@mailwoman/corpus 9.4.0 → 10.1.0

package/out/subvenue/harvest.d.ts

@@ -0,0 +1,117 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   @file This module defines harvestable source rows and reads them from files. Each pass extracts
+ *   attested surfaces and measures the identifier half of each designator.
+ *
+ *   {@link SubVenueHarvestRow} accepts rows from any source. An Overture Places row in the
+ *   `airport_terminal` category has the shape `{ designatorID, name }` and fits unchanged. The type is
+ *   declared here so the builder can read source rows without importing `@mailwoman/osm`. The corpus
+ *   package has no dependency on that package.
+ *
+ *   Two measurements shape {@link extractAttestedPhrases}. They show why it can run over raw OSM:
+ *
+ *   **Most feature names describe the venue.** In the Berlin extract, 2,060 matched features include a
+ *   `railway=platform` with the name `Stendaler Straße`. An `amenity=university` has the name `Hertie School`.
+ *   Among 250,116 Great Britain features with names, 6,003 (2.40%) contain a designator token. A name adds a surface
+ *   when it contains a phrase already in the surface index. `Terminal E (Untere Ebene)` qualifies. The full
+ *   venue name `Otto Lilienthal Flughafen Berlin Tegel` does not.
+ *
+ *   **The matched phrase determines which record receives the surface.** Every hit attributed to
+ *   `row.designatorID`, the rule that matched the feature, produced `west → platform`, `hall → platform`, and
+ *   `biggin → platform` on the GB extract. A bus stop tagged `public_transport=platform` can have the label
+ *   "Village Hall" or "West Kensington". Of 133 OSM-derived surfaces, 108 referred to records other than the
+ *   ones identified by the phrases. Attribution now follows the surface index. The row's designator stays in `context`.
+ *   A confound board can then distinguish `hall` found on a platform from `hall` found on a terminal.
+ */
+import { type SurfaceIndex } from "#subvenue/surfaces";
+import type { IdentifierShape, SubVenueSurface } from "#subvenue/table";
+/**
+ * One row of a harvestable source, as read back off jsonl or out of a layer database.
+ *
+ * This type accepts rows from any source.
+ * An Overture Places row in the `airport_terminal` category has the shape
+ * `{ designatorID, name }` and fits unchanged.
+ *
+ * The harvest function's hardcoded `osm:name` source stamp did not fit that input.
+ * See `overture-subvenue.ts` for the source handling.
+ *
+ * Declared locally so the builder does not import `@mailwoman/osm`
+ * (which `@mailwoman/corpus` does not depend on) just to name a shape it reads from a file.
+ */
+export interface SubVenueHarvestRow {
+    /**
+     * The designator the source's rule assigned to the feature.
+     *
+     * Not necessarily the record a matched phrase names.
+     * See this file's header on phrase attribution.
+     * Passed into {@link SubVenueSurface.context}.
+     */
+    designatorID: string;
+    /**
+     * `venue` (station, airport, campus) or `sub_venue`, when the source's rule assigns one.
+     *
+     * The harvest itself does not read it.
+     * The sub-venue recipe fills its venue slot from it.
+     */
+    tier?: string;
+    name?: string | null;
+    ref?: string | null;
+    localizedNames?: Record<string, string>;
+}
+/**
+ * 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 declare function classifyIdentifier(ref: string): string;
+/**
+ * Options for one harvest pass.
+ *
+ * They set the source stamp and region recorded on each surface.
+ *
+ * Both default to the OSM/unknown-region values wave 1 hardcoded, so an existing caller is unchanged.
+ */
+export interface HarvestOptions {
+    /**
+     * Source family: `osm` yields `osm:name` / `osm:name:<lang>`, `overture` yields `overture:name`.
+     */
+    source?: string;
+    /**
+     * ISO 3166-1 alpha-2 of the extract or partition.
+     *
+     * `""` when unknown.
+     */
+    region?: string;
+}
+/**
+ * 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 decides which record receives each hit.
+ * The row's designator is recorded as `context`.
+ *
+ * @returns Surfaces with real `observations` counts, so the lexicon can rank
+ * `terminal` above a phrase attested once.
+ */
+export declare function extractAttestedPhrases(rows: Iterable<SubVenueHarvestRow>, index: SurfaceIndex, options?: HarvestOptions): {
+    surfaces: SubVenueSurface[];
+    identifierShapes: IdentifierShape[];
+};
+/**
+ * Read a jsonl file of {@link SubVenueHarvestRow}s.
+ *
+ * The reader skips blank lines and rows it cannot parse.
+ * A malformed line does not stop lexicon generation.
+ */
+export declare function readSubVenueJSONL(path: string): Promise<SubVenueHarvestRow[]>;
+//# sourceMappingURL=harvest.d.ts.map

