@mailwoman/corpus 9.4.0 → 10.1.0

package/out/recipes/sub/venue/sources.js

@@ -0,0 +1,419 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ */
+import { readLocalJSONFile } from "@mailwoman/core/fs/readers";
+import { resolveModulePath } from "@mailwoman/core/module/resolvers";
+import { sample } from "@mailwoman/core/random";
+import { escapeRegExp } from "@mailwoman/core/strings/regexp";
+import { upperFirst } from "@mailwoman/normalize/case";
+import { DatabaseClient } from "@mailwoman/sqlite/client";
+import { classifyIdentifier, readSubVenueJSONL } from "#subvenue/harvest";
+import { SUBVENUE_PROMOTIONS } from "#subvenue/promotions";
+import { weightedPick } from "#synthesizers/utils";
+/**
+ * Resolves the path of the packaged sub-venue lexicon.
+ */
+export function defaultLexiconPath() {
+    return resolveModulePath("@mailwoman/corpus/data/sub-venue-lexicon.json");
+}
+/**
+ * Reads the sub-venue lexicon and throws on invalid JSON.
+ */
+export async function readSubVenueLexicon(path = defaultLexiconPath()) {
+    return await readLocalJSONFile(path);
+}
+const MAX_VENUE_NAME_LENGTH = 44;
+const MIN_NAME_LENGTH = 4;
+const MAX_ATTESTED_TOKENS = 4;
+/**
+ * Reports whether a name has usable length and contains a letter.
+ * It also rejects route punctuation.
+ *
+ * The check also rejects a single lowercase token because it represents a code in this source.
+ */
+export function isCleanName(name) {
+    if (name.length < MIN_NAME_LENGTH || name.length > MAX_VENUE_NAME_LENGTH)
+        return false;
+    if (/[/;,:()[\]<>|]/.test(name))
+        return false;
+    if (!/\p{L}/u.test(name))
+        return false;
+    if (!name.includes(" ") && name === name.toLowerCase())
+        return false;
+    return true;
+}
+const NON_VENUE_HEAD_WORDS = new Set([
+    "street",
+    "road",
+    "lane",
+    "avenue",
+    "drive",
+    "way",
+    "close",
+    "opposite",
+    "opp",
+    "adj",
+    "adjacent",
+    "outside",
+    "near",
+    "nr",
+    "stop",
+    "rue",
+    "boulevard",
+    "chemin",
+    "impasse",
+    "allée",
+    "allee",
+    "route",
+    "quai",
+    "place",
+    "calle",
+    "carrer",
+    "avenida",
+    "avinguda",
+    "carretera",
+    "plaza",
+    "paseo",
+    "camino",
+    "straße",
+    "strasse",
+    "weg",
+    "platz",
+    "gasse",
+]);
+const STREET_TAIL_WORDS = new Set([
+    "street",
+    "road",
+    "lane",
+    "avenue",
+    "drive",
+    "close",
+    "crescent",
+    "mews",
+    "terrace",
+    "walk",
+    "way",
+    "court",
+    "gardens",
+    "grove",
+    "rise",
+    "row",
+    "parade",
+    "esplanade",
+]);
+const GERMAN_STREET_TAIL = /(?:straße|strasse|weg|platz|gasse|allee|ring|damm)$/;
+/**
+ * Reports whether a clean name can fill the venue slot.
+ *
+ * The check rejects names that start with a street or stop word or end with a street type.
+ */
+export function isVenueSlotName(name) {
+    if (!isCleanName(name))
+        return false;
+    const words = name.toLowerCase().match(/[\p{L}]+/gu) ?? [];
+    if (!words.length)
+        return false;
+    if (NON_VENUE_HEAD_WORDS.has(words[0]))
+        return false;
+    const tail = words.at(-1);
+    return !STREET_TAIL_WORDS.has(tail) && !GERMAN_STREET_TAIL.test(tail);
+}
+/**
+ * Reports whether a lowercased name contains the phrase at word boundaries.
+ *
+ * The boundary check keeps `gate` from matching inside `Briggate`.
+ */
+export function containsPhrase(lowerName, phrase) {
+    let from = 0;
+    for (;;) {
+        const at = lowerName.indexOf(phrase, from);
+        if (at === -1)
+            return false;
+        const before = at === 0 ? "" : lowerName[at - 1];
+        const after = lowerName[at + phrase.length] ?? "";
+        if (!/[\p{L}\p{N}]/u.test(before) && !/[\p{L}\p{N}]/u.test(after))
+            return true;
+        from = at + 1;
+    }
+}
+const USABLE_IDENTIFIER_SHAPES = new Set([
+    "digit",
+    "letter",
+    "letter-digit",
+    "digit-letter",
+    "range",
+]);
+const SIGN_IDENTIFIER_ATOM = /^(?:[0-9]{1,3}|[A-Za-z]|[A-Za-z][0-9]{1,3}|[0-9]{1,3}[A-Za-z]{1,2})$/;
+/**
+ * Reports whether a value is a short sign identifier, such as `12`, `B` or `A3`,
+ * or a range of two identifiers joined by `-` or `/`.
+ *
+ * Shape classification by itself accepts some values that this check rejects, so callers apply both.
+ */
+export function isSignIdentifier(value) {
+    const parts = value.split(/[/-]/);
+    if (parts.length > 2 || parts.some((p) => !p))
+        return false;
+    return parts.every((part) => SIGN_IDENTIFIER_ATOM.test(part));
+}
+/**
+ * Reports whether a lowercased name uses a promoted phrase as `<phrase> <identifier>`,
+ * or as an English `<modifier> <phrase>` when the designator allows it.
+ *
+ * A bare mention of the phrase returns `false` because it does not show that the name is a sub-venue.
+ */
+export function hasPromotedShape(lowerName, promoted, modifiers = [], english = false) {
+    if (!containsPhrase(lowerName, promoted.phrase))
+        return false;
+    const escaped = escapeRegExp(promoted.phrase);
+    const withFollower = new RegExp(`(?:^|[^\\p{L}\\p{N}])${escaped}\\s+(\\S+)`, "u");
+    const follower = withFollower.exec(lowerName)?.[1];
+    if (follower && USABLE_IDENTIFIER_SHAPES.has(classifyIdentifier(follower)) && isSignIdentifier(follower))
+        return true;
+    if (promoted.identifierRequired)
+        return false;
+    if (!english || !promoted.modifierEligible)
+        return false;
+    return modifiers.some((modifier) => new RegExp(`(?:^|[^\\p{L}\\p{N}])${modifier}\\s+${escaped}(?![\\p{L}\\p{N}])`, "u").test(lowerName));
+}
+/**
+ * Applies {@link hasPromotedShape} to names of at most four tokens.
