@mailwoman/corpus 9.4.0 → 10.1.0

package/out/tools/postcode-triples.js

@@ -3,66 +3,30 @@
  * @license AGPL-3.0
  * @author Teffen Ellis, et al.
  *
- *   Extract `(postcode, locality, region, country)` tuples for the `trailing-region` slice.
+ * Extract `(postcode, locality, region, country)` tuples for the `trailing-region` recipe.
  *
- *   This step used to be a one-off. Its output survived — 17,908 rows at
- *   `$MAILWOMAN_DATA_ROOT/corpus/tuples/trailing-region-structured-tuples.jsonl` — but the code that produced it did
- *   not, so a note describing the join was the only record and it did not match what the databases contain. That is the
- *   reason this file exists: the slice is unbuildable for a new country without it.
- *
- *   ## Two sources, because one route does not reach every country
- *
- *   - **`postalcode-intl.db`** carries a real `parent_id` that resolves in the admin gazetteer. Measured share of rows
- *     with a parent: NL 97.5%, FR 90.7%, DE 66.1%, ES 34.9%, IT 27.4%; of those, 93.8–100% land on a `locality` or
- *     `localadmin`, and the region comes from that place's own ancestry. It is the ONLY slice with this — every
- *     `postalcode-geonames-*` and `postalcode-<cc>-overture.db` row reads `parent_id = 0`.
- *   - **GeoNames postal exports** carry the place and admin1 NAMES in columns 3 and 4, so there is nothing to join.
- *     `mailwoman corpus fetch geonames-postal` puts them on disk.
- *
- *   A nearest-locality-centroid join was measured as the general fallback and REJECTED: scored against the `parent_id`
- *   truth, it agreed NL 81.1% / DE 44.7% / ES 35.1% / FR 29.0% / IT 13.5%. A locality's centroid sits at its middle, so
- *   an edge postcode is routinely nearer a neighbouring town's centroid. Do not reach for it again.
- *
- *   ## The hub cap is PER COUNTRY, because a pooled one is a mixture
- *
- *   A few localities act as catch-all parents — `Schwedt/Oder` claims 9,222 DE postcodes against a DE median of 1. Left
- *   in, a handful of places dominate the slice. But the distribution differs so much by country that one threshold is
- *   not one rule: a p99 pooled across the five countries lands at 522, which keeps 100% of ES and IT, 83% of FR, 49% of
- *   NL and 19% of DE.
- *
- *   So the bound is a QUOTA and not a threshold. A threshold DELETES a locality that exceeds it, which removes exactly
- *   the largest cities — the places a parser most needs to have seen. A quota keeps every locality and bounds how many
- *   of its postcodes ride along, which is the balance the cap was reaching for without the deletion.
- *
- *   ## Placement is data, not a formatting choice
- *
- *   Each tuple is stamped with its country's {@link PostcodePlacement}. The same digits change tag with position, so a
- *   tuple that does not carry its placement teaches whichever convention the recipe happens to default to — see the
- *   recipe's header for the measurement.
+ * Each tuple is stamped with its country's {@link PostcodePlacement}, because the same digits change tag with position.
  */
+import { officialLanguagesAlpha3, regionLanguagesAlpha3 } from "@mailwoman/codex/country";
+import { foldName } from "@mailwoman/codex/normalize";
 import { dataRootPath } from "@mailwoman/core/data-root";
-import { pathExists } from "@mailwoman/core/fs/readers";
+import { assertPathExists } from "@mailwoman/core/fs/readers";
+import { wofDatabasePath } from "@mailwoman/resolver-wof-sqlite/paths";
 import { DatabaseClient } from "@mailwoman/sqlite/client";
-import { join } from "path-ts";
+import { PathBuilder } from "path-ts";
 import { TSVSpliterator } from "spliterator";
-import { GEONAMES_POSTAL_COLUMNS } from "#adapters/geonames-postal/adapter";
-import { escapeSQLString } from "#utils/parquet";
+import { GEONAMES_POSTAL_COLUMNS } from "#adapters/geonames/postal/adapter";
+import { escapeSQLString } from "#parquet/duckdb";
 /**
- * Where a country writes the postcode, and the locale tag its rows carry.
- *
- * A country is in this table only when a gauntlet board row ATTESTS its surface. An absent country is not an oversight
- * to be filled in by guessing: extracting it with the wrong placement teaches a convention that country does not use,
- * which is worse than not teaching it at all. `LEADING_POSTCODE_COUNTRIES` in `@mailwoman/neural`'s
- * `placetype-pair-prior.ts` draws the same line for the same reason.
+ * Records where a country writes the postcode.
+ * It also records the locale tag in its rows.
  *
- * AU and ZA are the worked examples of the bar. Both look like obvious additions and neither qualifies: the board's AU
- * rows are bare-city (`Melbourne`, `Sydney, Australia`) and carry no postcode at all, so nothing here says where AU
- * writes it; and ZA's `14 Long St, Green Point, Cape Town, 8001` carries no REGION, which this slice requires — a fact
- * its GeoNames export agrees with, at 100% place and 0% admin1.
+ * A country is in this table only when a gauntlet board row attests its surface, because extracting
+ * an absent country with the wrong placement teaches a convention that country does not use.
  */
 export const POSTCODE_CONVENTIONS = new Map([
-    // `Rue de l'Église, 3, 29217 Plougonvelin, Bretagne, France` and its siblings — `fr_structured`, `de_structured`,
-    // `es_structured`, `it_structured`, `pt_structured`, `mx_supermanzana`, `nl-op4-p-r-sloterdijk`.
+    // `Rue de l'Église, 3, 29217 Plougonvelin, Bretagne, France` and its siblings in
+    // the structured slices attest this leading placement.
     ["FR", { placement: "leading", locale: "fr-FR" }],
     ["DE", { placement: "leading", locale: "de-DE" }],
     ["ES", { placement: "leading", locale: "es-ES" }],
@@ -70,26 +34,30 @@ export const POSTCODE_CONVENTIONS = new Map([
     ["NL", { placement: "leading", locale: "nl-NL" }],
     ["PT", { placement: "leading", locale: "pt-PT" }],
     ["MX", { placement: "leading", locale: "es-MX" }],
-    // `…, Barcelona 6001, Anzoátegui, Venezuela` — the four `ve_city_postcode_trailing_state` rows. No postcode source
-    // on disk and GeoNames does not publish VE, so this entry currently yields nothing; it is here because the
-    // placement is what makes the absence legible.
+    // No postcode source on disk and GeoNames does not publish VE, so this entry yields no row.
+    // It is here because the placement makes the absence legible.
     ["VE", { placement: "after_locality", locale: "es-VE" }],
-    // `12 MG Road, Indiranagar, Bengaluru, Karnataka 560038, India` — three `in_*` rows, and `AGENTS.md` says the same
-    // ("en-IN is absent BECAUSE the PIN goes last"). The one trailing placement with real data behind it.
+    // `12 MG Road, Indiranagar, Bengaluru, Karnataka 560038, India` — three `in_*` rows,
+    // the one trailing placement with real data behind it.
     ["IN", { placement: "after_region", locale: "en-IN" }],
+    // Attested by the four `us_city_state_postcode` board rows; `localityColumn` keeps it from training counties.
+    ["US", { placement: "after_region", locale: "en-US", localityColumn: "place" }],
+    // The two `br_*` rows list locality, region and CEP in that order.
+    // The default `admin2` column is right here, because BR's export writes the
+    // municipality in column 3 and admin2 alike.
+    ["BR", { placement: "after_region", locale: "pt-BR" }],
 ]);
 /**
  * How many postcodes one locality may contribute.
  *
- * A quota rather than a cutoff — see the header. It bounds repetition without deleting a locality: a city with 9,222
- * postcodes contributes this many and stays in the slice.
+ * A quota rather than a cutoff: it bounds repetition without deleting a locality.
