/**
 * Thread Deduplication Service (Phase 3)
 *
 * Pure functions for detecting duplicate threads by embedding similarity,
 * token overlap, or normalized text equality. Zero I/O — all Supabase
 * and embedding calls live in the caller (create-thread.ts).
 *
 * Strategy:
 *   1. If embedding available: cosine similarity > 0.85 → duplicate
 *   2. Token overlap coefficient > 0.6 → duplicate (no API key needed)
 *      - Lowered to 0.4 when both threads share an issue prefix (e.g., OD-692:)
 *   3. Normalized text equality → duplicate
 *   4. If no existing threads: skip check
 */
import type { ThreadObject } from "../types/index.js";
export interface ThreadWithEmbedding {
    thread_id: string;
    text: string;
    embedding: number[] | null;
}
export interface DedupResult {
    is_duplicate: boolean;
    matched_thread_id: string | null;
    matched_text: string | null;
    similarity: number | null;
    method: "embedding" | "token_overlap" | "text_normalization" | "skipped";
}
export declare const DEDUP_SIMILARITY_THRESHOLD = 0.85;
export declare const TOKEN_OVERLAP_THRESHOLD = 0.6;
export declare const TOKEN_OVERLAP_ISSUE_PREFIX_THRESHOLD = 0.4;
/**
 * Check if new thread text is a semantic duplicate of any existing open thread.
 *
 * @param newText - Trimmed thread text
 * @param newEmbedding - Normalized embedding vector, or null if unavailable
 * @param existingThreads - Open threads with optional embeddings
 */
export declare function checkDuplicate(newText: string, newEmbedding: number[] | null, existingThreads: ThreadWithEmbedding[]): DedupResult;
/**
 * Cosine similarity between two normalized vectors.
 * Assumes vectors are already L2-normalized (dot product = cosine similarity).
 */
export declare function cosineSimilarity(a: number[], b: number[]): number;
/**
 * Normalize text for conservative text-only comparison.
 * Lowercase, collapse whitespace, trim, strip trailing punctuation.
 */
export declare function normalizeText(text: string): string;
/**
 * Tokenize text into content words for overlap comparison.
 * Lowercase, split on non-alphanumeric boundaries, remove stop words.
 */
export declare function tokenize(text: string): Set<string>;
/**
 * Overlap coefficient: |intersection| / min(|A|, |B|).
 * Handles the common case where one thread is a shorter variant of another.
 * Returns 0 if either set is empty.
 */
export declare function tokenOverlap(a: Set<string>, b: Set<string>): number;
/**
 * Extract issue prefix like "OD-692" or "PROJ-123" from thread text.
 * Returns null if no prefix found.
 */
export declare function extractIssuePrefix(text: string): string | null;
/**
 * Deduplicate a thread list by ID, normalized text, and token overlap.
 * First-seen wins. Skips empty-text threads. Does not mutate input.
 *
 * Applied at every thread loading/merging exit point to guarantee
 * no duplicates escape regardless of upstream logic.
 */
export declare function deduplicateThreadList(threads: ThreadObject[]): ThreadObject[];
//# sourceMappingURL=thread-dedup.d.ts.map