+ */
+export function matchesPromotedShape(lowerName, promoted, modifiers = [], english = false) {
+    if ((lowerName.match(/\S+/g) ?? []).length > MAX_ATTESTED_TOKENS)
+        return false;
+    return hasPromotedShape(lowerName, promoted, modifiers, english);
+}
+/**
+ * Returns the designator surfaces that may produce `unit` positives in a locale.
+ *
+ * English locales get every shipped designator.
+ * Every locale gets its promoted phrases.
+ *
+ * The function removes any pair that the ledger rejects for the locale.
+ */
+export function promotedSurfacesFor(locale, lexicon, promotions = SUBVENUE_PROMOTIONS) {
+    const language = locale.split("-")[0];
+    const rejected = new Set(promotions.filter((p) => p.decision === "reject" && p.locale === locale).map((p) => `${p.designatorID}|${p.phrase}`));
+    const out = new Map();
+    if (language === "en") {
+        for (const designator of lexicon.designators) {
+            if (!designator.shipped)
+                continue;
+            const key = `${designator.id}|${designator.id}`;
+            if (rejected.has(key))
+                continue;
+            out.set(key, {
+                designatorID: designator.id,
+                phrase: designator.id,
+                surface: upperFirst(designator.id),
+                identifierRequired: false,
+                modifierEligible: designator.modifierEligible,
+            });
+        }
+    }
+    for (const promotion of promotions) {
+        if (promotion.decision !== "promote" || promotion.locale !== locale)
+            continue;
+        const designator = lexicon.designators.find((d) => d.id === promotion.designatorID);
+        const key = `${promotion.designatorID}|${promotion.phrase}`;
+        if (rejected.has(key))
+            continue;
+        out.set(key, {
+            designatorID: promotion.designatorID,
+            phrase: promotion.phrase,
+            surface: upperFirst(promotion.phrase),
+            identifierRequired: promotion.shape === "identifier-required",
+            modifierEligible: Boolean(designator?.modifierEligible) && promotion.shape !== "identifier-required",
+        });
+    }
+    return [...out.values()].toSorted((a, b) => a.phrase.localeCompare(b.phrase));
+}
+/**
+ * Returns the phrases that the promotion ledger rejects for a locale.
+ */
+export function rejectedPhrasesFor(locale, promotions = SUBVENUE_PROMOTIONS) {
+    return promotions.filter((p) => p.decision === "reject" && p.locale === locale).map((p) => p.phrase);
+}
+const POOLED_IDENTIFIER_DESIGNATORS = ["gate", "terminal", "campus"];
+const MIN_OWN_SHAPE_OBSERVATIONS = 20;
+/**
+ * Builds a region's identifier model from the lexicon's `identifierShapes`.
+ *
+ * The pooled fallback combines the gate, terminal and campus distributions.
+ */
+export function buildIdentifierModel(lexicon, region) {
+    const byDesignator = new Map();
+    const pooled = [];
+    for (const row of lexicon.identifierShapes) {
+        if (row.region !== region)
+            continue;
+        if (!USABLE_IDENTIFIER_SHAPES.has(row.shape))
+            continue;
+        const examples = row.examples.filter((e) => isSignIdentifier(e));
+        if (!examples.length)
+            continue;
+        const bucket = { shape: row.shape, observations: row.observations, examples };
+        const list = byDesignator.get(row.designatorID);
+        if (list) {
+            list.push(bucket);
+        }
+        else {
+            byDesignator.set(row.designatorID, [bucket]);
+        }
+        if (POOLED_IDENTIFIER_DESIGNATORS.includes(row.designatorID)) {
+            pooled.push(bucket);
+        }
+    }
+    return { byDesignator, pooled };
+}
+/**
+ * Samples an identifier for a designator, weighting shapes by observation count.
+ *
+ * A designator with fewer than 20 observations of its own uses the pooled distribution.
+ */
+export function sampleIdentifier(model, designatorID, random) {
+    const own = model.byDesignator.get(designatorID) ?? [];
+    const ownTotal = own.reduce((sum, b) => sum + b.observations, 0);
+    const buckets = ownTotal >= MIN_OWN_SHAPE_OBSERVATIONS ? own : model.pooled;
+    if (!buckets.length)
+        return null;
+    const bucket = weightedPick(buckets, random, (b) => b.observations, { inclusive: false });
+    return sample(bucket.examples, random);
+}
+/**
+ * Empty name pools for a source that a leg lacks.
+ */
+export const EMPTY_NAME_POOLS = {
+    venues: [],
+    attested: [],
+    rejectedVenues: [],
+    longerNames: [],
+    unpromotedShapes: [],
+};
+function isLongerProperName(low, name, designatorPhrases) {
+    return designatorPhrases.some((phrase) => containsPhrase(low, phrase) && !low.startsWith(phrase) && !/\d/.test(name));
+}
+/**
+ * Reads an OSM sub-venue extract and sorts its names into name pools.
+ */
+export async function readExtractPools(path, query) {
+    const rows = await readSubVenueJSONL(path);
+    const venues = new Set();
+    const attested = new Set();
+    const rejectedVenues = new Set();
+    const longerNames = new Set();
+    const unpromotedShapes = new Set();
+    for (const row of rows) {
+        const name = (row.name ?? "").trim();
+        if (!name || !isCleanName(name))
+            continue;
+        const low = name.toLowerCase();
+        const venueSlot = isVenueSlotName(name);
+        if (row.tier === "venue" && venueSlot) {
+            venues.add(name);
+        }
+        for (const promoted of query.promoted) {
+            if (!containsPhrase(low, promoted.phrase))
+                continue;
+            if (!hasPromotedShape(low, promoted, query.modifiers, query.english)) {
+                if (venueSlot) {
+                    unpromotedShapes.add(name);
+                }
+            }
+            else if (matchesPromotedShape(low, promoted, query.modifiers, query.english)) {
+                attested.add(name);
+            }
+        }
+        if (!venueSlot)
+            continue;
+        if (query.rejectedPhrases.some((phrase) => containsPhrase(low, phrase))) {
+            rejectedVenues.add(name);
+        }
+        if (isLongerProperName(low, name, query.designatorPhrases)) {
+            longerNames.add(name);
+        }
+    }
+    return {
+        venues: [...venues],
+        attested: [...attested],
+        rejectedVenues: [...rejectedVenues],
+        longerNames: [...longerNames],
+        unpromotedShapes: [...unpromotedShapes],
+    };
+}
+const POI_VENUE_CATEGORIES = [
+    "airport",
+    "airport_terminal",
+    "train_station",
+    "hospital",
+    "college_university",
+];
+const POI_CONFOUND_CATEGORIES = [
+    "shoe_store",
+    "furniture_store",
+    "home_decor_store",
+    "town_hall",
+    "martial_arts_club",
+    "chicken_wings_restaurant",
+    "fire_station",
+];
+/**
+ * Reads the venue and confound name pools for one country from `poi.db`.
