@push.rocks/smartmongo 3.0.0 → 4.1.1

package/ts/tsmdb/storage/FileStorageAdapter.ts

@@ -1,6 +1,17 @@
 import * as plugins from '../tsmdb.plugins.js';
 import type { IStorageAdapter } from './IStorageAdapter.js';
 import type { IStoredDocument, IOpLogEntry, Document } from '../types/interfaces.js';
+import { calculateDocumentChecksum, verifyChecksum } from '../utils/checksum.js';
+
+/**
+ * File storage adapter options
+ */
+export interface IFileStorageAdapterOptions {
+  /** Enable checksum verification for data integrity */
+  enableChecksums?: boolean;
+  /** Throw error on checksum mismatch (default: false, just log warning) */
+  strictChecksums?: boolean;
+}
 
 /**
  * File-based storage adapter for TsmDB
@@ -11,9 +22,13 @@ export class FileStorageAdapter implements IStorageAdapter {
   private opLogCounter = 0;
   private initialized = false;
   private fs = new plugins.smartfs.SmartFs(new plugins.smartfs.SmartFsProviderNode());
+  private enableChecksums: boolean;
+  private strictChecksums: boolean;
 
-  constructor(basePath: string) {
+  constructor(basePath: string, options?: IFileStorageAdapterOptions) {
     this.basePath = basePath;
+    this.enableChecksums = options?.enableChecksums ?? false;
+    this.strictChecksums = options?.strictChecksums ?? false;
   }
 
   // ============================================================================
@@ -68,6 +83,45 @@ export class FileStorageAdapter implements IStorageAdapter {
     return doc;
   }
 
+  /**
+   * Verify document checksum and handle errors
+   */
+  private verifyDocumentChecksum(doc: any): boolean {
+    if (!this.enableChecksums || !doc._checksum) {
+      return true;
+    }
+
+    const isValid = verifyChecksum(doc);
+    if (!isValid) {
+      const errorMsg = `Checksum mismatch for document ${doc._id}`;
+      if (this.strictChecksums) {
+        throw new Error(errorMsg);
+      } else {
+        console.warn(`WARNING: ${errorMsg}`);
+      }
+    }
+    return isValid;
+  }
+
+  /**
+   * Add checksum to document before storing
+   */
+  private prepareDocumentForStorage(doc: any): any {
+    if (!this.enableChecksums) {
+      return doc;
+    }
+    const checksum = calculateDocumentChecksum(doc);
+    return { ...doc, _checksum: checksum };
+  }
+
+  /**
+   * Remove internal checksum field before returning to user
+   */
+  private cleanDocumentForReturn(doc: any): IStoredDocument {
+    const { _checksum, ...cleanDoc } = doc;
+    return this.restoreObjectIds(cleanDoc);
+  }
+
   // ============================================================================
   // Initialization
   // ============================================================================
@@ -233,7 +287,9 @@ export class FileStorageAdapter implements IStorageAdapter {
       throw new Error(`Duplicate key error: _id ${idStr}`);
     }
 
-    docs.push(storedDoc);
+    // Add checksum if enabled
+    const docToStore = this.prepareDocumentForStorage(storedDoc);
+    docs.push(docToStore);
     await this.writeJsonFile(collPath, docs);
     return storedDoc;
   }
@@ -258,7 +314,9 @@ export class FileStorageAdapter implements IStorageAdapter {
       }
 
       existingIds.add(idStr);
-      docs.push(storedDoc);
+      // Add checksum if enabled
+      const docToStore = this.prepareDocumentForStorage(storedDoc);
+      docs.push(docToStore);
       results.push(storedDoc);
     }
 
@@ -270,10 +328,33 @@ export class FileStorageAdapter implements IStorageAdapter {
     await this.createCollection(dbName, collName);
     const collPath = this.getCollectionPath(dbName, collName);
     const docs = await this.readJsonFile<any[]>(collPath, []);
-    return docs.map(doc => this.restoreObjectIds(doc));
+    return docs.map(doc => {
+      // Verify checksum if enabled
+      this.verifyDocumentChecksum(doc);
+      // Clean and return document without internal checksum field
+      return this.cleanDocumentForReturn(doc);
+    });
+  }
+
+  async findByIds(dbName: string, collName: string, ids: Set<string>): Promise<IStoredDocument[]> {
+    await this.createCollection(dbName, collName);
+    const collPath = this.getCollectionPath(dbName, collName);
+    const docs = await this.readJsonFile<any[]>(collPath, []);
+    const results: IStoredDocument[] = [];
+    for (const doc of docs) {
+      // Verify checksum if enabled
+      this.verifyDocumentChecksum(doc);
+      // Clean and restore document
+      const cleaned = this.cleanDocumentForReturn(doc);
+      if (ids.has(cleaned._id.toHexString())) {
+        results.push(cleaned);
+      }
+    }
+    return results;
   }
 