  */
 export const DEFAULT_LOCALITY_QUOTA = 24;
 /**
  * Take at most `quota` tuples per locality, in the order they arrive.
  *
- * Order matters and is the caller's to choose: both readers below walk their source in id / file order, which is stable
- * across runs, so the same quota selects the same rows.
+ * Both readers walk their source in id/file order.
+ * That order is stable across runs, so the same quota selects the same rows.
  */
 export function applyLocalityQuota(triples, quota = DEFAULT_LOCALITY_QUOTA) {
     const seen = new Map();
@@ -105,48 +73,101 @@ export function applyLocalityQuota(triples, quota = DEFAULT_LOCALITY_QUOTA) {
     return kept;
 }
 /**
- * Take at most `budget` tuples per COUNTRY, in the order they arrive.
+ * Take at most `budget` tuples per country, in the order they arrive.
  *
- * A per-locality quota bounds how often one place repeats; it cannot bound a country. IN has 128,152 distinct
- * localities, so even at a quota of ONE it contributes 63,533 rows against 39,790 from the other seven combined — the
- * slice would teach the trailing surface as an Indian fact rather than a general one, and at 103,323 rows it would take
- * 30% of an 8,000-step run's sample budget at three reps per row.
- *
- * Applied AFTER {@link applyLocalityQuota}, so a country's budget is spent on breadth (many localities) rather than on
- * one city's postcode list.
+ * Applied after {@link applyLocalityQuota}, and spent BY region in rounds so a cap
+ * smaller than the source still reaches every region rather than one corner.
+ * `subKey` adds a second round-robin dimension so the budget spreads across `(region, subKey)`.
  */
-export function applyCountryBudget(triples, budget) {
-    const spent = new Map();
-    const kept = [];
+export function applyCountryBudget(triples, budget, subKey) {
+    const byRegion = new Map();
+    const order = [];
     for (const triple of triples) {
         const cap = typeof budget === "number" ? budget : budget.get(triple.cc);
         if (cap === undefined)
             continue;
-        const n = spent.get(triple.cc) ?? 0;
-        if (n >= cap)
-            continue;
-        spent.set(triple.cc, n + 1);
-        kept.push(triple);
+        const key = `${triple.cc} ${triple.region ?? ""}${subKey ? ` ${subKey(triple)}` : ""}`;
+        const bucket = byRegion.get(key);
+        if (bucket) {
+            bucket.push(triple);
+        }
+        else {
+            byRegion.set(key, [triple]);
+            order.push(key);
+        }
+    }
+    const spent = new Map();
+    const kept = [];
+    const deepest = Math.max(0, ...[...byRegion.values()].map((bucket) => bucket.length));
+    for (let round = 0; round < deepest; round++) {
+        for (const key of order) {
+            const triple = byRegion.get(key)[round];
+            if (!triple)
+                continue;
+            const cap = typeof budget === "number" ? budget : budget.get(triple.cc);
+            const n = spent.get(triple.cc) ?? 0;
+            if (n >= cap)
+                continue;
+            spent.set(triple.cc, n + 1);
+            kept.push(triple);
+        }
     }
     return kept;
 }
 /**
- * Read triples out of `postalcode-intl.db` by following each postcode's `parent_id` into the admin gazetteer and that
- * place's ancestry to a region.
+ * The surfaces a region is written as: its preferred names in the official language(s),
+ * then the region's co-official languages, then the gazetteer's own `spr.name`
+ * (the English exonym), deduplicated with order kept.
+ *
+ * An input that uses only `spr.name` teaches stripped exonyms (`Balearic Islands`, `Cordoba`)
+ * against the forms a user writes.
+ */
+export function regionWrittenForms(sprName, names) {
+    const out = [];
+    for (const raw of [...names.official, ...names.coOfficial, ...sprName.split(BILINGUAL_JOINED)]) {
+        const name = raw.trim().replace(PROVINCE_GENERIC, "");
+        if (name && !out.includes(name)) {
+            out.push(name);
+        }
+    }
+    return out;
+}
+/**
+ * The separator Who's On First joins a region's two co-official names with.
+ *
+ * Spaces are required on both sides, because a slash with no surrounding spaces appears inside single names.
+ */
+const BILINGUAL_JOINED = / \/ /;
+/**
+ * The provincial generic Who's On First keeps in a province's Catalan and Asturian
+ * preferred names, dropped so the surface dedupes against the Castilian one.
+ */
+const PROVINCE_GENERIC = /^prov[ií]ncia (?:de |d')/iu;
+/**
+ * The surface a locality is written as: the official-language preferred name matching
+ * `spr.name` folded, else the first official-language name, else `spr.name`.
+ */
+export function localityWrittenForm(sprName, names) {
+    const folded = foldName(sprName);
+    return names.official.find((name) => foldName(name) === folded) ?? names.official[0] ?? sprName;
+}
+/**
+ * Read triples out of `postalcode-intl.db` by following each postcode's `parent_id`
+ * into the admin gazetteer and that place's ancestry to a region.
  *
- * A postcode whose parent does not resolve, or whose parent has no region ancestor, is DROPPED rather than emitted with
- * a blank — the recipe already skips a tuple with no region, and a blank here would hide how much of the source is
- * actually reachable.
+ * A postcode whose parent does not resolve, or whose parent has no region ancestor,
+ * is dropped rather than emitted with a blank.
  */
 export async function readTriplesFromParentJoin(countries, options = {}) {
-    const postcodeDB = options.postcodeDB ?? String(dataRootPath("wof", "postalcode-intl.db"));
-    const adminDB = options.adminDB ?? String(dataRootPath("wof", "admin-global-priority-importance.db"));
-    if (!(await pathExists(postcodeDB)) || !(await pathExists(adminDB)))
-        return [];
+    const postcodeDB = options.postcodeDB ?? wofDatabasePath("postalcode-intl.db");
+    const adminDB = options.adminDB ?? wofDatabasePath("admin-global-priority-importance.db");
+    await assertPathExists(postcodeDB, `readTriplesFromParentJoin: no postcode database at ${postcodeDB}`);
+    await assertPathExists(adminDB, `readTriplesFromParentJoin: no admin gazetteer at ${adminDB}`);
     using db = new DatabaseClient(adminDB, { readOnly: true });
-    db.exec(`ATTACH DATABASE '${escapeSQLString(postcodeDB)}' AS pc`);
+    db.exec(`ATTACH DATABASE '${escapeSQLString(postcodeDB.toString())}' AS pc`);
     const statement = db.prepare(`
-		SELECT p.name AS postcode, a.name AS locality, r.name AS region, c.name AS country, p.country AS cc
+		SELECT p.name AS postcode, a.id AS locality_id, a.name AS locality, r.id AS region_id, r.name AS region,
+		       c.name AS country, p.country AS cc
 		FROM pc.spr p
 		JOIN spr a ON a.id = p.parent_id AND a.placetype IN ('locality', 'localadmin')
 		JOIN ancestors anc ON anc.id = a.id AND anc.ancestor_placetype = 'region'
@@ -156,44 +177,124 @@ export async function readTriplesFromParentJoin(countries, options = {}) {
 		WHERE p.country = ? AND p.parent_id > 0
 		ORDER BY p.id
 	`);
+    const surfaces = createSurfaceReader(db);
     const out = [];
     for (const cc of countries) {
         const convention = POSTCODE_CONVENTIONS.get(cc);
         if (!convention)
             continue;
         for (const row of statement.all(cc)) {
-            if (!row["postcode"] || !row["locality"] || !row["region"])
+            if (!row.postcode || !row.locality || !row.region)
                 continue;
-            out.push({
-                postcode: row["postcode"],
-                locality: row["locality"],
-                region: row["region"],
-                country: row["country"] ?? "",
-                cc,
-                locale: convention.locale,
-                postcodePlacement: convention.placement,
-            });
+            const locality = surfaces.locality(cc, row.locality_id, row.locality);
+            for (const region of surfaces.region(cc, row.region_id, row.region)) {
+                out.push({
+                    postcode: row.postcode,
+                    locality,
+                    region,
+                    country: row.country ?? "",
+                    cc,
+                    locale: convention.locale,
+                    postcodePlacement: convention.placement,
+                });
+            }
         }
     }
     return out;
 }
 /**
- * A predicate answering whether a name is a LOCALITY the admin gazetteer knows, for one country.