+ *
+ * The database covers only a few countries, so an empty result may reflect missing coverage.
+ */
+export function readPOIPools(dbPath, country, query) {
+    using db = new DatabaseClient(dbPath, { readOnly: true });
+    const codes = db.prepare("select id, category from poi_category_codes").all();
+    const byName = new Map(codes.map((c) => [c.category, c.id]));
+    const venueIDs = POI_VENUE_CATEGORIES.map((c) => byName.get(c)).filter((id) => id != null);
+    const confoundIDs = POI_CONFOUND_CATEGORIES.map((c) => byName.get(c)).filter((id) => id != null);
+    const wanted = [...venueIDs, ...confoundIDs];
+    if (!wanted.length)
+        throw new Error(`poi.db at ${dbPath} has none of the expected categories`);
+    const rows = db
+        .prepare(`select name, category_id from poi where country = ? and name is not null and category_id in (${wanted.map(() => "?").join(",")})`)
+        .all(country, ...wanted);
+    const venueSet = new Set(venueIDs);
+    const venues = new Set();
+    const rejectedVenues = new Set();
+    const longerNames = new Set();
+    for (const row of rows) {
+        const name = row.name.trim();
+        if (!isVenueSlotName(name))
+            continue;
+        const low = name.toLowerCase();
+        if (venueSet.has(row.category_id)) {
+            venues.add(name);
+        }
+        if (query.rejectedPhrases.some((phrase) => containsPhrase(low, phrase))) {
+            rejectedVenues.add(name);
+        }
+        if (isLongerProperName(low, name, query.designatorPhrases)) {
+            longerNames.add(name);
+        }
+    }
+    return {
+        venues: [...venues],
+        attested: [],
+        rejectedVenues: [...rejectedVenues],
+        longerNames: [...longerNames],
+        unpromotedShapes: [],
+    };
+}
+/**
+ * Concatenates two sources' name pools.
+ */
+export function mergeNamePools(a, b) {
+    return {
+        venues: [...a.venues, ...b.venues],
+        attested: [...a.attested, ...b.attested],
+        rejectedVenues: [...a.rejectedVenues, ...b.rejectedVenues],
+        longerNames: [...a.longerNames, ...b.longerNames],
+        unpromotedShapes: [...a.unpromotedShapes, ...b.unpromotedShapes],
+    };
+}
+//# sourceMappingURL=sources.js.map

package/out/recipes/sub/venue/sources.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"sources.js","sourceRoot":"","sources":["../../../../lib/recipes/sub/venue/sources.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAEH,OAAO,EAAE,iBAAiB,EAAE,MAAM,4BAA4B,CAAA;AAC9D,OAAO,EAAE,iBAAiB,EAAE,MAAM,kCAAkC,CAAA;AACpE,OAAO,EAAE,MAAM,EAAE,MAAM,wBAAwB,CAAA;AAC/C,OAAO,EAAE,YAAY,EAAE,MAAM,gCAAgC,CAAA;AAC7D,OAAO,EAAE,UAAU,EAAE,MAAM,2BAA2B,CAAA;AAEtD,OAAO,EAAE,cAAc,EAAE,MAAM,0BAA0B,CAAA;AAGzD,OAAO,EAAE,kBAAkB,EAAE,iBAAiB,EAAE,MAAM,mBAAmB,CAAA;AACzE,OAAO,EAAE,mBAAmB,EAA0B,MAAM,sBAAsB,CAAA;AAGlF,OAAO,EAAE,YAAY,EAAE,MAAM,qBAAqB,CAAA;AAElD;;GAEG;AACH,MAAM,UAAU,kBAAkB;IACjC,OAAO,iBAAiB,CAAC,+CAA+C,CAAC,CAAA;AAC1E,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,KAAK,UAAU,mBAAmB,CAAC,IAAI,GAAW,kBAAkB,EAAE;IAC5E,OAAO,MAAM,iBAAiB,CAAuB,IAAI,CAAC,CAAA;AAC3D,CAAC;AAED,MAAM,qBAAqB,GAAG,EAAE,CAAA;AAEhC,MAAM,eAAe,GAAG,CAAC,CAAA;AAEzB,MAAM,mBAAmB,GAAG,CAAC,CAAA;AAE7B;;;;;GAKG;AACH,MAAM,UAAU,WAAW,CAAC,IAAY;IACvC,IAAI,IAAI,CAAC,MAAM,GAAG,eAAe,IAAI,IAAI,CAAC,MAAM,GAAG,qBAAqB;QAAE,OAAO,KAAK,CAAA;IAEtF,IAAI,gBAAgB,CAAC,IAAI,CAAC,IAAI,CAAC;QAAE,OAAO,KAAK,CAAA;IAE7C,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC;QAAE,OAAO,KAAK,CAAA;IAEtC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,IAAI,KAAK,IAAI,CAAC,WAAW,EAAE;QAAE,OAAO,KAAK,CAAA;IAEpE,OAAO,IAAI,CAAA;AACZ,CAAC;AAED,MAAM,oBAAoB,GAAwB,IAAI,GAAG,CAAC;IACzD,QAAQ;IACR,MAAM;IACN,MAAM;IACN,QAAQ;IACR,OAAO;IACP,KAAK;IACL,OAAO;IACP,UAAU;IACV,KAAK;IACL,KAAK;IACL,UAAU;IACV,SAAS;IACT,MAAM;IACN,IAAI;IACJ,MAAM;IAEN,KAAK;IACL,WAAW;IACX,QAAQ;IACR,SAAS;IACT,OAAO;IACP,OAAO;IACP,OAAO;IACP,MAAM;IACN,OAAO;IAEP,OAAO;IACP,QAAQ;IACR,SAAS;IACT,UAAU;IACV,WAAW;IACX,OAAO;IACP,OAAO;IACP,QAAQ;IAER,QAAQ;IACR,SAAS;IACT,KAAK;IACL,OAAO;IACP,OAAO;CACP,CAAC,CAAA;AAEF,MAAM,iBAAiB,GAAwB,IAAI,GAAG,CAAC;IACtD,QAAQ;IACR,MAAM;IACN,MAAM;IACN,QAAQ;IACR,OAAO;IACP,OAAO;IACP,UAAU;IACV,MAAM;IACN,SAAS;IACT,MAAM;IACN,KAAK;IACL,OAAO;IACP,SAAS;IACT,OAAO;IACP,MAAM;IACN,KAAK;IACL,QAAQ;IACR,WAAW;CACX,CAAC,CAAA;AAEF,MAAM,kBAAkB,GAAG,qDAAqD,CAAA;AAEhF;;;;GAIG;AACH,MAAM,UAAU,eAAe,CAAC,IAAY;IAC3C,IAAI,CAAC,WAAW,CAAC,IAAI,CAAC;QAAE,OAAO,KAAK,CAAA;IACpC,MAAM,KAAK,GAAG,IAAI,CAAC,WAAW,EAAE,CAAC,KAAK,CAAC,YAAY,CAAC,IAAI,EAAE,CAAA;IAE1D,IAAI,CAAC,KAAK,CAAC,MAAM;QAAE,OAAO,KAAK,CAAA;IAE/B,IAAI,oBAAoB,CAAC,GAAG,CAAC,KAAK,CAAC,CAAC,CAAE,CAAC;QAAE,OAAO,KAAK,CAAA;IACrD,MAAM,IAAI,GAAG,KAAK,CAAC,EAAE,CAAC,CAAC,CAAC,CAAE,CAAA;IAE1B,OAAO,CAAC,iBAAiB,CAAC,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,kBAAkB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;AACtE,CAAC;AAiBD;;;;GAIG;AACH,MAAM,UAAU,cAAc,CAAC,SAAiB,EAAE,MAAc;IAC/D,IAAI,IAAI,GAAG,CAAC,CAAA;IAEZ,SAAS,CAAC;QACT,MAAM,EAAE,GAAG,SAAS,CAAC,OAAO,CAAC,MAAM,EAAE,IAAI,CAAC,CAAA;QAE1C,IAAI,EAAE,KAAK,CAAC,CAAC;YAAE,OAAO,KAAK,CAAA;QAC3B,MAAM,MAAM,GAAG,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,SAAS,CAAC,EAAE,GAAG,CAAC,CAAE,CAAA;QACjD,MAAM,KAAK,GAAG,SAAS,CAAC,EAAE,GAAG,MAAM,CAAC,MAAM,CAAC,IAAI,EAAE,CAAA;QAEjD,IAAI,CAAC,eAAe,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,eAAe,CAAC,IAAI,CAAC,KAAK,CAAC;YAAE,OAAO,IAAI,CAAA;QAC9E,IAAI,GAAG,EAAE,GAAG,CAAC,CAAA;IACd,CAAC;AACF,CAAC;AAED,MAAM,wBAAwB,GAAwB,IAAI,GAAG,CAAC;IAC7D,OAAO;IACP,QAAQ;IACR,cAAc;IACd,cAAc;IACd,OAAO;CACP,CAAC,CAAA;AAEF,MAAM,oBAAoB,GAAG,sEAAsE,CAAA;AAEnG;;;;;GAKG;AACH,MAAM,UAAU,gBAAgB,CAAC,KAAa;IAC7C,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,MAAM,CAAC,CAAA;IAEjC,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC;QAAE,OAAO,KAAK,CAAA;IAE3D,OAAO,KAAK,CAAC,KAAK,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,oBAAoB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAA;AAC9D,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,gBAAgB,CAC/B,SAAiB,EACjB,QAAyB,EACzB,SAAS,GAAsB,EAAE,EACjC,OAAO,GAAG,KAAK;IAEf,IAAI,CAAC,cAAc,CAAC,SAAS,EAAE,QAAQ,CAAC,MAAM,CAAC;QAAE,OAAO,KAAK,CAAA;IAE7D,MAAM,OAAO,GAAG,YAAY,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAA;IAC7C,MAAM,YAAY,GAAG,IAAI,MAAM,CAAC,wBAAwB,OAAO,YAAY,EAAE,GAAG,CAAC,CAAA;IACjF,MAAM,QAAQ,GAAG,YAAY,CAAC,IAAI,CAAC,SAAS,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAElD,IAAI,QAAQ,IAAI,wBAAwB,CAAC,GAAG,CAAC,kBAAkB,CAAC,QAAQ,CAAC,CAAC,IAAI,gBAAgB,CAAC,QAAQ,CAAC;QAAE,OAAO,IAAI,CAAA;IAErH,IAAI,QAAQ,CAAC,kBAAkB;QAAE,OAAO,KAAK,CAAA;IAE7C,IAAI,CAAC,OAAO,IAAI,CAAC,QAAQ,CAAC,gBAAgB;QAAE,OAAO,KAAK,CAAA;IAExD,OAAO,SAAS,CAAC,IAAI,CAAC,CAAC,QAAQ,EAAE,EAAE,CAClC,IAAI,MAAM,CAAC,wBAAwB,QAAQ,OAAO,OAAO,oBAAoB,EAAE,GAAG,CAAC,CAAC,IAAI,CAAC,SAAS,CAAC,CACnG,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,oBAAoB,CACnC,SAAiB,EACjB,QAAyB,EACzB,SAAS,GAAsB,EAAE,EACjC,OAAO,GAAG,KAAK;IAEf,IAAI,CAAC,SAAS,CAAC,KAAK,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM,GAAG,mBAAmB;QAAE,OAAO,KAAK,CAAA;IAE9E,OAAO,gBAAgB,CAAC,SAAS,EAAE,QAAQ,EAAE,SAAS,EAAE,OAAO,CAAC,CAAA;AACjE,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,mBAAmB,CAClC,MAAc,EACd,OAA6B,EAC7B,UAAU,GAAiC,mBAAmB;IAE9D,MAAM,QAAQ,GAAG,MAAM,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAE,CAAA;IAEtC,MAAM,QAAQ,GAAG,IAAI,GAAG,CACvB,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,QAAQ,KAAK,QAAQ,IAAI,CAAC,CAAC,MAAM,KAAK,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,CAAC,YAAY,IAAI,CAAC,CAAC,MAAM,EAAE,CAAC,CACpH,CAAA;IAED,MAAM,GAAG,GAAG,IAAI,GAAG,EAA2B,CAAA;IAE9C,IAAI,QAAQ,KAAK,IAAI,EAAE,CAAC;QACvB,KAAK,MAAM,UAAU,IAAI,OAAO,CAAC,WAAW,EAAE,CAAC;YAC9C,IAAI,CAAC,UAAU,CAAC,OAAO;gBAAE,SAAQ;YACjC,MAAM,GAAG,GAAG,GAAG,UAAU,CAAC,EAAE,IAAI,UAAU,CAAC,EAAE,EAAE,CAAA;YAE/C,IAAI,QAAQ,CAAC,GAAG,CAAC,GAAG,CAAC;gBAAE,SAAQ;YAE/B,GAAG,CAAC,GAAG,CAAC,GAAG,EAAE;gBACZ,YAAY,EAAE,UAAU,CAAC,EAAE;gBAC3B,MAAM,EAAE,UAAU,CAAC,EAAE;gBACrB,OAAO,EAAE,UAAU,CAAC,UAAU,CAAC,EAAE,CAAC;gBAClC,kBAAkB,EAAE,KAAK;gBACzB,gBAAgB,EAAE,UAAU,CAAC,gBAAgB;aAC7C,CAAC,CAAA;QACH,CAAC;IACF,CAAC;IAED,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE,CAAC;QACpC,IAAI,SAAS,CAAC,QAAQ,KAAK,SAAS,IAAI,SAAS,CAAC,MAAM,KAAK,MAAM;YAAE,SAAQ;QAC7E,MAAM,UAAU,GAAG,OAAO,CAAC,WAAW,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,KAAK,