ocr-provenance-mcp 1.0.0

package/dist/services/storage/database/converters.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"converters.js","sourceRoot":"","sources":["../../../../src/services/storage/database/converters.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAqBH,yDAAyD;AACzD,MAAM,uBAAuB,GAA8B,CAAC,SAAS,EAAE,YAAY,EAAE,UAAU,EAAE,QAAQ,CAAC,CAAC;AAE3G,yDAAyD;AACzD,MAAM,sBAAsB,GAAsB;IAChD,UAAU,EAAE,YAAY,EAAE,OAAO,EAAE,OAAO,EAAE,iBAAiB;IAC7D,YAAY,EAAE,WAAW,EAAE,YAAY,EAAE,YAAY,EAAE,WAAW;CACnE,CAAC;AAEF,oDAAoD;AACpD,MAAM,kBAAkB,GAAyB,CAAC,SAAS,EAAE,YAAY,EAAE,UAAU,EAAE,QAAQ,CAAC,CAAC;AAEjG;;;GAGG;AACH,SAAS,YAAY,CAAmB,KAAa,EAAE,WAAyB,EAAE,SAAiB,EAAE,EAAU;IAC7G,IAAI,CAAC,WAAW,CAAC,QAAQ,CAAC,KAAU,CAAC,EAAE,CAAC;QACtC,MAAM,IAAI,KAAK,CAAC,WAAW,SAAS,KAAK,KAAK,eAAe,EAAE,mBAAmB,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IAC9G,CAAC;IACD,OAAO,KAAU,CAAC;AACpB,CAAC;AAED;;GAEG;AACH,SAAS,qBAAqB,CAAC,EAAU,EAAE,GAAW;IACpD,IAAI,CAAC;QACH,OAAO,IAAI,CAAC,KAAK,CAAC,GAAG,CAA4B,CAAC;IACpD,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,OAAO,CAAC,KAAK,CACX,wDAAwD,EAAE,KAAK,GAAG,GAAG,EACrE,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CACvD,CAAC;QACF,OAAO,EAAE,YAAY,EAAE,IAAI,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC;IAC3C,CAAC;AACH,CAAC;AAED;;;GAGG;AACH,SAAS,aAAa,CAAC,EAAU,EAAE,GAAW;IAC5C,IAAI,CAAC;QACH,OAAO,IAAI,CAAC,KAAK,CAAC,GAAG,CAAuB,CAAC;IAC/C,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,OAAO,CAAC,KAAK,CACX,+CAA+C,EAAE,KAAK,GAAG,GAAG,EAC5D,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CACvD,CAAC;QACF,OAAO,IAAI,CAAC;IACd,CAAC;AACH,CAAC;AAED;;;GAGG;AACH,SAAS,sBAAsB,CAAC,EAAU,EAAE,GAAW;IACrD,IAAI,CAAC;QACH,OAAO,IAAI,CAAC,KAAK,CAAC,GAAG,CAAsB,CAAC;IAC9C,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACf,OAAO,CAAC,KAAK,CACX,qDAAqD,EAAE,KAAK,GAAG,GAAG,EAClE,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CACvD,CAAC;QACF,OAAO,IAAI,CAAC;IACd,CAAC;AACH,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,aAAa,CAAC,GAAgB;IAC5C,OAAO;QACL,EAAE,EAAE,GAAG,CAAC,EAAE;QACV,SAAS,EAAE,GAAG,CAAC,SAAS;QACxB,SAAS,EAAE,GAAG,CAAC,SAAS;QACxB,SAAS,EAAE,GAAG,CAAC,SAAS;QACxB,SAAS,EAAE,GAAG,CAAC,SAAS;QACxB,SAAS,EAAE,GAAG,CAAC,SAAS;QACxB,MAAM,EAAE,YAAY,CAAC,GAAG,CAAC,MAAM,EAAE,uBAAuB,EAAE,gBAAgB,EAAE,GAAG,CAAC,EAAE,CAAC;QACnF,UAAU,EAAE,GAAG,CAAC,UAAU;QAC1B,aAAa,EAAE,GAAG,CAAC,aAAa;QAChC,UAAU,EAAE,GAAG,CAAC,UAAU;QAC1B,WAAW,EAAE,GAAG,CAAC,WAAW;QAC5B,gBAAgB,EAAE,GAAG,CAAC,gBAAgB;QACtC,aAAa,EAAE,GAAG,CAAC,aAAa;QAChC,SAAS,EAAE,GAAG,CAAC,SAAS,IAAI,IAAI;QAChC,UAAU,EAAE,GAAG,CAAC,UAAU,IAAI,IAAI;QAClC,WAAW,EAAE,GAAG,CAAC,WAAW,IAAI,IAAI;QACpC,eAAe,EAAE,GAAG,CAAC,eAAe,IAAI,IAAI;KAC7C,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,cAAc,CAAC,GAAiB;IAC9C,OAAO;QACL,EAAE,EAAE,GAAG,CAAC,EAAE;QACV,aAAa,EAAE,GAAG,CAAC,aAAa;QAChC,WAAW,EAAE,GAAG,CAAC,WAAW;QAC5B,cAAc,EAAE,GAAG,CAAC,cAAc;QAClC,WAAW,EAAE,GAAG,CAAC,WAAW;QAC5B,kBAAkB,EAAE,GAAG,CAAC,kBAAkB;QAC1C,YAAY,EAAE,GAAG,CAAC,YAAgD;QAClE,mBAAmB,EAAE,GAAG,CAAC,mBAAmB;QAC5C,UAAU,EAAE,GAAG,CAAC,UAAU;QAC1B,UAAU,EAAE,GAAG,CAAC,UAAU;QAC1B,YAAY,EAAE,GAAG,CAAC,YAAY;QAC9B,qBAAqB,EAAE,GAAG,CAAC,qBAAqB;QAChD,uBAAuB,EAAE,GAAG,CAAC,uBAAuB;QACpD,sBAAsB,EAAE,GAAG,CAAC,sBAAsB;QAClD,WAAW,EAAE,GAAG,CAAC,WAAW,IAAI,IAAI;QACpC,WAAW,EAAE,GAAG,CAAC,WAAW,IAAI,IAAI;KACrC,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,UAAU,CAAC,GAAa;IACtC,OAAO;QACL,EAAE,EAAE,GAAG,CAAC,EAAE;QACV,WAAW,EAAE,GAAG,CAAC,WAAW;QAC5B,aAAa,EAAE,GAAG,CAAC,aAAa;QAChC,IAAI,EAAE,GAAG,CAAC,IAAI;QACd,SAAS,EAAE,GAAG,CAAC,SAAS;QACxB,WAAW,EAAE,GAAG,CAAC,WAAW