/**
 * @typedef {{ mode: "convert" | "markdown" | "text" | "blob", format: string }} DocClass
 */
/**
 * Classify an ingest by filename extension (or an explicit `format` override) into a routing `mode`
 * and the `format` tag stored on the row. `convert`/`markdown`/`text` are chunkable (body-searchable);
 * `blob` is stored byte-exact with only its filename indexed. `format` is the lowercased extension
 * ("md" canonicalizes "markdown", "txt" canonicalizes "text"; a blob with no extension is "bin").
 * @param {string} [filename]
 * @param {string} [explicit]  caller override, e.g. "pdf" | "docx" | "csv"
 * @returns {DocClass}
 */
export function classifyDocument(filename?: string, explicit?: string): DocClass;
/**
 * @typedef {object} DocSegmentOptions
 * @property {"convert"|"markdown"|"text"} mode  routing from {@link classifyDocument} — "convert" runs
 *   a parser (pdf/docx), "markdown" segments md text directly, "text" packs flat plaintext
 *   (txt/text/log/csv). (Blobs never reach here.)
 * @property {string} format    the resolved format tag ("pdf" | "docx" | "md" | "txt" | "log" | "csv")
 * @property {number} [maxSize]   byte cap; over → reject before parse (default 10 MB)
 * @property {number} [maxPages]  page cap for PDF (default 2000)
 * @property {number} [parseTimeoutMs]  wall-clock parse cap (default 30 s)
 */
/**
 * Convert a chunkable document buffer (pdf/docx/md) to recall-ready markdown segments, BOUNDED and
 * failing with a CLEAR, specific error (never a crash; never an empty/garbage unit). The conversion +
 * segmentation half of {@link LiteCtx#ingest}; storage is the caller's. Routing (`mode`/`format`)
 * comes from {@link classifyDocument} — blobs are handled by the caller and never reach here.
 * @param {Uint8Array} buffer  the document bytes
 * @param {DocSegmentOptions} opts
 * @returns {Promise<{ format: string, segments: string[] }>}
 */
export function documentToSegments(buffer: Uint8Array, opts: DocSegmentOptions): Promise<{
    format: string;
    segments: string[];
}>;
export const DEFAULT_MAX_SIZE: number;
export type DocClass = {
    mode: "convert" | "markdown" | "text" | "blob";
    format: string;
};
export type DocSegmentOptions = {
    /**
     * routing from {@link classifyDocument} — "convert" runs
     * a parser (pdf/docx), "markdown" segments md text directly, "text" packs flat plaintext
     * (txt/text/log/csv). (Blobs never reach here.)
     */
    mode: "convert" | "markdown" | "text";
    /**
     * the resolved format tag ("pdf" | "docx" | "md" | "txt" | "log" | "csv")
     */
    format: string;
    /**
     * byte cap; over → reject before parse (default 10 MB)
     */
    maxSize?: number | undefined;
    /**
     * page cap for PDF (default 2000)
     */
    maxPages?: number | undefined;
    /**
     * wall-clock parse cap (default 30 s)
     */
    parseTimeoutMs?: number | undefined;
};