name taking a
- *     `<designator> <identifier>` or `<modifier> <designator>` shape.
- *   - **confound** — everything else, and the `confoundNote` says WHAT it is. A count with no note is
- *     not a census; "3,274 hits" told nobody that 3,205 of them were bus stops named after a village
- *     hall.
- *
- *   The instrument is `context` on {@link SubVenueSurface} plus the name-shape split; both are
- *   reproducible from the committed extracts, and `PROVENANCE.md` records the commands.
- *
- *   ── A rejection is a deliverable ─────────────────────────────────────────────────────────────────
- *   Rejections stay in this table forever. They are what stops the next contributor re-proposing
- *   `hall` for en-GB from the same Wikidata concept that proposed it the first time.
- */
-/**
- * One curation decision.
- */
-export interface SubVenuePromotion {
-    /**
-     * The {@link SubVenueDesignator.id} the phrase is a surface of.
-     */
-    designatorID: string;
-    /**
-     * The exact normalized surface being decided on — `halle`, not `hall`, when the German form is at issue.
-     */
-    phrase: string;
-    /**
-     * BCP-47-ish `<lang>-<REGION>`. The region half is matched against a surface's `region` and the language half against
-     * its `lang`, so `de-DE` reaches both a German Wikidata label (region-free) and a German extract's untagged name.
-     */
-    locale: string;
-    decision: "promote" | "reject";
-    /**
-     * Set when the board's verdict is valid ONLY under a syntactic shape — the machine-readable half of a shape-separable
-     * confound. `identifier-required` means the phrase is promoted solely in `<phrase> <identifier>` position (Halle 8);
-     * bare occurrences stay unpromoted. A consumer that reads promotions MUST honour this field: the de-DE `halle` board
-     * (2026-08-05 five-whys review) is the founding case — its 168-hit confound includes the CITY Halle (Saale), and the
-     * 32/32-real enumeration that justified promotion was an enumeration of the identifier-bearing shape, not of the
-     * phrase. Absent = the promotion holds in any shape.
-     */
-    shape?: "identifier-required";
-    /**
-     * Occurrences judged genuine venue-interior structure in that locale's census.
-     */
-    real: number;
-    /**
-     * Occurrences judged confound.
-     */
-    confound: number;
-    /**
-     * What the confound half IS. Required prose — a bare number is not a board.
-     */
-    confoundNote: string;
-    /**
-     * Which extract or dataset the census was taken over, and when.
-     */
-    census: string;
-}
-/**
- * The ledger. Sorted by designator, then locale, then phrase in the emitted artifact; source order here is grouped by
- * designator for reading.
- *
- * ── A rejection of a SHIPPED designator is ADVISORY ──────────────────────────────────────────────
- * `neural/venue-structure.ts` ships `wing`, `terminal`, `concourse` and six more as a FLAT English vocabulary with no
- * locale gate. Nothing in this table can un-ship them: the `wing` / en-US rejection below tells a slice author which
- * locale to exclude from a generated line, and does not stop the span proposer from firing on "Red Wing". Giving the