;QAC5B,eAAe,EAAE,GAAG,CAAC,eAAe;QACpC,aAAa,EAAE,GAAG,CAAC,aAAa;QAChC,WAAW,EAAE,GAAG,CAAC,WAAW;QAC5B,UAAU,EAAE,GAAG,CAAC,UAAU;QAC1B,gBAAgB,EAAE,GAAG,CAAC,gBAAgB;QACtC,YAAY,EAAE,GAAG,CAAC,YAAY;QAC9B,aAAa,EAAE,GAAG,CAAC,aAAa;QAChC,UAAU,EAAE,GAAG,CAAC,UAAU;QAC1B,gBAAgB,EAAE,GAAG,CAAC,gBAAqD;QAC3E,WAAW,EAAE,GAAG,CAAC,WAAW;QAC5B,iBAAiB,EAAE,GAAG,CAAC,iBAAiB,IAAI,IAAI;QAChD,eAAe,EAAE,GAAG,CAAC,eAAe;QACpC,aAAa,EAAE,GAAG,CAAC,aAAa;QAChC,YAAY,EAAE,GAAG,CAAC,YAAY;QAC9B,aAAa,EAAE,GAAG,CAAC,aAAa;QAChC,SAAS,EAAE,GAAG,CAAC,SAAS;QACxB,iBAAiB,EAAE,GAAG,CAAC,iBAAiB;KACzC,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,cAAc,CAAC,GAAiB;IAC9C,OAAO;QACL,EAAE,EAAE,GAAG,CAAC,EAAE;QACV,QAAQ,EAAE,GAAG,CAAC,QAAQ;QACtB,QAAQ,EAAE,GAAG,CAAC,QAAQ;QACtB,aAAa,EAAE,GAAG,CAAC,aAAa;QAChC,WAAW,EAAE,GAAG,CAAC,WAAW;QAC5B,aAAa,EAAE,GAAG,CAAC,aAAa;QAChC,oBAAoB,EAAE,GAAG,CAAC,oBAAoB;QAC9C,gBAAgB,EAAE,GAAG,CAAC,gBAAgB;QACtC,gBAAgB,EAAE,GAAG,CAAC,gBAAgB;QACtC,gBAAgB,EAAE,GAAG,CAAC,gBAAgB;QACtC,WAAW,EAAE,GAAG,CAAC,WAAW;QAC5B,UAAU,EAAE,GAAG,CAAC,UAAU;QAC1B,eAAe,EAAE,GAAG,CAAC,eAAe;QACpC,aAAa,EAAE,GAAG,CAAC,aAAa;QAChC,WAAW,EAAE,GAAG,CAAC,WAAW;QAC5B,YAAY,EAAE,GAAG,CAAC,YAAY;QAC9B,UAAU,EAAE,GAAG,CAAC,UAAU;QAC1B,aAAa,EAAE,GAAG,CAAC,aAAa;QAChC,SAAS,EAAE,GAAG,CAAC,SAA+C;QAC9D,cAAc,EAAE,GAAG,CAAC,cAAyB;QAC7C,UAAU,EAAE,GAAG,CAAC,UAAU,IAAI,EAAE;QAChC,aAAa,EAAE,GAAG,CAAC,aAAa;QAChC,YAAY,EAAE,GAAG,CAAC,YAAY;QAC9B,UAAU,EAAE,GAAG,CAAC,UAAU;QAC1B,sBAAsB,EAAE,GAAG,CAAC,sBAAsB;KACnD,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,eAAe,CAAC,GAAkB;IAChD,OAAO;QACL,EAAE,EAAE,GAAG,CAAC,EAAE;QACV,IAAI,EAAE,YAAY,CAAC,GAAG,CAAC,IAAI,EAAE,sBAAsB,EAAE,gBAAgB,EAAE,GAAG,CAAC,EAAE,CAAmB;QAChG,UAAU,EAAE,GAAG,CAAC,UAAU;QAC1B,YAAY,EAAE,GAAG,CAAC,YAAY;QAC9B,sBAAsB,EAAE,GAAG,CAAC,sBAAsB;QAClD,uBAAuB,EAAE,GAAG,CAAC,uBAAuB;QACpD,WAAW,EAAE,GAAG,CAAC,WAAyB;QAC1C,WAAW,EAAE,GAAG,CAAC,WAAW;QAC5B,SAAS,EAAE,GAAG,CAAC,SAAS;QACxB,gBAAgB,EAAE,GAAG,CAAC,gBAAgB;QACtC,QAAQ,EAAE,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,aAAa,CAAC,GAAG,CAAC,EAAE,EAAE,GAAG,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,IAAI;QACnE,YAAY,EAAE,GAAG,CAAC,YAAY;QAC9B,UAAU,EAAE,GAAG,CAAC,UAAU;QAC1B,SAAS,EAAE,GAAG,CAAC,SAAS;QACxB,SAAS,EAAE,GAAG,CAAC,SAAS;QACxB,iBAAiB,EAAE,GAAG,CAAC,iBAAiB;QACxC,iBAAiB,EAAE,qBAAqB,CAAC,GAAG,CAAC,EAAE,EAAE,GAAG,CAAC,iBAAiB,CAAC;QACvE,sBAAsB,EAAE,GAAG,CAAC,sBAAsB;QAClD,wBAAwB,EAAE,GAAG,CAAC,wBAAwB;QACtD,SAAS,EAAE,GAAG,CAAC,SAAS;QACxB,UAAU,EAAE,GAAG,CAAC,UAAU;QAC1B,WAAW,EAAE,GAAG,CAAC,WAAW;QAC5B,UAAU,EAAE,GAAG,CAAC,UAAU;KAC3B,CAAC;AACJ,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,UAAU,CAAC,GAAa;IACtC,OAAO;QACL,EAAE,EAAE,GAAG,CAAC,EAAE;QACV,WAAW,EAAE,GAAG,CAAC,WAAW;QAC5B,aAAa,EAAE,GAAG,CAAC,aAAa;QAChC,WAAW,EAAE,GAAG,CAAC,WAAW;QAC5B,YAAY,EAAE;YACZ,CAAC,EAAE,GAAG,CAAC,MAAM;YACb,CAAC,EAAE,GAAG,CAAC,MAAM;YACb,KAAK,EAAE,GAAG,CAAC,UAAU;YACrB,MAAM,EAAE,GAAG,CAAC,WAAW;SACxB;QACD,WAAW,EAAE,GAAG,CAAC,WAAW;QAC5B,MAAM,EAAE,GAAG,CAAC,MAAM;QAClB,UAAU,EAAE;YACV,KAAK,EAAE,GAAG,CAAC,KAAK;YAChB,MAAM,EAAE,GAAG,CAAC,MAAM;SACnB;QACD,cAAc,EAAE,GAAG,CAAC,cAAc;QAClC,SAAS,EAAE,GAAG,CAAC,SAAS;QACxB,UAAU,EAAE,YAAY,CAAC,GAAG,CAAC,UAAU,EAAE,kBAAkB,EAAE,WAAW,EAAE,GAAG,CAAC,EAAE,CAAC;QACjF,eAAe,EAAE,GAAG,CAAC,eAAe;QACpC,mBAAmB,EAAE,GAAG,CAAC,mBAAmB;YAC1C,CAAC,CAAC,sBAAsB,CAAC,GAAG,CAAC,EAAE,EAAE,GAAG,CAAC,mBAAmB,CAAC;YACzD,CAAC,CAAC,IAAI;QACR,gBAAgB,EAAE,GAAG,CAAC,gBAAgB;QACtC,SAAS,EAAE,GAAG,CAAC,SAAS;QACxB,cAAc,EAAE,GAAG,CAAC,cAAc;QAClC,gBAAgB,EAAE,GAAG,CAAC,gBAAgB;QACtC,eAAe,EAAE,GAAG,CAAC,eAAe;QACpC,YAAY,EAAE,GAAG,CAAC,YAAY;QAC9B,aAAa,EAAE,GAAG,CAAC,aAAa;QAChC,UAAU,EAAE,GAAG,CAAC,UAAU;QAC1B,aAAa,EAAE,GAAG,CAAC,aAAa;QAChC,UAAU,EAAE,GAAG,CAAC,UAAU,IAAI,IAAI;QAClC,gBAAgB,EAAE,OAAO,CAAC,GAAG,CAAC,gBAAgB,CAAC;QAC/C,YAAY,EAAE,GAAG,CAAC,YAAY,IAAI,IAAI;KACvC,CAAC;AACJ,CAAC"}

