/**
 * @copyright Sister Software
 * @license AGPL-3.0
 * @author Teffen Ellis, et al.
 *
 *   Build per-script parquet shards from the DeepSeek-generated transliteration JSONL and emit the
 *   corpus-v0.4.0 MANIFEST that combines them with the existing kryptonite + v0.3.0 shards.
 *
 *   Sibling to `shard-kryptonite.ts`. The two modules share the same composition pattern: take a
 *   base MANIFEST, append new shards, write a combined MANIFEST. Differences specific to
 *   transliteration:
 *
 *   - One JSONL contains rows from N target scripts (source = `deepseek-translit-<slug>`). We bucket by
 *       `source` and write one shard per script so `audit.ts` can attribute each shard to its
 *       synthetic source without relying on filename-prefix inference.
 *   - Each shard is written to `train/part-translit-<slug>.parquet` (distinct from kryptonite's
 *       `part-0000.parquet`, which v0.4.0's first builder already produced).
 *   - Inherits the path-canonicalization fix flagged in Thread B's postmortem: v0.3.0 shard paths are
 *       rewritten from `$MAILWOMAN_DATA_ROOT/...` to `/data/...` in the combined MANIFEST so
 *       all paths share one container-friendly form.
 *
 *   See docs/engineering/reference/CORPUS_V0_4_0_GENERATION.md for prompts, model, and the
 *   reproducibility contract.
 *
 *   Invoke via `mailwoman corpus shard translit \
 *   --jsonl /data/corpus/versioned/v0.4.0/transliteration/canonical-transliteration.jsonl \
 *   --base-manifest /data/corpus/versioned/v0.4.0/corpus-v0.4.0/MANIFEST.json \
 *   --out-dir /data/corpus/versioned/v0.4.0`
 */
export interface ShardTranslitOptions {
    jsonl: string;
    baseManifest: string;
    outDir: string;
    /**
     * Default `"0.4.0"`.
     */
    corpusVersion?: string;
    /**
     * Default `"/data/"`.
     */
    canonicalPathPrefix?: string;
    /**
     * Prefix the base manifest's shard paths currently carry, to be rewritten to
     * {@link ShardTranslitOptions.canonicalPathPrefix}. Defaults to `mailwomanDataRoot()` with a trailing slash — the
     * root that WROTE those paths. Pass it explicitly when translating a manifest generated under a different
     * `$MAILWOMAN_DATA_ROOT` than the one you are running with.
     */
    legacyPathPrefix?: string;
}
export declare function buildTranslitShard(options: ShardTranslitOptions, report?: (line: string) => void): Promise<void>;
//# sourceMappingURL=shard-translit.d.ts.map