+ * Bind {@link regionWrittenForms} and {@link localityWrittenForm} to an open gazetteer.
  *
- * The parent-join reader gets this for free — its query restricts the parent to `locality`/`localadmin`, so every name
- * it emits is one by construction. The GeoNames reader has no such guarantee, and the gap is large enough to matter:
- * sampling 400 rows per country against the gazetteer, the share of GeoNames place names that name a locality we know
- * is PT 75%, IN 62%, **MX 41%**. The Mexican misses are colonias — `Zona Centro`, `San Fernando INFONAVIT`, `FOVISSSTE
- * 3a Sección` — and a row teaching one of those as `locality` trains the locality/dependent_locality boundary in the
- * wrong direction. Dropping the row instead costs coverage and teaches nothing false, which is the better of the two.
+ * The languages come from the codex instead of the names table's language list.
+ * The table's "preferred" name may use a language not spoken in the region and may refer to the parent.
+ */
+function createSurfaceReader(db) {
+    const preferredStatement = db.prepare(`SELECT language, name FROM names WHERE id = ? AND privateuse = 'preferred' AND language IS NOT NULL ORDER BY language, name`);
+    const preferredByID = new Map();
+    const preferredNames = (id) => {
+        let byLanguage = preferredByID.get(id);
+        if (!byLanguage) {
+            byLanguage = new Map();
+            for (const row of preferredStatement.all(id)) {
+                const list = byLanguage.get(row.language) ?? [];
+                list.push(row.name);
+                byLanguage.set(row.language, list);
+            }
+            preferredByID.set(id, byLanguage);
+        }
+        return byLanguage;
+    };
+    const namesIn = (byLanguage, languages) => languages.flatMap((language) => byLanguage.get(language) ?? []);
+    return {
+        region(cc, id, sprName) {
+            const officialLanguages = officialLanguagesAlpha3(cc);
+            const official = namesIn(preferredNames(id), officialLanguages);
+            // The co-official table is keyed by the region's name in the first official language.
+            // A region the names table has no such name for is looked up by its `spr.name`,
+            // which for a monolingual country is the same string.
+            const regionLanguages = regionLanguagesAlpha3(cc, official[0] ?? sprName);
+            const coOfficialLanguages = regionLanguages.filter((language) => !officialLanguages.includes(language));
+            return regionWrittenForms(sprName, { official, coOfficial: namesIn(preferredNames(id), coOfficialLanguages) });
+        },
+        locality(cc, id, sprName) {
+            const official = namesIn(preferredNames(id), officialLanguagesAlpha3(cc));
+            return localityWrittenForm(sprName, { official, coOfficial: [] });
+        },
+    };
+}
+/**
+ * Read `(locality, region, country)` pairs for a country straight from the
+ * admin gazetteer, with no postcode.
  *
- * Returns a predicate that answers `true` for everything when the gazetteer is not on disk, so a checkout without it
- * builds the same rows it did before rather than silently emitting none.
+ * Produces one pair per region surface, following {@link readTriplesFromParentJoin}.
+ * The reader synthesizes no postcode.
+ */
+export async function readPairsFromAdmin(countries, options = {}) {
+    const adminDB = options.adminDB ?? wofDatabasePath("admin-global-priority-importance.db");
+    await assertPathExists(adminDB, `readPairsFromAdmin: no admin gazetteer at ${adminDB}`);
+    using db = new DatabaseClient(adminDB, { readOnly: true });
+    const statement = db.prepare(`
+		SELECT a.id AS locality_id, a.name AS locality, r.id AS region_id, r.name AS region, c.name AS country
+		FROM spr a
+		JOIN ancestors anc ON anc.id = a.id AND anc.ancestor_placetype = 'region'
+		JOIN spr r ON r.id = anc.ancestor_id
+		JOIN ancestors cnc ON cnc.id = a.id AND cnc.ancestor_placetype = 'country'
+		JOIN spr c ON c.id = cnc.ancestor_id
+		WHERE a.country = ? AND a.placetype IN ('locality', 'localadmin') AND a.is_current != 0 AND a.is_deprecated = 0
+		ORDER BY a.id
+	`);
+    const surfaces = createSurfaceReader(db);
+    const out = [];
+    for (const cc of countries) {
+        const locale = options.locale?.(cc) ?? "und";
+        for (const row of statement.all(cc)) {
+            if (!row.locality || !row.region)
+                continue;
+            const locality = surfaces.locality(cc, row.locality_id, row.locality);
+            for (const region of surfaces.region(cc, row.region_id, row.region)) {
+                // A pair whose region repeats its locality adds no signal about the boundary this recipe measures.
+                // The recipe drops that pair, so filtering it here keeps the country budget for rows that remain.
+                if (region === locality)
+                    continue;
+                out.push({ locality, region, country: row.country ?? "", cc, locale });
+            }
+        }
+    }
+    return out;
+}
+/**
+ * A predicate answering whether a name is a locality the admin gazetteer knows, for one country.
+ *
+ * The parent-join reader supplies this field directly.
+ * The GeoNames reader requires this correction because treating a colonia as `locality`
+ * teaches the wrong locality/dependent_locality boundary.
+ *
+ * @throws When the gazetteer is not on disk, because a predicate that accepted every
+ * name would emit unfiltered rows as though the filter had run.
  */
 export async function createKnownLocalityCheck(country, adminDB) {
-    const path = adminDB ?? String(dataRootPath("wof", "admin-global-priority-importance.db"));
-    if (!(await pathExists(path)))
-        return () => true;
+    const path = adminDB ?? wofDatabasePath("admin-global-priority-importance.db");
+    await assertPathExists(path, `createKnownLocalityCheck: no admin gazetteer at ${path}`);
     using db = new DatabaseClient(path, { readOnly: true });
     const names = new Set();
     for (const row of db
@@ -203,68 +304,53 @@ export async function createKnownLocalityCheck(country, adminDB) {
             names.add(row.name.toLowerCase());
         }
     }
-    // An empty set means the gazetteer has no localities for this country at all, which is a coverage fact about the
-    // gazetteer rather than a verdict on the source — so check nothing rather than drop everything.
+    // An empty set is a coverage fact about the gazetteer rather than a verdict on the source,
+    // so accept every name rather than dropping everything.
     if (!names.size)
         return () => true;
     return (name) => names.has(name.toLowerCase());
 }
 /**
- * Read triples straight out of a GeoNames `<CC>.txt` export — no join, the names are columns 3 and 4.
- *
- * COLUMN 3 IS NOT THE LOCALITY. It is the finest-grained named place for the code, and for `560001` that is `Mahatma
- * Gandhi Road` — a STREET — while the city, `Bengaluru`, is column 5 (admin2). MX is the same shape (`Roma Norte` is a
- * colonia inside `Cuauhtémoc`) and so is PT (`Abrigada` inside `Alenquer`). Reading column 3 as the locality is how the
- * v4.8.0 slice came to teach street names as cities. So `admin2` is the locality, `admin1` the region, and column 3 the
- * DEPENDENT locality — which is also the left context the slice needs.
