/**
 * @copyright Sister Software
 * @license AGPL-3.0
 * @author Teffen Ellis, et al.
 *
 *   @file Builds the sub-venue designator lexicon from Wikidata, OSM and Overture fetch outputs.
 */
import { type PathBuilderLike } from "path-ts";
import { type SubVenueHarvestRow } from "#subvenue/harvest";
import { type SubVenuePromotion } from "#subvenue/promotions";
import { type SubVenueLexiconSource, type SubVenueLexiconTable, type SubVenueSurface } from "#subvenue/table";
export * from "#subvenue/harvest";
export * from "#tools/sub/venue/head-nouns";
export * from "#subvenue/surfaces";
export * from "#subvenue/table";
export * from "#tools/sub/venue/wikidata";
/**
 * Returns a copy of the surfaces with `curated: true` set on each surface that a promotion matches.
 *
 * A promotion matches a surface with the same record id and phrase whose language
 * is the promotion's language or `und`.
 * An extract surface must also come from the promotion's region.
 *
 * A surface without a region, such as a Wikidata label, matches only when no rejection
 * exists for the same designator, phrase and language in any region.
 * For example, `pier` is promoted for en-GB and rejected for en-US,
 * so the region-free English `pier` stays uncurated.
 *
 * Rejections never mark a surface.
 * They stay in the table as a record of the decision.
 */
export declare function applyPromotions(surfaces: readonly SubVenueSurface[], promotions: readonly SubVenuePromotion[]): SubVenueSurface[];
/**
 * One harvest input with the source and region stamped on its surfaces.
 */
export interface SubVenueHarvest {
    rows: readonly SubVenueHarvestRow[];
    source?: string;
    region?: string;
}
/**
 * The parsed inputs of {@link buildSubVenueLexicon}, which never reads the filesystem.
 */
export interface BuildSubVenueLexiconInput {
    /**
     * The raw `designator-labels.json` SPARQL response, or `null` to build the seed-only table.
     */
    wikidata: unknown | null;
    /**
     * The harvest inputs in contribution order.
     *
     * Each harvest can match only phrases that exist before it runs, including those from earlier harvests.
     */
    harvests: readonly SubVenueHarvest[];
    /**
     * Provenance rows that the caller copies from the fetch manifests.
     */
    sources: readonly SubVenueLexiconSource[];
    /**
     * Defaults to {@link SUBVENUE_PROMOTIONS}.
     *
     * An empty array builds the uncurated table.
     */
    promotions?: readonly SubVenuePromotion[];
}
/**
 * Builds the lexicon table deterministically, so identical inputs produce identical output.
 *
 * Three ordering rules apply:
 *
 * 1. Seed surfaces come first, so `terminal` indexes to the `terminal` designator
 *    instead of a Wikidata alias.
 * 2. Head nouns are derived after Wikidata and before the harvests, because a harvest
 *    can find only phrases such as `ターミナル` that already exist as surfaces.
 * 3. Promotions apply last, so they can curate a surface from any source.
 */
export declare function buildSubVenueLexicon(input: BuildSubVenueLexiconInput): SubVenueLexiconTable;
/**
 * Serializes the table as pretty-printed JSON.
 *
 * Run `oxfmt` over the written file before committing, because committed JSON must match oxfmt's output.
 */
export declare function serializeSubVenueLexicon(table: SubVenueLexiconTable): string;
/**
 * One OSM extract to harvest, with the ISO country code of its rows.
 */
export interface SubVenueExtractInput {
    path: string;
    region: string;
}
/**
 * Options for {@link generateSubVenueLexicon}.
 */
export interface GenerateSubVenueLexiconOptions {
    /**
     * The output directory of `mailwoman corpus fetch wikidata-subvenue`.
     *
     * Without it, the function builds the seed-only table.
     */
    wikidataDir?: PathBuilderLike;
    /**
     * OSM extract JSONL files, one per region.
     */
    extracts?: readonly SubVenueExtractInput[];
    /**
     * Overture rows that the caller already read with `readOvertureSubVenues`.
     *
     * The caller opens `poi.db`, so this function does not depend on the database.
     */
    overtureRows?: readonly (SubVenueHarvestRow & {
        country: string;
    })[];
    /**
     * The `poi.db` layer vintage.
     * The builder uses it only when `overtureRows` is non-empty.
     */
    overtureVintage?: string;
    /**
     * The path of the written table.
     */
    outPath: string;
}
/**
 * Reads the fetch outputs, builds the table with {@link buildSubVenueLexicon}, and writes it.
 *
 * Run `oxfmt` over `outPath` afterwards, because committed JSON must match oxfmt's output.
 */
export declare function generateSubVenueLexicon(options: GenerateSubVenueLexiconOptions): Promise<SubVenueLexiconTable>;
//# sourceMappingURL=lexicon.d.ts.map