import { type KeywordExtractorOptions } from "./KeywordExtractor.js";
import type { KeywordResult } from "./strategies.js";
/**
 * Minimal document shape for pipeline-oriented extraction.
 */
export interface InputDocument<TMetadata = Record<string, unknown>> {
    id: string;
    body: string;
    language?: string;
    title?: string;
    metadata?: TMetadata;
}
/**
 * Extraction result paired with the original document identity and metadata.
 */
export interface DocumentKeywordResult<TMetadata = Record<string, unknown>> {
    id: string;
    language: string;
    title?: string;
    metadata?: TMetadata;
    keywords: KeywordResult[];
}
/**
 * Context passed to pre-extraction document text hooks.
 */
export interface DocumentTextHookContext<TMetadata = unknown> {
    document: InputDocument<TMetadata>;
    language: string;
    includeTitleInText: boolean;
}
/**
 * Context passed to post-extraction keyword hooks.
 */
export interface DocumentKeywordHookContext<TMetadata = unknown> {
    document: InputDocument<TMetadata>;
    language: string;
    text: string;
}
/**
 * Extraction options for document helpers.
 */
export interface DocumentExtractionOptions extends KeywordExtractorOptions {
    includeTitleInText?: boolean;
    beforeExtractText?: (text: string, context: DocumentTextHookContext) => string;
    afterExtractKeywords?: (keywords: KeywordResult[], context: DocumentKeywordHookContext) => KeywordResult[];
}
/**
 * Extracts keyword details from a single document.
 */
export declare function extractFromDocument<TMetadata = Record<string, unknown>>(document: InputDocument<TMetadata>, options?: DocumentExtractionOptions): DocumentKeywordResult<TMetadata>;
/**
 * Extracts keyword details for an iterable of documents.
 */
export declare function extractFromDocuments<TMetadata = Record<string, unknown>>(documents: Iterable<InputDocument<TMetadata>>, options?: DocumentExtractionOptions): Array<DocumentKeywordResult<TMetadata>>;
/**
 * Extracts keyword details for an async stream of documents.
 */
export declare function extractFromDocumentStream<TMetadata = Record<string, unknown>>(documents: AsyncIterable<InputDocument<TMetadata>>, options?: DocumentExtractionOptions): AsyncGenerator<DocumentKeywordResult<TMetadata>>;
/**
 * Stable JSON serialization for a document extraction result.
 */
export declare function serializeDocumentKeywordResult<TMetadata = Record<string, unknown>>(result: DocumentKeywordResult<TMetadata>): string;
/**
 * Stable JSON serialization for an ordered collection of document extraction results.
 */
export declare function serializeDocumentKeywordResults<TMetadata = Record<string, unknown>>(results: Iterable<DocumentKeywordResult<TMetadata>>): string;
/**
 * Estimates the UTF-8 byte size of a stable serialized payload.
 */
export declare function estimateSerializedBytes(value: unknown): number;
//# sourceMappingURL=document.d.ts.map