- * shipped vocabulary a per-locale gate is the largest thing the slice will want that does not exist yet.
- */
-export declare const SUBVENUE_PROMOTIONS: readonly SubVenuePromotion[];
-//# sourceMappingURL=promotions.d.ts.map

package/out/tools/sub/venue/promotions.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"promotions.d.ts","sourceRoot":"","sources":["../../../../lib/tools/sub/venue/promotions.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAqCG;AAEH;;GAEG;AACH,MAAM,WAAW,iBAAiB;IACjC;;OAEG;IACH,YAAY,EAAE,MAAM,CAAA;IACpB;;OAEG;IACH,MAAM,EAAE,MAAM,CAAA;IACd;;;OAGG;IACH,MAAM,EAAE,MAAM,CAAA;IACd,QAAQ,EAAE,SAAS,GAAG,QAAQ,CAAA;IAC9B;;;;;;;OAOG;IACH,KAAK,CAAC,EAAE,qBAAqB,CAAA;IAC7B;;OAEG;IACH,IAAI,EAAE,MAAM,CAAA;IACZ;;OAEG;IACH,QAAQ,EAAE,MAAM,CAAA;IAChB;;OAEG;IACH,YAAY,EAAE,MAAM,CAAA;IACpB;;OAEG;IACH,MAAM,EAAE,MAAM,CAAA;CACd;AAED;;;;;;;;;GASG;AACH,eAAO,MAAM,mBAAmB,EAAE,SAAS,iBAAiB,EA2O3D,CAAA"}

package/out/tools/sub/venue/promotions.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"promotions.js","sourceRoot":"","sources":["../../../../lib/tools/sub/venue/promotions.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAqCG;AA+CH;;;;;;;;;GASG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAAiC;IAChE,mGAAmG;IACnG,6FAA6F;IAC7F,4EAA4E;IAC5E;QACC,YAAY,EAAE,MAAM;QACpB,MAAM,EAAE,MAAM;QACd,MAAM,EAAE,OAAO;QACf,QAAQ,EAAE,SAAS;QACnB,IAAI,EAAE,EAAE;QACR,QAAQ,EAAE,CAAC;QACX,YAAY,EACX,0GAA0G;YAC1G,4GAA4G;YAC5G,2GAA2G;YAC3G,sGAAsG;YACtG,4GAA4G;YAC5G,6GAA6G;YAC7G,uBAAuB;QACxB,MAAM,EAAE,wEAAwE;KAChF;IACD;QACC,YAAY,EAAE,MAAM;QACpB,MAAM,EAAE,MAAM;QACd,MAAM,EAAE,OAAO;QACf,QAAQ,EAAE,QAAQ;QAClB,IAAI,EAAE,CAAC;QACP,QAAQ,EAAE,IAAI;QACd,YAAY,EACX,kFAAkF;YAClF,8GAA8G;YAC9G,0FAA0F;QAC3F,MAAM,EAAE,0DAA0D;KAClE;IACD;QACC,YAAY,EAAE,MAAM;QACpB,MAAM,EAAE,MAAM;QACd,MAAM,EAAE,OAAO;QACf,QAAQ,EAAE,QAAQ;QAClB,IAAI,EAAE,CAAC;QACP,QAAQ,EAAE,EAAE;QACZ,YAAY,EACX,sGAAsG;YACtG,gBAAgB;QACjB,MAAM,EAAE,wEAAwE;KAChF;IAED,mGAAmG;IACnG,8FAA8F;IAC9F,sEAAsE;IACtE;QACC,YAAY,EAAE,MAAM;QACpB,MAAM,EAAE,MAAM;QACd,MAAM,EAAE,OAAO;QACf,QAAQ,EAAE,QAAQ;QAClB,IAAI,EAAE,CAAC;QACP,QAAQ,EAAE,IAAI;QACd,YAAY,EACX,6GAA6G;YAC7G,8GAA8G;YAC9G,6GAA6G;YAC7G,4GAA4G;YAC5G,aAAa;QACd,MAAM,EAAE,iDAAiD;KACzD;IACD;QACC,YAAY,EAAE,MAAM;QACpB,MAAM,EAAE,MAAM;QACd,MAAM,EAAE,OAAO;QACf,QAAQ,EAAE,QAAQ;QAClB,IAAI,EAAE,IAAI;QACV,QAAQ,EAAE,MAAM;QAChB,YAAY,EACX,yGAAyG;YACzG,2GAA2G;YAC3G,+GAA+G;YAC/G,UAAU;QACX,MAAM,EAAE,0DAA0D;KAClE;IACD;QACC,YAAY,EAAE,MAAM;QACpB,MAAM,EAAE,OAAO;QACf,MAAM,EAAE,OAAO;QACf,QAAQ,EAAE,SAAS;QACnB,KAAK,EAAE,qBAAqB;QAC5B,IAAI,EAAE,EAAE;QACR,QAAQ,EAAE,GAAG;QACb,YAAY,EACX,6GAA6G;YAC7G,qGAAqG;YACrG,yGAAyG;YACzG,0GAA0G;YAC1G,0GAA0G;YAC1G,6GAA6G;YAC7G,gFAAgF;QACjF,MAAM,EAAE,iFAAiF;KACzF;IACD;QACC,YAAY,EAAE,MAAM;QACpB,MAAM,EAAE,MAAM;QACd,MAAM,EAAE,OAAO;QACf,QAAQ,EAAE,SAAS;QACnB,IAAI,EAAE,EAAE;QACR,QAAQ,EAAE,CAAC;QACX,YAAY,EACX,2GAA2G;YAC3G,4GAA4G;YAC5G,2GAA2G;YAC3G,0FAA0F;QAC3F,MAAM,EAAE,+EAA+E;KACvF;IAED,mGAAmG;IACnG,kGAAkG;IAClG,wEAAwE;IACxE;QACC,YAAY,EAAE,MAAM;QACpB,MAAM,EAAE,WAAW;QACnB,MAAM,EAAE,OAAO;QACf,QAAQ,EAAE,SAAS;QACnB,IAAI,EAAE,EAAE;QACR,QAAQ,EAAE,CAAC;QACX,YAAY,EACX,yGAAyG;YACzG,2GAA2G;YAC3G,uGAAuG;YACvG,oGAAoG;YACpG,yGAAyG;QAC1G,MAAM,EAAE,4EAA4E;KACpF;IACD;QACC,YAAY,EAAE,MAAM;QACpB,MAAM,EAAE,OAAO;QACf,MAAM,EAAE,OAAO;QACf,QAAQ,EAAE,QAAQ;QAClB,IAAI,EAAE,EAAE;QACR,QAAQ,EAAE,GAAG;QACb,YAAY,EACX,yGAAyG;YACzG,sGAAsG;YACtG,2GAA2G;YAC3G,0GAA0G;YAC1G,2EAA2E;QAC5E,MAAM,EAAE,gFAAgF;KACxF;IAED,mGAAmG;IACnG,kGAAkG;IAClG,yEAAyE;IACzE;QACC,YAAY,EAAE,MAAM;QACpB,MAAM,EAAE,MAAM;QACd,MAAM,EAAE,OAAO;QACf,QAAQ,EAAE,SAAS;QACnB,IAAI,EAAE,GAAG;QACT,QAAQ,EAAE,EAAE;QACZ,YAAY,EACX,4GAA4G;YAC5G,8GAA8G;YAC9G,8GAA8G;YAC9G,0GAA0G;QAC3G,MAAM,EAAE,+CAA+C;KACvD;IACD;QACC,YAAY,EAAE,MAAM;QACpB,MAAM,EAAE,MAAM;QACd,MAAM,EAAE,OAAO;QACf,QAAQ,EAAE,QAAQ;QAClB,IAAI,EAAE,GAAG;QACT,QAAQ,EAAE,IAAI;QACd,YAAY,EACX,+GAA+G;YAC/G,8GAA8G;YAC9G,wCAAwC;QACzC,MAAM,EAAE,0DAA0D;KAClE;IAED,mGAAmG;IACnG,oGAAoG;IACpG,gDAAgD;IAChD;QACC,YAAY,EAAE,UAAU;QACxB,MAAM,EAAE,UAAU;QAClB,MAAM,EAAE,OAAO;QACf,QAAQ,EAAE,SAAS;QACnB,IAAI,EAAE,GAAG;QACT,QAAQ,EAAE,CAAC;QACX,YAAY,EACX,4GAA4G;YAC5G,0GAA0G;YAC1G,+GAA+G;YAC/G,sCAAsC;QACvC,MAAM,EAAE,4DAA4D;KACpE;IACD;QACC,YAAY,EAAE,UAAU;QACxB,MAAM,EAAE,UAAU;QAClB,MAAM,EAAE,OAAO;QACf,QAAQ,EAAE,SAAS;QACnB,IAAI,EAAE,EAAE;QACR,QAAQ,EAAE,CAAC;QACX,YAAY,EACX,4GAA4G;YAC5G,4FAA4F;QAC7F,MAAM,EAAE,yCAAyC;KACjD;IACD;QACC,YAAY,EAAE,UAAU;QACxB,MAAM,EAAE,UAAU;QAClB,MAAM,EAAE,OAAO;QACf,QAAQ,EAAE,SAAS;QACnB,IAAI,EAAE,GAAG;QACT,QAAQ,EAAE,CAAC;QACX,YAAY,EACX,uGAAuG;YACvG,wGAAwG;YACxG,2FAA2F;QAC5F,MAAM,EAAE,4EAA4E;KACpF;IACD;QACC,YAAY,EAAE,UAAU;QACxB,MAAM,EAAE,OAAO;QACf,MAAM,EAAE,OAAO;QACf,QAAQ,EAAE,SAAS;QACnB,IAAI,EAAE,IAAI;QACV,QAAQ,EAAE,CAAC;QACX,YAAY,EACX,qGAAqG;YACrG,iGAAiG;YACjG,uFAAuF;YACvF,mFAAmF;YACnF,4GAA4G;YAC5G,2CAA2C;QAC5C,MAAM,EAAE,8DAA8D;KACtE;CACD,CAAA"}

package/out/tools/sub/venue/surfaces.d.ts

@@ -1,45 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   @file Surface-form primitives for the sub-venue lexicon: the normalizer every phrase passes
- *   through, the phrase → record index built over the surfaces accumulated so far, and the operator
- *   that decides whether a feature's name contains a known designator.
- *
- *   The two matching rules here are script-conditional, and both narrowings are required — see
- *   {@link NON_SPACING_SCRIPT} and {@link nameContainsSurfaces}.
- */
-import type { SubVenueSurface } from "#tools/sub/venue/table";
-/**
- * Normalize a surface for the table: trim, collapse internal whitespace, and lowercase ONLY when the string is entirely
- * in a bicameral script. `ターミナルビル` and `航站楼` pass through untouched; `Flughafenterminal` folds.
- */
-export declare function normalizeSurface(text: string): string;
-/**
- * A phrase → record index, keyed on the normalized phrase. Built by {@link buildSurfaceIndex} from the surfaces present
- * before the harvest runs, and the reason a matched phrase can be attributed to the record it actually names.
- */
-export type SurfaceIndex = ReadonlyMap<string, {
-    recordID: string;
-    recordKind: "designator" | "modifier";
-}>;
-/**
- * Index the surfaces accumulated so far by phrase. First writer wins, so a seed record beats a Wikidata alias that
- * happens to collide — `terminal` stays the `terminal` designator even though it is also an Italian alias for it.
- */
-export declare function buildSurfaceIndex(surfaces: readonly SubVenueSurface[]): SurfaceIndex;
-/**
- * Every known phrase found in `name`, as whole-token runs for spacing scripts and as substrings for non-spacing ones.