   async findById(dbName: string, collName: string, id: plugins.bson.ObjectId): Promise<IStoredDocument | null> {
+    // Use findAll which already handles checksum verification
     const docs = await this.findAll(dbName, collName);
     const idStr = id.toHexString();
     return docs.find(d => d._id.toHexString() === idStr) || null;
@@ -291,7 +372,9 @@ export class FileStorageAdapter implements IStorageAdapter {
 
     if (idx === -1) return false;
 
-    docs[idx] = doc;
+    // Add checksum if enabled
+    const docToStore = this.prepareDocumentForStorage(doc);
+    docs[idx] = docToStore;
     await this.writeJsonFile(collPath, docs);
     return true;
   }

package/ts/tsmdb/storage/IStorageAdapter.ts

@@ -90,6 +90,12 @@ export interface IStorageAdapter {
    */
   findAll(dbName: string, collName: string): Promise<IStoredDocument[]>;
 
+  /**
+   * Find documents by a set of _id strings (hex format)
+   * Used for index-accelerated queries
+   */
+  findByIds(dbName: string, collName: string, ids: Set<string>): Promise<IStoredDocument[]>;
+
   /**
    * Find a document by _id
    */

package/ts/tsmdb/storage/MemoryStorageAdapter.ts

@@ -196,6 +196,18 @@ export class MemoryStorageAdapter implements IStorageAdapter {
     return Array.from(collection.values());
   }
 
+  async findByIds(dbName: string, collName: string, ids: Set<string>): Promise<IStoredDocument[]> {
+    const collection = this.ensureCollection(dbName, collName);
+    const results: IStoredDocument[] = [];
+    for (const id of ids) {
+      const doc = collection.get(id);
+      if (doc) {
+        results.push(doc);
+      }
+    }
+    return results;
+  }
+
   async findById(dbName: string, collName: string, id: plugins.bson.ObjectId): Promise<IStoredDocument | null> {
     const collection = this.ensureCollection(dbName, collName);
     return collection.get(id.toHexString()) || null;