package/dist/services/storage/database/document-operations.d.ts

@@ -0,0 +1,145 @@
+/**
+ * Document operations for DatabaseService
+ *
+ * Handles all CRUD operations for documents including
+ * insert, get, list, update, and delete with cascade.
+ */
+import Database from 'better-sqlite3';
+import { Document, DocumentStatus } from '../../../models/document.js';
+import { ListDocumentsOptions } from './types.js';
+/**
+ * Encode a cursor from a (created_at, id) tuple.
+ * Uses base64url encoding for URL-safe transport.
+ *
+ * @param createdAt - ISO 8601 timestamp
+ * @param id - Document UUID
+ * @returns Base64url-encoded cursor string
+ */
+export declare function encodeCursor(createdAt: string, id: string): string;
+/**
+ * Decode a cursor back to a (created_at, id) tuple.
+ *
+ * @param cursor - Base64url-encoded cursor string
+ * @returns Decoded cursor with created_at and id
+ * @throws Error if cursor is invalid or malformed
+ */
+export declare function decodeCursor(cursor: string): {
+    created_at: string;
+    id: string;
+};
+/**
+ * Insert a new document
+ *
+ * @param db - Database connection
+ * @param doc - Document data (created_at will be generated)
+ * @param updateMetadataCounts - Callback to update metadata counts
+ * @returns string - The document ID
+ */
+export declare function insertDocument(db: Database.Database, doc: Omit<Document, 'created_at'>, updateMetadataCounts: () => void): string;
+/**
+ * Get a document by ID
+ *
+ * @param db - Database connection
+ * @param id - Document ID
+ * @returns Document | null - The document or null if not found
+ */
+export declare function getDocument(db: Database.Database, id: string): Document | null;
+/**
+ * Get a document by file path
+ *
+ * @param db - Database connection
+ * @param filePath - Full file path
+ * @returns Document | null - The document or null if not found
+ */
+export declare function getDocumentByPath(db: Database.Database, filePath: string): Document | null;
+/**
+ * Get a document by file hash
+ *
+ * @param db - Database connection
+ * @param fileHash - SHA-256 file hash
+ * @returns Document | null - The document or null if not found
+ */
+export declare function getDocumentByHash(db: Database.Database, fileHash: string): Document | null;
+/**
+ * Result from listDocuments when cursor-based pagination is used
+ */
+export interface ListDocumentsResult {
+    documents: Document[];
+    /** Cursor for the next page (null if no more results) */
+    next_cursor: string | null;
+}
+/**
+ * List documents with optional filtering.
+ *
+ * Supports both offset-based and cursor-based pagination:
+ * - When `cursor` is provided, uses keyset pagination (WHERE created_at < cursor.created_at
+ *   OR (created_at = cursor.created_at AND id < cursor.id)) and ignores offset.
+ * - When `cursor` is absent, uses traditional LIMIT/OFFSET.
+ *
+ * @param db - Database connection
+ * @param options - Optional filter options (status, limit, offset, cursor)
+ * @returns Document[] - Array of documents (backward-compatible)
+ */
+export declare function listDocuments(db: Database.Database, options?: ListDocumentsOptions): Document[];
+/**
+ * List documents with cursor-based pagination support.
+ *
+ * Returns both the documents and a next_cursor for fetching the next page.
