/**
 * Pre-mask harness: substitute deterministically-detected structured PII with
 * neutral sentinels *before* the contextual model runs.
 *
 * The heuristic layer (SSN / CREDIT_CARD / IP_ADDRESS / EMAIL / URL) is
 * validator- or pattern-backed and runs synchronously. Rather than show the
 * model raw card/SSN/IP digits or email/URL strings — which it never needs to
 * classify and which only add noise — we replace each heuristic span with a
 * fixed sentinel token (`[SSN]`, `[CREDIT_CARD]`, `[IP_ADDRESS]`, ...). The model
 * is trained on text masked the exact same way, so the train-time and
 * inference-time input distributions match by construction.
 *
 * Offsets: the masked string is paired with a per-character map back to the raw
 * input, so a span the model reports in masked coordinates can be projected to
 * exact raw offsets. Sentinel characters map onto their source span's raw range,
 * so if the model happens to label part of a sentinel the projection lands
 * inside the heuristic span (which then wins the merge on its score-1 weight).
 *
 * The heuristic spans themselves are returned with their original raw offsets,
 * so final redaction and the session table are unaffected: only the text handed
 * to the model is masked, never the text that gets placeholdered.
 */
import type { PiiLabel, Span } from "./types";
/** A masked copy of the input plus a map from each masked char to raw offsets. */
export interface PremaskResult {
    /** Input with heuristic spans replaced by sentinel tokens. */
    readonly masked: string;
    /** `rawStart[i]` is the raw offset of the source of `masked[i]`. */
    readonly rawStart: number[];
    /** `rawEnd[i]` is the raw offset just past that source. */
    readonly rawEnd: number[];
}
/** The sentinel substituted for a premasked label. Stable across train/serve. */
export declare function sentinelFor(label: PiiLabel): string;
/**
 * Replace each (disjoint) heuristic span in `raw` with its label sentinel,
 * recording per masked-character raw offsets. Spans are merged first so they are
 * non-overlapping; verbatim runs between spans map 1:1 to raw, and sentinel
 * characters map onto the whole span they stand in for.
 */
export declare function premask(raw: string, spans: readonly Span[]): PremaskResult;
/**
 * Project a span reported in masked coordinates back onto the raw input. Returns
 * `null` for an empty/degenerate span. The raw `text` is sliced from `raw` so
 * callers always carry the real substring for rehydration.
 */
export declare function projectMaskedSpan(span: Span, raw: string, map: PremaskResult): Span | null;
//# sourceMappingURL=premask.d.ts.map