/**
 * Vocabulary Dump (counts-only)
 *
 * Inventory of the vocabulary available in an AAC pageset, aggregated by
 * source and part of speech. Only counts are emitted — never word lists —
 * so the output is safe to embed in privacy-preserving reports.
 *
 * Sources counted:
 *   buttons              — labelled buttons (the static on-board vocabulary)
 *   wordLists            — page WordLists feeding dynamic AutoContent cells
 *                          (Grid 3 `<WordList>`; absent in other formats)
 *   predictionDictionaries — prediction wordlists attached to prediction cells
 *                          (Grid 3 `Prediction.PredictThis` dictionaries)
 *   wordForms            — smart-grammar inflections generated by
 *                          MetricsCalculator (only when a MetricsResult is
 *                          supplied)
 *
 * Non-Grid 3 formats simply report zeros for the Grid 3-specific sources.
 */
import type { AACTree } from '../../../types/aac';
import type { MetricsResult } from './types';
/** Counts for one vocabulary source. */
export interface VocabularySourceCounts {
    /** Total entries found in this source (before deduplication). */
    entries: number;
    /** Unique single words after normalisation. */
    uniqueWords: number;
    /** Unique multi-word phrases after normalisation (e.g. "thank you"). */
    uniquePhrases: number;
    /** Unique entries per part-of-speech tag ('Unknown' when untagged). */
    byPartOfSpeech: Record<string, number>;
}
/** Aggregated, counts-only vocabulary inventory. */
export interface VocabularySummary {
    totalBoards: number;
    totalButtons: number;
    /** Vocabulary carried by labelled buttons. */
    buttons: VocabularySourceCounts;
    /** Page WordLists (dynamic content cells; Grid 3). */
    wordLists: VocabularySourceCounts & {
        lists: number;
    };
    /** Prediction dictionaries attached to prediction cells (Grid 3). */
    predictionDictionaries: VocabularySourceCounts & {
        buttonsWithDictionaries: number;
    };
    /** Smart-grammar inflections; null when no MetricsResult was supplied. */
    wordForms: (VocabularySourceCounts & {
        parentButtons: number;
    }) | null;
    /** Unique entries across all sources combined. */
    combined: {
        uniqueEntries: number;
        uniqueWords: number;
        uniquePhrases: number;
    };
}
/** Full vocabulary dump document (schema-versioned). */
export interface VocabularyDump {
    schema: 'aac-vocabulary-dump/v1';
    generatedAt: string;
    source: {
        format?: string;
        name?: string;
        locale?: string;
    };
    summary: VocabularySummary;
}
export interface VocabularyDumpOptions {
    /**
     * Precomputed metrics (MetricsCalculator.analyze). Enables the
     * smart-grammar word-form counts. Run analyze() BEFORE dumpVocabulary():
     * analyze() expands morphological predictions on the tree.
     */
    metrics?: MetricsResult;
}
/**
 * Count the vocabulary available in a pageset, by source and part of speech.
 * Emits counts only — no word lists.
 *
 * @example
 * const processor = new GridsetProcessor();
 * const tree = await processor.loadIntoTree('my.gridset');
 * const metrics = new MetricsCalculator().analyze(tree);
 * const dump = dumpVocabulary(tree, { metrics });
 * console.log(dump.summary.wordLists.lists, 'wordlists');
 */
export declare function dumpVocabulary(tree: AACTree, options?: VocabularyDumpOptions): VocabularyDump;