SAAS,CAAC,YAAY,CAAC,CAAA;QACnF,MAAM,GAAG,GAAG,GAAG,SAAS,CAAC,YAAY,IAAI,SAAS,CAAC,MAAM,EAAE,CAAA;QAE3D,IAAI,QAAQ,CAAC,GAAG,CAAC,GAAG,CAAC;YAAE,SAAQ;QAE/B,GAAG,CAAC,GAAG,CAAC,GAAG,EAAE;YACZ,YAAY,EAAE,SAAS,CAAC,YAAY;YACpC,MAAM,EAAE,SAAS,CAAC,MAAM;YACxB,OAAO,EAAE,UAAU,CAAC,SAAS,CAAC,MAAM,CAAC;YACrC,kBAAkB,EAAE,SAAS,CAAC,KAAK,KAAK,qBAAqB;YAE7D,gBAAgB,EAAE,OAAO,CAAC,UAAU,EAAE,gBAAgB,CAAC,IAAI,SAAS,CAAC,KAAK,KAAK,qBAAqB;SACpG,CAAC,CAAA;IACH,CAAC;IAED,OAAO,CAAC,GAAG,GAAG,CAAC,MAAM,EAAE,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,aAAa,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAA;AAC9E,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,kBAAkB,CACjC,MAAc,EACd,UAAU,GAAiC,mBAAmB;IAE9D,OAAO,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,QAAQ,KAAK,QAAQ,IAAI,CAAC,CAAC,MAAM,KAAK,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,CAAA;AACrG,CAAC;AAgBD,MAAM,6BAA6B,GAAsB,CAAC,MAAM,EAAE,UAAU,EAAE,QAAQ,CAAC,CAAA;AAEvF,MAAM,0BAA0B,GAAG,EAAE,CAAA;AAErC;;;;GAIG;AACH,MAAM,UAAU,oBAAoB,CAAC,OAA6B,EAAE,MAAc;IACjF,MAAM,YAAY,GAAG,IAAI,GAAG,EAAyB,CAAA;IACrD,MAAM,MAAM,GAAkB,EAAE,CAAA;IAEhC,KAAK,MAAM,GAAG,IAAI,OAAO,CAAC,gBAAgB,EAAE,CAAC;QAC5C,IAAI,GAAG,CAAC,MAAM,KAAK,MAAM;YAAE,SAAQ;QAEnC,IAAI,CAAC,wBAAwB,CAAC,GAAG,CAAC,GAAG,CAAC,KAAK,CAAC;YAAE,SAAQ;QACtD,MAAM,QAAQ,GAAG,GAAG,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,gBAAgB,CAAC,CAAC,CAAC,CAAC,CAAA;QAEhE,IAAI,CAAC,QAAQ,CAAC,MAAM;YAAE,SAAQ;QAC9B,MAAM,MAAM,GAAgB,EAAE,KAAK,EAAE,GAAG,CAAC,KAAK,EAAE,YAAY,EAAE,GAAG,CAAC,YAAY,EAAE,QAAQ,EAAE,CAAA;QAC1F,MAAM,IAAI,GAAG,YAAY,CAAC,GAAG,CAAC,GAAG,CAAC,YAAY,CAAC,CAAA;QAE/C,IAAI,IAAI,EAAE,CAAC;YACV,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;QAClB,CAAC;aAAM,CAAC;YACP,YAAY,CAAC,GAAG,CAAC,GAAG,CAAC,YAAY,EAAE,CAAC,MAAM,CAAC,CAAC,CAAA;QAC7C,CAAC;QAED,IAAI,6BAA6B,CAAC,QAAQ,CAAC,GAAG,CAAC,YAAY,CAAC,EAAE,CAAC;YAC9D,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;QACpB,CAAC;IACF,CAAC;IAED,OAAO,EAAE,YAAY,EAAE,MAAM,EAAE,CAAA;AAChC,CAAC;AAED;;;;GAIG;AACH,MAAM,UAAU,gBAAgB,CAAC,KAAsB,EAAE,YAAoB,EAAE,MAAoB;IAClG,MAAM,GAAG,GAAG,KAAK,CAAC,YAAY,CAAC,GAAG,CAAC,YAAY,CAAC,IAAI,EAAE,CAAA;IACtD,MAAM,QAAQ,GAAG,GAAG,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,CAAC,EAAE,EAAE,CAAC,GAAG,GAAG,CAAC,CAAC,YAAY,EAAE,CAAC,CAAC,CAAA;IAChE,MAAM,OAAO,GAAG,QAAQ,IAAI,0BAA0B,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,KAAK,CAAC,MAAM,CAAA;IAE3E,IAAI,CAAC,OAAO,CAAC,MAAM;QAAE,OAAO,IAAI,CAAA;IAEhC,MAAM,MAAM,GAAG,YAAY,CAAC,OAAO,EAAE,MAAM,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,YAAY,EAAE,EAAE,SAAS,EAAE,KAAK,EAAE,CAAC,CAAA;IAEzF,OAAO,MAAM,CAAC,MAAM,CAAC,QAAQ,EAAE,MAAM,CAAC,CAAA;AACvC,CAAC;AAyCD;;GAEG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAc;IAC1C,MAAM,EAAE,EAAE;IACV,QAAQ,EAAE,EAAE;IACZ,cAAc,EAAE,EAAE;IAClB,WAAW,EAAE,EAAE;IACf,gBAAgB,EAAE,EAAE;CACpB,CAAA;AAaD,SAAS,kBAAkB,CAAC,GAAW,EAAE,IAAY,EAAE,iBAAoC;IAC1F,OAAO,iBAAiB,CAAC,IAAI,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,cAAc,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,UAAU,CAAC,MAAM,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,CAAA;AACtH,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,KAAK,UAAU,gBAAgB,CAAC,IAAY,EAAE,KAAgB;IACpE,MAAM,IAAI,GAAG,MAAM,iBAAiB,CAAC,IAAI,CAAC,CAAA;IAC1C,MAAM,MAAM,GAAG,IAAI,GAAG,EAAU,CAAA;IAChC,MAAM,QAAQ,GAAG,IAAI,GAAG,EAAU,CAAA;IAClC,MAAM,cAAc,GAAG,IAAI,GAAG,EAAU,CAAA;IACxC,MAAM,WAAW,GAAG,IAAI,GAAG,EAAU,CAAA;IACrC,MAAM,gBAAgB,GAAG,IAAI,GAAG,EAAU,CAAA;IAE1C,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QACxB,MAAM,IAAI,GAAG,CAAC,GAAG,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,IAAI,EAAE,CAAA;QAEpC,IAAI,CAAC,IAAI,IAAI,CAAC,WAAW,CAAC,IAAI,CAAC;YAAE,SAAQ;QACzC,MAAM,GAAG,GAAG,IAAI,CAAC,WAAW,EAAE,CAAA;QAC9B,MAAM,SAAS,GAAG,eAAe,CAAC,IAAI,CAAC,CAAA;QAEvC,IAAI,GAAG,CAAC,IAAI,KAAK,OAAO,IAAI,SAAS,EAAE,CAAC;YACvC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;QACjB,CAAC;QAED,KAAK,MAAM,QAAQ,IAAI,KAAK,CAAC,QAAQ,EAAE,CAAC;YACvC,IAAI,CAAC,cAAc,CAAC,GAAG,EAAE,QAAQ,CAAC,MAAM,CAAC;gBAAE,SAAQ;YAEnD,IAAI,CAAC,gBAAgB,CAAC,GAAG,EAAE,QAAQ,EAAE,KAAK,CAAC,SAAS,EAAE,KAAK,CAAC,OAAO,CAAC,EAAE,CAAC;gBACtE,IAAI,SAAS,EAAE,CAAC;oBACf,gBAAgB,