+ *
+ * @param db - Database connection
+ * @param options - Optional filter options (status, limit, offset, cursor)
+ * @returns ListDocumentsResult with documents and next_cursor
+ */
+export declare function listDocumentsWithCursor(db: Database.Database, options?: ListDocumentsOptions): ListDocumentsResult;
+/**
+ * Update document status
+ *
+ * @param db - Database connection
+ * @param id - Document ID
+ * @param status - New status
+ * @param errorMessage - Optional error message (for 'failed' status)
+ * @param updateMetadataModified - Callback to update metadata modified timestamp
+ */
+export declare function updateDocumentStatus(db: Database.Database, id: string, status: DocumentStatus, errorMessage: string | undefined, updateMetadataModified: () => void): void;
+/**
+ * Update document when OCR completes
+ *
+ * @param db - Database connection
+ * @param id - Document ID
+ * @param pageCount - Number of pages processed
+ * @param ocrCompletedAt - ISO 8601 completion timestamp
+ * @param updateMetadataModified - Callback to update metadata modified timestamp
+ */
+export declare function updateDocumentOCRComplete(db: Database.Database, id: string, pageCount: number, ocrCompletedAt: string, updateMetadataModified: () => void): void;
+/**
+ * Update document metadata (title, author, subject) from OCR extraction
+ *
+ * @param db - Database connection
+ * @param id - Document ID
+ * @param metadata - Metadata fields to update (null values are ignored via COALESCE)
+ * @param updateMetadataModified - Callback to update metadata modified timestamp
+ */
+export declare function updateDocumentMetadata(db: Database.Database, id: string, metadata: {
+    docTitle?: string | null;
+    docAuthor?: string | null;
+    docSubject?: string | null;
+}, updateMetadataModified: () => void): void;
+/**
+ * Delete a document and all related data (CASCADE DELETE)
+ *
+ * @param db - Database connection
+ * @param id - Document ID to delete
+ * @param updateMetadataCounts - Callback to update metadata counts
+ */
+export declare function deleteDocument(db: Database.Database, id: string, updateMetadataCounts: () => void): void;
+/**
+ * Clean all derived data for a document, keeping the document record and its DOCUMENT-level provenance.
+ *
+ * Deletes: vec_embeddings, embeddings, images, chunks, ocr_results, and non-root provenance records.
+ * This is used by retry_failed to reset a document to a clean "pending" state.
+ *
+ * @param db - Database connection
+ * @param documentId - Document ID to clean
+ */
+export declare function cleanDocumentDerivedData(db: Database.Database, documentId: string): void;
+//# sourceMappingURL=document-operations.d.ts.map

package/dist/services/storage/database/document-operations.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"document-operations.d.ts","sourceRoot":"","sources":["../../../../src/services/storage/database/document-operations.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,QAAQ,MAAM,gBAAgB,CAAC;AAEtC,OAAO,EAAE,QAAQ,EAAE,cAAc,EAAE,MAAM,6BAA6B,CAAC;AACvE,OAAO,EAAiD,oBAAoB,EAAE,MAAM,YAAY,CAAC;AASjG;;;;;;;GAOG;AACH,wBAAgB,YAAY,CAAC,SAAS,EAAE,MAAM,EAAE,EAAE,EAAE,MAAM,GAAG,MAAM,CAElE;AAED;;;;;;GAMG;AACH,wBAAgB,YAAY,CAAC,MAAM,EAAE,MAAM,GAAG;IAAE,UAAU,EAAE,MAAM,CAAC;IAAC,EAAE,EAAE,MAAM,CAAA;CAAE,CAU/E;AAED;;;;;;;GAOG;AACH,wBAAgB,cAAc,CAC5B,EAAE,EAAE,QAAQ,CAAC,QAAQ,EACrB,GAAG,EAAE,IAAI,CAAC,QAAQ,EAAE,YAAY,CAAC,EACjC,oBAAoB,EAAE,MAAM,IAAI,GAC/B,MAAM,CAiCR;AAED;;;;;;GAMG;AACH,wBAAgB,WAAW,CAAC,EAAE,EAAE,QAAQ,CAAC,QAAQ,EAAE,EAAE,EAAE,MAAM,GAAG,QAAQ,GAAG,IAAI,CAI9E;AAED;;;;;;GAMG;AACH,wBAAgB,iBAAiB,CAAC,EAAE,EAAE,QAAQ,CAAC,QAAQ,EAAE,QAAQ,EAAE,MAAM,GAAG,QAAQ,GAAG,IAAI,CAI1F;AAED;;;;;;GAMG;AACH,wBAAgB,iBAAiB,CAAC,EAAE,EAAE,QAAQ,CAAC,QAAQ,EAAE,QAAQ,EAAE,MAAM,GAAG,QAAQ,GAAG,IAAI,CAI1F;AAED;;GAEG;AACH,MAAM,WAAW,mBAAmB;IAClC,SAAS,EAAE,QAAQ,EAAE,CAAC;IACtB,yDAAyD;IACzD,WAAW,EAAE,MAAM,GAAG,IAAI,CAAC;CAC5B;AAED;;;;;;;;;;;GAWG;AACH,wBAAgB,aAAa,CAAC,EAAE,EAAE,QAAQ,CAAC,QAAQ,EAAE,OAAO,CAAC,EAAE,oBAAoB,GAAG,QAAQ,EAAE,CAG/F;AAED;;;;;;;;GAQG;AACH,wBAAgB,uBAAuB,CACrC,EAAE,EAAE,QAAQ,CAAC,QAAQ,EACrB,OAAO,CAAC,EAAE,oBAAoB,GAC7B,mBAAmB,CA6CrB;AAED;;;;;;;;GAQG;AACH,wBAAgB,oBAAoB,CAClC,EAAE,EAAE,QAAQ,CAAC,QAAQ,EACrB,EAAE,EAAE,MAAM,EACV,MAAM,EAAE,cAAc,EACtB,YAAY,EAAE,MAAM,GAAG,SAAS,EAChC,sBAAsB,EAAE,MAAM,IAAI,GACjC,IAAI,CAgBN;AAED;;;;;;;;GAQG;AACH,wBAAgB,yBAAyB,CACvC,EAAE,EAAE,QAAQ,CAAC,QAAQ,EACrB,EAAE,EAAE,MAAM,EACV,SAAS,EAAE,MAAM,EACjB,cAAc,EAAE,MAAM,EACtB,sBAAsB,EAAE,MAAM,IAAI,GACjC,IAAI,CAgBN;AAED;;;;;;;GAOG;AACH,wBAAgB,sBAAsB,CACpC,EAAE,EAAE,QAAQ,CAAC,QAAQ,EACrB,EAAE,EAAE,MAAM,EACV,QAAQ,EAAE;IAAE,QAAQ,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IAAC,SAAS,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;IAAC,UAAU,CAAC,EAAE,MAAM,GAAG,IAAI,CAAA;CAAE,EAC7F,sBAAsB,EAAE,MAAM,IAAI,GACjC,IAAI,CAkBN;AAmPD;;;;;;GAMG;AACH,wBAAgB,cAAc,CAC5B,EAAE,EAAE,QAAQ,CAAC,QAAQ,EACrB,EAAE,EAAE,MAAM,EACV,oBAAoB,EAAE,MAAM,IAAI,GAC/B,IAAI,CA4DN;AAED;;;;;;;;GAQG;AACH,wBAAgB,wBAAwB,CAAC,EAAE,EAAE,QAAQ,CAAC,QAAQ,EAAE,UAAU,EAAE,MAAM,GAAG,IAAI,CAuCxF"}

package/dist/services/storage/database/document-operations.js

@@ -0,0 +1,498 @@
+/**
+ * Document operations for DatabaseService
+ *
+ * Handles all CRUD operations for documents including
+ * insert, get, list, update, and delete with cascade.