package/out/subvenue/harvest.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"harvest.d.ts","sourceRoot":"","sources":["../../lib/subvenue/harvest.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;GA2BG;AAMH,OAAO,EAAwB,KAAK,YAAY,EAAE,MAAM,oBAAoB,CAAA;AAC5E,OAAO,KAAK,EAAE,eAAe,EAAE,eAAe,EAAE,MAAM,iBAAiB,CAAA;AAEvE;;;;;;;;;;;;GAYG;AACH,MAAM,WAAW,kBAAkB;IAClC;;;;;;OAMG;IACH,YAAY,EAAE,MAAM,CAAA;IACpB;;;;;OAKG;IACH,IAAI,CAAC,EAAE,MAAM,CAAA;IACb,IAAI,CAAC,EAAE,MAAM,GAAG,IAAI,CAAA;IACpB,GAAG,CAAC,EAAE,MAAM,GAAG,IAAI,CAAA;IACnB,cAAc,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;CACvC;AAED;;;;;;GAMG;AACH,wBAAgB,kBAAkB,CAAC,GAAG,EAAE,MAAM,GAAG,MAAM,CActD;AAgBD;;;;;;GAMG;AACH,MAAM,WAAW,cAAc;IAC9B;;OAEG;IACH,MAAM,CAAC,EAAE,MAAM,CAAA;IACf;;;;OAIG;IACH,MAAM,CAAC,EAAE,MAAM,CAAA;CACf;AAED;;;;;;;;;;;;;;GAcG;AACH,wBAAgB,sBAAsB,CACrC,IAAI,EAAE,QAAQ,CAAC,kBAAkB,CAAC,EAClC,KAAK,EAAE,YAAY,EACnB,OAAO,GAAE,cAAmB,GAC1B;IAAE,QAAQ,EAAE,eAAe,EAAE,CAAC;IAAC,gBAAgB,EAAE,eAAe,EAAE,CAAA;CAAE,CAiFtE;AAED;;;;;GAKG;AACH,wBAAsB,iBAAiB,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CAAC,kBAAkB,EAAE,CAAC,CAenF"}

