@mailwoman/corpus 9.3.0 → 10.0.0

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

@@ -0,0 +1,266 @@
+/**
+ * @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.
+ */
+/**
+ * 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 const SUBVENUE_PROMOTIONS = [
+    // ── wing ────────────────────────────────────────────────────────────────────────────────────────
+    // The designator the corpus task's board rests on (`West Wing` is the one modifier case that
+    // already parses), and the cleanest per-locale split in the whole exercise.
+    {
+        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 slice 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 ────────────────────────────────────────────────────────────────────────────────────────
+    // Wave 1 flagged `hall` as per-locale poison on a GB number alone. Two locales measured; both
+    // reject, and for DIFFERENT reasons, which is the part worth keeping.
+    {
+        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 and is deliberately NOT modifier-eligible. These decide its localized
+    // forms, and the two land on opposite sides for a reason worth keeping.
+    {
+        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 ────────────────────────────────────────────────────────────────────────────────────────
+    // New in wave 2, and the second clean per-locale split: promotable in en-GB, killed in en-US by a
+    // furniture chain whose name is exactly the designator+identifier shape.
+    {
+        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 already ships. These promote the LOCALIZED forms, which is what a non-English
+    // slice line needs and what wave 1 had none of.
+    {
+        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/tools/sub/venue/promotions.js.map

@@ -0,0 +1 @@
+{"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

@@ -0,0 +1,45 @@
+/**
+ * @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

@@ -0,0 +1 @@
+{"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.map

@@ -0,0 +1 @@
+{"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

@@ -0,0 +1,229 @@
+/**
+ * @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

@@ -0,0 +1 @@
+{"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"}

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

@@ -0,0 +1 @@
+{"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

@@ -0,0 +1,26 @@
+/**
+ * @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

@@ -0,0 +1 @@
+{"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"}