/**
 * Content Deduplicator for MAMA OS Standalone
 *
 * Prevents duplicate content injection into system prompts by using
 * SHA-256 content hashing and realpath normalization. Handles symlinks,
 * duplicate file paths, and identical content from different sources.
 */
/**
 * Represents a unique content entry after deduplication.
 */
export interface ContentEntry {
    /** Original file path as provided */
    path: string;
    /** Resolved real path (symlinks resolved) */
    realPath: string;
    /** The content string */
    content: string;
    /** Semantic distance from query (lower = closer match) */
    distance: number;
    /** SHA-256 hash prefix of the content */
    hash: string;
}
/**
 * Deduplicates content entries by SHA-256 hash and realpath normalization.
 *
 * When two entries share the same content hash, the entry with the smaller
 * distance (closer semantic match) is preferred. This ensures that even if
 * the same file is referenced via different paths (e.g., symlinks), only
 * the most relevant instance is kept.
 *
 * @example
 * ```typescript
 * const dedup = new ContentDeduplicator();
 * dedup.add('/path/to/file.ts', 'const x = 1;', 0.3);
 * dedup.add('/symlink/to/file.ts', 'const x = 1;', 0.5); // duplicate, ignored (higher distance)
 * dedup.add('/other/file.ts', 'const y = 2;', 0.1);
 * const entries = dedup.getEntries(); // 2 entries, sorted by distance
 * ```
 */
export declare class ContentDeduplicator {
    private seenHashes;
    /**
     * Add content for deduplication.
     *
     * Returns true if the content is new (not a duplicate). On hash collision,
     * the entry with the closest distance is preferred.
     *
     * @param path - File path of the content source
     * @param content - Raw content string
     * @param distance - Semantic distance from query (lower = better)
     * @returns true if content was added (new), false if duplicate
     */
    add(path: string, content: string, distance: number): boolean;
    /**
     * Get all unique entries sorted by distance (closest first).
     *
     * @returns Array of deduplicated content entries ordered by ascending distance
     */
    getEntries(): ContentEntry[];
    /**
     * Reset the deduplicator for reuse.
     * Clears all tracked hashes and entries.
     */
    reset(): void;
    /**
     * Hash raw content with SHA-256, returning first 16 hex characters for efficiency.
     *
     * 16 hex chars = 64 bits of entropy, sufficient for collision avoidance
     * in typical prompt injection scenarios (< 10,000 entries).
     *
     * @param content - Raw content string to hash
     * @returns First 16 hex characters of SHA-256 digest
     */
    private hashContent;
    /**
     * Resolve the real path of a file, following symlinks.
     *
     * Falls back to the original path if realpath resolution fails
     * (e.g., file doesn't exist yet, permission denied).
     *
     * @param filePath - File path to resolve
     * @returns Resolved real path or original path on error
     */
    private safeRealpath;
}
//# sourceMappingURL=content-dedup.d.ts.map