package/ts/tsmdb/storage/WAL.ts

@@ -0,0 +1,375 @@
+import * as plugins from '../tsmdb.plugins.js';
+import type { Document, IStoredDocument } from '../types/interfaces.js';
+
+/**
+ * WAL entry operation types
+ */
+export type TWalOperation = 'insert' | 'update' | 'delete' | 'checkpoint' | 'begin' | 'commit' | 'abort';
+
+/**
+ * WAL entry structure
+ */
+export interface IWalEntry {
+  /** Log Sequence Number - monotonically increasing */
+  lsn: number;
+  /** Timestamp of the operation */
+  timestamp: number;
+  /** Operation type */
+  operation: TWalOperation;
+  /** Database name */
+  dbName: string;
+  /** Collection name */
+  collName: string;
+  /** Document ID (hex string) */
+  documentId: string;
+  /** Document data (BSON serialized, base64 encoded) */
+  data?: string;
+  /** Previous document data for updates (for rollback) */
+  previousData?: string;
+  /** Transaction ID if part of a transaction */
+  txnId?: string;
+  /** CRC32 checksum of the entry (excluding this field) */
+  checksum: number;
+}
+
+/**
+ * Checkpoint record
+ */
+interface ICheckpointRecord {
+  lsn: number;
+  timestamp: number;
+  lastCommittedLsn: number;
+}
+
+/**
+ * Write-Ahead Log (WAL) for durability and crash recovery
+ *
+ * The WAL ensures durability by writing operations to a log file before
+ * they are applied to the main storage. On crash recovery, uncommitted
+ * operations can be replayed to restore the database to a consistent state.
+ */
+export class WAL {
+  private walPath: string;
+  private currentLsn: number = 0;
+  private lastCheckpointLsn: number = 0;
+  private entries: IWalEntry[] = [];
+  private isInitialized: boolean = false;
+  private fs = new plugins.smartfs.SmartFs(new plugins.smartfs.SmartFsProviderNode());
+
+  // In-memory uncommitted entries per transaction
+  private uncommittedTxns: Map<string, IWalEntry[]> = new Map();
+
+  // Checkpoint interval (number of entries between checkpoints)
+  private checkpointInterval: number = 1000;
+
+  constructor(walPath: string, options?: { checkpointInterval?: number }) {
+    this.walPath = walPath;
+    if (options?.checkpointInterval) {
+      this.checkpointInterval = options.checkpointInterval;
+    }
+  }
+
+  /**
+   * Initialize the WAL, loading existing entries and recovering if needed
+   */
+  async initialize(): Promise<{ recoveredEntries: IWalEntry[] }> {
+    if (this.isInitialized) {
+      return { recoveredEntries: [] };
+    }
+
+    // Ensure WAL directory exists
+    const walDir = this.walPath.substring(0, this.walPath.lastIndexOf('/'));
+    if (walDir) {
+      await this.fs.directory(walDir).recursive().create();
+    }
+
+    // Try to load existing WAL
+    const exists = await this.fs.file(this.walPath).exists();
+    if (exists) {
+      const content = await this.fs.file(this.walPath).encoding('utf8').read();
+      const lines = (content as string).split('\n').filter(line => line.trim());
+
+      for (const line of lines) {
+        try {
+          const entry = JSON.parse(line) as IWalEntry;
+          // Verify checksum
+          if (this.verifyChecksum(entry)) {
+            this.entries.push(entry);
+            if (entry.lsn > this.currentLsn) {
+              this.currentLsn = entry.lsn;
+            }
+            if (entry.operation === 'checkpoint') {
+              this.lastCheckpointLsn = entry.lsn;
+            }
+          }
+        } catch {
+          // Skip corrupted entries
+          console.warn('Skipping corrupted WAL entry');
+        }
+      }
+    }
+
+    this.isInitialized = true;
+
+    // Return entries after last checkpoint that need recovery
+    const recoveredEntries = this.entries.filter(
+      e => e.lsn > this.lastCheckpointLsn &&
+           (e.operation === 'insert' || e.operation === 'update' || e.operation === 'delete')
+    );
+
+    return { recoveredEntries };
+  }
+
+  /**
+   * Log an insert operation
+   */
+  async logInsert(dbName: string, collName: string, doc: IStoredDocument, txnId?: string): Promise<number> {
+    return this.appendEntry({
+      operation: 'insert',
+      dbName,
+      collName,
+      documentId: doc._id.toHexString(),
+      data: this.serializeDocument(doc),
+      txnId,
+    });
+  }
+
+  /**
+   * Log an update operation
+   */
+  async logUpdate(
+    dbName: string,
+    collName: string,
+    oldDoc: IStoredDocument,
+    newDoc: IStoredDocument,
+    txnId?: string
+  ): Promise<number> {
+    return this.appendEntry({
+      operation: 'update',
+      dbName,
+      collName,
+      documentId: oldDoc._id.toHexString(),
+      data: this.serializeDocument(newDoc),
+      previousData: this.serializeDocument(oldDoc),
+      txnId,
+    });
+  }
+
+  /**
+   * Log a delete operation
+   */
+  async logDelete(dbName: string, collName: string, doc: IStoredDocument, txnId?: string): Promise<number> {
+    return this.appendEntry({