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;gBAC3B,CAAC;YACF,CAAC;iBAAM,IAAI,oBAAoB,CAAC,GAAG,EAAE,QAAQ,EAAE,KAAK,CAAC,SAAS,EAAE,KAAK,CAAC,OAAO,CAAC,EAAE,CAAC;gBAChF,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;YACnB,CAAC;QACF,CAAC;QAED,IAAI,CAAC,SAAS;YAAE,SAAQ;QAExB,IAAI,KAAK,CAAC,eAAe,CAAC,IAAI,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,cAAc,CAAC,GAAG,EAAE,MAAM,CAAC,CAAC,EAAE,CAAC;YACzE,cAAc,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;QACzB,CAAC;QAED,IAAI,kBAAkB,CAAC,GAAG,EAAE,IAAI,EAAE,KAAK,CAAC,iBAAiB,CAAC,EAAE,CAAC;YAC5D,WAAW,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;QACtB,CAAC;IACF,CAAC;IAED,OAAO;QACN,MAAM,EAAE,CAAC,GAAG,MAAM,CAAC;QACnB,QAAQ,EAAE,CAAC,GAAG,QAAQ,CAAC;QACvB,cAAc,EAAE,CAAC,GAAG,cAAc,CAAC;QACnC,WAAW,EAAE,CAAC,GAAG,WAAW,CAAC;QAC7B,gBAAgB,EAAE,CAAC,GAAG,gBAAgB,CAAC;KACvC,CAAA;AACF,CAAC;AAED,MAAM,oBAAoB,GAAsB;IAC/C,SAAS;IACT,kBAAkB;IAClB,eAAe;IACf,UAAU;IACV,oBAAoB;CACpB,CAAA;AAED,MAAM,uBAAuB,GAAsB;IAClD,YAAY;IACZ,iBAAiB;IACjB,kBAAkB;IAClB,WAAW;IACX,mBAAmB;IACnB,0BAA0B;IAC1B,cAAc;CACd,CAAA;AAED;;;;GAIG;AACH,MAAM,UAAU,YAAY,CAAC,MAAuB,EAAE,OAAe,EAAE,KAAgB;IACtF,MAAM,EAAE,GAAG,IAAI,cAAc,CAAc,MAAM,EAAE,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,CAAA;IAEtE,MAAM,KAAK,GAAG,EAAE,CAAC,OAAO,CAAC,6CAA6C,CAAC,CAAC,GAAG,EAGzE,CAAA;IAEF,MAAM,MAAM,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,QAAQ,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAC5D,MAAM,QAAQ,GAAG,oBAAoB,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,EAAE,EAAgB,EAAE,CAAC,EAAE,IAAI,IAAI,CAAC,CAAA;IACxG,MAAM,WAAW,GAAG,uBAAuB,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,EAAE,EAAgB,EAAE,CAAC,EAAE,IAAI,IAAI,CAAC,CAAA;IAC9G,MAAM,MAAM,GAAG,CAAC,GAAG,QAAQ,EAAE,GAAG,WAAW,CAAC,CAAA;IAE5C,IAAI,CAAC,MAAM,CAAC,MAAM;QAAE,MAAM,IAAI,KAAK,CAAC,aAAa,MAAM,sCAAsC,CAAC,CAAA;IAE9F,MAAM,IAAI,GAAG,EAAE;SACb,OAAO,CACP,gGAAgG,MAAM,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,CAClI;SACA,GAAG,CAAC,OAAO,EAAE,GAAG,MAAM,CAAiD,CAAA;IAEzE,MAAM,QAAQ,GAAG,IAAI,GAAG,CAAC,QAAQ,CAAC,CAAA;IAClC,MAAM,MAAM,GAAG,IAAI,GAAG,EAAU,CAAA;IAChC,MAAM,cAAc,GAAG,IAAI,GAAG,EAAU,CAAA;IACxC,MAAM,WAAW,GAAG,IAAI,GAAG,EAAU,CAAA;IAErC,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QACxB,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,CAAC,IAAI,EAAE,CAAA;QAE5B,IAAI,CAAC,eAAe,CAAC,IAAI,CAAC;YAAE,SAAQ;QACpC,MAAM,GAAG,GAAG,IAAI,CAAC,WAAW,EAAE,CAAA;QAE9B,IAAI,QAAQ,CAAC,GAAG,CAAC,GAAG,CAAC,WAAW,CAAC,EAAE,CAAC;YACnC,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;QACjB,CAAC;QAED,IAAI,KAAK,CAAC,eAAe,CAAC,IAAI,CAAC,CAAC,MAAM,EAAE,EAAE,CAAC,cAAc,CAAC,GAAG,EAAE,MAAM,CAAC,CAAC,EAAE,CAAC;YACzE,cAAc,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;QACzB,CAAC;QAED,IAAI,kBAAkB,CAAC,GAAG,EAAE,IAAI,EAAE,KAAK,CAAC,iBAAiB,CAAC,EAAE,CAAC;YAC5D,WAAW,CAAC,GAAG,CAAC,IAAI,CAAC,CAAA;QACtB,CAAC;IACF,CAAC;IAED,OAAO;QACN,MAAM,EAAE,CAAC,GAAG,MAAM,CAAC;QACnB,QAAQ,EAAE,EAAE;QACZ,cAAc,EAAE,CAAC,GAAG,cAAc,CAAC;QACnC,WAAW,EAAE,CAAC,GAAG,WAAW,CAAC;QAC7B,gBAAgB,EAAE,EAAE;KACpB,CAAA;AACF,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,cAAc,CAAC,CAAY,EAAE,CAAY;IACxD,OAAO;QACN,MAAM,EAAE,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,GAAG,CAAC,CAAC,MAAM,CAAC;QAClC,QAAQ,EAAE,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC,QAAQ,CAAC;QACxC,cAAc,EAAE,CAAC,GAAG,CAAC,CAAC,cAAc,EAAE,GAAG,CAAC,CAAC,cAAc,CAAC;QAC1D,WAAW,EAAE,CAAC,GAAG,CAAC,CAAC,WAAW,EAAE,GAAG,CAAC,CAAC,WAAW,CAAC;QACjD,gBAAgB,EAAE,CAAC,GAAG,CAAC,CAAC,gBAAgB,EAAE,GAAG,CAAC,CAAC,gBAAgB,CAAC;KAChE,CAAA;AACF,CAAC"}

package/out/recipes/sub/venue.d.ts

@@ -0,0 +1,176 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ *   Generates rows that tag venue-interior strings such as `Terminal 5` or `North Gate` as `unit` beside a `venue`,
+ *   with confound rows as negatives.
+ */
+import { type CorpusRecipe } from "#recipes/scaffold";
+import { type IdentifierModel, type PromotedSurface } from "#recipes/sub/venue/sources";
+export * from "#recipes/sub/venue/sources";
+export * from "#recipes/sub/venue/context";
+export * from "#recipes/sub/venue/render";
+/**
+ * One locale's leg of the recipe.
+ *
