/**
 * scanner — recursive `.md` walker for `cleo docs import`.
 *
 * Walks a directory tree from a single root, returning one entry per
 * markdown file discovered. Each entry carries the absolute path, the
 * project-relative source-dir prefix used by the type classifier, and a
 * SHA-256 of the file's content so the dedup gate can decide whether
 * the bytes are already in the docs SSoT.
 *
 * Excludes well-known transient/build directories by default
 * (`node_modules`, `.git`, `dist`, `coverage`, `build`).
 *
 * @epic T9628 (Saga T9625)
 * @task T9710 (ST-MIG-1a)
 */
/**
 * Document import classification used by the type classifier.
 *
 * CLI-level concept distinct from {@link DocKind} — `research`, `note`,
 * and `spec` ride in `meta.importType` while the blob backend stores
 * them as `agent-output`. `adr` maps 1:1 to the existing DocKind.
 */
export type DocImportType = 'research' | 'adr' | 'note' | 'spec';
/** A single markdown file discovered during scanning. */
export interface ScannedFile {
    /** Absolute filesystem path of the markdown file. */
    readonly absPath: string;
    /** Path relative to the scan root (e.g. `.cleo/research/foo.md`). */
    readonly relPath: string;
    /** SHA-256 of the file bytes, hex encoded. */
    readonly contentSha: string;
    /** Classified import type per the source-dir rules table. */
    readonly suggestedType: DocImportType;
    /** Raw file content (UTF-8). Cached to avoid re-reading during import. */
    readonly content: string;
}
/** Options for {@link scanDirectory}. */
export interface ScanOptions {
    /** Absolute filesystem path to scan recursively. */
    readonly root: string;
    /**
     * Directory names to skip during the walk. Default:
     * `node_modules`, `.git`, `dist`, `coverage`, `build`.
     */
    readonly excludeDirs?: ReadonlySet<string>;
    /**
     * Override for the default type classifier. When provided this overrides
     * the source-dir rules table for the given `relPath`.
     */
    readonly classify?: (relPath: string) => DocImportType;
}
/**
 * Default directory names skipped by the scanner.
 */
export declare const DEFAULT_EXCLUDE_DIRS: ReadonlySet<string>;
/**
 * Default type classifier — see ADR-073 / T9639 spec for the rules table:
 *
 * ```
 * .cleo/adrs/         → adr
 * .cleo/research/     → research
 * .cleo/agent-outputs → note
 * .cleo/rcasd/        → research   (T9791 — added so rcasd dirs classify
 *                                   when the scan root is the project root)
 * docs/specs/         → spec
 * docs/               → spec (catch-all)
 * *                   → note (default)
 * ```
 *
 * NOTE: This function expects {@link relPath} to start with the source-dir
 * prefix (e.g. `.cleo/adrs/`). When the scanner is invoked with a scan root
 * INSIDE one of these source dirs (e.g. `scanRoot = .cleo/adrs/`), every
 * file's relPath will be missing that prefix and default to `note`. Callers
 * SHOULD pass a {@link makeClassifierForScanRoot} closure to recover the
 * correct classification in that scenario.
 *
 * @param relPath - Path relative to the scan root.
 * @returns Classified import type.
 */
export declare function classifyByRelPath(relPath: string): DocImportType;
/**
 * Map a canonical source-dir name to its {@link DocImportType}.
 *
 * Source-dir aware classification: when `cleo docs import .cleo/adrs` is run,
 * the scanner sees relPaths like `ADR-001.md` (no `.cleo/adrs/` prefix) so
 * {@link classifyByRelPath} would fall through to `note`. This map records
 * the canonical mapping by source-dir name so the CLI can pre-resolve the
 * classification before scanning.
 *
 * Keys are project-relative POSIX paths without trailing slash.
 *
 * @task T9791
 */
export declare const SOURCE_DIR_TO_TYPE: ReadonlyMap<string, DocImportType>;
/**
 * Detect the source-dir for an absolute or project-relative scan root and
 * build a classifier closure that returns the matching {@link DocImportType}
 * for every file it sees.
 *
 * When the scan root does not match any canonical source-dir the returned
 * closure falls back to {@link classifyByRelPath} so callers retain the
 * pre-T9791 behaviour.
 *
 * @param scanRoot   - Absolute (or project-relative) path passed to the scanner.
 * @param projectRoot - Absolute project root for resolving the project-relative
 *                      form of {@link scanRoot}.
 * @returns A classifier that resolves the type for any scanned file under
 *          {@link scanRoot}.
 *
 * @task T9791
 */
export declare function makeClassifierForScanRoot(scanRoot: string, projectRoot: string): (relPath: string) => DocImportType;
/**
 * Recursively walk `root` and return every `.md` file with its sha + type.
 *
 * Symbolic links are skipped (we only follow regular files and directories)
 * to avoid pathological cycles. Files that fail to read are silently
 * skipped — callers can detect this via the `errorCount` they maintain
 * around the scan.
 *
 * @param options - Scan options including the root directory.
 * @returns One {@link ScannedFile} per discovered `.md` file.
 */
export declare function scanDirectory(options: ScanOptions): Promise<ScannedFile[]>;
//# sourceMappingURL=scanner.d.ts.map