- *
- * WHICH COUNTRIES THIS READER CAN SERVE. It needs `admin2` (the city) and `admin1` (the region), and a country can
- * publish one without the other. Measured 2026-08-23 across the exports on disk:
- *
- * | country    | rows      | admin1 | admin2 | usable here                     |
- * | ---------- | --------: | -----: | -----: | ------------------------------- |
- * | PT, MX, IN | 145k–207k | 100%   | 100%   | yes                             |
- * | BD, LK     | 1.3k–1.8k | 100%   | 100%   | yes                             |
- * | PH         | 2,317     | 88%    | 88%    | yes, 88% of rows                |
- * | PK         | 2,563     | 100%   | **0%** | NO — no city column             |
- * | TH         | 903       | 100%   | **1%** | NO — effectively no city column |
- * | ID         | 81,058    | **0%** | 0%     | NO — no region either           |
- * | ZA         | 3,920     | **0%** | —      | NO                              |
+ * Read triples straight out of a GeoNames `<CC>.txt` export, with no join.
  *
- * A country in the NO rows yields zero from this reader, and that is the correct outcome rather than a gap to route
- * around: taking column 3 as the locality is what made the v4.8.0 slice train `Mahatma Gandhi Road` as a city. If one
- * of them is wanted, it needs a city column from somewhere else, not a relaxed mapping.
+ * `admin2` supplies the locality, `admin1` supplies the region and column 3 supplies the dependent locality.
+ * A locality value from column 3 taught `Mahatma Gandhi Road` as a city.
  *
- * NOT PUBLISHED AT ALL by GeoNames, checked the same day: VE, VN, NP, MM, KH. Those are acquisition questions, and for
- * VE specifically OpenAddresses 404s too — see the arc retrospective.
+ * A country whose export lacks admin1 or admin2 yields zero from this reader.
+ * That result is correct and needs no fallback.
  *
- * Three source properties a caller cannot see from a row count, all handled here. Hyphen-format countries publish each
- * code TWICE (`3750-000` and `3750000`, exactly 2.00× for PT and PL), so the first surface of a code wins and its twin
- * is dropped. Some countries populate the place but not admin1 — ZA is 100% place, 0% region — which yields nothing
- * this slice can use, so those rows are dropped rather than emitted with a blank region. And the "place name" is often
- * a SUB-locality, which {@link createKnownLocalityCheck} filters.
+ * Hyphen-format countries publish each code twice, so the first surface of a code wins.
  */
 export async function readTriplesFromGeonames(country, path, countryName, options = {}) {
     const convention = POSTCODE_CONVENTIONS.get(country);
-    if (!convention || !(await pathExists(path)))
+    if (!convention)
         return [];
+    await assertPathExists(path, `readTriplesFromGeonames: no GeoNames postal export at ${path}`);
     const isKnownLocality = options.isKnownLocality ?? (await createKnownLocalityCheck(country));
     const out = [];
     const seen = new Set();
+    // Which column the city sits in is a property of the country's export —
+    // see {@link GeonamesLocalityColumn}; the other column becomes the dependent locality,
+    // a sub-locality for PT/MX/IN and a county for the US.
+    const localityIsPlace = convention.localityColumn === "place";
     for await (const cells of TSVSpliterator.fromAsync(path, { header: false })) {
-        const postcode = (cells[GEONAMES_POSTAL_COLUMNS.postcode] ?? "").trim();
-        const dependentLocality = (cells[GEONAMES_POSTAL_COLUMNS.place] ?? "").trim();
-        const locality = (cells[GEONAMES_POSTAL_COLUMNS.admin2Name] ?? "").trim();
-        const region = (cells[GEONAMES_POSTAL_COLUMNS.admin1Name] ?? "").trim();
+        const postcode = cells[GEONAMES_POSTAL_COLUMNS.postcode] ?? "";
+        const place = cells[GEONAMES_POSTAL_COLUMNS.place] ?? "";
+        const admin2 = cells[GEONAMES_POSTAL_COLUMNS.admin2Name] ?? "";
+        const locality = localityIsPlace ? place : admin2;
+        const dependentLocality = localityIsPlace ? "" : place;
+        const region = cells[GEONAMES_POSTAL_COLUMNS.admin1Name] ?? "";
         if (!postcode || !locality || !region)
             continue;
-        // The check applies to the LOCALITY — admin2 — not to the fine-grained name, which is expected to be a street or
-        // a colonia and is emitted as the dependent locality rather than dropped.
+        // The check applies to the locality.
+        // For PT/MX/IN, the other column may hold a street or colonia.
+        // The reader emits that value as the dependent locality instead of dropping it.
         if (!isKnownLocality(locality))
             continue;
-        // A dependent locality that merely repeats its parent teaches a doubled segment, not a boundary.
+        // A dependent locality that merely repeats its parent teaches a doubled segment rather than a boundary.
         const dep = dependentLocality && dependentLocality !== locality ? dependentLocality : "";