+ * `positiveShare` and `negativeShare` are relative weights that the run normalizes across legs.
+ */
+export interface SubVenueLeg {
+    locale: string;
+    country: string;
+    /**
+     * The ISO 3166-1 alpha-2 key into the lexicon's `identifierShapes`.
+     *
+     * Gate and terminal numbering differs by country, so each leg samples identifiers from its own region.
+     */
+    region: string;
+    /**
+     * The OSM extract filename under `--extracts-dir`.
+     *
+     * A leg without an extract draws its venue and confound pools from `poi.db`.
+     */
+    extract?: string;
+    /**
+     * Whether this leg may use the English `<modifier> <designator>` form.
+     *
+     * The modifier list is English, so non-English legs emit only designator-plus-identifier forms.
+     */
+    english: boolean;
+    positiveShare: number;
+    negativeShare: number;
+    /**
+     * Postcode prefixes that restrict the leg's address context.
+     *
+     * The ca-ES leg uses the prefixes of the Catalan-speaking provinces
+     * because the OpenAddresses region strings are unreliable.
+     */
+    postcodePrefixes?: readonly string[];
+}
+/**
+ * The locale legs and their shares.
+ *
+ * The English legs account for the largest shares because only English has the modifier form
+ * and most eval confound rows are GB or US addresses.
+ * The en-US leg accounts for the largest negative share because its confound pool is the largest.
+ *
+ * The JP corpus builder processes Japanese sub-venue rows.
+ * It uses a different label set.
+ */
+export declare const SUBVENUE_LEGS: readonly SubVenueLeg[];
+/**
+ * The region whose identifier distribution the en-US leg uses.
+ *
+ * The en-US leg has no OSM extract.
+ * Its `poi.db` contains venue names without refs.
+ * The recipe borrows the identifier distribution from GB.
+ */
+export declare const US_IDENTIFIER_REGION_BORROWED_FROM = "GB";
+/**
+ * The suggested `--count` for this recipe.
+ *
+ * The training sampler drops a source from its multinomial once the source runs out of rows,
+ * so an undersized output receives less than its configured weight.
+ * At a source weight of 12.0 in a one-million-row epoch, the sampler draws
+ * about 77,000 rows from this output.
+ * The value 120,000 leaves headroom above that.
+ */
+export declare const RECOMMENDED_ROW_COUNT = 120000;
+/**
+ * Lowercase surfaces from the eval board
+ * in `packages/mailwoman/tools/eval-harness/fixtures/venue-structure-confounds.jsonl`.
+ *
+ * The recipe drops any row that contains one of these and counts it in `contaminated`.
+ * The list is copied by hand because `@mailwoman/corpus` cannot read the board's fixture at run time.
+ * Update it when the board changes.
+ */
+export declare const BOARD_RESERVED_SURFACES: readonly string[];
+/**
+ * Returns whether the row text contains any surface in {@link BOARD_RESERVED_SURFACES}.
+ */
+export declare function isBoardReserved(raw: string): boolean;
+/**
+ * A sub-venue string with the form and designator that produced it.
+ */
+export interface SubVenueForm {
+    text: string;
+    form: "designator-identifier" | "modifier-designator" | "attested";
+    designatorID: string;
+}
+/**
+ * Builds one sub-venue string for a leg, or returns `null` when it cannot.
+ *
+ * A promotion with `shape: "identifier-required"` always renders as `<Phrase> <identifier>`.
+ * German `Halle` also serves as a place name.
+ *
+ * The identifier distinguishes the two readings.
+ * The guard lives here so that every caller inherits it.
+ */
+export declare function buildSubVenueForm(leg: SubVenueLeg, promoted: readonly PromotedSurface[], model: IdentifierModel, modifiers: readonly string[], attested: readonly string[], random: () => number): SubVenueForm | null;
+/**
+ * An alias of {@link buildSubVenueForm}.
+ *
+ * @deprecated Use {@link buildSubVenueForm}.
+ */
+export declare const buildPositiveForms: typeof buildSubVenueForm;
+/**
+ * The confound classes that the recipe emits as negatives.
+ */
+export declare const NegativeClass: {
+    /**
+     * A surface rejected for the locale in the venue slot, such as `Red Wing Shoes`.