- *
- * Token-boundary matching for Latin script, not substring: `Nordterminal` is a real German compound in which `terminal`
- * is a suffix, and a substring test would also fire on `Terminalstraße`. The compound case is a genuine miss and it is
- * the right miss — admitting suffix matches would fire on every `-hall`/`-gate` compound in Germanic and Nordic street
- * naming, which is exactly the confound class `Briggate`/`Kirkgate` represents.
- *
- * For Han/Kana names that rule finds nothing at all, because the script has no word boundaries: `第1ターミナル` splits into
- * one token that matches no surface. There the LONGEST known substring is the correct operator, and the compound
- * objection does not transfer — Japanese has no `-gate` street-name suffix class.
- */
-export declare function nameContainsSurfaces(name: string, index: SurfaceIndex): string[];
-//# sourceMappingURL=surfaces.d.ts.map

package/out/tools/sub/venue/surfaces.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"surfaces.d.ts","sourceRoot":"","sources":["../../../../lib/tools/sub/venue/surfaces.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAEH,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,wBAAwB,CAAA;AAiB7D;;;GAGG;AACH,wBAAgB,gBAAgB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAIrD;AAED;;;GAGG;AACH,MAAM,MAAM,YAAY,GAAG,WAAW,CAAC,MAAM,EAAE;IAAE,QAAQ,EAAE,MAAM,CAAC;IAAC,UAAU,EAAE,YAAY,GAAG,UAAU,CAAA;CAAE,CAAC,CAAA;AAE3G;;;GAGG;AACH,wBAAgB,iBAAiB,CAAC,QAAQ,EAAE,SAAS,eAAe,EAAE,GAAG,YAAY,CASpF;AAED;;;;;;;;;;;GAWG;AACH,wBAAgB,oBAAoB,CAAC,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,YAAY,GAAG,MAAM,EAAE,CAqBhF"}

package/out/tools/sub/venue/surfaces.js

@@ -1,77 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   @file Surface-form primitives for the sub-venue lexicon: the normalizer every phrase passes
- *   through, the phrase → record index built over the surfaces accumulated so far, and the operator
- *   that decides whether a feature's name contains a known designator.
- *
- *   The two matching rules here are script-conditional, and both narrowings are required — see
- *   {@link NON_SPACING_SCRIPT} and {@link nameContainsSurfaces}.
- */
-/**
- * Scripts whose case is meaningful to fold. Everything else is left as written — see {@link SubVenueSurface.phrase}.
- */
-const CASE_FOLDING_SCRIPT = /^[\p{Script=Latin}\p{Script=Cyrillic}\p{Script=Greek}\d\s\p{P}]+$/u;
-/**
- * Scripts written without spaces between words, where a token split cannot find a designator and a SUBSTRING test is
- * the correct operator. Han, Hiragana, Katakana; Hangul is excluded because Korean does space its words.
- *
- * The Germanic-compound argument that keeps {@link nameContainsSurfaces} token-bounded for Latin script does not
- * transfer here — there is no `-gate`/`-hall` street-name suffix class in Japanese, and `第1ターミナル` is unreachable by any
- * token split. Measured on the Japan extract: see the harvest counts in `corpus/data/PROVENANCE.md`.
- */
-const NON_SPACING_SCRIPT = /[\p{Script=Han}\p{Script=Hiragana}\p{Script=Katakana}]/u;
-/**
- * Normalize a surface for the table: trim, collapse internal whitespace, and lowercase ONLY when the string is entirely
- * in a bicameral script. `ターミナルビル` and `航站楼` pass through untouched; `Flughafenterminal` folds.
- */
-export function normalizeSurface(text) {
-    const trimmed = text.trim().replaceAll(/\s+/gu, " ");
-    return CASE_FOLDING_SCRIPT.test(trimmed) ? trimmed.toLowerCase() : trimmed;
-}
-/**
- * Index the surfaces accumulated so far by phrase. First writer wins, so a seed record beats a Wikidata alias that
- * happens to collide — `terminal` stays the `terminal` designator even though it is also an Italian alias for it.
- */
-export function buildSurfaceIndex(surfaces) {
-    const index = new Map();
-    for (const surface of surfaces) {
-        if (index.has(surface.phrase))
-            continue;
-        index.set(surface.phrase, { recordID: surface.recordID, recordKind: surface.recordKind });
-    }
-    return index;
-}
-/**
- * Every known phrase found in `name`, as whole-token runs for spacing scripts and as substrings for non-spacing ones.
- *
- * Token-boundary matching for Latin script, not substring: `Nordterminal` is a real German compound in which `terminal`
- * is a suffix, and a substring test would also fire on `Terminalstraße`. The compound case is a genuine miss and it is
- * the right miss — admitting suffix matches would fire on every `-hall`/`-gate` compound in Germanic and Nordic street
- * naming, which is exactly the confound class `Briggate`/`Kirkgate` represents.
- *
- * For Han/Kana names that rule finds nothing at all, because the script has no word boundaries: `第1ターミナル` splits into
- * one token that matches no surface. There the LONGEST known substring is the correct operator, and the compound
- * objection does not transfer — Japanese has no `-gate` street-name suffix class.