package/out/subvenue/harvest.js

@@ -0,0 +1,172 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   @file This module defines harvestable source rows and reads them from files. Each pass extracts
+ *   attested surfaces and measures the identifier half of each designator.
+ *
+ *   {@link SubVenueHarvestRow} accepts rows from any source. An Overture Places row in the
+ *   `airport_terminal` category has the shape `{ designatorID, name }` and fits unchanged. The type is
+ *   declared here so the builder can read source rows without importing `@mailwoman/osm`. The corpus
+ *   package has no dependency on that package.
+ *
+ *   Two measurements shape {@link extractAttestedPhrases}. They show why it can run over raw OSM:
+ *
+ *   **Most feature names describe the venue.** In the Berlin extract, 2,060 matched features include a
+ *   `railway=platform` with the name `Stendaler Straße`. An `amenity=university` has the name `Hertie School`.
+ *   Among 250,116 Great Britain features with names, 6,003 (2.40%) contain a designator token. A name adds a surface
+ *   when it contains a phrase already in the surface index. `Terminal E (Untere Ebene)` qualifies. The full
+ *   venue name `Otto Lilienthal Flughafen Berlin Tegel` does not.
+ *
+ *   **The matched phrase determines which record receives the surface.** Every hit attributed to
+ *   `row.designatorID`, the rule that matched the feature, produced `west → platform`, `hall → platform`, and
+ *   `biggin → platform` on the GB extract. A bus stop tagged `public_transport=platform` can have the label
+ *   "Village Hall" or "West Kensington". Of 133 OSM-derived surfaces, 108 referred to records other than the
+ *   ones identified by the phrases. Attribution now follows the surface index. The row's designator stays in `context`.
+ *   A confound board can then distinguish `hall` found on a platform from `hall` found on a terminal.
+ */
+import { readLocalTextFile } from "@mailwoman/core/fs/readers";
+import { parseJSONStrict } from "@mailwoman/core/json";
+import { TextSpliterator } from "spliterator";
+import { nameContainsSurfaces } from "#subvenue/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]+$/.test(value))
+        return "range";
+    return "other";
+}
+/**
+ * How many real `ref` values each {@link IdentifierShape} keeps.
+ *
+ * Eight rather than "all" and not one.
+ * The field exists so a recipe author can see what a class actually contains.
+ *
+ * GB's `other` class turned out to be semicolon multi-values (`1.2.3`, `13.14`),
+ * which one example would have hidden and which the class name does not say.
+ *
+ * Eight fits a terminal line and covers the variety inside every class the GB extract produced.
+ * The count lives in `observations`; this is a sample rather than a census.
+ */
+const IDENTIFIER_EXAMPLES_PER_SHAPE = 8;
+/**
+ * Harvest attested phrases and identifier shapes out of a source's rows.
+ *
+ * `index` gates the name harvest.
+ * A name contributes only when it contains a phrase already in the table.
+ *
+ * That filter is the whole reason this function is safe to run over raw OSM:
+ * see this file's header for the Berlin measurement that motivated it.
+ *
+ * The index decides which record receives each hit.
+ * The row's designator is recorded as `context`.
+ *
+ * @returns Surfaces with real `observations` counts, so the lexicon can rank
+ * `terminal` above a phrase attested once.
+ */
+export function extractAttestedPhrases(rows, index, options = {}) {
+    const source = options.source ?? "osm";
+    const region = options.region ?? "";
+    /**
+     * `phrase\0lang\0source` → { count, context }.
+     */
+    const surfaceCounts = new Map();
+    /**
+     * `designatorID\0shape` → { count, examples }.
+     */
+    const shapes = new Map();
+    const note = (phrase, lang, sourceTag, context) => {
+        const key = `${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 has 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.
+ *
+ * The reader skips blank lines and rows it cannot parse.
+ * A malformed line does not stop lexicon generation.
+ */
+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))) {
+        try {
+            out.push(parseJSONStrict(line));
+        }
+        catch {
+            continue;
+        }
+    }
+    return out;
+}
+//# sourceMappingURL=harvest.js.map

package/out/subvenue/harvest.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"harvest.js","sourceRoot":"","sources":["../../lib/subvenue/harvest.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;GA2BG;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,oBAAoB,CAAA;AAqC5E;;;;;;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;;;;;;;;;;;GAWG;AACH,MAAM,6BAA6B,GAAG,CAAC,CAAA;AAsBvC;;;;;;;;;;;;;;GAcG;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,kDAAkD;gBAClD,+FAA+F;gBAC/F,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;;;;;GAKG;AACH,MAAM,CAAC,KAAK,UAAU,iBAAiB,CAAC,IAAY;IACnD,MAAM,GAAG,GAAyB,EAAE,CAAA;IAEpC,+CAA+C;IAC/C,6FAA6F;IAC7F,+FAA+F;IAC/F,KAAK,MAAM,IAAI,IAAI,eAAe,CAAC,IAAI,CAAC,MAAM,iBAAiB,CAAC,IAAI,CAAC,CAAC,EAAE,CAAC;QACxE,IAAI,CAAC;YACJ,GAAG,CAAC,IAAI,CAAC,eAAe,CAAqB,IAAI,CAAC,CAAC,CAAA;QACpD,CAAC;QAAC,MAAM,CAAC;YACR,SAAQ;QACT,CAAC;IACF,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC"}