+ */
+import { v4 as uuidv4 } from 'uuid';
+import { DatabaseError, DatabaseErrorCode } from './types.js';
+import { runWithForeignKeyCheck } from './helpers.js';
+import { rowToDocument } from './converters.js';
+import { computeHash } from '../../../utils/hash.js';
+// ═══════════════════════════════════════════════════════════════════════════════
+// CURSOR-BASED PAGINATION HELPERS
+// ═══════════════════════════════════════════════════════════════════════════════
+/**
+ * Encode a cursor from a (created_at, id) tuple.
+ * Uses base64url encoding for URL-safe transport.
+ *
+ * @param createdAt - ISO 8601 timestamp
+ * @param id - Document UUID
+ * @returns Base64url-encoded cursor string
+ */
+export function encodeCursor(createdAt, id) {
+    return Buffer.from(JSON.stringify({ created_at: createdAt, id })).toString('base64url');
+}
+/**
+ * Decode a cursor back to a (created_at, id) tuple.
+ *
+ * @param cursor - Base64url-encoded cursor string
+ * @returns Decoded cursor with created_at and id
+ * @throws Error if cursor is invalid or malformed
+ */
+export function decodeCursor(cursor) {
+    try {
+        const decoded = JSON.parse(Buffer.from(cursor, 'base64url').toString('utf-8'));
+        if (typeof decoded.created_at !== 'string' || typeof decoded.id !== 'string') {
+            throw new Error('Invalid cursor format: missing created_at or id');
+        }
+        return { created_at: decoded.created_at, id: decoded.id };
+    }
+    catch (error) {
+        throw new Error(`Invalid cursor: ${error instanceof Error ? error.message : String(error)}`);
+    }
+}
+/**
+ * Insert a new document
+ *
+ * @param db - Database connection
+ * @param doc - Document data (created_at will be generated)
+ * @param updateMetadataCounts - Callback to update metadata counts
+ * @returns string - The document ID
+ */
+export function insertDocument(db, doc, updateMetadataCounts) {
+    const created_at = new Date().toISOString();
+    const stmt = db.prepare(`
+    INSERT INTO documents (
+      id, file_path, file_name, file_hash, file_size, file_type,
+      status, page_count, provenance_id, created_at, modified_at,
+      ocr_completed_at, error_message
+    ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+  `);
+    runWithForeignKeyCheck(stmt, [
+        doc.id,
+        doc.file_path,
+        doc.file_name,
+        doc.file_hash,
+        doc.file_size,
+        doc.file_type,
+        doc.status,
+        doc.page_count,
+        doc.provenance_id,
+        created_at,
+        doc.modified_at,
+        doc.ocr_completed_at,
+        doc.error_message,
+    ], `inserting document: provenance_id "${doc.provenance_id}" does not exist`);
+    updateMetadataCounts();
+    return doc.id;
+}
+/**
+ * Get a document by ID
+ *
+ * @param db - Database connection
+ * @param id - Document ID
+ * @returns Document | null - The document or null if not found
+ */
+export function getDocument(db, id) {
+    const stmt = db.prepare('SELECT * FROM documents WHERE id = ?');
+    const row = stmt.get(id);
+    return row ? rowToDocument(row) : null;
+}
+/**
+ * Get a document by file path
+ *
+ * @param db - Database connection
+ * @param filePath - Full file path
+ * @returns Document | null - The document or null if not found
+ */
+export function getDocumentByPath(db, filePath) {
+    const stmt = db.prepare('SELECT * FROM documents WHERE file_path = ?');
+    const row = stmt.get(filePath);
+    return row ? rowToDocument(row) : null;
+}
+/**
+ * Get a document by file hash
+ *
+ * @param db - Database connection
+ * @param fileHash - SHA-256 file hash
+ * @returns Document | null - The document or null if not found
+ */
+export function getDocumentByHash(db, fileHash) {
+    const stmt = db.prepare('SELECT * FROM documents WHERE file_hash = ?');
+    const row = stmt.get(fileHash);
+    return row ? rowToDocument(row) : null;
+}
+/**
+ * List documents with optional filtering.
+ *
+ * Supports both offset-based and cursor-based pagination:
+ * - When `cursor` is provided, uses keyset pagination (WHERE created_at < cursor.created_at
+ *   OR (created_at = cursor.created_at AND id < cursor.id)) and ignores offset.
+ * - When `cursor` is absent, uses traditional LIMIT/OFFSET.
+ *
+ * @param db - Database connection
+ * @param options - Optional filter options (status, limit, offset, cursor)
+ * @returns Document[] - Array of documents (backward-compatible)
+ */
+export function listDocuments(db, options) {
+    const result = listDocumentsWithCursor(db, options);
+    return result.documents;
+}
+/**
+ * List documents with cursor-based pagination support.