- */
-export function nameContainsSurfaces(name, index) {
-    const normalized = normalizeSurface(name);
-    const hits = new Set();
-    for (const token of normalized.split(/[\s,()/]+/u)) {
-        const stripped = token.replaceAll(/^[^\p{L}\p{N}]+|[^\p{L}\p{N}]+$/gu, "");
-        if (stripped && index.has(stripped)) {
-            hits.add(stripped);
-        }
-    }
-    if (NON_SPACING_SCRIPT.test(normalized)) {
-        for (const [phrase] of index) {
-            if (NON_SPACING_SCRIPT.test(phrase) && normalized.includes(phrase)) {
-                hits.add(phrase);
-            }
-        }
-    }
-    return [...hits];
-}
-//# sourceMappingURL=surfaces.js.map

package/out/tools/sub/venue/surfaces.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"surfaces.js","sourceRoot":"","sources":["../../../../lib/tools/sub/venue/surfaces.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAIH;;GAEG;AACH,MAAM,mBAAmB,GAAG,oEAAoE,CAAA;AAEhG;;;;;;;GAOG;AACH,MAAM,kBAAkB,GAAG,yDAAyD,CAAA;AAEpF;;;GAGG;AACH,MAAM,UAAU,gBAAgB,CAAC,IAAY;IAC5C,MAAM,OAAO,GAAG,IAAI,CAAC,IAAI,EAAE,CAAC,UAAU,CAAC,OAAO,EAAE,GAAG,CAAC,CAAA;IAEpD,OAAO,mBAAmB,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC,OAAO,CAAA;AAC3E,CAAC;AAQD;;;GAGG;AACH,MAAM,UAAU,iBAAiB,CAAC,QAAoC;IACrE,MAAM,KAAK,GAAG,IAAI,GAAG,EAAuE,CAAA;IAE5F,KAAK,MAAM,OAAO,IAAI,QAAQ,EAAE,CAAC;QAChC,IAAI,KAAK,CAAC,GAAG,CAAC,OAAO,CAAC,MAAM,CAAC;YAAE,SAAQ;QACvC,KAAK,CAAC,GAAG,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,QAAQ,EAAE,OAAO,CAAC,QAAQ,EAAE,UAAU,EAAE,OAAO,CAAC,UAAU,EAAE,CAAC,CAAA;IAC1F,CAAC;IAED,OAAO,KAAK,CAAA;AACb,CAAC;AAED;;;;;;;;;;;GAWG;AACH,MAAM,UAAU,oBAAoB,CAAC,IAAY,EAAE,KAAmB;IACrE,MAAM,UAAU,GAAG,gBAAgB,CAAC,IAAI,CAAC,CAAA;IACzC,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAA;IAE9B,KAAK,MAAM,KAAK,IAAI,UAAU,CAAC,KAAK,CAAC,YAAY,CAAC,EAAE,CAAC;QACpD,MAAM,QAAQ,GAAG,KAAK,CAAC,UAAU,CAAC,mCAAmC,EAAE,EAAE,CAAC,CAAA;QAE1E,IAAI,QAAQ,IAAI,KAAK,CAAC,GAAG,CAAC,QAAQ,CAAC,EAAE,CAAC;YACrC,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAA;QACnB,CAAC;IACF,CAAC;IAED,IAAI,kBAAkB,CAAC,IAAI,CAAC,UAAU,CAAC,EAAE,CAAC;QACzC,KAAK,MAAM,CAAC,MAAM,CAAC,IAAI,KAAK,EAAE,CAAC;YAC9B,IAAI,kBAAkB,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,UAAU,CAAC,QAAQ,CAAC,MAAM,CAAC,EAAE,CAAC;gBACpE,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,CAAA;YACjB,CAAC;QACF,CAAC;IACF,CAAC;IAED,OAAO,CAAC,GAAG,IAAI,CAAC,CAAA;AACjB,CAAC"}

package/out/tools/sub/venue/table.d.ts

@@ -1,229 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   @file The sub-venue lexicon's record schema and the shipped vocabulary it is seeded from — the
- *   emitted artifact's shape plus the designator, modifier and Wikidata-concept tables every build
- *   starts out holding.
- *
- *   The shape follows `@mailwoman/poi-taxonomy`'s `taxonomy.json` idiom exactly — typed records plus a
- *   FLAT phrase array keyed back to a record id, which is what makes a longest-match phrase index cheap
- *   to build over it. {@link SubVenueSurface} is this table's `SynonymEntry`.
- *
- *   The seeds DUPLICATE `neural/venue-structure.ts` knowingly: `@mailwoman/corpus` does not depend on
- *   `@mailwoman/neural` (the dependency runs the other way for the training path, and pulling
- *   onnxruntime into a corpus build to read three string arrays would be absurd), so the shipped
- *   vocabulary is re-declared here. That is a drift surface and it is stated rather than hidden —
- *   `sub-venue-lexicon.test.ts` pins the seed's contents literally, so a change in either place fails a
- *   test rather than passing silently.