-        // The bare twin of a punctuated code carries no new fact, and keeping both doubles the country's weight.
+        // The bare twin of a punctuated code adds no new fact.
+        // Both entries would double the country's weight.
         const key = `${postcode.replaceAll("-", "")} ${locality} ${dep}`;
         if (seen.has(key))
             continue;
@@ -286,7 +372,6 @@ export async function readTriplesFromGeonames(country, path, countryName, option
  * Resolve a GeoNames export path under the standard fetch out-root.
  */
 export function geonamesPostalPath(country, sourcesRoot) {
-    const root = sourcesRoot ?? String(dataRootPath("corpus", "sources"));
-    return join(root, "geonames-postal", `${country.toUpperCase()}.txt`);
+    return PathBuilder.from(sourcesRoot ?? dataRootPath("corpus", "sources"))("geonames-postal", `${country.toUpperCase()}.txt`);
 }
 //# sourceMappingURL=postcode-triples.js.map

package/out/tools/postcode-triples.js.map

@@ -1 +1 @@
-{"version":3,"file":"postcode-triples.js","sourceRoot":"","sources":["../../lib/tools/postcode-triples.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAyCG;AAEH,OAAO,EAAE,YAAY,EAAE,MAAM,2BAA2B,CAAA;AACxD,OAAO,EAAE,UAAU,EAAE,MAAM,4BAA4B,CAAA;AAEvD,OAAO,EAAE,cAAc,EAAE,MAAM,0BAA0B,CAAA;AACzD,OAAO,EAAE,IAAI,EAAE,MAAM,SAAS,CAAA;AAC9B,OAAO,EAAE,cAAc,EAAE,MAAM,aAAa,CAAA;AAE5C,OAAO,EAAE,uBAAuB,EAAE,MAAM,mCAAmC,CAAA;AAE3E,OAAO,EAAE,eAAe,EAAE,MAAM,gBAAgB,CAAA;AAsBhD;;;;;;;;;;;;GAYG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAA0E,IAAI,GAAG,CAAC;IAClH,kHAAkH;IAClH,iGAAiG;IACjG,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IACjD,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IACjD,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IACjD,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IACjD,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IACjD,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IACjD,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IACjD,mHAAmH;IACnH,2GAA2G;IAC3G,+CAA+C;IAC/C,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,gBAAgB,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IACxD,mHAAmH;IACnH,sGAAsG;IACtG,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,cAAc,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;CACtD,CAAC,CAAA;AAEF;;;;;GAKG;AACH,MAAM,CAAC,MAAM,sBAAsB,GAAG,EAAE,CAAA;AAExC;;;;;GAKG;AACH,MAAM,UAAU,kBAAkB,CACjC,OAAkC,EAClC,QAAgB,sBAAsB;IAEtC,MAAM,IAAI,GAAG,IAAI,GAAG,EAAkB,CAAA;IACtC,MAAM,IAAI,GAAqB,EAAE,CAAA;IAEjC,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;QAC9B,MAAM,GAAG,GAAG,GAAG,MAAM,CAAC,EAAE,IAAI,MAAM,CAAC,QAAQ,EAAE,CAAA;QAC7C,MAAM,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;QAE5B,IAAI,CAAC,IAAI,KAAK;YAAE,SAAQ;QAExB,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,GAAG,CAAC,CAAC,CAAA;QACpB,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;IAClB,CAAC;IAED,OAAO,IAAI,CAAA;AACZ,CAAC;AAED;;;;;;;;;;GAUG;AACH,MAAM,UAAU,kBAAkB,CACjC,OAAkC,EAClC,MAA4C;IAE5C,MAAM,KAAK,GAAG,IAAI,GAAG,EAAkB,CAAA;IACvC,MAAM,IAAI,GAAqB,EAAE,CAAA;IAEjC,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;QAC9B,MAAM,GAAG,GAAG,OAAO,MAAM,KAAK,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC,CAAA;QAEvE,IAAI,GAAG,KAAK,SAAS;YAAE,SAAQ;QAE/B,MAAM,CAAC,GAAG,KAAK,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC,IAAI,CAAC,CAAA;QAEnC,IAAI,CAAC,IAAI,GAAG;YAAE,SAAQ;QAEtB,KAAK,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,CAAC,CAAA;QAC3B,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;IAClB,CAAC;IAED,OAAO,IAAI,CAAA;AACZ,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,CAAC,KAAK,UAAU,yBAAyB,CAC9C,SAA4B,EAC5B,UAAqD,EAAE;IAEvD,MAAM,UAAU,GAAG,OAAO,CAAC,UAAU,IAAI,MAAM,CAAC,YAAY,CAAC,KAAK,EAAE,oBAAoB,CAAC,CAAC,CAAA;IAC1F,MAAM,OAAO,GAAG,OAAO,CAAC,OAAO,IAAI,MAAM,CAAC,YAAY,CAAC,KAAK,EAAE,qCAAqC,CAAC,CAAC,CAAA;IAErG,IAAI,CAAC,CAAC,MAAM,UAAU,CAAC,UAAU,CAAC,CAAC,IAAI,CAAC,CAAC,MAAM,UAAU,CAAC,OAAO,CAAC,CAAC;QAAE,OAAO,EAAE,CAAA;IAE9E,MAAM,EAAE,GAAG,IAAI,cAAc,CAAc,OAAO,EAAE,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,CAAA;IAEvE,EAAE,CAAC,IAAI,CAAC,oBAAoB,eAAe,CAAC,UAAU,CAAC,SAAS,CAAC,CAAA;IAEjE,MAAM,SAAS,GAAG,EAAE,CAAC,OAAO,CAAC;;;;;;;;;;EAU5B,CAAC,CAAA;IAEF,MAAM,GAAG,GAAqB,EAAE,CAAA;IAEhC,KAAK,MAAM,EAAE,IAAI,SAAS,EAAE,CAAC;QAC5B,MAAM,UAAU,GAAG,oBAAoB,CAAC,GAAG,CAAC,EAAE,CAAC,CAAA;QAE/C,IAAI,CAAC,UAAU;YAAE,SAAQ;QAEzB,KAAK,MAAM,GAAG,IAAI,SAAS,CAAC,GAAG,CAAC,EAAE,CAAkC,EAAE,CAAC;YACtE,IAAI,CAAC,GAAG,CAAC,UAAU,CAAC,IAAI,CAAC,GAAG,CAAC,UAAU,CAAC,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC;gBAAE,SAAQ;YAEpE,GAAG,CAAC,IAAI,CAAC;gBACR,QAAQ,EAAE,GAAG,CAAC,UAAU,CAAC;gBACzB,QAAQ,EAAE,GAAG,CAAC,UAAU,CAAC;gBACzB,MAAM,EAAE,GAAG,CAAC,QAAQ,CAAC;gBACrB,OAAO,EAAE,GAAG,CAAC,SAAS,CAAC,IAAI,EAAE;gBAC7B,EAAE;gBACF,MAAM,EAAE,UAAU,CAAC,MAAM;gBACzB,iBAAiB,EAAE,UAAU,CAAC,SAAS;aACvC,CAAC,CAAA;QACH,CAAC;IACF,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC;AAED;;;;;;;;;;;;GAYG;AACH,MAAM,CAAC,KAAK,UAAU,wBAAwB,CAAC,OAAe,EAAE,OAAgB;IAC/E,MAAM,IAAI,GAAG,OAAO,IAAI,MAAM,CAAC,YAAY,CAAC,KAAK,EAAE,qCAAqC,CAAC,CAAC,CAAA;IAE1F,IAAI,CAAC,CAAC,MAAM,UAAU,CAAC,IAAI,CAAC,CAAC;QAAE,OAAO,GAAG,EAAE,CAAC,IAAI,CAAA;IAEhD,MAAM,EAAE,GAAG,IAAI,cAAc,CAAc,IAAI,EAAE,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,CAAA;IAEpE,MAAM,KAAK,GAAG,IAAI,GAAG,EAAU,CAAA;IAE/B,KAAK,MAAM,GAAG,IAAI,EAAE;SAClB,OAAO,CAAC,oFAAoF,CAAC;SAC7F,GAAG,CAAC,OAAO,CAAmC,EAAE,CAAC;QAClD,IAAI,GAAG,CAAC,IAAI,EAAE,CAAC;YACd,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,WAAW,EAAE,CAAC,CAAA;QAClC,CAAC;IACF,CAAC;IAED,iHAAiH;IACjH,gGAAgG;IAChG,IAAI,CAAC,KAAK,CAAC,IAAI;QAAE,OAAO,GAAG,EAAE,CAAC,IAAI,CAAA;IAElC,OAAO,CAAC,IAAY,EAAE,EAAE,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,WAAW,EAAE,CAAC,CAAA;AACvD,CAAC;AAED;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAkCG;AACH,MAAM,CAAC,KAAK,UAAU,uBAAuB,CAC5C,OAAe,EACf,IAAY,EACZ,WAAmB,EACnB,UAA2D,EAAE;IAE7D,MAAM,UAAU,GAAG,oBAAoB,CAAC,GAAG,CAAC,OAAO,CAAC,CAAA;IAEpD,IAAI,CAAC,UAAU,IAAI,CAAC,CAAC,MAAM,UAAU,CAAC,IAAI,CAAC,CAAC;QAAE,OAAO,EAAE,CAAA;IAEvD,MAAM,eAAe,GAAG,OAAO,CAAC,eAAe,IAAI,CAAC,MAAM,wBAAwB,CAAC,OAAO,CAAC,CAAC,CAAA;IAC5F,MAAM,GAAG,GAAqB,EAAE,CAAA;IAChC,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAA;IAE9B,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,cAAc,CAAC,SAAS,CAAC,IAAI,EAAE,EAAE,MAAM,EAAE,KAAK,EAAE,CAA4B,EAAE,CAAC;QACxG,MAAM,QAAQ,GAAG,CAAC,KAAK,CAAC,uBAAuB,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;QACvE,MAAM,iBAAiB,GAAG,CAAC,KAAK,CAAC,uBAAuB,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;QAC7E,MAAM,QAAQ,GAAG,CAAC,KAAK,CAAC,uBAAuB,CAAC,UAAU,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;QACzE,MAAM,MAAM,GAAG,CAAC,KAAK,CAAC,uBAAuB,CAAC,UAAU,CAAC,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;QAEvE,IAAI,CAAC,QAAQ,IAAI,CAAC,QAAQ,IAAI,CAAC,MAAM;YAAE,SAAQ;QAE/C,iHAAiH;QACjH,0EAA0E;QAC1E,IAAI,CAAC,eAAe,CAAC,QAAQ,CAAC;YAAE,SAAQ;QAExC,iGAAiG;QACjG,MAAM,GAAG,GAAG,iBAAiB,IAAI,iBAAiB,KAAK,QAAQ,CAAC,CAAC,CAAC,iBAAiB,CAAC,CAAC,CAAC,EAAE,CAAA;QAExF,yGAAyG;QACzG,MAAM,GAAG,GAAG,GAAG,QAAQ,CAAC,UAAU,CAAC,GAAG,EAAE,EAAE,CAAC,IAAI,QAAQ,IAAI,GAAG,EAAE,CAAA;QAEhE,IAAI,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC;YAAE,SAAQ;QAE3B,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;QAEb,GAAG,CAAC,IAAI,CAAC;YACR,QAAQ;YACR,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,iBAAiB,EAAE,GAAG,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;YAC1C,QAAQ;YACR,MAAM;YACN,OAAO,EAAE,WAAW;YACpB,EAAE,EAAE,OAAO;YACX,MAAM,EAAE,UAAU,CAAC,MAAM;YACzB,iBAAiB,EAAE,UAAU,CAAC,SAAS;SACvC,CAAC,CAAA;IACH,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,kBAAkB,CAAC,OAAe,EAAE,WAAoB;IACvE,MAAM,IAAI,GAAG,WAAW,IAAI,MAAM,CAAC,YAAY,CAAC,QAAQ,EAAE,SAAS,CAAC,CAAC,CAAA;IAErE,OAAO,IAAI,CAAC,IAAI,EAAE,iBAAiB,EAAE,GAAG,OAAO,CAAC,WAAW,EAAE,MAAM,CAAC,CAAA;AACrE,CAAC"}
+{"version":3,"file":"postcode-triples.js","sourceRoot":"","sources":["../../lib/tools/postcode-triples.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAEH,OAAO,EAAE,uBAAuB,EAAE,qBAAqB,EAAE,MAAM,0BAA0B,CAAA;AACzF,OAAO,EAAE,QAAQ,EAAE,MAAM,4BAA4B,CAAA;AACrD,OAAO,EAAE,YAAY,EAAE,MAAM,2BAA2B,CAAA;AACxD,OAAO,EAAE,gBAAgB,EAAE,MAAM,4BAA4B,CAAA;AAC7D,OAAO,EAAE,eAAe,EAAE,MAAM,sCAAsC,CAAA;AAEtE,OAAO,EAAE,cAAc,EAAE,MAAM,0BAA0B,CAAA;AACzD,OAAO,EAAE,WAAW,EAAwB,MAAM,SAAS,CAAA;AAC3D,OAAO,EAAE,cAAc,EAAE,MAAM,aAAa,CAAA;AAE5C,OAAO,EAAE,uBAAuB,EAAE,MAAM,mCAAmC,CAAA;AAC3E,OAAO,EAAE,eAAe,EAAE,MAAM,iBAAiB,CAAA;AA+BjD;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAG7B,IAAI,GAAG,CAAC;IACX,iFAAiF;IACjF,uDAAuD;IACvD,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IACjD,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IACjD,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IACjD,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IACjD,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IACjD,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IACjD,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IACjD,4FAA4F;IAC5F,8DAA8D;IAC9D,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,gBAAgB,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IACxD,qFAAqF;IACrF,uDAAuD;IACvD,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,cAAc,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;IACtD,8GAA8G;IAC9G,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,cAAc,EAAE,MAAM,EAAE,OAAO,EAAE,cAAc,EAAE,OAAO,EAAE,CAAC;IAC/E,mEAAmE;IACnE,4EAA4E;IAC5E,6CAA6C;IAC7C,CAAC,IAAI,EAAE,EAAE,SAAS,EAAE,cAAc,EAAE,MAAM,EAAE,OAAO,EAAE,CAAC;CACtD,CAAC,CAAA;AAEF;;;;GAIG;AACH,MAAM,CAAC,MAAM,sBAAsB,GAAG,EAAE,CAAA;AAExC;;;;;GAKG;AACH,MAAM,UAAU,kBAAkB,CACjC,OAAqB,EACrB,KAAK,GAAW,sBAAsB;IAEtC,MAAM,IAAI,GAAG,IAAI,GAAG,EAAkB,CAAA;IACtC,MAAM,IAAI,GAAQ,EAAE,CAAA;IAEpB,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;QAC9B,MAAM,GAAG,GAAG,GAAG,MAAM,CAAC,EAAE,IAAI,MAAM,CAAC,QAAQ,EAAE,CAAA;QAC7C,MAAM,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;QAE5B,IAAI,CAAC,IAAI,KAAK;YAAE,SAAQ;QAExB,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,GAAG,CAAC,CAAC,CAAA;QACpB,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;IAClB,CAAC;IAED,OAAO,IAAI,CAAA;AACZ,