@mailwoman/corpus 9.4.0 → 10.1.0

package/out/tools/sub-venue/table.js

@@ -1,108 +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.
- */
-/**
- * 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 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 const LexiconTier = {
-    SubVenue: "subvenue",
-    Venue: "venue",
-};
-/**
- * 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 const SHIPPED_DESIGNATOR_SEED = [
-    { id: "arcade", modifierEligible: true, provenance: ["wof:placetype"] },
-    { id: "building", modifierEligible: false, provenance: ["wof:placetype"] },
-    { id: "campus", modifierEligible: true, provenance: ["wof:placetype"] },
-    { id: "concourse", modifierEligible: true, provenance: ["wof:placetype"] },
-    { id: "enclosure", modifierEligible: false, provenance: ["wof:placetype"] },
-    { id: "gate", modifierEligible: false, provenance: ["osm:aeroway=gate"] },
-    { id: "installation", modifierEligible: false, provenance: ["wof:placetype"] },
-    { id: "terminal", modifierEligible: true, provenance: ["osm:aeroway=terminal"] },
-    { id: "wing", modifierEligible: true, provenance: ["wof:placetype"] },
-];
-/**
- * The shipped positional modifiers, re-declared from `neural/venue-structure.ts`'s `VENUE_STRUCTURE_MODIFIERS`.
- */
-export const SHIPPED_MODIFIER_SEED = [
-    "central",
-    "east",
-    "front",
-    "inner",
-    "lower",
-    "main",
-    "north",
-    "outer",
-    "rear",
-    "south",
-    "upper",
-    "west",
-];
-/**
- * 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 const PROPOSED_DESIGNATORS = [
-    { id: "airport", tier: LexiconTier.Venue, provenance: ["osm:aeroway=aerodrome"] },
-    { id: "hall", tier: LexiconTier.SubVenue, provenance: ["wikidata:Q240854"] },
-    { id: "pier", tier: LexiconTier.SubVenue, provenance: ["overture:pier"] },
-    { id: "platform", tier: LexiconTier.SubVenue, provenance: ["osm:public_transport=platform", "osm:railway=platform"] },
-    { id: "satellite", tier: LexiconTier.SubVenue, provenance: ["wikidata:Q15990706"] },
-    { id: "station", tier: LexiconTier.Venue, provenance: ["osm:railway=station"] },
-];
-/**
- * `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 const CONCEPT_QIDS = {
-    terminal: "Q849706",
-    gate: "Q247739",
-    concourse: "Q862212",
-    campus: "Q209465",
-    building: "Q41176",
-    arcade: "Q186637",
-    hall: "Q240854",
-    satellite: "Q15990706",
-};
-//# sourceMappingURL=table.js.map

package/out/tools/sub-venue/table.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"table.js","sourceRoot":"","sources":["../../../lib/tools/sub-venue/table.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;GAmBG;AAIH;;;;;GAKG;AACH,MAAM,CAAC,MAAM,wBAAwB,GAAG,OAAO,CAAA;AAE/C;;;GAGG;AACH,MAAM,CAAC,MAAM,WAAW,GAAG;IAC1B,QAAQ,EAAE,UAAU;IACpB,KAAK,EAAE,OAAO;CACL,CAAA;AAgKV;;;;;;;;GAQG;AACH,MAAM,CAAC,MAAM,uBAAuB,GAI/B;IACJ,EAAE,EAAE,EAAE,QAAQ,EAAE,gBAAgB,EAAE,IAAI,EAAE,UAAU,EAAE,CAAC,eAAe,CAAC,EAAE;IACvE,EAAE,EAAE,EAAE,UAAU,EAAE,gBAAgB,EAAE,KAAK,EAAE,UAAU,EAAE,CAAC,eAAe,CAAC,EAAE;IAC1E,EAAE,EAAE,EAAE,QAAQ,EAAE,gBAAgB,EAAE,IAAI,EAAE,UAAU,EAAE,CAAC,eAAe,CAAC,EAAE;IACvE,EAAE,EAAE,EAAE,WAAW,EAAE,gBAAgB,EAAE,IAAI,EAAE,UAAU,EAAE,CAAC,eAAe,CAAC,EAAE;IAC1E,EAAE,EAAE,EAAE,WAAW,EAAE,gBAAgB,EAAE,KAAK,EAAE,UAAU,EAAE,CAAC,eAAe,CAAC,EAAE;IAC3E,EAAE,EAAE,EAAE,MAAM,EAAE,gBAAgB,EAAE,KAAK,EAAE,UAAU,EAAE,CAAC,kBAAkB,CAAC,EAAE;IACzE,EAAE,EAAE,EAAE,cAAc,EAAE,gBAAgB,EAAE,KAAK,EAAE,UAAU,EAAE,CAAC,eAAe,CAAC,EAAE;IAC9E,EAAE,EAAE,EAAE,UAAU,EAAE,gBAAgB,EAAE,IAAI,EAAE,UAAU,EAAE,CAAC,sBAAsB,CAAC,EAAE;IAChF,EAAE,EAAE,EAAE,MAAM,EAAE,gBAAgB,EAAE,IAAI,EAAE,UAAU,EAAE,CAAC,eAAe,CAAC,EAAE;CACrE,CAAA;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,qBAAqB,GAAsB;IACvD,SAAS;IACT,MAAM;IACN,OAAO;IACP,OAAO;IACP,OAAO;IACP,MAAM;IACN,OAAO;IACP,OAAO;IACP,MAAM;IACN,OAAO;IACP,OAAO;IACP,MAAM;CACN,CAAA;AAED;;;;;;;;;;;GAWG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAI5B;IACJ,EAAE,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,WAAW,CAAC,KAAK,EAAE,UAAU,EAAE,CAAC,uBAAuB,CAAC,EAAE;IACjF,EAAE,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,WAAW,CAAC,QAAQ,EAAE,UAAU,EAAE,CAAC,kBAAkB,CAAC,EAAE;IAC5E,EAAE,EAAE,EAAE,MAAM,EAAE,IAAI,EAAE,WAAW,CAAC,QAAQ,EAAE,UAAU,EAAE,CAAC,eAAe,CAAC,EAAE;IACzE,EAAE,EAAE,EAAE,UAAU,EAAE,IAAI,EAAE,WAAW,CAAC,QAAQ,EAAE,UAAU,EAAE,CAAC,+BAA+B,EAAE,sBAAsB,CAAC,EAAE;IACrH,EAAE,EAAE,EAAE,WAAW,EAAE,IAAI,EAAE,WAAW,CAAC,QAAQ,EAAE,UAAU,EAAE,CAAC,oBAAoB,CAAC,EAAE;IACnF,EAAE,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,WAAW,CAAC,KAAK,EAAE,UAAU,EAAE,CAAC,qBAAqB,CAAC,EAAE;CAC/E,CAAA;AAED;;;;GAIG;AACH,MAAM,CAAC,MAAM,YAAY,GAAqC;IAC7D,QAAQ,EAAE,SAAS;IACnB,IAAI,EAAE,SAAS;IACf,SAAS,EAAE,SAAS;IACpB,MAAM,EAAE,SAAS;IACjB,QAAQ,EAAE,QAAQ;IAClB,MAAM,EAAE,SAAS;IACjB,IAAI,EAAE,SAAS;IACf,SAAS,EAAE,WAAW;CACtB,CAAA"}