- */
-import type { SubVenuePromotion } from "#tools/sub/venue/promotions";
-/**
- * This table's own data version. Bump when the source vintages or the build semantics change.
- *
- * `0.2.0` — wave 2: per-region attestation, the phrase-attribution fix (finding 3 above), derived head nouns, the
- * Overture source, and the `promotions[]` receipts section.
- */
-export declare const SUBVENUE_LEXICON_VERSION = "0.2.0";
-/**
- * Which side of the containment relation a designator names. Mirrors `@mailwoman/osm/sdk`'s `SubVenueTier`; re-declared
- * for the same dependency-direction reason as the seed.
- */
-export declare const LexiconTier: {
-    readonly SubVenue: "subvenue";
-    readonly Venue: "venue";
-};
-export type LexiconTier = (typeof LexiconTier)[keyof typeof LexiconTier];
-/**
- * One designator record — a venue-interior (or containing-venue) structural noun.
- */
-export interface SubVenueDesignator {
-    /**
-     * Canonical id, lowercase English. Matches `neural/venue-structure.ts`'s `VENUE_STRUCTURE_DESIGNATORS` wherever the
-     * two overlap.
-     */
-    id: string;
-    tier: LexiconTier;
-    /**
-     * Whether this designator may be preceded by a {@link SubVenueModifier} — the `North Terminal` shape.
-     *
-     * A SUBSET, and the exclusions are required: `gate` and `building` form ordinary STREET names in exactly this shape
-     * ("East Gate" is a real GB street, "Building Society Place" is a real street), so admitting them turns a correct
-     * street parse into a sub-venue one. Setting this true means claiming no street is named `<modifier> <id>`. Check
-     * before you do.
-     */
-    modifierEligible: boolean;
-    /**
-     * Whether the shipped span proposer already recognizes this designator. `false` means the lexicon proposes it and
-     * nothing consumes it yet.
-     */
-    shipped: boolean;
-    /**
-     * Where the term comes from, one entry per attesting source: `wof:placetype`, `osm:aeroway=terminal`,
-     * `wikidata:Q849706`, `overture:airport_terminal`. Sorted, so a regenerate is stable.
-     */
-    provenance: string[];
-}
-/**
- * One positional modifier — the `North`/`Upper`/`Main` half of `North Terminal`.
- */
-export interface SubVenueModifier {
-    id: string;
-    shipped: boolean;
-    provenance: string[];
-}
-/**
- * One surface form: a phrase, the record it names, and where it was attested.
- *
- * This is the table's `SynonymEntry` — the flat array a phrase index is built over.
- */
-export interface SubVenueSurface {
-    /**
-     * The phrase, lowercased for Latin-script languages and left as written otherwise — case-folding is meaningless for
-     * Han and Kana, which the script guard excludes.
-     *
-     * It does NOT exclude Turkish: `İ` (U+0130) is `\p{Script=Latin}`, so the guard admits it and `toLowerCase` folds it
-     * to `i` plus a combining dot above. A Turkish surface therefore round-trips through a form its own locale would not
-     * write. Nothing shipped depends on that today; a Turkish designator would.
-     */
-    phrase: string;
-    /**
-     * The {@link SubVenueDesignator.id} or {@link SubVenueModifier.id} this phrase is a surface of.
-     */
-    recordID: string;
-    /**
-     * Which record table `recordID` points into.
-     */
-    recordKind: "designator" | "modifier";
-    /**
-     * BCP-47-ish language subtag as the source wrote it (`en`, `ja`, `zh-Hant`, `pt-BR`), or `und` when the source gave
-     * an untagged default name.
-     */
-    lang: string;
-    /**
-     * ISO 3166-1 alpha-2 of the DATA the phrase was attested in, `""` for vocabulary sources that attest a term's
-     * existence rather than its use anywhere. This is the axis promotion is decided on: `hall` is attested 3,274 times in
-     * `GB` and every promotion of it lives or dies on a per-region census, never a global one.
-     */
-    region: string;
-    /**
-     * `wikidata:label`, `wikidata:alt`, `osm:name`, `osm:name:<lang>`, `overture:name`, `derived:head-noun`, or `seed`.
-     */
-    source: string;
-    /**
-     * Whether a human has approved this surface for parsing use IN ITS REGION. Everything machine-derived starts `false`
-     * and is flipped only by a matching {@link SubVenuePromotion}. A consumer that gates a parse MUST filter on this — see
-     * `sub-venue-lexicon.ts`'s module docstring for what a promotion decides and why it is per-locale.
-     */
-    curated: boolean;
-    /**
-     * How many source features attested this exact phrase, when the source counts (OSM, Overture). `0` for vocabulary
-     * sources, which attest a term's EXISTENCE rather than its frequency.
-     */
-    observations: number;
-    /**
-     * The rule-assigned designator of the FEATURES that carried this phrase, with a count each — `platform:3205
-     * campus:49` for GB's `hall`. Empty for vocabulary sources.
