/**
 * @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