/**
 * Deterministic evaluation corpus.
 *
 * Reproducibility rules this file exists to enforce:
 *  - The corpus is derived from git-tracked files only, so a clean checkout at
 *    a given commit produces exactly the same document set.
 *  - The list is sorted by path, so ordering never depends on filesystem walk
 *    order (which differs between macOS and Linux).
 *  - Every run records a corpusHash over (path, sha256) pairs. Two scoreboard
 *    rows with different corpusHash values are NOT comparable.
 *
 * @module v1/cli/knowledge/eval/corpus
 */
export interface CorpusDoc {
    /** Repo-relative POSIX path — the stable document id used in the golden set. */
    id: string;
    absPath: string;
    bytes: number;
    sha256: string;
}
export interface Corpus {
    docs: CorpusDoc[];
    corpusHash: string;
    repoRoot: string;
    /** Paths matched by the include rules but dropped, with the reason. */
    excluded: Array<{
        id: string;
        reason: string;
    }>;
    /**
     * Byte-identical documents collapsed to one content unit.
     *
     * This repo ships a verbatim mirror of `.claude/**` inside the CLI package
     * (an npm-packaging requirement), so ~200 documents exist twice. Without
     * this, a retriever that returns the identical twin of the gold document
     * scores a MISS for returning the correct content — which measures path
     * luck, not retrieval. Identical bytes are one document.
     */
    canonicalOf: Map<string, string>;
    /** Number of distinct content units (<= docs.length). */
    contentUnits: number;
    duplicateGroups: number;
    /** AppleDouble `._` resource forks found. Must be zero. */
    appleDoubleCount: number;
}
/** AppleDouble resource forks. Per-SEGMENT anchored, so it catches
 *  `docs/concepts/._memory.md` and not merely a leading dot at the repo root. */
export declare const APPLEDOUBLE_RE: RegExp;
/** The repository top level. Without this the corpus silently differs
 *  depending on which subdirectory you ran from — and two scoreboard rows
 *  would be incomparable for a reason nobody would think to check. */
export declare function resolveRepoRoot(start: string): string;
export declare function listTrackedFiles(repoRoot: string): string[];
export declare function buildCorpus(startDir: string): Corpus;
export declare function readDoc(doc: CorpusDoc): string;
//# sourceMappingURL=corpus.d.ts.map