/**
 * The `zotero://` ref grammar as regex sources, shared by the host parser
 * (`refs.ts`) and the client's text scanners (the Sources panel extracts refs
 * embedded in call-block args and results). One source so the two halves
 * cannot drift: a client-side copy that lags the host grammar silently drops
 * refs (the group-library forms) the tools legitimately emit. The captures
 * are named, so consumers read them by meaning instead of a positional index
 * that shifts with the grammar.
 * @module dsh-zotero/ref-grammar
 */
/** The object-key source: 8 uppercase alphanumerics. */
export declare const REF_KEY_SOURCE = "[A-Z0-9]{8}";
/**
 * The `links.attachment.href` key extractor: `/items/<KEY>` with a path
 * boundary, unanchored so query strings and fragments pass through.
 */
export declare const ATTACHMENT_HREF_PATTERN: RegExp;
/**
 * One canonical Zotero relation path (`/users/<id>/items/<KEY>` or
 * `/groups/<id>/items/<KEY>`), trailing content allowed for the relation
 * forms Zotero serves.
 */
export declare const ZOTERO_USER_ITEM_PATH_PATTERN: RegExp;
/** The group-library form of {@link ZOTERO_USER_ITEM_PATH_PATTERN}. */
export declare const ZOTERO_GROUP_ITEM_PATH_PATTERN: RegExp;
/** The five object kinds the canonical Zotero ref grammar admits. */
export type RefKind = 'item' | 'attachment' | 'annotation' | 'collection' | 'search';
/** Full ref grammar, anchored: the whole string must be exactly one ref. */
export declare const REF_PATTERN: RegExp;
/**
 * One ref embedded in longer text (tool args, results, prose): the
 * key-bearing prefix, unanchored. Provenance qualifiers are not matched,
 * because key extraction only needs the object identity.
 */
export declare const REF_IN_TEXT_PATTERN: RegExp;
export declare function isItemTypeIdentifier(value: string): boolean;
//# sourceMappingURL=ref-grammar.d.ts.map