/**
 * Near-duplicate body signatures — MinHash + LSH primitives.
 *
 * Single source of truth for signature constants and the digest tail used by
 * both tree-sitter adapters and the TypeScript inventory path. Signatures are
 * derived from the same canonical body string that feeds `bodyHash`.
 */
import { type BodyDigest } from './body-digest.js';
/** Number of MinHash values per body signature. */
export declare const NEAR_DUP_SIGNATURE_K = 128;
/** Default LSH band count — co-tuned with {@link NEAR_DUP_LSH_ROWS}. */
export declare const NEAR_DUP_LSH_BANDS = 8;
/** Rows per band; `bands × rows === k`. Knee ≈ 0.878 at threshold 0.85. */
export declare const NEAR_DUP_LSH_ROWS = 16;
/**
 * Signature ALGORITHM version. Bump on ANY change to shingling, hashing, or the
 * permutation scheme that alters signature VALUES. Feeds the `sig=` cache-key
 * segment so catalogs built with an older algorithm invalidate — mixing old- and
 * new-algorithm signatures across an incremental build (some occurrences cached,
 * some re-walked) would corrupt every cross-occurrence Jaccard estimate.
 *
 * v1 = k independent SHA-256 hashes per (shingle, seed). v2 = one SHA-256 base
 * hash per shingle + k cheap 32-bit mixers (~k× fewer hashes, identical MinHash
 * semantics).
 */
export declare const NEAR_DUP_SIGNATURE_VERSION = 2;
/** Digest including hash, size, and optional near-duplicate MinHash signature. */
export type BodyDigestWithSignature = BodyDigest;
/**
 * Character k-grams from canonical body text. Bodies shorter than `gramSize`
 * yield a single shingle of the whole string when non-empty.
 */
export declare function shingle(canonical: string, gramSize?: number): ReadonlySet<string>;
/**
 * Deterministic MinHash signature over char shingles. Stable across runs and
 * machines (SHA-256 base hash + fixed permutation seeds).
 *
 * Each shingle is hashed ONCE with SHA-256 (the only expensive step); the k
 * MinHash values are then derived by mixing that base hash with k fixed 32-bit
 * seeds. This is ~k× fewer SHA-256 computations than hashing every
 * (shingle, seed) pair while preserving MinHash semantics (each of the k mixers
 * is an independent hash of the shingle universe, so the per-position min
 * estimates Jaccard exactly as before). Algorithm {@link NEAR_DUP_SIGNATURE_VERSION}.
 */
export declare function bodySignature(canonical: string, k?: number): readonly number[];
/** MinHash Jaccard estimate: equal-position fraction. */
export declare function estimateJaccard(a: readonly number[], b: readonly number[]): number;
/** LSH band hashes for candidate-pair generation. */
export declare function lshBandHashes(signature: readonly number[], bands: number, rows: number): readonly string[];
/**
 * Hash + size + signature for an already-normalized canonical body string.
 * Skips `signature` when the canonical text yields no shingles.
 */
export declare function digestCanonicalBody(canonical: string): BodyDigestWithSignature;
//# sourceMappingURL=near-duplicate-signature.d.ts.map