package/out/subvenue/promotions.d.ts

@@ -0,0 +1,65 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   The hand-authored ledger of per-locale decisions on sub-venue designator surfaces.
+ */
+/**
+ * One curation decision about a designator phrase in one locale.
+ *
+ * Decisions are per locale because the same token can be a designator in one language
+ * and a common confound in another.
+ */
+export interface SubVenuePromotion {
+    /**
+     * The {@link SubVenueDesignator.id} that the phrase belongs to.
+     */
+    designatorID: string;
+    /**
+     * The exact normalized phrase, such as `halle` for the German form of `hall`.
+     */
+    phrase: string;
+    /**
+     * A `<lang>-<region>` locale.
+     *
+     * The language part matches a surface's `lang`, and the region part matches its `region`.
+     */
+    locale: string;
+    decision: "promote" | "reject";
+    /**
+     * A shape restriction on a promotion.
+     *
+     * `identifier-required` promotes the phrase only as `<phrase> <identifier>`, such as `Halle 8`.
+     * Consumers must honor it, because the bare German `Halle` is also a city.
+     * Without this field, the promotion holds in any shape.
+     */
+    shape?: "identifier-required";
+    /**
+     * The count of census hits that are real venue-interior structure.
+     */
+    real: number;
+    /**
+     * The count of census hits that are confounds.
+     */
+    confound: number;
+    /**
+     * A description of what the confound hits are.
+     */
+    confoundNote: string;
+    /**
+     * The extract or dataset that the census counted, plus the census date.
+     */
+    census: string;
+}
+/**
+ * Every curation decision, grouped by designator.
+ *
+ * This ledger is the only input that sets `curated: true` on a machine-derived surface.
+ * Rejections stay in the ledger so the system can reject duplicate proposals until new evidence arrives.
+ *
+ * The span proposer in `packages/neural/lib/venue-structure.ts` ships its English designators
+ * without a locale gate, so a rejection here guides recipes only and does not stop the proposer.
+ */
+export declare const SUBVENUE_PROMOTIONS: readonly SubVenuePromotion[];
+//# sourceMappingURL=promotions.d.ts.map

package/out/subvenue/promotions.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"promotions.d.ts","sourceRoot":"","sources":["../../lib/subvenue/promotions.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAEH;;;;;GAKG;AACH,MAAM,WAAW,iBAAiB;IACjC;;OAEG;IACH,YAAY,EAAE,MAAM,CAAA;IACpB;;OAEG;IACH,MAAM,EAAE,MAAM,CAAA;IACd;;;;OAIG;IACH,MAAM,EAAE,MAAM,CAAA;IACd,QAAQ,EAAE,SAAS,GAAG,QAAQ,CAAA;IAC9B;;;;;;OAMG;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;;;;;;;;GAQG;AACH,eAAO,MAAM,mBAAmB,EAAE,SAAS,iBAAiB,EAmO3D,CAAA"}