package/out/tools/sub-venue/wikidata.d.ts

@@ -1,26 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   @file The Wikidata leg of the sub-venue lexicon — turning the designator-label SPARQL payload into
- *   surfaces.
- *
- *   Reads an already-parsed payload rather than issuing a query: the acquisition lives in
- *   `fetch/wikidata-subvenue.ts`, and keeping the conversion a pure function of parsed input is what
- *   lets the builder stay deterministic and testable with no network and no fixture on disk.
- *
- *   Wikidata gives a CONCEPT NAME per language, not a designator as addressed. Q849706's Spanish label
- *   is `terminal aeroportuaria`; the addressed form is `Terminal`. So every surface produced here lands
- *   `curated: false`, and the head-noun derivation is what proposes the addressed form from it.
- */
-import { type SubVenueSurface } from "#tools/sub-venue/table";
-/**
- * Turn the Wikidata designator-label payload into surfaces.
- *
- * A row is dropped when its language tag is empty (an untagged literal, which Wikidata occasionally carries), when the
- * QID maps to no designator in {@link CONCEPT_QIDS}, or when the normalized phrase is empty. Everything that survives
- * lands `curated: false` — see the module docstring.
- */
-export declare function surfacesFromWikidata(payload: unknown, conceptQIDs?: Readonly<Record<string, string>>): SubVenueSurface[];
-//# sourceMappingURL=wikidata.d.ts.map

package/out/tools/sub-venue/wikidata.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"wikidata.d.ts","sourceRoot":"","sources":["../../../lib/tools/sub-venue/wikidata.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;GAeG;AAGH,OAAO,EAAgB,KAAK,eAAe,EAAE,MAAM,wBAAwB,CAAA;AAgB3E;;;;;;GAMG;AACH,wBAAgB,oBAAoB,CACnC,OAAO,EAAE,OAAO,EAChB,WAAW,GAAE,QAAQ,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAgB,GAC1D,eAAe,EAAE,CAwCnB"}

