/**
 * @copyright Sister Software
 * @license AGPL-3.0
 * @author Teffen Ellis, et al.
 *
 *   Corpus linter. Compares a new shard against pre-computed corpus statistics and flags patterns
 *   that would cause the class of failure we hit with v0.6.2's "5th Avenue Theatre" adversarial
 *   venue templates.
 *
 *   Per DeepSeek turn 9 design (2026-05-29). v1 checks:
 *
 *   1. **Token-label distribution outliers.** For each token in the new shard, compare the shard's
 *        majority label to the corpus's majority label. Flag when the corpus has a
 *        confidently-established majority (>66%) AND the shard's majority differs AND both have
 *        non-trivial counts (shard ≥ 50, corpus ≥ 200).
 *   2. **Label-vacuum tokens.** Token labeled with a tag that has ZERO instances in the corpus for that
 *        token, despite the token being well-represented in the corpus. Stronger signal than #1 —
 *        we're introducing a novel association, not shifting a distribution.
 *   3. **Bigram-label collisions.** Identical (token_bigram, label_bigram) appears in shard while the
 *        same token_bigram has a DIFFERENT majority label_bigram in the corpus. The "5th Avenue"
 *        with [B-venue, I-venue] vs corpus's [B-house_number, I-street] case.
 *   4. **Common-form anti-pattern rules.** Applies `lint-rules.json` — token-regex → forbidden-labels
 *        mappings — flagging matches.
 *   5. **Basic sanity.** Truncated rows (tokens.length !== labels.length), all-O rows >90% of shard.
 *
 *   Output: markdown report on stdout, optional JSON sidecar via `outJSON`. The command exits 0 if
 *   no errors, 1 if any errors (warnings don't gate). Per the design, the MANIFEST entry for a
 *   flagged shard should require `lint_acknowledged: true` before training consumes it.
 *
 *   Usage: mailwoman dev lint corpus-shard\
 *   --shard <new-shard.parquet>\
 *   --stats <corpus-stats.json>\
 *   [--rules <rules.json>]\
 *   [--out-md /tmp/lint-report.md]\
 *   [--out-json /tmp/lint-report.json]
 */
/**
 * Options for {@linkcode lintCorpusShard}.
 */
export interface LintCorpusShardOptions {
    /**
     * The new shard parquet to lint.
     */
    shardPath: string;
    /**
     * Pre-computed corpus stats JSON (see `corpus-stats.ts`).
     */
    statsPath: string;
    /**
     * Anti-pattern rules JSON. Default: the `lint-rules.json` beside this module.
     */
    rulesPath?: string;
    /**
     * Write the markdown report here as well as stdout.
     */
    outMd?: string;
    /**
     * Write a JSON sidecar of the flags + summary here.
     */
    outJSON?: string;
}
/**
 * One lint flag emitted by a check.
 */
export interface LintShardFlag {
    check: string;
    severity: "error" | "warn";
    token?: string;
    bigram?: string;
    shardLabel?: string;
    corpusLabel?: string;
    shardCount?: number;
    corpusCount?: number;
    detail: string;
    ruleID?: string;
}
/**
 * Findings summary returned by {@linkcode lintCorpusShard}.
 */
export interface LintCorpusShardSummary {
    errors: number;
    warnings: number;
    findings: LintShardFlag[];
    /**
     * The rendered markdown report (also printed to stdout).
     */
    report: string;
}
/**
 * Lint a shard against corpus stats + the anti-pattern rules; print the markdown report to stdout.
 */
export declare function lintCorpusShard(options: LintCorpusShardOptions, report?: (line: string) => void): Promise<LintCorpusShardSummary>;
//# sourceMappingURL=lint-shard.d.ts.map