-     *
-     * This is the confound axis. A `hall` on a `platform` row is a British bus stop named after a village hall; a `hall`
-     * on a `terminal` row is a real German departure hall. Without it, a surface's `observations` count is a magnitude
-     * with no sign — see the repo's "meaning of zero" rule, which applies just as hard to a large number.
-     */
-    context: Record<string, number>;
-}
-/**
- * The measured shape of a designator's identifier half — what follows `Gate`/`Terminal` in real data.
- *
- * Derived from OSM `ref` values, NOT from names, and that is why the artifact has a section for it at all. Every one of
- * Berlin's 26 `aeroway=gate` features is unnamed and carries only a `ref`: `13`, `6`, `0/1`, `14/15`, `16-18`. So `Gate
- * A12` is a RENDERING (`<designator> <ref>`) rather than a string anyone has written down, and a slice that wants to
- * generate the designator+identifier form needs the identifier DISTRIBUTION, not a list of phrases.
- */
-export interface IdentifierShape {
-    designatorID: string;
-    /**
-     * ISO 3166-1 alpha-2 of the extract this distribution was measured in. Per-region because the shapes differ: GB gates
-     * are 70% bare digits, Japanese platform refs are overwhelmingly bare digits with a different range, and a slice that
-     * generates `Gate <ref>` for a French address should sample France's distribution.
-     */
-    region: string;
-    /**
-     * A coarse class: `digit` (`5`), `letter` (`B`), `letter-digit` (`A12`), `digit-letter` (`2F`), `range` (`16-18`,
-     * `0/1`), or `other`.
-     */
-    shape: string;
-    observations: number;
-    /**
-     * Up to eight real values, sorted, so a slice author can see what the class actually contains.
-     */
-    examples: string[];
-}
-/**
- * One input source's provenance, copied off its fetch manifest.
- */
-export interface SubVenueLexiconSource {
-    id: string;
-    origin: string;
-    license: string;
-    retrieved: string;
-    rows: number;
-}
-/**
- * The committed table.
- */
-export interface SubVenueLexiconTable {
-    version: string;
-    sources: SubVenueLexiconSource[];
-    designators: SubVenueDesignator[];
-    modifiers: SubVenueModifier[];
-    surfaces: SubVenueSurface[];
-    identifierShapes: IdentifierShape[];
-    /**
-     * Every curation decision taken against this table, promotion AND rejection, each with the census that backs it. A
-     * rejection is as required as a promotion: it is what stops the next reader re-proposing `hall` for en-GB.
-     */
-    promotions: SubVenuePromotion[];
-}
-/**
- * The vocabulary that already ships in `neural/venue-structure.ts`, re-declared. See the module docstring for why this
- * duplication exists.
- *
- * `tier` is added here (the shipped list has no such field): the seven WOF placetypes plus `terminal`/`gate` are all
- * venue-INTERIOR, except `campus` and `building`, which name a whole venue as often as a part of one. They are marked
- * `subvenue` anyway, because that is the role the span proposer uses them in — `Building 43, Googleplex` is a unit
- * inside a venue.
- */
-export declare const SHIPPED_DESIGNATOR_SEED: ReadonlyArray<{
-    id: string;
-    modifierEligible: boolean;
-    provenance: string[];
-}>;
-/**
- * The shipped positional modifiers, re-declared from `neural/venue-structure.ts`'s `VENUE_STRUCTURE_MODIFIERS`.
- */
-export declare const SHIPPED_MODIFIER_SEED: readonly string[];
-/**
- * Designators the lexicon ADDS beyond what ships, each with the source that attests it.
- *
- * `platform`, `station` and `airport` come from the OSM extractor's rule table and are the rail/aviation venue-side
- * vocabulary the corpus line needs. `hall` and `satellite` come from Wikidata concepts and from
- * `wof-osm-placetype-map.mdx`'s own "plausible additions" note, which lists `hall` explicitly. `pier` joins them in
- * wave 2 on 282 Overture attestations in the `pier` category plus 162 in the GB extract — the corpus task names `Pier
- * C` as a target shape, so the record has to exist before a slice can generate it.
- *
- * None is `modifierEligible`: that claim needs a confound board per term AND per locale, and `sub-venue-promotions.ts`
- * is where those live. A promotion marks a SURFACE usable; it does not widen the modifier grammar.
- */
-export declare const PROPOSED_DESIGNATORS: ReadonlyArray<{
-    id: string;
-    tier: LexiconTier;
-    provenance: string[];
-}>;
-/**
- * `designatorID` → Wikidata QID, mirroring `fetch/wikidata-subvenue.ts`'s `SUBVENUE_CONCEPTS`. Re-declared here so the
- * builder stays a pure function over PARSED input rather than reaching into a fetch module for a constant; the test
- * pins the two against each other.