package/out/tools/sub-venue/wikidata.js

@@ -1,62 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   @file The Wikidata leg of the sub-venue lexicon — turning the designator-label SPARQL payload into
- *   surfaces.
- *
- *   Reads an already-parsed payload rather than issuing a query: the acquisition lives in
- *   `fetch/wikidata-subvenue.ts`, and keeping the conversion a pure function of parsed input is what
- *   lets the builder stay deterministic and testable with no network and no fixture on disk.
- *
- *   Wikidata gives a CONCEPT NAME per language, not a designator as addressed. Q849706's Spanish label
- *   is `terminal aeroportuaria`; the addressed form is `Terminal`. So every surface produced here lands
- *   `curated: false`, and the head-noun derivation is what proposes the addressed form from it.
- */
-import { normalizeSurface } from "#tools/sub-venue/surfaces";
-import { CONCEPT_QIDS } from "#tools/sub-venue/table";
-/**
- * Turn the Wikidata designator-label payload into surfaces.
- *
- * A row is dropped when its language tag is empty (an untagged literal, which Wikidata occasionally carries), when the
- * QID maps to no designator in {@link CONCEPT_QIDS}, or when the normalized phrase is empty. Everything that survives
- * lands `curated: false` — see the module docstring.
- */
-export function surfacesFromWikidata(payload, conceptQIDs = CONCEPT_QIDS) {
-    const byQID = new Map(Object.entries(conceptQIDs).map(([id, qid]) => [qid, id]));
-    const envelope = payload;
-    const seen = new Set();
-    const out = [];
-    for (const binding of envelope.results?.bindings ?? []) {
-        const qid = binding.item?.value?.split("/").pop();
-        const recordID = qid ? byQID.get(qid) : undefined;
-        const lang = binding.lang?.value;
-        const raw = binding.label?.value;
-        if (!recordID || !lang || !raw)
-            continue;
-        const phrase = normalizeSurface(raw);
-        if (!phrase)
-            continue;
-        const source = binding.kind?.value === "alt" ? "wikidata:alt" : "wikidata:label";
-        // A concept can carry the same string as both a label and an alias, and across dialect subtags
-        // (`zh`, `zh-cn`, `zh-hans` all say 航站楼). Key the dedupe on the tuple that identifies a row.
-        const key = `${phrase}\0${recordID}\0${lang}\0${source}`;
-        if (seen.has(key))
-            continue;
-        seen.add(key);
-        out.push({
-            phrase,
-            recordID,
-            recordKind: "designator",
-            lang,
-            region: "",
-            source,
-            curated: false,
-            observations: 0,
-            context: {},
-        });
-    }
-    return out;
-}
-//# sourceMappingURL=wikidata.js.map

package/out/tools/sub-venue/wikidata.js.map

@@ -1 +0,0 @@
-{"version":3,"file":"wikidata.js","sourceRoot":"","sources":["../../../lib/tools/sub-venue/wikidata.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;GAeG;AAEH,OAAO,EAAE,gBAAgB,EAAE,MAAM,2BAA2B,CAAA;AAC5D,OAAO,EAAE,YAAY,EAAwB,MAAM,wBAAwB,CAAA;AAgB3E;;;;;;GAMG;AACH,MAAM,UAAU,oBAAoB,CACnC,OAAgB,EAChB,cAAgD,YAAY;IAE5D,MAAM,KAAK,GAAG,IAAI,GAAG,CAAC,MAAM,CAAC,OAAO,CAAC,WAAW,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,GAAG,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC,CAAC,CAAA;IAChF,MAAM,QAAQ,GAAG,OAAyB,CAAA;IAC1C,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAA;IAC9B,MAAM,GAAG,GAAsB,EAAE,CAAA;IAEjC,KAAK,MAAM,OAAO,IAAI,QAAQ,CAAC,OAAO,EAAE,QAAQ,IAAI,EAAE,EAAE,CAAC;QACxD,MAAM,GAAG,GAAG,OAAO,CAAC,IAAI,EAAE,KAAK,EAAE,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,CAAA;QACjD,MAAM,QAAQ,GAAG,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,SAAS,CAAA;QACjD,MAAM,IAAI,GAAG,OAAO,CAAC,IAAI,EAAE,KAAK,CAAA;QAChC,MAAM,GAAG,GAAG,OAAO,CAAC,KAAK,EAAE,KAAK,CAAA;QAEhC,IAAI,CAAC,QAAQ,IAAI,CAAC,IAAI,IAAI,CAAC,GAAG;YAAE,SAAQ;QAExC,MAAM,MAAM,GAAG,gBAAgB,CAAC,GAAG,CAAC,CAAA;QAEpC,IAAI,CAAC,MAAM;YAAE,SAAQ;QAErB,MAAM,MAAM,GAAG,OAAO,CAAC,IAAI,EAAE,KAAK,KAAK,KAAK,CAAC,CAAC,CAAC,cAAc,CAAC,CAAC,CAAC,gBAAgB,CAAA;QAChF,+FAA+F;QAC/F,6FAA6F;QAC7F,MAAM,GAAG,GAAG,GAAG,MAAM,KAAK,QAAQ,KAAK,IAAI,KAAK,MAAM,EAAE,CAAA;QAExD,IAAI,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC;YAAE,SAAQ;QAC3B,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;QAEb,GAAG,CAAC,IAAI,CAAC;YACR,MAAM;YACN,QAAQ;YACR,UAAU,EAAE,YAAY;YACxB,IAAI;YACJ,MAAM,EAAE,EAAE;YACV,MAAM;YACN,OAAO,EAAE,KAAK;YACd,YAAY,EAAE,CAAC;YACf,OAAO,EAAE,EAAE;SACX,CAAC,CAAA;IACH,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC"}