package/out/subvenue/promotions.js

@@ -0,0 +1,226 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   The hand-authored ledger of per-locale decisions on sub-venue designator surfaces.
+ */
+/**
+ * Every curation decision, grouped by designator.
+ *
+ * This ledger is the only input that sets `curated: true` on a machine-derived surface.
+ * Rejections stay in the ledger so the system can reject duplicate proposals until new evidence arrives.
+ *
+ * The span proposer in `packages/neural/lib/venue-structure.ts` ships its English designators
+ * without a locale gate, so a rejection here guides recipes only and does not stop the proposer.
+ */
+export const SUBVENUE_PROMOTIONS = [
+    // wing
+    {
+        designatorID: "wing",
+        phrase: "wing",
+        locale: "en-GB",
+        decision: "promote",
+        real: 23,
+        confound: 6,
+        confoundNote: "Wing Close, Wing Road and Wing Tollgate House — the Buckinghamshire village of Wing — plus Cambian Wing " +
+            "College. All six are the <wing> <word> shape; every real hit is a hospital wing (Bexley Wing, Sorby Wing, " +
+            "Maternity Wing, West Wing, Bedford Hospital North Wing). CAVEAT, and it limits how far this generalizes: " +
+            "24 of the 29 hits sit on a public_transport=platform, because British bus stops are named after the " +
+            "hospital wing they serve. The extractor maps no building wings at all — `aile` in France is 0 hits, `ala` " +
+            "in Spain 0, `flügel` in Germany 0 — so this evidence is a property of British stop naming, not of a source " +
+            "that has wings in it.",
+        census: "great-britain.jsonl 2026-08-05, 29 name hits across name + name:<lang>",
+    },
+    {
+        designatorID: "wing",
+        phrase: "wing",
+        locale: "en-US",
+        decision: "reject",
+        real: 4,
+        confound: 3354,
+        confoundNote: "Red Wing, the boot brand, is 676 shoe_store rows; chicken wings are another 759 " +
+            "(chicken_wings_restaurant 394, chicken_restaurant 365) plus 166 general restaurants. The four real hits are " +
+            "campus_building rows. This is the same word as en-GB's clean 23 and it is unusable here.",
+        census: "poi.db 2026-05-20.0, full 13,681,698-row scan 2026-08-05",
+    },
+    {
+        designatorID: "wing",
+        phrase: "wing",
+        locale: "fr-FR",
+        decision: "reject",
+        real: 0,
+        confound: 26,
+        confoundNote: "Wing Chun and Wing Tsun martial-arts clubs, 15 of 26. Nothing in the French partition of the layer is a " +
+            "building wing.",
+        census: "poi.db 2026-05-20.0, full scan 2026-08-05 (FR partition, 721,352 rows)",
+    },
+    // hall
+    {
+        designatorID: "hall",
+        phrase: "hall",
+        locale: "en-GB",
+        decision: "reject",
+        real: 0,
+        confound: 3273,
+        confoundNote: "3,204 of 3,273 hits sit on a public_transport=platform: a British bus stop named after the village hall it " +
+            "stands outside. The distribution is Village Hall, Town Hall, British Legion Hall, and then Hall Lane / Hall " +
+            "Road / SPEKE HALL ROAD — streets. The 52 that take a <hall> <identifier> shape are bus-stop position codes " +
+            "(Village Hall (o/s), Town Hall (3)), and all 16 <modifier> <hall> hits are streets (East Hall Lane, North " +
+            "Hall Road).",
+        census: "great-britain.jsonl 2026-08-05, 3,273 name hits",
+    },
+    {
+        designatorID: "hall",
+        phrase: "hall",
+        locale: "en-US",
+        decision: "reject",
+        real: 2095,
+        confound: 27_081,
+        confoundNote: "City Hall (5,066 town_hall rows), Kingdom Hall of Jehovah's Witnesses (2,312), event halls (2,622) and " +
+            "dormitory halls (1,989). Every one names a whole VENUE, not interior structure, so promoting the surface " +
+            "would tag them `unit`. The 2,095 real hits are US campus buildings (UAA Cuddy Hall) and they are outnumbered " +
+            "13 to 1.",
+        census: "poi.db 2026-05-20.0, full 13,681,698-row scan 2026-08-05",
+    },
+    {
+        designatorID: "hall",
+        phrase: "halle",
+        locale: "de-DE",
+        decision: "promote",
+        shape: "identifier-required",
+        real: 32,
+        confound: 168,
+        confoundNote: "The confound is 97 hits of the CITY Halle — Halle (Saale), population 240,000, plus Halle (Westf) — and 68 " +
+            "village and event halls (PHOENIX Halle, Urexweiler Halle), which are whole venues. That is a worse " +
+            "confound than wave 1 assumed when it called `Halle 2` the German counter-example to en-GB: the word is " +
+            "also a locality. It is promoted anyway because the confound is SEPARABLE BY SHAPE, and that was checked " +
+            "rather than assumed — dumping all 32 <halle> <identifier> hits returns numbered factory, trade-fair and " +
+            "airport halls (VW Halle 42, Audi GVZ Halle G, Messe West Halle 8, Flughafen Halle 7, Speyer Halle 101) and " +
+            "not one instance of the city. Never modifier-eligible and never proposed bare.",
+        census: "germany.jsonl 2026-08-05 (403,863 rows), 200 hits; all 32 shape hits enumerated",
+    },
+    {
+        designatorID: "hall",
+        phrase: "hall",
+        locale: "fr-FR",
+        decision: "promote",
+        real: 35,
+        confound: 5,
+        confoundNote: "Five: an English `Town Hall` and `Multipurpose hall` on French name:en tags, plus `Manning Hall`. French " +
+            "uses `hall` as the airport designator directly — 19 hits are Charles de Gaulle (Terminal 2E - Hall K/L/M, " +
+            "Terminal 1 - Hall A/B, Hall D) and 16 more are building halls (Hall des Sources, Hall du Marché, Hôpital " +
+            "Hall Sud). The word that is poison in en-GB is the ordinary designator one country away.",