+     */
+    readonly RejectedVenue: "rejected-venue";
+    /**
+     * A longer proper name that contains a designator, tagged whole as `venue`,
+     * such as `Lochaline Ferry Terminal`.
+     */
+    readonly LongerName: "longer-name";
+    /**
+     * A real street whose name contains a designator token, such as `Pier Road`.
+     */
+    readonly DesignatorStreet: "designator-street";
+    /**
+     * A real street with the `<modifier> <designator>` shape, such as `East Gate`.
+     */
+    readonly ModifierDesignatorStreet: "modifier-designator-street";
+    /**
+     * A GB single-token `-gate` street, such as `Moorgate`.
+     */
+    readonly GateSuffixStreet: "gate-suffix-street";
+    /**
+     * A promoted phrase in a shape that its promotion excludes, such as `Halle Rosengarten`.
+     *
+     * See `LegPools.unpromotedShapes`.
+     */
+    readonly UnpromotedShape: "unpromoted-shape";
+};
+/**
+ * One of the {@link NegativeClass} values.
+ */
+export type NegativeClass = (typeof NegativeClass)[keyof typeof NegativeClass];
+/**
+ * Per-leg composition counts that the run prints.
+ */
+export interface SubVenueLegStats {
+    locale: string;
+    positives: number;
+    negatives: number;
+    byForm: Record<string, number>;
+    byDesignator: Record<string, number>;
+    byNegativeClass: Record<string, number>;
+    byRegister: Record<string, number>;
+    poolSizes: Record<string, number>;
+}
+/**
+ * Splits `total` across relative `shares` by the largest-remainder method, so the parts sum to `total`.
+ */
+export declare function allocate(total: number, shares: readonly number[]): number[];
+/**
+ * Generates sub-venue positives and confound negatives across the {@link SUBVENUE_LEGS}.
+ *
+ * Only (designator, locale) pairs that the promotion ledger accepted produce positives.
+ * The recipe requires `--count`.
+ */
+export declare const subVenueRecipe: CorpusRecipe;
+//# sourceMappingURL=venue.d.ts.map

package/out/recipes/sub/venue.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"venue.d.ts","sourceRoot":"","sources":["../../../lib/recipes/sub/venue.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAUH,OAAO,EAAkB,KAAK,YAAY,EAAE,MAAM,mBAAmB,CAAA;AAIrE,OAAO,EAIN,KAAK,eAAe,EAIpB,KAAK,eAAe,EAOpB,MAAM,4BAA4B,CAAA;AAKnC,cAAc,4BAA4B,CAAA;AAC1C,cAAc,4BAA4B,CAAA;AAC1C,cAAc,2BAA2B,CAAA;AAIzC;;;;GAIG;AACH,MAAM,WAAW,WAAW;IAC3B,MAAM,EAAE,MAAM,CAAA;IACd,OAAO,EAAE,MAAM,CAAA;IACf;;;;OAIG;IACH,MAAM,EAAE,MAAM,CAAA;IACd;;;;OAIG;IACH,OAAO,CAAC,EAAE,MAAM,CAAA;IAChB;;;;OAIG;IACH,OAAO,EAAE,OAAO,CAAA;IAChB,aAAa,EAAE,MAAM,CAAA;IACrB,aAAa,EAAE,MAAM,CAAA;IACrB;;;;;OAKG;IACH,gBAAgB,CAAC,EAAE,SAAS,MAAM,EAAE,CAAA;CACpC;AAED;;;;;;;;;GASG;AACH,eAAO,MAAM,aAAa,EAAE,SAAS,WAAW,EAgD/C,CAAA;AAED;;;;;;GAMG;AACH,eAAO,MAAM,kCAAkC,OAAO,CAAA;AAMtD;;;;;;;;GAQG;AACH,eAAO,MAAM,qBAAqB,SAAU,CAAA;AA4B5C;;;;;;;GAOG;AACH,eAAO,MAAM,uBAAuB,EAAE,SAAS,MAAM,EAkCpD,CAAA;AAED;;GAEG;AACH,wBAAgB,eAAe,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO,CAIpD;AAQD;;GAEG;AACH,MAAM,WAAW,YAAY;IAC5B,IAAI,EAAE,MAAM,CAAA;IACZ,IAAI,EAAE,uBAAuB,GAAG,qBAAqB,GAAG,UAAU,CAAA;IAClE,YAAY,EAAE,MAAM,CAAA;CACpB;AAED;;;;;;;;GAQG;AACH,wBAAgB,iBAAiB,CAChC,GAAG,EAAE,WAAW,EAChB,QAAQ,EAAE,SAAS,eAAe,EAAE,EACpC,KAAK,EAAE,eAAe,EACtB,SAAS,EAAE,SAAS,MAAM,EAAE,EAC5B,QAAQ,EAAE,SAAS,MAAM,EAAE,EAC3B,MAAM,EAAE,MAAM,MAAM,GAClB,YAAY,GAAG,IAAI,CAiCrB;AAED;;;;GAIG;AACH,eAAO,MAAM,kBAAkB,0BAAoB,CAAA;AAMnD;;GAEG;AACH,eAAO,MAAM,aAAa;IACzB;;OAEG;aACH,aAAa,EAAE,gBAAgB;IAC/B;;;OAGG;aACH,UAAU,EAAE,aAAa;IACzB;;OAEG;aACH,gBAAgB,EAAE,mBAAmB;IACrC;;OAEG;aACH,wBAAwB,EAAE,4BAA4B;IACtD;;OAEG;aACH,gBAAgB,EAAE,oBAAoB;IACtC;;;;OAIG;aACH,eAAe,EAAE,kBAAkB;CAC1B,CAAA;AAEV;;GAEG;AACH,MAAM,MAAM,aAAa,GAAG,CAAC,OAAO,aAAa,CAAC,CAAC,MAAM,OAAO,aAAa,CAAC,CAAA;AAiC9E;;GAEG;AACH,MAAM,WAAW,gBAAgB;IAChC,MAAM,EAAE,MAAM,CAAA;IACd,SAAS,EAAE,MAAM,CAAA;IACjB,SAAS,EAAE,MAAM,CAAA;IACjB,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;IAC9B,YAAY,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;IACpC,eAAe,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;IACvC,UAAU,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;IAClC,SAAS,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;CACjC;AAkND;;GAEG;AACH,wBAAgB,QAAQ,CAAC,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,SAAS,MAAM,EAAE,GAAG,MAAM,EAAE,CAiB3E;AA4DD;;;;;GAKG;AACH,eAAO,MAAM,cAAc,EAAE,YAiG5B,CAAA"}