package/out/tools/sub-venue-lexicon.d.ts

@@ -1,156 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   @file Build the sub-venue designator lexicon (#35) — the vocabulary a corpus slice and, eventually,
- *   the span proposer read to recognize `Terminal 5`, `North Terminal`, `Concourse B`, `ターミナル1` as
- *   venue-INTERIOR structure. This is the assembly: the record schema lives in `sub-venue/table.ts`, the
- *   implementation in its siblings, and the curation decisions in `sub-venue-promotions.ts`.
- *
- *   Reads the fetch outputs (`mailwoman corpus fetch wikidata-subvenue`, a JSONL of
- *   `@mailwoman/osm/sdk`'s `SubVenueSourceRow`s per region, and the Overture slice of `poi.db` via
- *   `overture-subvenue.ts`) and emits one committed JSON table.
- *
- *   ── Determinism ──────────────────────────────────────────────────────────────────────────────────
- *   {@link buildSubVenueLexicon} is a PURE function of its inputs with a stable sort on every array, so
- *   a regenerate against the same fetch outputs is byte-identical. No timestamp is emitted for the same
- *   reason `taxonomy.json` carries none — a clock in the artifact makes every regenerate a diff.
- *   Vintages live in `sources[]`, taken from the fetch manifests.
- *
- *   ── Where the stages live ────────────────────────────────────────────────────────────────────────
- *   Each stage carries the measurements that shaped it; the order they run in is
- *   {@link buildSubVenueLexicon}'s own docstring, and it is required.
- *
- *   - `sub-venue/table.ts` — the emitted record schema plus the shipped seed vocabulary.
- *   - `sub-venue/surfaces.ts` — phrase normalization, the phrase → record index, and the name-match
- *     operator that gates the harvest.
- *   - `sub-venue/wikidata.ts` — the designator-label SPARQL payload turned into surfaces.
- *   - `sub-venue/head-nouns.ts` — the addressed form derived from an encyclopaedic label.
- *   - `sub-venue/harvest.ts` — the harvestable row shape, its JSONL reader, and the attestation pass.
- *
- *   ── What `curated: false` means, and how a surface stops being it ────────────────────────────────
- *   Every machine-derived surface lands `curated: false`, and {@link SubVenueLexiconTable} consumers
- *   that gate parsing MUST filter to `curated: true`. A surface becomes curated ONLY by matching a
- *   {@link SubVenuePromotion} in `sub-venue-promotions.ts` — a per-designator, per-LOCALE decision
- *   carrying the census that backs it. Promotion is per-locale because the same token is a designator in
- *   one language and a disaster in another: `hall` is `Halle 2` at Frankfurt and `Village Hall` at 3,205
- *   British bus stops.
- */
-import { type SubVenuePromotion } from "#tools/sub-venue-promotions";
-import { type SubVenueHarvestRow } from "#tools/sub-venue/harvest";
-import { type SubVenueLexiconSource, type SubVenueLexiconTable, type SubVenueSurface } from "#tools/sub-venue/table";
-export * from "#tools/sub-venue/harvest";
-export * from "#tools/sub-venue/head-nouns";
-export * from "#tools/sub-venue/surfaces";
-export * from "#tools/sub-venue/table";
-export * from "#tools/sub-venue/wikidata";
-/**
- * Apply the curation decisions to a surface list, IN PLACE on a copy.
- *
- * A promotion binds `(designatorID, phrase, locale)`. A surface matches when it names the same record with the same
- * phrase and its language is the locale's language OR the untagged `und` — the default `name` tag carries no language,
- * and a German extract's untagged `Halle 2` is German.
- *
- * REGION is the subtle half. A surface attested in an extract carries that extract's region and matches only its own
- * locale. A surface with `region: ""` is region-FREE — a Wikidata label or a derived head noun — and a promotion
- * reaches it only when no REJECTION exists for the same designator, phrase and language anywhere else. That guard is
- * not decoration: `pier` is promoted for en-GB and rejected for en-US, and without it the en-GB decision would curate
- * the region-free English surface and hand `Pier 1 Imports` the promotion en-US was refused. Where no rejection
- * competes — `terminal` in `es`, `ターミナル` in `ja` — the region-free surface is the whole point, since a language's
- * designator does not stop at a border.
- *
- * Rejections mark nothing themselves. They exist in `promotions[]` as the record of a decision taken, so the next
- * reader meets en-GB `hall`'s 3,204 bus stops before re-proposing it, not after.
- */
-export declare function applyPromotions(surfaces: readonly SubVenueSurface[], promotions: readonly SubVenuePromotion[]): SubVenueSurface[];
-/**
- * One harvestable input: rows plus the stamp they carry into the table.
- */
-export interface SubVenueHarvest {
-    rows: readonly SubVenueHarvestRow[];
-    source?: string;
-    region?: string;
-}
-/**
- * Everything {@link buildSubVenueLexicon} needs, already parsed. Keeping the builder off the filesystem is what makes it
- * deterministic and testable without fixtures on disk.
- */
-export interface BuildSubVenueLexiconInput {
-    /**
-     * The raw `designator-labels.json` SPARQL envelope, or `null` to build the seed-only table.
-     */
-    wikidata: unknown | null;
-    /**
-     * Every harvestable source, in the order they should contribute. Order matters only for the surface INDEX: a source
-     * can match a phrase an earlier source introduced, never a later one.
-     */
-    harvests: readonly SubVenueHarvest[];
-    /**
-     * Provenance rows, copied off the fetch manifests by the caller.
-     */
-    sources: readonly SubVenueLexiconSource[];
-    /**
-     * Curation decisions. Defaults to the committed {@link SUBVENUE_PROMOTIONS}; pass an empty array to build the
-     * pre-curation table (which is what the promotion census itself is taken against).
-     */
-    promotions?: readonly SubVenuePromotion[];
-}
-/**
- * Build the lexicon table. PURE and deterministic — same inputs, byte-identical output.
- *
- * Order of operations is required in three places:
- *
- * 1. Seed surfaces are inserted before anything else, so `terminal` indexes to the `terminal` designator rather than to
- *    whichever Wikidata alias sorts first.
- * 2. Head nouns are derived AFTER Wikidata and BEFORE the harvests, because `ターミナル` has to exist as a surface before a
- *    Japanese extract can be searched for it. That ordering is the entire reason the Japan harvest finds anything — see
- *    `PROVENANCE.md`.
- * 3. Promotions are applied LAST, over the union, so a decision can promote a surface whichever source produced it.
- */
-export declare function buildSubVenueLexicon(input: BuildSubVenueLexiconInput): SubVenueLexiconTable;
-/**
- * Serialize the table the way the committed artifact stores it: pretty-printed, trailing newline. Run `oxfmt` over the
- * result before committing — repo law is that committed JSON is oxfmt-clean, which `JSON.stringify` cannot reproduce.
- */
-export declare function serializeSubVenueLexicon(table: SubVenueLexiconTable): string;
-/**
- * One OSM extract to harvest: the JSONL path plus the ISO country its rows describe.
- */
-export interface SubVenueExtractInput {
-    path: string;
-    region: string;
-}
-export interface GenerateSubVenueLexiconOptions {
-    /**
-     * Directory holding the `mailwoman corpus fetch wikidata-subvenue` output. Omit to build the seed-only table.
-     */
-    wikidataDir?: string;
-    /**
-     * OSM extract JSONLs, one per region.
-     */
-    extracts?: readonly SubVenueExtractInput[];
-    /**
-     * Already-read Overture rows (`readOvertureSubVenues`), grouped by the caller. Kept as a parameter rather than a path
-     * so this function stays free of a 3.9 GB database dependency — the CLI opens `poi.db`, this assembles.
-     */
-    overtureRows?: readonly (SubVenueHarvestRow & {
-        country: string;
-    })[];
-    /**
-     * `poi.db`'s layer vintage, for `sources[]`. Only read when `overtureRows` is non-empty.
-     */
-    overtureVintage?: string;
-    /**
-     * Where the table is written.
-     */
-    outPath: string;
-}
-/**
- * Read the fetch outputs, build the table, and write it.
- *
- * The IO half only — every decision lives in {@link buildSubVenueLexicon}, which is pure. Run `oxfmt` over `outPath`
- * afterwards; repo law is that committed JSON is oxfmt-clean.
- */
-export declare function generateSubVenueLexicon(options: GenerateSubVenueLexiconOptions): Promise<SubVenueLexiconTable>;
-//# sourceMappingURL=sub-venue-lexicon.d.ts.map