+        census: "france.jsonl 2026-08-05 (251,260 rows), 40 hits; all 19 shape hits enumerated",
+    },
+    // gate: the English surface ships without modifier eligibility.
+    // These entries cover its localized forms.
+    {
+        designatorID: "gate",
+        phrase: "flugsteig",
+        locale: "de-DE",
+        decision: "promote",
+        real: 19,
+        confound: 0,
+        confoundNote: "None, at full-country scale. 18 of 19 hits are <flugsteig> <identifier> (Flugsteig 10, Flugsteig 9/9A, " +
+            "Terminal 1 Flugsteig B) and the last is bare; context is 12 terminal features and 7 gates. The word is a " +
+            "pure aviation term with no street, locality or business collision in German. Frankfurt's own name:en " +
+            "renders Flugsteig J as `Pier J`, which is a second attestation of the same feature under a second " +
+            "designator. A 11-of-11 read on the Hessen extract was the hypothesis; this is the count that earned it.",
+        census: "germany.jsonl 2026-08-05 (403,863 rows), 19 hits across name + name:<lang>",
+    },
+    {
+        designatorID: "gate",
+        phrase: "porte",
+        locale: "fr-FR",
+        decision: "reject",
+        real: 19,
+        confound: 927,
+        confoundNote: "946 hits, of which 894 are Paris city gates and the Métro stations named after them — Porte d'Orléans, " +
+            "Porte Dauphine, Porte de Champerret, Porte de Saint-Cloud. Unlike `halle`, the confound REACHES the " +
+            "designator+identifier shape: 17 of the 36 hits in that bucket are Porte Saint-Martin, Porte Saint-Denis, " +
+            "Porte Notre-Dame. Shape cannot separate it, which is the same test that killed `pier` for en-US. The 19 " +
+            "real hits are ferry and cruise gates (Porte 4 : Ferrys, Berliet Porte C).",
+        census: "france.jsonl 2026-08-05 (251,260 rows), 946 hits; all 36 shape hits enumerated",
+    },
+    // pier
+    {
+        designatorID: "pier",
+        phrase: "pier",
+        locale: "en-GB",
+        decision: "promote",
+        real: 120,
+        confound: 44,
+        confoundNote: "Pier Road, Pier Street, Pier Avenue, Pier Terrace — 44 street names, all of them the <pier> <street-type> " +
+            "shape, which the designator+identifier rule cannot reach. The real side is 98 named piers (Eastbourne Pier, " +
+            "Cadogan Pier), 10 in <pier> <identifier> shape (Pier 1, Pier 2, `Terminal 3, Pier 6`), 10 bare and 2 with a " +
+            "modifier. Promoted as a SURFACE only; not modifier-eligible, on the same reasoning that excludes `gate`.",
+        census: "great-britain.jsonl 2026-08-05, 164 name hits",
+    },
+    {
+        designatorID: "pier",
+        phrase: "pier",
+        locale: "en-US",
+        decision: "reject",
+        real: 278,
+        confound: 2330,
+        confoundNote: "Pier 1 Imports and its franchises are 341 home_decor_store plus 217 furniture_store rows, and they carry the " +
+            "<pier> <identifier> shape EXACTLY — so unlike en-GB the confound is not separable by shape. Another 159 are " +
+            "seafood restaurants and 144 are parks.",
+        census: "poi.db 2026-05-20.0, full 13,681,698-row scan 2026-08-05",
+    },
+    // terminal: the English surface ships.
+    // These entries cover its localized forms.
+    {
+        designatorID: "terminal",
+        phrase: "terminal",
+        locale: "es-ES",
+        decision: "promote",
+        real: 190,
+        confound: 0,
+        confoundNote: "None. Spanish borrows the English word unchanged and uses it in every shape the task targets: Terminal A, " +
+            "Terminal T1, Terminal 4S, plus named terminals (Terminal Marítima de Santander). No Spanish street type " +
+            "collides with it. The Wikidata label `terminal aeroportuaria` is the encyclopaedic form and stays uncurated; " +
+            "this promotes the derived head noun.",
+        census: "spain.jsonl 2026-08-05, 190 hits across name + name:<lang>",
+    },
+    {
+        designatorID: "terminal",
+        phrase: "terminal",
+        locale: "ca-ES",
+        decision: "promote",
+        real: 15,
+        confound: 0,
+        confoundNote: "None. Catalan is the same borrowing (Terminal 1, Terminal Sortides, Terminal Arribades) and needs its own " +
+            "entry because a surface's language subtag is `ca`, which an es-ES decision does not reach.",
+        census: "spain.jsonl 2026-08-05, 15 name:ca hits",
+    },
+    {
+        designatorID: "terminal",
+        phrase: "terminal",
+        locale: "fr-FR",
+        decision: "promote",
+        real: 169,
+        confound: 0,
+        confoundNote: "None. 99 hits are <terminal> <identifier> (Aéroport CDG Terminal 2F, Terminal 2G, Lyon Saint-Exupéry " +
+            "Terminal 1), 51 are named terminals (Terminal de Grande-Bretagne, Terminal Croisières) and 13 trail a " +
+            "venue name. French borrows the word unchanged and no French street type collides with it.",
+        census: "france.jsonl 2026-08-05 (251,260 rows), 169 hits across name + name:<lang>",
+    },
+    {
+        designatorID: "terminal",
+        phrase: "ターミナル",
+        locale: "ja-JP",
+        decision: "promote",
+        real: 1213,
+        confound: 2,
+        confoundNote: "Two hits are `ターミナル前` — a bus stop named 'in front of the terminal', which is arguably still real. " +
+            "Japanese has no street-name class containing the word. 246 hits take the Japanese target shape " +
+            "<identifier><designator> (第2ターミナル, 羽田空港第1ターミナル); 957 are compounds naming a terminal " +
+            "(バスターミナル, フェリーターミナル, 成田空港第一ターミナル); 10 are <designator> <identifier>. This is the " +
+            "surface the corpus task named as missing, and it exists only because the head-noun derivation produced it " +
+            "from five compound Wikidata labels first.",
+        census: "japan.jsonl 2026-08-05, 1,215 hits across name + name:<lang>",
+    },
+];
+//# sourceMappingURL=promotions.js.map