+ *
+ * Returns both the documents and a next_cursor for fetching the next page.
+ *
+ * @param db - Database connection
+ * @param options - Optional filter options (status, limit, offset, cursor)
+ * @returns ListDocumentsResult with documents and next_cursor
+ */
+export function listDocumentsWithCursor(db, options) {
+    const conditions = [];
+    const params = [];
+    if (options?.status) {
+        conditions.push('status = ?');
+        params.push(options.status);
+    }
+    // Cursor-based pagination: keyset filtering
+    if (options?.cursor) {
+        const decoded = decodeCursor(options.cursor);
+        conditions.push('(created_at < ? OR (created_at = ? AND id < ?))');
+        params.push(decoded.created_at, decoded.created_at, decoded.id);
+    }
+    let query = 'SELECT * FROM documents';
+    if (conditions.length > 0) {
+        query += ' WHERE ' + conditions.join(' AND ');
+    }
+    query += ' ORDER BY created_at DESC, id DESC';
+    const limit = options?.limit ?? 10000;
+    query += ' LIMIT ?';
+    params.push(limit);
+    // Only apply OFFSET when NOT using cursor-based pagination
+    if (!options?.cursor && options?.offset !== undefined) {
+        query += ' OFFSET ?';
+        params.push(options.offset);
+    }
+    const stmt = db.prepare(query);
+    const rows = stmt.all(...params);
+    const documents = rows.map(rowToDocument);
+    // Compute next_cursor from the last row
+    let next_cursor = null;
+    if (documents.length > 0 && documents.length === limit) {
+        const lastDoc = documents[documents.length - 1];
+        next_cursor = encodeCursor(lastDoc.created_at, lastDoc.id);
+    }
+    return { documents, next_cursor };
+}
+/**
+ * Update document status
+ *
+ * @param db - Database connection
+ * @param id - Document ID
+ * @param status - New status
+ * @param errorMessage - Optional error message (for 'failed' status)
+ * @param updateMetadataModified - Callback to update metadata modified timestamp
+ */
+export function updateDocumentStatus(db, id, status, errorMessage, updateMetadataModified) {
+    const modified_at = new Date().toISOString();
+    const stmt = db.prepare(`
+    UPDATE documents
+    SET status = ?, error_message = ?, modified_at = ?
+    WHERE id = ?
+  `);
+    const result = stmt.run(status, errorMessage ?? null, modified_at, id);
+    if (result.changes === 0) {
+        throw new DatabaseError(`Document "${id}" not found`, DatabaseErrorCode.DOCUMENT_NOT_FOUND);
+    }
+    updateMetadataModified();
+}
+/**
+ * Update document when OCR completes
+ *
+ * @param db - Database connection
+ * @param id - Document ID
+ * @param pageCount - Number of pages processed
+ * @param ocrCompletedAt - ISO 8601 completion timestamp
+ * @param updateMetadataModified - Callback to update metadata modified timestamp
+ */
+export function updateDocumentOCRComplete(db, id, pageCount, ocrCompletedAt, updateMetadataModified) {
+    const modified_at = new Date().toISOString();
+    const stmt = db.prepare(`
+    UPDATE documents
+    SET status = 'processing', page_count = ?, ocr_completed_at = ?, modified_at = ?
+    WHERE id = ?
+  `);
+    const result = stmt.run(pageCount, ocrCompletedAt, modified_at, id);
+    if (result.changes === 0) {
+        throw new DatabaseError(`Document "${id}" not found`, DatabaseErrorCode.DOCUMENT_NOT_FOUND);
+    }
+    updateMetadataModified();
+}
+/**
+ * Update document metadata (title, author, subject) from OCR extraction
+ *
+ * @param db - Database connection
+ * @param id - Document ID
+ * @param metadata - Metadata fields to update (null values are ignored via COALESCE)
+ * @param updateMetadataModified - Callback to update metadata modified timestamp
+ */
+export function updateDocumentMetadata(db, id, metadata, updateMetadataModified) {
+    const modified_at = new Date().toISOString();
+    const stmt = db.prepare(`
+    UPDATE documents
+    SET doc_title = COALESCE(?, doc_title),
+        doc_author = COALESCE(?, doc_author),
+        doc_subject = COALESCE(?, doc_subject),
+        modified_at = ?
+    WHERE id = ?
+  `);
+    const result = stmt.run(metadata.docTitle ?? null, metadata.docAuthor ?? null, metadata.docSubject ?? null, modified_at, id);
+    if (result.changes > 0)
+        updateMetadataModified();
+}
+/**
+ * Shared cleanup: delete all derived records for a document.
+ *
+ * Deletion order (FK-safe):
+ *   1. vec_embeddings (no inbound FKs)
+ *   2. NULL images.vlm_embedding_id (break circular FK with embeddings)
+ *   3. Re-queue orphaned images from other documents (VLM dedup)
+ *   4. embeddings (covers chunk, VLM, and extraction types in one pass)
+ *   5. images (safe after embeddings.image_id references gone)
+ *   6. chunks
+ *   7. extractions (before ocr_results: extractions.ocr_result_id -> ocr_results)
+ *   8. ocr_results
+ *   9. FTS metadata count updates (ids 1, 2, 3)
+ *
+ * @returns The number of embedding IDs deleted (for logging)
+ */
+function deleteDerivedRecords(db, documentId, caller) {
+    // M-3: Count embeddings first, then use subquery DELETE instead of loading all IDs
+    const embeddingCount = db.prepare('SELECT COUNT(*) as cnt FROM embeddings WHERE document_id = ?').get(documentId).cnt;
+    // Delete from vec_embeddings using a single subquery
+    db.prepare('DELETE FROM vec_embeddings WHERE embedding_id IN (SELECT id FROM embeddings WHERE document_id = ?)').run(documentId);
+    // Break circular FK: images.vlm_embedding_id → embeddings ↔ embeddings.image_id → images
+    // NULL out vlm_embedding_id on THIS document's images so embeddings can be deleted
+    db.prepare('UPDATE images SET vlm_embedding_id = NULL WHERE document_id = ?').run(documentId);
+    // Re-queue OTHER documents' images that shared embeddings via VLM dedup.
+    // Setting vlm_status='pending' ensures they get re-processed instead of
+    // silently remaining 'complete' but invisible to search (orphaned).