package/out/tools/sub-venue-lexicon.d.ts.map

@@ -1 +0,0 @@
-{"version":3,"file":"sub-venue-lexicon.d.ts","sourceRoot":"","sources":["../../lib/tools/sub-venue-lexicon.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAsCG;AAOH,OAAO,EAAuB,KAAK,iBAAiB,EAAE,MAAM,6BAA6B,CAAA;AACzF,OAAO,EAA6C,KAAK,kBAAkB,EAAE,MAAM,0BAA0B,CAAA;AAG7G,OAAO,EAQN,KAAK,qBAAqB,EAC1B,KAAK,oBAAoB,EAEzB,KAAK,eAAe,EAEpB,MAAM,wBAAwB,CAAA;AAG/B,cAAc,0BAA0B,CAAA;AACxC,cAAc,6BAA6B,CAAA;AAC3C,cAAc,2BAA2B,CAAA;AACzC,cAAc,wBAAwB,CAAA;AACtC,cAAc,2BAA2B,CAAA;AAEzC;;;;;;;;;;;;;;;;;GAiBG;AACH,wBAAgB,eAAe,CAC9B,QAAQ,EAAE,SAAS,eAAe,EAAE,EACpC,UAAU,EAAE,SAAS,iBAAiB,EAAE,GACtC,eAAe,EAAE,CA6BnB;AAED;;GAEG;AACH,MAAM,WAAW,eAAe;IAC/B,IAAI,EAAE,SAAS,kBAAkB,EAAE,CAAA;IACnC,MAAM,CAAC,EAAE,MAAM,CAAA;IACf,MAAM,CAAC,EAAE,MAAM,CAAA;CACf;AAED;;;GAGG;AACH,MAAM,WAAW,yBAAyB;IACzC;;OAEG;IACH,QAAQ,EAAE,OAAO,GAAG,IAAI,CAAA;IACxB;;;OAGG;IACH,QAAQ,EAAE,SAAS,eAAe,EAAE,CAAA;IACpC;;OAEG;IACH,OAAO,EAAE,SAAS,qBAAqB,EAAE,CAAA;IACzC;;;OAGG;IACH,UAAU,CAAC,EAAE,SAAS,iBAAiB,EAAE,CAAA;CACzC;AAED;;;;;;;;;;;GAWG;AACH,wBAAgB,oBAAoB,CAAC,KAAK,EAAE,yBAAyB,GAAG,oBAAoB,CAgI3F;AAED;;;GAGG;AACH,wBAAgB,wBAAwB,CAAC,KAAK,EAAE,oBAAoB,GAAG,MAAM,CAE5E;AAYD;;GAEG;AACH,MAAM,WAAW,oBAAoB;IACpC,IAAI,EAAE,MAAM,CAAA;IACZ,MAAM,EAAE,MAAM,CAAA;CACd;AAED,MAAM,WAAW,8BAA8B;IAC9C;;OAEG;IACH,WAAW,CAAC,EAAE,MAAM,CAAA;IACpB;;OAEG;IACH,QAAQ,CAAC,EAAE,SAAS,oBAAoB,EAAE,CAAA;IAC1C;;;OAGG;IACH,YAAY,CAAC,EAAE,SAAS,CAAC,kBAAkB,GAAG;QAAE,OAAO,EAAE,MAAM,CAAA;KAAE,CAAC,EAAE,CAAA;IACpE;;OAEG;IACH,eAAe,CAAC,EAAE,MAAM,CAAA;IACxB;;OAEG;IACH,OAAO,EAAE,MAAM,CAAA;CACf;AAED;;;;;GAKG;AACH,wBAAsB,uBAAuB,CAAC,OAAO,EAAE,8BAA8B,GAAG,OAAO,CAAC,oBAAoB,CAAC,CAyEpH"}