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,kBAAkB,CACjC,OAAqB,EACrB,MAA4C,EAC5C,MAA8B;IAE9B,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAe,CAAA;IACvC,MAAM,KAAK,GAAa,EAAE,CAAA;IAE1B,KAAK,MAAM,MAAM,IAAI,OAAO,EAAE,CAAC;QAC9B,MAAM,GAAG,GAAG,OAAO,MAAM,KAAK,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC,CAAA;QAEvE,IAAI,GAAG,KAAK,SAAS;YAAE,SAAQ;QAE/B,MAAM,GAAG,GAAG,GAAG,MAAM,CAAC,EAAE,IAAI,MAAM,CAAC,MAAM,IAAI,EAAE,GAAG,MAAM,CAAC,CAAC,CAAC,IAAI,MAAM,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,EAAE,CAAA;QACtF,MAAM,MAAM,GAAG,QAAQ,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;QAEhC,IAAI,MAAM,EAAE,CAAC;YACZ,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;QACpB,CAAC;aAAM,CAAC;YACP,QAAQ,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,MAAM,CAAC,CAAC,CAAA;YAC3B,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;QAChB,CAAC;IACF,CAAC;IAED,MAAM,KAAK,GAAG,IAAI,GAAG,EAAkB,CAAA;IACvC,MAAM,IAAI,GAAQ,EAAE,CAAA;IACpB,MAAM,OAAO,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,GAAG,CAAC,GAAG,QAAQ,CAAC,MAAM,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAA;IAErF,KAAK,IAAI,KAAK,GAAG,CAAC,EAAE,KAAK,GAAG,OAAO,EAAE,KAAK,EAAE,EAAE,CAAC;QAC9C,KAAK,MAAM,GAAG,IAAI,KAAK,EAAE,CAAC;YACzB,MAAM,MAAM,GAAG,QAAQ,CAAC,GAAG,CAAC,GAAG,CAAE,CAAC,KAAK,CAAC,CAAA;YAExC,IAAI,CAAC,MAAM;gBAAE,SAAQ;YAErB,MAAM,GAAG,GAAG,OAAO,MAAM,KAAK,QAAQ,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,CAAE,CAAA;YACxE,MAAM,CAAC,GAAG,KAAK,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC,IAAI,CAAC,CAAA;YAEnC,IAAI,CAAC,IAAI,GAAG;gBAAE,SAAQ;YAEtB,KAAK,CAAC,GAAG,CAAC,MAAM,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,CAAC,CAAA;YAC3B,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;QAClB,CAAC;IACF,CAAC;IAED,OAAO,IAAI,CAAA;AACZ,CAAC;AAiBD;;;;;;;GAOG;AACH,MAAM,UAAU,kBAAkB,CAAC,OAAe,EAAE,KAAqB;IACxE,MAAM,GAAG,GAAa,EAAE,CAAA;IAExB,KAAK,MAAM,GAAG,IAAI,CAAC,GAAG,KAAK,CAAC,QAAQ,EAAE,GAAG,KAAK,CAAC,UAAU,EAAE,GAAG,OAAO,CAAC,KAAK,CAAC,gBAAgB,CAAC,CAAC,EAAE,CAAC;QAChG,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,EAAE,CAAC,OAAO,CAAC,gBAAgB,EAAE,EAAE,CAAC,CAAA;QAErD,IAAI,IAAI,IAAI,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;YACjC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACf,CAAC;IACF,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC;AAED;;;;GAIG;AACH,MAAM,gBAAgB,GAAG,MAAM,CAAA;AAE/B;;;GAGG;AACH,MAAM,gBAAgB,GAAG,4BAA4B,CAAA;AAErD;;;GAGG;AACH,MAAM,UAAU,mBAAmB,CAAC,OAAe,EAAE,KAAqB;IACzE,MAAM,MAAM,GAAG,QAAQ,CAAC,OAAO,CAAC,CAAA;IAEhC,OAAO,KAAK,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,QAAQ,CAAC,IAAI,CAAC,KAAK,MAAM,CAAC,IAAI,KAAK,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,OAAO,CAAA;AAChG,CAAC;AAED;;;;;;GAMG;AACH,MAAM,CAAC,KAAK,UAAU,yBAAyB,CAC9C,SAA4B,EAC5B,OAAO,GAAgE,EAAE;IAEzE,MAAM,UAAU,GAAG,OAAO,CAAC,UAAU,IAAI,eAAe,CAAC,oBAAoB,CAAC,CAAA;IAC9E,MAAM,OAAO,GAAG,OAAO,CAAC,OAAO,IAAI,eAAe,CAAC,qCAAqC,CAAC,CAAA;IAEzF,MAAM,gBAAgB,CAAC,UAAU,EAAE,sDAAsD,UAAU,EAAE,CAAC,CAAA;IACtG,MAAM,gBAAgB,CAAC,OAAO,EAAE,oDAAoD,OAAO,EAAE,CAAC,CAAA;IAE9F,MAAM,EAAE,GAAG,IAAI,cAAc,CAAc,OAAO,EAAE,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,CAAA;IAEvE,EAAE,CAAC,IAAI,CAAC,oBAAoB,eAAe,CAAC,UAAU,CAAC,QAAQ,EAAE,CAAC,SAAS,CAAC,CAAA;IAE5E,MAAM,SAAS,GAAG,EAAE,CAAC,OAAO,CAAC;;;;;;;;;;;EAW5B,CAAC,CAAA;IAEF,MAAM,QAAQ,GAAG,mBAAmB,CAAC,EAAE,CAAC,CAAA;IACxC,MAAM,GAAG,GAAqB,EAAE,CAAA;IAEhC,KAAK,MAAM,EAAE,IAAI,SAAS,EAAE,CAAC;QAC5B,MAAM,UAAU,GAAG,oBAAoB,CAAC,GAAG,CAAC,EAAE,CAAC,CAAA;QAE/C,IAAI,CAAC,UAAU;YAAE,SAAQ;QAEzB,KAAK,MAAM,GAAG,IAAI,SAAS,CAAC,GAAG,CAAC,EAAE,CAOhC,EAAE,CAAC;YACJ,IAAI,CAAC,GAAG,CAAC,QAAQ,IAAI,CAAC,GAAG,CAAC,QAAQ,IAAI,CAAC,GAAG,CAAC,MAAM;gBAAE,SAAQ;YAE3D,MAAM,QAAQ,GAAG,QAAQ,CAAC,QAAQ,CAAC,EAAE,EAAE,GAAG,CAAC,WAAW,EAAE,GAAG,CAAC,QAAQ,CAAC,CAAA;YAErE,KAAK,MAAM,MAAM,IAAI,QAAQ,CAAC,MAAM,CAAC,EAAE,EAAE,GAAG,CAAC,SAAS,EAAE,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC;gBACrE,GAAG,CAAC,IAAI,CAAC;oBACR,QAAQ,EAAE,GAAG,CAAC,QAAQ;oBACtB,QAAQ;oBACR,MAAM;oBACN,OAAO,EAAE,GAAG,CAAC,OAAO,IAAI,EAAE;oBAC1B,EAAE;oBACF,MAAM,EAAE,UAAU,CAAC,MAAM;oBACzB,iBAAiB,EAAE,UAAU,CAAC,SAAS;iBACvC,CAAC,CAAA;YACH,CAAC;QACF,CAAC;IACF,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC;AAmBD;;;;;GAKG;AACH,SAAS,mBAAmB,CAAC,EAA+B;IAC3D,MAAM,kBAAkB,GAAG,EAAE,CAAC,OAAO,CACpC,6HAA6H,CAC7H,CAAA;IAED,MAAM,aAAa,GAAG,IAAI,GAAG,EAAiC,CAAA;IAE9D,MAAM,cAAc,GAAG,CAAC,EAAU,EAAyB,EAAE;QAC5D,IAAI,UAAU,GAAG,aAAa,CAAC,GAAG,CAAC,EAAE,CAAC,CAAA;QAEtC,IAAI,CAAC,UAAU,EAAE,CAAC;YACjB,UAAU,GAAG,IAAI,GAAG,EAAE,CAAA;YAEtB,KAAK,MAAM,GAAG,IAAI,kBAAkB,CAAC,GAAG,CAAC,EAAE,CAA8C,EAAE,CAAC;gBAC3F,MAAM,IAAI,GAAG,UAAU,CAAC,GAAG,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAA;gBAE/C,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;gBACnB,UAAU,CAAC,GAAG,CAAC,GAAG,CAAC,QAAQ,EAAE,IAAI,CAAC,CAAA;YACnC,CAAC;YAED,aAAa,CAAC,GAAG,CAAC,EAAE,EAAE,UAAU,CAAC,CAAA;QAClC,CAAC;QAED,OAAO,UAAU,CAAA;IAClB,CAAC,CAAA;IAED,MAAM,OAAO,GAAG,CAAC,UAAiC,EAAE,SAA4B,EAAY,EAAE,CAC7F,SAAS,CAAC,OAAO,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,UAAU,CAAC,GAAG,