- */
-export declare const CONCEPT_QIDS: Readonly<Record<string, string>>;
-//# sourceMappingURL=table.d.ts.map

package/out/tools/sub/venue/table.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"table.d.ts","sourceRoot":"","sources":["../../../../lib/tools/sub/venue/table.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;GAmBG;AAEH,OAAO,KAAK,EAAE,iBAAiB,EAAE,MAAM,6BAA6B,CAAA;AAEpE;;;;;GAKG;AACH,eAAO,MAAM,wBAAwB,UAAU,CAAA;AAE/C;;;GAGG;AACH,eAAO,MAAM,WAAW;aACvB,QAAQ,EAAE,UAAU;aACpB,KAAK,EAAE,OAAO;CACL,CAAA;AAEV,MAAM,MAAM,WAAW,GAAG,CAAC,OAAO,WAAW,CAAC,CAAC,MAAM,OAAO,WAAW,CAAC,CAAA;AAExE;;GAEG;AACH,MAAM,WAAW,kBAAkB;IAClC;;;OAGG;IACH,EAAE,EAAE,MAAM,CAAA;IACV,IAAI,EAAE,WAAW,CAAA;IACjB;;;;;;;OAOG;IACH,gBAAgB,EAAE,OAAO,CAAA;IACzB;;;OAGG;IACH,OAAO,EAAE,OAAO,CAAA;IAChB;;;OAGG;IACH,UAAU,EAAE,MAAM,EAAE,CAAA;CACpB;AAED;;GAEG;AACH,MAAM,WAAW,gBAAgB;IAChC,EAAE,EAAE,MAAM,CAAA;IACV,OAAO,EAAE,OAAO,CAAA;IAChB,UAAU,EAAE,MAAM,EAAE,CAAA;CACpB;AAED;;;;GAIG;AACH,MAAM,WAAW,eAAe;IAC/B;;;;;;;OAOG;IACH,MAAM,EAAE,MAAM,CAAA;IACd;;OAEG;IACH,QAAQ,EAAE,MAAM,CAAA;IAChB;;OAEG;IACH,UAAU,EAAE,YAAY,GAAG,UAAU,CAAA;IACrC;;;OAGG;IACH,IAAI,EAAE,MAAM,CAAA;IACZ;;;;OAIG;IACH,MAAM,EAAE,MAAM,CAAA;IACd;;OAEG;IACH,MAAM,EAAE,MAAM,CAAA;IACd;;;;OAIG;IACH,OAAO,EAAE,OAAO,CAAA;IAChB;;;OAGG;IACH,YAAY,EAAE,MAAM,CAAA;IACpB;;;;;;;OAOG;IACH,OAAO,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;CAC/B;AAED;;;;;;;GAOG;AACH,MAAM,WAAW,eAAe;IAC/B,YAAY,EAAE,MAAM,CAAA;IACpB;;;;OAIG;IACH,MAAM,EAAE,MAAM,CAAA;IACd;;;OAGG;IACH,KAAK,EAAE,MAAM,CAAA;IACb,YAAY,EAAE,MAAM,CAAA;IACpB;;OAEG;IACH,QAAQ,EAAE,MAAM,EAAE,CAAA;CAClB;AAED;;GAEG;AACH,MAAM,WAAW,qBAAqB;IACrC,EAAE,EAAE,MAAM,CAAA;IACV,MAAM,EAAE,MAAM,CAAA;IACd,OAAO,EAAE,MAAM,CAAA;IACf,SAAS,EAAE,MAAM,CAAA;IACjB,IAAI,EAAE,MAAM,CAAA;CACZ;AAED;;GAEG;AACH,MAAM,WAAW,oBAAoB;IACpC,OAAO,EAAE,MAAM,CAAA;IACf,OAAO,EAAE,qBAAqB,EAAE,CAAA;IAChC,WAAW,EAAE,kBAAkB,EAAE,CAAA;IACjC,SAAS,EAAE,gBAAgB,EAAE,CAAA;IAC7B,QAAQ,EAAE,eAAe,EAAE,CAAA;IAC3B,gBAAgB,EAAE,eAAe,EAAE,CAAA;IACnC;;;OAGG;IACH,UAAU,EAAE,iBAAiB,EAAE,CAAA;CAC/B;AAED;;;;;;;;GAQG;AACH,eAAO,MAAM,uBAAuB,EAAE,aAAa,CAAC;IACnD,EAAE,EAAE,MAAM,CAAA;IACV,gBAAgB,EAAE,OAAO,CAAA;IACzB,UAAU,EAAE,MAAM,EAAE,CAAA;CACpB,CAUA,CAAA;AAED;;GAEG;AACH,eAAO,MAAM,qBAAqB,EAAE,SAAS,MAAM,EAalD,CAAA;AAED;;;;;;;;;;;GAWG;AACH,eAAO,MAAM,oBAAoB,EAAE,aAAa,CAAC;IAChD,EAAE,EAAE,MAAM,CAAA;IACV,IAAI,EAAE,WAAW,CAAA;IACjB,UAAU,EAAE,MAAM,EAAE,CAAA;CACpB,CAOA,CAAA;AAED;;;;GAIG;AACH,eAAO,MAAM,YAAY,EAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CASzD,CAAA"}