package/out/tools/sub-venue-lexicon.js

@@ -1,283 +0,0 @@
-/**
- * @copyright Sister Software
- * @license AGPL-3.0
- * @author Teffen Ellis, et al.
- *
- *   @file Build the sub-venue designator lexicon (#35) — the vocabulary a corpus slice and, eventually,
- *   the span proposer read to recognize `Terminal 5`, `North Terminal`, `Concourse B`, `ターミナル1` as
- *   venue-INTERIOR structure. This is the assembly: the record schema lives in `sub-venue/table.ts`, the
- *   implementation in its siblings, and the curation decisions in `sub-venue-promotions.ts`.
- *
- *   Reads the fetch outputs (`mailwoman corpus fetch wikidata-subvenue`, a JSONL of
- *   `@mailwoman/osm/sdk`'s `SubVenueSourceRow`s per region, and the Overture slice of `poi.db` via
- *   `overture-subvenue.ts`) and emits one committed JSON table.
- *
- *   ── Determinism ──────────────────────────────────────────────────────────────────────────────────
- *   {@link buildSubVenueLexicon} is a PURE function of its inputs with a stable sort on every array, so
- *   a regenerate against the same fetch outputs is byte-identical. No timestamp is emitted for the same
- *   reason `taxonomy.json` carries none — a clock in the artifact makes every regenerate a diff.
- *   Vintages live in `sources[]`, taken from the fetch manifests.
- *
- *   ── Where the stages live ────────────────────────────────────────────────────────────────────────
- *   Each stage carries the measurements that shaped it; the order they run in is
- *   {@link buildSubVenueLexicon}'s own docstring, and it is required.
- *
- *   - `sub-venue/table.ts` — the emitted record schema plus the shipped seed vocabulary.
- *   - `sub-venue/surfaces.ts` — phrase normalization, the phrase → record index, and the name-match
- *     operator that gates the harvest.
- *   - `sub-venue/wikidata.ts` — the designator-label SPARQL payload turned into surfaces.
- *   - `sub-venue/head-nouns.ts` — the addressed form derived from an encyclopaedic label.
- *   - `sub-venue/harvest.ts` — the harvestable row shape, its JSONL reader, and the attestation pass.
- *
- *   ── What `curated: false` means, and how a surface stops being it ────────────────────────────────
- *   Every machine-derived surface lands `curated: false`, and {@link SubVenueLexiconTable} consumers
- *   that gate parsing MUST filter to `curated: true`. A surface becomes curated ONLY by matching a
- *   {@link SubVenuePromotion} in `sub-venue-promotions.ts` — a per-designator, per-LOCALE decision
- *   carrying the census that backs it. Promotion is per-locale because the same token is a designator in
- *   one language and a disaster in another: `hall` is `Halle 2` at Frankfurt and `Village Hall` at 3,205
- *   British bus stops.
- */
-import { readLocalJSONFile, statPath } from "@mailwoman/core/fs/readers";
-import { writeLocalFile } from "@mailwoman/core/fs/writers";
-import { isoDate } from "@mailwoman/core/utils";
-import { basename, join } from "path-ts";
-import { SUBVENUE_PROMOTIONS } from "#tools/sub-venue-promotions";
-import { extractAttestedPhrases, readSubVenueJSONL } from "#tools/sub-venue/harvest";
-import { deriveHeadNounSurfaces } from "#tools/sub-venue/head-nouns";
-import { buildSurfaceIndex } from "#tools/sub-venue/surfaces";
-import { CONCEPT_QIDS, LexiconTier, PROPOSED_DESIGNATORS, SHIPPED_DESIGNATOR_SEED, SHIPPED_MODIFIER_SEED, SUBVENUE_LEXICON_VERSION, } from "#tools/sub-venue/table";
-import { surfacesFromWikidata } from "#tools/sub-venue/wikidata";
-export * from "#tools/sub-venue/harvest";
-export * from "#tools/sub-venue/head-nouns";
-export * from "#tools/sub-venue/surfaces";
-export * from "#tools/sub-venue/table";
-export * from "#tools/sub-venue/wikidata";
-/**
- * Apply the curation decisions to a surface list, IN PLACE on a copy.
- *
- * A promotion binds `(designatorID, phrase, locale)`. A surface matches when it names the same record with the same
- * phrase and its language is the locale's language OR the untagged `und` — the default `name` tag carries no language,
- * and a German extract's untagged `Halle 2` is German.
- *
- * REGION is the subtle half. A surface attested in an extract carries that extract's region and matches only its own
- * locale. A surface with `region: ""` is region-FREE — a Wikidata label or a derived head noun — and a promotion
- * reaches it only when no REJECTION exists for the same designator, phrase and language anywhere else. That guard is
- * not decoration: `pier` is promoted for en-GB and rejected for en-US, and without it the en-GB decision would curate
- * the region-free English surface and hand `Pier 1 Imports` the promotion en-US was refused. Where no rejection
- * competes — `terminal` in `es`, `ターミナル` in `ja` — the region-free surface is the whole point, since a language's
- * designator does not stop at a border.
- *
- * Rejections mark nothing themselves. They exist in `promotions[]` as the record of a decision taken, so the next