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC,CAAA;IAEhE,OAAO;QACN,MAAM,CAAC,EAAE,EAAE,EAAE,EAAE,OAAO;YACrB,MAAM,iBAAiB,GAAG,uBAAuB,CAAC,EAAE,CAAC,CAAA;YACrD,MAAM,QAAQ,GAAG,OAAO,CAAC,cAAc,CAAC,EAAE,CAAC,EAAE,iBAAiB,CAAC,CAAA;YAC/D,sFAAsF;YACtF,gFAAgF;YAChF,sDAAsD;YACtD,MAAM,eAAe,GAAG,qBAAqB,CAAC,EAAE,EAAE,QAAQ,CAAC,CAAC,CAAC,IAAI,OAAO,CAAC,CAAA;YACzE,MAAM,mBAAmB,GAAG,eAAe,CAAC,MAAM,CAAC,CAAC,QAAQ,EAAE,EAAE,CAAC,CAAC,iBAAiB,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC,CAAA;YAEvG,OAAO,kBAAkB,CAAC,OAAO,EAAE,EAAE,QAAQ,EAAE,UAAU,EAAE,OAAO,CAAC,cAAc,CAAC,EAAE,CAAC,EAAE,mBAAmB,CAAC,EAAE,CAAC,CAAA;QAC/G,CAAC;QACD,QAAQ,CAAC,EAAE,EAAE,EAAE,EAAE,OAAO;YACvB,MAAM,QAAQ,GAAG,OAAO,CAAC,cAAc,CAAC,EAAE,CAAC,EAAE,uBAAuB,CAAC,EAAE,CAAC,CAAC,CAAA;YAEzE,OAAO,mBAAmB,CAAC,OAAO,EAAE,EAAE,QAAQ,EAAE,UAAU,EAAE,EAAE,EAAE,CAAC,CAAA;QAClE,CAAC;KACD,CAAA;AACF,CAAC;AAOD;;;;;;GAMG;AACH,MAAM,CAAC,KAAK,UAAU,kBAAkB,CACvC,SAA4B,EAC5B,OAAO,GAAmE,EAAE;IAE5E,MAAM,OAAO,GAAG,OAAO,CAAC,OAAO,IAAI,eAAe,CAAC,qCAAqC,CAAC,CAAA;IAEzF,MAAM,gBAAgB,CAAC,OAAO,EAAE,6CAA6C,OAAO,EAAE,CAAC,CAAA;IAEvF,MAAM,EAAE,GAAG,IAAI,cAAc,CAAc,OAAO,EAAE,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,CAAA;IAEvE,MAAM,SAAS,GAAG,EAAE,CAAC,OAAO,CAAC;;;;;;;;;EAS5B,CAAC,CAAA;IAEF,MAAM,QAAQ,GAAG,mBAAmB,CAAC,EAAE,CAAC,CAAA;IACxC,MAAM,GAAG,GAAgB,EAAE,CAAA;IAE3B,KAAK,MAAM,EAAE,IAAI,SAAS,EAAE,CAAC;QAC5B,MAAM,MAAM,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,CAAC,IAAI,KAAK,CAAA;QAE5C,KAAK,MAAM,GAAG,IAAI,SAAS,CAAC,GAAG,CAAC,EAAE,CAMhC,EAAE,CAAC;YACJ,IAAI,CAAC,GAAG,CAAC,QAAQ,IAAI,CAAC,GAAG,CAAC,MAAM;gBAAE,SAAQ;YAE1C,MAAM,QAAQ,GAAG,QAAQ,CAAC,QAAQ,CAAC,EAAE,EAAE,GAAG,CAAC,WAAW,EAAE,GAAG,CAAC,QAAQ,CAAC,CAAA;YAErE,KAAK,MAAM,MAAM,IAAI,QAAQ,CAAC,MAAM,CAAC,EAAE,EAAE,GAAG,CAAC,SAAS,EAAE,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC;gBACrE,mGAAmG;gBACnG,kGAAkG;gBAClG,IAAI,MAAM,KAAK,QAAQ;oBAAE,SAAQ;gBAEjC,GAAG,CAAC,IAAI,CAAC,EAAE,QAAQ,EAAE,MAAM,EAAE,OAAO,EAAE,GAAG,CAAC,OAAO,IAAI,EAAE,EAAE,EAAE,EAAE,MAAM,EAAE,CAAC,CAAA;YACvE,CAAC;QACF,CAAC;IACF,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC;AAED;;;;;;;;;GASG;AACH,MAAM,CAAC,KAAK,UAAU,wBAAwB,CAC7C,OAAe,EACf,OAAyB;IAEzB,MAAM,IAAI,GAAG,OAAO,IAAI,eAAe,CAAC,qCAAqC,CAAC,CAAA;IAE9E,MAAM,gBAAgB,CAAC,IAAI,EAAE,mDAAmD,IAAI,EAAE,CAAC,CAAA;IAEvF,MAAM,EAAE,GAAG,IAAI,cAAc,CAAc,IAAI,EAAE,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,CAAA;IAEpE,MAAM,KAAK,GAAG,IAAI,GAAG,EAAU,CAAA;IAE/B,KAAK,MAAM,GAAG,IAAI,EAAE;SAClB,OAAO,CAAC,oFAAoF,CAAC;SAC7F,GAAG,CAAC,OAAO,CAAmC,EAAE,CAAC;QAClD,IAAI,GAAG,CAAC,IAAI,EAAE,CAAC;YACd,KAAK,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,CAAC,WAAW,EAAE,CAAC,CAAA;QAClC,CAAC;IACF,CAAC;IAED,2FAA2F;IAC3F,wDAAwD;IACxD,IAAI,CAAC,KAAK,CAAC,IAAI;QAAE,OAAO,GAAG,EAAE,CAAC,IAAI,CAAA;IAElC,OAAO,CAAC,IAAY,EAAE,EAAE,CAAC,KAAK,CAAC,GAAG,CAAC,IAAI,CAAC,WAAW,EAAE,CAAC,CAAA;AACvD,CAAC;AAED;;;;;;;;;;GAUG;AACH,MAAM,CAAC,KAAK,UAAU,uBAAuB,CAC5C,OAAe,EACf,IAAqB,EACrB,WAAmB,EACnB,OAAO,GAAoD,EAAE;IAE7D,MAAM,UAAU,GAAG,oBAAoB,CAAC,GAAG,CAAC,OAAO,CAAC,CAAA;IAEpD,IAAI,CAAC,UAAU;QAAE,OAAO,EAAE,CAAA;IAE1B,MAAM,gBAAgB,CAAC,IAAI,EAAE,yDAAyD,IAAI,EAAE,CAAC,CAAA;IAE7F,MAAM,eAAe,GAAG,OAAO,CAAC,eAAe,IAAI,CAAC,MAAM,wBAAwB,CAAC,OAAO,CAAC,CAAC,CAAA;IAC5F,MAAM,GAAG,GAAqB,EAAE,CAAA;IAChC,MAAM,IAAI,GAAG,IAAI,GAAG,EAAU,CAAA;IAE9B,wEAAwE;IACxE,uFAAuF;IACvF,uDAAuD;IACvD,MAAM,eAAe,GAAG,UAAU,CAAC,cAAc,KAAK,OAAO,CAAA;IAE7D,IAAI,KAAK,EAAE,MAAM,KAAK,IAAI,cAAc,CAAC,SAAS,CAAC,IAAI,EAAE,EAAE,MAAM,EAAE,KAAK,EAAE,CAA4B,EAAE,CAAC;QACxG,MAAM,QAAQ,GAAG,KAAK,CAAC,uBAAuB,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAA;QAC9D,MAAM,KAAK,GAAG,KAAK,CAAC,uBAAuB,CAAC,KAAK,CAAC,IAAI,EAAE,CAAA;QACxD,MAAM,MAAM,GAAG,KAAK,CAAC,uBAAuB,CAAC,UAAU,CAAC,IAAI,EAAE,CAAA;QAC9D,MAAM,QAAQ,GAAG,eAAe,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,MAAM,CAAA;QACjD,MAAM,iBAAiB,GAAG,eAAe,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,KAAK,CAAA;QACtD,MAAM,MAAM,GAAG,KAAK,CAAC,uBAAuB,CAAC,UAAU,CAAC,IAAI,EAAE,CAAA;QAE9D,IAAI,CAAC,QAAQ,IAAI,CAAC,QAAQ,IAAI,CAAC,MAAM;YAAE,SAAQ;QAE/C,qCAAqC;QACrC,+DAA+D;QAC/D,gFAAgF;QAChF,IAAI,CAAC,eAAe,CAAC,QAAQ,CAAC;YAAE,SAAQ;QAExC,wGAAwG;QACxG,MAAM,GAAG,GAAG,iBAAiB,IAAI,iBAAiB,KAAK,QAAQ,CAAC,CAAC,CAAC,iBAAiB,CAAC,CAAC,CAAC,EAAE,CAAA;QAExF,uDAAuD;QACvD,kDAAkD;QAClD,MAAM,GAAG,GAAG,GAAG,QAAQ,CAAC,UAAU,CAAC,GAAG,EAAE,EAAE,CAAC,IAAI,QAAQ,IAAI,GAAG,EAAE,CAAA;QAEhE,IAAI,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC;YAAE,SAAQ;QAE3B,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,CAAA;QAEb,GAAG,CAAC,IAAI,CAAC;YACR,QAAQ;YACR,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,iBAAiB,EAAE,GAAG,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;YAC1C,QAAQ;YACR,MAAM;YACN,OAAO,EAAE,WAAW;YACpB,EAAE,EAAE,OAAO;YACX,MAAM,EAAE,UAAU,CAAC,MAAM;YACzB,iBAAiB,EAAE,UAAU,CAAC,SAAS;SACvC,CAAC,CAAA;IACH,CAAC;IAED,OAAO,GAAG,CAAA;AACX,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,kBAAkB,CAAC,OAAe,EAAE,WAA6B;IAChF,OAAO,WAAW,CAAC,IAAI,CAAC,WAAW,IAAI,YAAY,CAAC,QAAQ,EAAE,SAAS,CAAC,CAAC,CACxE,iBAAiB,EACjB,GAAG,OAAO,CAAC,WAAW,EAAE,MAAM,CAC9B,CAAA;AACF,CAAC"}