package/out/subvenue/promotions.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"promotions.js","sourceRoot":"","sources":["../../lib/subvenue/promotions.ts"],"names":[],"mappings":"AAAA;;;;;;GAMG;AAkDH;;;;;;;;GAQG;AACH,MAAM,CAAC,MAAM,mBAAmB,GAAiC;IAChE,OAAO;IACP;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,0GAA0G;YAC1G,gBAAgB;QACjB,MAAM,EAAE,wEAAwE;KAChF;IAED,OAAO;IACP;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,gEAAgE;IAChE,2CAA2C;IAC3C;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,OAAO;IACP;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,uCAAuC;IACvC,2CAA2C;IAC3C;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/subvenue/surfaces.d.ts

@@ -0,0 +1,39 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * @file Surface-form primitives for the sub-venue lexicon. They normalize phrases and index records by phrase.
+ * They also decide whether a feature name contains a known designator.
+ *
+ * The two matching rules depend on script. Both narrowings are required.
+ */
+import type { SubVenueSurface } from "#subvenue/table";
+/**
+ * Normalize a surface for the table by trimming, collapsing internal whitespace,
+ * then lowercasing only when the string is entirely in a bicameral script.
+ */
+export declare function normalizeSurface(text: string): string;
+/**
+ * A phrase-to-record index keyed by the normalized phrase.
+ * It attributes a matched phrase to its record.
+ */
+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.
+ */
+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.
+ *
+ * Latin matching is token-bounded rather than substring, so `Nordterminal` is a deliberate miss.
+ * Han and Kana have no word boundaries, so the longest known substring is the correct operator there.
+ */
+export declare function nameContainsSurfaces(name: string, index: SurfaceIndex): string[];
+//# sourceMappingURL=surfaces.d.ts.map

package/out/subvenue/surfaces.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"surfaces.d.ts","sourceRoot":"","sources":["../../lib/subvenue/surfaces.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAIH,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,iBAAiB,CAAA;AAoBtD;;;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;;;;GAIG;AACH,wBAAgB,iBAAiB,CAAC,QAAQ,EAAE,SAAS,eAAe,EAAE,GAAG,YAAY,CASpF;AAED;;;;;;GAMG;AACH,wBAAgB,oBAAoB,CAAC,IAAI,EAAE,MAAM,EAAE,KAAK,EAAE,YAAY,GAAG,MAAM,EAAE,CAqBhF"}