+    const orphanedImages = db
+        .prepare(`
+    SELECT id, document_id FROM images
+    WHERE vlm_embedding_id IN (SELECT id FROM embeddings WHERE document_id = ?)
+    AND document_id != ?
+  `)
+        .all(documentId, documentId);
+    if (orphanedImages.length > 0) {
+        console.error(`[WARN] ${caller} "${documentId}": re-queuing ${orphanedImages.length} images from other documents ` +
+            `that shared VLM embeddings (document_ids: ${[...new Set(orphanedImages.map((i) => i.document_id))].join(', ')})`);
+        db.prepare(`
+      UPDATE images SET vlm_embedding_id = NULL, vlm_status = 'pending'
+      WHERE vlm_embedding_id IN (SELECT id FROM embeddings WHERE document_id = ?)
+      AND document_id != ?
+    `).run(documentId, documentId);
+    }
+    // Delete entity_tags referencing entities about to be deleted (polymorphic FK, no CASCADE)
+    try {
+        db.prepare(`
+      DELETE FROM entity_tags WHERE
+        (entity_type = 'document' AND entity_id = ?)
+        OR (entity_type = 'chunk' AND entity_id IN (SELECT id FROM chunks WHERE document_id = ?))
+        OR (entity_type = 'image' AND entity_id IN (SELECT id FROM images WHERE document_id = ?))
+        OR (entity_type = 'extraction' AND entity_id IN (SELECT id FROM extractions WHERE document_id = ?))
+    `).run(documentId, documentId, documentId, documentId);
+    }
+    catch (e) {
+        const msg = e instanceof Error ? e.message : String(e);
+        if (!msg.includes('no such table'))
+            throw e;
+        console.error('[document-operations] entity_tags table not found, skipping:', msg);
+    }
+    // Delete from embeddings (safe: images.vlm_embedding_id already NULLed)
+    db.prepare('DELETE FROM embeddings WHERE document_id = ?').run(documentId);
+    // Delete from images (safe: embeddings.image_id references gone)
+    db.prepare('DELETE FROM images WHERE document_id = ?').run(documentId);
+    // Decrement cluster document_count before removing assignments
+    db.prepare(`UPDATE clusters SET document_count = document_count - 1
+     WHERE id IN (SELECT cluster_id FROM document_clusters WHERE document_id = ? AND cluster_id IS NOT NULL)`).run(documentId);
+    // Delete document-cluster assignments
+    db.prepare('DELETE FROM document_clusters WHERE document_id = ?').run(documentId);
+    // Delete comparisons referencing this document
+    db.prepare('DELETE FROM comparisons WHERE document_id_1 = ? OR document_id_2 = ?').run(documentId, documentId);
+    // Delete form_fills linked to this document via source_file_hash
+    // (form_fills has no document_id FK — it joins through source_file_hash)
+    try {
+        const docRow = db.prepare('SELECT file_hash FROM documents WHERE id = ?').get(documentId);
+        if (docRow) {
+            db.prepare('DELETE FROM form_fills WHERE source_file_hash = ?').run(docRow.file_hash);
+        }
+    }
+    catch (e) {
+        const msg = e instanceof Error ? e.message : String(e);
+        if (!msg.includes('no such table'))
+            throw e;
+        console.error('[document-operations] form_fills table not found, skipping:', msg);
+    }
+    // Delete from chunks
+    db.prepare('DELETE FROM chunks WHERE document_id = ?').run(documentId);
+    // Delete from extractions (BEFORE ocr_results: extractions.ocr_result_id REFERENCES ocr_results(id))
+    db.prepare('DELETE FROM extractions WHERE document_id = ?').run(documentId);
+    // Delete from ocr_results (safe now that extractions are gone)
+    db.prepare('DELETE FROM ocr_results WHERE document_id = ?').run(documentId);
+    // Delete uploaded_files whose provenance_id references this document's provenance chain.
+    // Must happen before provenance cleanup (callers delete provenance after this function).
+    // uploaded_files has provenance_id NOT NULL REFERENCES provenance(id).
+    try {
+        const docForProv = db.prepare('SELECT provenance_id FROM documents WHERE id = ?').get(documentId);
+        if (docForProv) {
+            db.prepare('DELETE FROM uploaded_files WHERE provenance_id IN (SELECT id FROM provenance WHERE root_document_id = ?)').run(docForProv.provenance_id);
+        }
+    }
+    catch (e) {
+        const msg = e instanceof Error ? e.message : String(e);
+        if (!msg.includes('no such table'))
+            throw e;
+        console.error('[document-operations] uploaded_files table not found, skipping:', msg);
+    }
+    // Update FTS metadata counts after chunk/embedding deletion
+    try {
+        const chunkCount = db.prepare('SELECT COUNT(*) as cnt FROM chunks').get()
+            .cnt;
+        db.prepare(`
+      UPDATE fts_index_metadata SET chunks_indexed = ?, last_rebuild_at = ?
+      WHERE id = 1
+    `).run(chunkCount, new Date().toISOString());
+        // Update VLM FTS metadata if table exists
+        const vlmCount = db.prepare('SELECT COUNT(*) as cnt FROM embeddings WHERE image_id IS NOT NULL').get().cnt;
+        db.prepare(`
+      UPDATE fts_index_metadata SET chunks_indexed = ?, last_rebuild_at = ?
+      WHERE id = 2
+    `).run(vlmCount, new Date().toISOString());
+        // Update extractions FTS metadata (id=3)
+        const extCount = db.prepare('SELECT COUNT(*) as cnt FROM extractions').get().cnt;
+        db.prepare(`
+      UPDATE fts_index_metadata SET chunks_indexed = ?, last_rebuild_at = ?
+      WHERE id = 3
+    `).run(extCount, new Date().toISOString());
+    }
+    catch (e) {
+        // Only ignore "no such table" errors from older schemas pre-v4
+        const msg = e instanceof Error ? e.message : String(e);
+        if (!msg.includes('no such table')) {
+            throw e;
+        }
+        console.error('[document-operations] fts_index_metadata table not found, skipping FTS update:', msg);
+    }
+    return embeddingCount;
+}
+/**
+ * Get or create the synthetic ORPHANED_ROOT provenance record.