- * reader meets en-GB `hall`'s 3,204 bus stops before re-proposing it, not after.
- */
-export function applyPromotions(surfaces, promotions) {
-    const language = (locale) => locale.split("-")[0];
-    const region = (locale) => locale.split("-")[1] ?? "";
-    const rejectedLanguages = new Set(promotions
-        .filter((promotion) => promotion.decision === "reject")
-        .map((promotion) => `${promotion.designatorID} ${promotion.phrase} ${language(promotion.locale)}`));
-    const promoted = promotions.filter((promotion) => promotion.decision === "promote");
-    return surfaces.map((surface) => {
-        if (surface.curated)
-            return surface;
-        const hit = promoted.some((promotion) => {
-            if (promotion.designatorID !== surface.recordID || promotion.phrase !== surface.phrase)
-                return false;
-            const lang = language(promotion.locale);
-            if (surface.lang !== lang && surface.lang !== "und")
-                return false;
-            if (surface.region === region(promotion.locale))
-                return true;
-            return surface.region === "" && !rejectedLanguages.has(`${promotion.designatorID} ${promotion.phrase} ${lang}`);
-        });
-        return hit ? { ...surface, curated: true } : surface;
-    });
-}
-/**
- * Build the lexicon table. PURE and deterministic — same inputs, byte-identical output.
- *
- * Order of operations is required in three places:
- *
- * 1. Seed surfaces are inserted before anything else, so `terminal` indexes to the `terminal` designator rather than to
- *    whichever Wikidata alias sorts first.
- * 2. Head nouns are derived AFTER Wikidata and BEFORE the harvests, because `ターミナル` has to exist as a surface before a
- *    Japanese extract can be searched for it. That ordering is the entire reason the Japan harvest finds anything — see
- *    `PROVENANCE.md`.
- * 3. Promotions are applied LAST, over the union, so a decision can promote a surface whichever source produced it.
- */
-export function buildSubVenueLexicon(input) {
-    const designators = SHIPPED_DESIGNATOR_SEED.map((seed) => ({
-        id: seed.id,
-        tier: LexiconTier.SubVenue,
-        modifierEligible: seed.modifierEligible,
-        shipped: true,
-        provenance: [...seed.provenance],
-    }));
-    const byID = new Map(designators.map((d) => [d.id, d]));
-    for (const proposed of PROPOSED_DESIGNATORS) {
-        const existing = byID.get(proposed.id);
-        if (existing) {
-            existing.provenance = [...new Set([...existing.provenance, ...proposed.provenance])];
-            continue;
-        }
-        const record = {
-            id: proposed.id,
-            tier: proposed.tier,
-            modifierEligible: false,
-            shipped: false,
-            provenance: [...proposed.provenance],
-        };
-        designators.push(record);
-        byID.set(record.id, record);
-    }
-    // A Wikidata concept id is provenance for the designator it names, whether or not the concept
-    // contributed a usable surface.
-    for (const [id, qid] of Object.entries(CONCEPT_QIDS)) {
-        const record = byID.get(id);
-        if (record) {
-            record.provenance = [...new Set([...record.provenance, `wikidata:${qid}`])];
-        }
-    }
-    const modifiers = SHIPPED_MODIFIER_SEED.map((id) => ({
-        id,
-        shipped: true,
-        provenance: ["codex:directionals"],
-    }));
-    const surfaces = [
-        ...designators.map((d) => ({
-            phrase: d.id,
-            recordID: d.id,
-            recordKind: "designator",
-            lang: "en",
-            region: "",
-            source: "seed",
-            // The English designator IS the shipped vocabulary — curated by construction.
-            curated: d.shipped,
-            observations: 0,
-            context: {},
-        })),
-        ...modifiers.map((m) => ({
-            phrase: m.id,
-            recordID: m.id,
-            recordKind: "modifier",
-            lang: "en",
-            region: "",
-            source: "seed",
-            curated: true,
-            observations: 0,
-            context: {},
-        })),
-    ];
-    if (input.wikidata) {
-        surfaces.push(...surfacesFromWikidata(input.wikidata));
-    }
-    surfaces.push(...deriveHeadNounSurfaces(surfaces));
-    const identifierShapes = [];
-    for (const harvest of input.harvests) {
-        const attested = extractAttestedPhrases(harvest.rows, buildSurfaceIndex(surfaces), {
-            source: harvest.source,
-            region: harvest.region,
-        });
-        surfaces.push(...attested.surfaces);
-        identifierShapes.push(...attested.identifierShapes);
-    }
-    const promotions = [...(input.promotions ?? SUBVENUE_PROMOTIONS)];
-    const curated = applyPromotions(surfaces, promotions);
-    // Deterministic order everywhere. `localeCompare` matches the tie-break discipline
-    // `generate-taxonomy.ts` and `build-brands.ts` already use.