+ * Used to re-parent provenance records when their original document is deleted
+ * but surviving clusters still reference them (P1.4).
+ *
+ * @param db - Database connection
+ * @returns The ID of the ORPHANED_ROOT provenance record
+ */
+function getOrCreateOrphanedRoot(db) {
+    const existing = db
+        .prepare("SELECT id FROM provenance WHERE root_document_id = 'ORPHANED_ROOT' AND type = 'DOCUMENT' LIMIT 1")
+        .get();
+    if (existing) {
+        return existing.id;
+    }
+    // Create synthetic orphaned root provenance
+    const id = uuidv4();
+    const now = new Date().toISOString();
+    const contentHash = computeHash('ORPHANED_ROOT');
+    db.prepare(`
+    INSERT INTO provenance (
+      id, type, created_at, processed_at, source_type, source_id,
+      root_document_id, content_hash, input_hash, processor,
+      processor_version, processing_params, parent_id, parent_ids,
+      chain_depth, chain_path
+    ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+  `).run(id, 'DOCUMENT', now, now, 'FILE', null, 'ORPHANED_ROOT', contentHash, null, 'system', '1.0.0', '{}', null, '[]', 0, '["DOCUMENT"]');
+    return id;
+}
+/**
+ * Delete a document and all related data (CASCADE DELETE)
+ *
+ * @param db - Database connection
+ * @param id - Document ID to delete
+ * @param updateMetadataCounts - Callback to update metadata counts
+ */
+export function deleteDocument(db, id, updateMetadataCounts) {
+    // First check document exists (outside transaction - read-only)
+    const doc = getDocument(db, id);
+    if (!doc) {
+        throw new DatabaseError(`Document "${id}" not found`, DatabaseErrorCode.DOCUMENT_NOT_FOUND);
+    }
+    // H-5: Wrap entire cascade delete in a transaction so a crash mid-sequence
+    // cannot leave the database in an inconsistent state.
+    const runInTransaction = db.transaction(() => {
+        deleteDerivedRecords(db, id, 'deleteDocument');
+        // Delete the document itself BEFORE provenance
+        // (document has FK to provenance via provenance_id)
+        db.prepare('DELETE FROM documents WHERE id = ?').run(id);
+        // Delete from provenance - must delete in reverse chain_depth order
+        // due to self-referential FKs on source_id and parent_id
+        // NOTE: root_document_id stores the document's provenance_id, NOT document id
+        const provenanceIds = db
+            .prepare('SELECT id FROM provenance WHERE root_document_id = ? ORDER BY chain_depth DESC')
+            .all(doc.provenance_id);
+        // P1.4: Get or create orphaned root provenance for re-parenting
+        const orphanedRootId = getOrCreateOrphanedRoot(db);
+        // Pre-clear self-referencing FKs (parent_id, source_id) on provenance records being deleted.
+        // Within the same chain_depth, parent provenance may appear before child provenance in the
+        // iteration order, causing FK violations. NULLing these first breaks the circular references.
+        const clearSelfRefStmt = db.prepare('UPDATE provenance SET parent_id = NULL, source_id = NULL WHERE id = ?');
+        for (const { id: provId } of provenanceIds) {
+            clearSelfRefStmt.run(provId);
+        }
+        const deleteProvStmt = db.prepare('DELETE FROM provenance WHERE id = ?');
+        const clusterRefCheck = db.prepare('SELECT COUNT(*) as cnt FROM clusters WHERE provenance_id = ?');
+        const reparentProvStmt = db.prepare('UPDATE provenance SET source_id = NULL, parent_id = ?, root_document_id = ? WHERE id = ?');
+        for (const { id: provId } of provenanceIds) {
+            // Skip CLUSTERING provenance still referenced by clusters (NOT NULL FK).
+            // Re-parent to orphaned root so provenance chain is preserved (P1.4).
+            // These are cleaned up when the cluster run is deleted.
+            const clusterRefs = clusterRefCheck.get(provId).cnt;
+            if (clusterRefs > 0) {
+                reparentProvStmt.run(orphanedRootId, 'ORPHANED_ROOT', provId);
+                continue;
+            }
+            deleteProvStmt.run(provId);
+        }
+        // Update metadata counts inside transaction for atomicity
+        updateMetadataCounts();
+    });
+    runInTransaction();
+}
+/**
+ * Clean all derived data for a document, keeping the document record and its DOCUMENT-level provenance.
+ *
+ * Deletes: vec_embeddings, embeddings, images, chunks, ocr_results, and non-root provenance records.
+ * This is used by retry_failed to reset a document to a clean "pending" state.
+ *
+ * @param db - Database connection
+ * @param documentId - Document ID to clean
+ */
+export function cleanDocumentDerivedData(db, documentId) {
+    // Validate document exists (outside transaction - read-only)
+    const doc = getDocument(db, documentId);
+    if (!doc) {
+        throw new DatabaseError(`Document "${documentId}" not found`, DatabaseErrorCode.DOCUMENT_NOT_FOUND);
+    }
+    // H-5: Wrap cleanup in a transaction so partial deletes cannot leave
+    // the database in an inconsistent state.
+    let embeddingCount = 0;
+    let provenanceCount = 0;
+    const runInTransaction = db.transaction(() => {
+        embeddingCount = deleteDerivedRecords(db, documentId, 'cleanDocumentDerivedData');
+        // Delete non-root provenance records (keep DOCUMENT-level provenance at chain_depth=0)
+        // root_document_id stores the document's provenance_id, NOT document id
+        const nonRootProvIds = db
+            .prepare('SELECT id FROM provenance WHERE root_document_id = ? AND chain_depth > 0 ORDER BY chain_depth DESC')
+            .all(doc.provenance_id);
+        const deleteProvStmt = db.prepare('DELETE FROM provenance WHERE id = ?');
+        for (const { id: provId } of nonRootProvIds) {
+            deleteProvStmt.run(provId);
+        }
+        provenanceCount = nonRootProvIds.length;
+    });
+    runInTransaction();
+    console.error(`[INFO] Cleaned derived data for document ${documentId}: ${embeddingCount} embeddings, ${provenanceCount} provenance records removed`);
+}
+//# sourceMappingURL=document-operations.js.map