-    designators.sort((a, b) => a.id.localeCompare(b.id));
-    modifiers.sort((a, b) => a.id.localeCompare(b.id));
-    curated.sort((a, b) => a.phrase.localeCompare(b.phrase) ||
-        a.recordID.localeCompare(b.recordID) ||
-        a.lang.localeCompare(b.lang) ||
-        a.region.localeCompare(b.region) ||
-        a.source.localeCompare(b.source));
-    return {
-        version: SUBVENUE_LEXICON_VERSION,
-        sources: input.sources.toSorted((a, b) => a.id.localeCompare(b.id)),
-        designators,
-        modifiers,
-        surfaces: curated,
-        identifierShapes: identifierShapes.toSorted((a, b) => a.designatorID.localeCompare(b.designatorID) ||
-            a.region.localeCompare(b.region) ||
-            a.shape.localeCompare(b.shape)),
-        promotions: promotions.toSorted((a, b) => a.designatorID.localeCompare(b.designatorID) ||
-            a.locale.localeCompare(b.locale) ||
-            a.phrase.localeCompare(b.phrase)),
-    };
-}
-/**
- * Serialize the table the way the committed artifact stores it: pretty-printed, trailing newline. Run `oxfmt` over the
- * result before committing — repo law is that committed JSON is oxfmt-clean, which `JSON.stringify` cannot reproduce.
- */
-export function serializeSubVenueLexicon(table) {
-    return JSON.stringify(table, null, 2) + "\n";
-}
-/**
- * Read the fetch outputs, build the table, and write it.
- *
- * The IO half only — every decision lives in {@link buildSubVenueLexicon}, which is pure. Run `oxfmt` over `outPath`
- * afterwards; repo law is that committed JSON is oxfmt-clean.
- */
-export async function generateSubVenueLexicon(options) {
-    const sources = [];
-    const harvests = [];
-    let wikidata = null;
-    if (options.wikidataDir) {
-        wikidata = await readLocalJSONFile(join(options.wikidataDir, "designator-labels.json"));
-        const manifest = await readLocalJSONFile(join(options.wikidataDir, "MANIFEST.json"));
-        const labelFile = manifest.files?.find((f) => f.filename === "designator-labels.json");
-        sources.push({
-            id: "wikidata",
-            origin: manifest.endpoint ?? "https://query.wikidata.org/sparql",
-            license: manifest.license ?? "CC0",
-            // The DATE only. A full ISO timestamp would make every re-fetch a diff in the committed artifact for
-            // no information a reader of a vocabulary table can act on.
-            retrieved: (manifest.downloaded_at ?? "").slice(0, 10),
-            rows: labelFile?.rows ?? 0,
-        });
-    }
-    for (const extract of options.extracts ?? []) {
-        const rows = await readSubVenueJSONL(extract.path);
-        harvests.push({ rows, source: "osm", region: extract.region });
-        sources.push({
-            id: `osm:${extract.region.toLowerCase()}`,
-            // The extract's NAME, never its path. `AGENTS.md` forbids re-hardcoding the lab data root
-            // anywhere, and a committed artifact carrying `/mnt/playpen/...` would do exactly that while
-            // telling a reader on another machine nothing. `great-britain` identifies the Geofabrik region,
-            // which is the fact that matters.
-            origin: `OpenStreetMap via Geofabrik (${basename(extract.path, ".jsonl")})`,
-            license: "ODbL (OpenStreetMap)",
-            // The extract's mtime — when the rows were produced. `corpus/AGENTS.md`'s standing warning that
-            // a file's mtime is not its DATA's vintage applies to a downloaded archive; this file is a build
-            // output of ours, so its mtime is exactly the right number.
-            retrieved: isoDate((await statPath(extract.path)).mtime),
-            rows: rows.length,
-        });
-    }
-    if (options.overtureRows?.length) {
-        // Overture rows carry their own country, so they are harvested per REGION rather than in one pass —
-        // a `region` on the surface is the axis promotion is decided on and a mixed-country bucket would
-        // make it meaningless.
-        const byCountry = new Map();
-        for (const row of options.overtureRows) {
-            const bucket = byCountry.get(row.country) ?? [];
-            bucket.push(row);
-            byCountry.set(row.country, bucket);
-        }
-        for (const [country, rows] of [...byCountry].toSorted((a, b) => a[0].localeCompare(b[0]))) {
-            harvests.push({ rows, source: "overture", region: country });
-        }
-        sources.push({
-            id: "overture",
-            origin: "Overture Maps Foundation places, via the poi.db spatial layer",
-            license: "CDLA-Permissive-2.0",
-            retrieved: options.overtureVintage ?? "",
-            rows: options.overtureRows.length,
-        });
-    }
-    const table = buildSubVenueLexicon({ wikidata, harvests, sources });
-    await writeLocalFile(serializeSubVenueLexicon(table), options.outPath);
-    return table;
-}
-//# sourceMappingURL=sub-venue-lexicon.js.map