+      operation: 'delete',
+      dbName,
+      collName,
+      documentId: doc._id.toHexString(),
+      previousData: this.serializeDocument(doc),
+      txnId,
+    });
+  }
+
+  /**
+   * Log transaction begin
+   */
+  async logBeginTransaction(txnId: string): Promise<number> {
+    this.uncommittedTxns.set(txnId, []);
+    return this.appendEntry({
+      operation: 'begin',
+      dbName: '',
+      collName: '',
+      documentId: '',
+      txnId,
+    });
+  }
+
+  /**
+   * Log transaction commit
+   */
+  async logCommitTransaction(txnId: string): Promise<number> {
+    this.uncommittedTxns.delete(txnId);
+    return this.appendEntry({
+      operation: 'commit',
+      dbName: '',
+      collName: '',
+      documentId: '',
+      txnId,
+    });
+  }
+
+  /**
+   * Log transaction abort
+   */
+  async logAbortTransaction(txnId: string): Promise<number> {
+    this.uncommittedTxns.delete(txnId);
+    return this.appendEntry({
+      operation: 'abort',
+      dbName: '',
+      collName: '',
+      documentId: '',
+      txnId,
+    });
+  }
+
+  /**
+   * Get entries to roll back for an aborted transaction
+   */
+  getTransactionEntries(txnId: string): IWalEntry[] {
+    return this.entries.filter(e => e.txnId === txnId);
+  }
+
+  /**
+   * Create a checkpoint - marks a consistent point in the log
+   */
+  async checkpoint(): Promise<number> {
+    const lsn = await this.appendEntry({
+      operation: 'checkpoint',
+      dbName: '',
+      collName: '',
+      documentId: '',
+    });
+    this.lastCheckpointLsn = lsn;
+
+    // Truncate old entries (keep only entries after checkpoint)
+    await this.truncate();
+
+    return lsn;
+  }
+
+  /**
+   * Truncate the WAL file, removing entries before the last checkpoint
+   */
+  private async truncate(): Promise<void> {
+    // Keep entries after last checkpoint
+    const newEntries = this.entries.filter(e => e.lsn >= this.lastCheckpointLsn);
+    this.entries = newEntries;
+
+    // Rewrite the WAL file
+    const lines = this.entries.map(e => JSON.stringify(e)).join('\n');
+    await this.fs.file(this.walPath).encoding('utf8').write(lines);
+  }
+
+  /**
+   * Get current LSN
+   */
+  getCurrentLsn(): number {
+    return this.currentLsn;
+  }
+
+  /**
+   * Get entries after a specific LSN (for recovery)
+   */
+  getEntriesAfter(lsn: number): IWalEntry[] {
+    return this.entries.filter(e => e.lsn > lsn);
+  }
+
+  /**
+   * Close the WAL
+   */
+  async close(): Promise<void> {
+    if (this.isInitialized) {
+      // Final checkpoint before close
+      await this.checkpoint();
+    }
+    this.isInitialized = false;
+  }
+
+  // ============================================================================
+  // Private Methods
+  // ============================================================================
+
+  private async appendEntry(
+    partial: Omit<IWalEntry, 'lsn' | 'timestamp' | 'checksum'>
+  ): Promise<number> {
+    await this.initialize();
+
+    this.currentLsn++;
+    const entry: IWalEntry = {
+      ...partial,
+      lsn: this.currentLsn,
+      timestamp: Date.now(),
+      checksum: 0, // Will be calculated
+    };
+
+    // Calculate checksum
+    entry.checksum = this.calculateChecksum(entry);
+
+    // Track in transaction if applicable
+    if (partial.txnId && this.uncommittedTxns.has(partial.txnId)) {
+      this.uncommittedTxns.get(partial.txnId)!.push(entry);
+    }
+
+    // Add to in-memory log
+    this.entries.push(entry);
+
+    // Append to file (append mode for durability)
+    await this.fs.file(this.walPath).encoding('utf8').append(JSON.stringify(entry) + '\n');
+
+    // Check if we need a checkpoint
+    if (this.entries.length - this.lastCheckpointLsn >= this.checkpointInterval) {
+      await this.checkpoint();
+    }
+
+    return entry.lsn;
+  }
+
+  private serializeDocument(doc: Document): string {
+    // Serialize document to BSON and encode as base64
+    const bsonData = plugins.bson.serialize(doc);
+    return Buffer.from(bsonData).toString('base64');
+  }
+
+  private deserializeDocument(data: string): Document {
+    // Decode base64 and deserialize from BSON
+    const buffer = Buffer.from(data, 'base64');
+    return plugins.bson.deserialize(buffer);
+  }
+
+  private calculateChecksum(entry: IWalEntry): number {
+    // Simple CRC32-like checksum
+    const str = JSON.stringify({
+      lsn: entry.lsn,
+      timestamp: entry.timestamp,
+      operation: entry.operation,
+      dbName: entry.dbName,
+      collName: entry.collName,
+      documentId: entry.documentId,
+      data: entry.data,
+      previousData: entry.previousData,
+      txnId: entry.txnId,
+    });
+
+    let crc = 0xFFFFFFFF;
+    for (let i = 0; i < str.length; i++) {
+      crc ^= str.charCodeAt(i);
+      for (let j = 0; j < 8; j++) {
+        crc = (crc >>> 1) ^ (crc & 1 ? 0xEDB88320 : 0);
+      }
+    }
+    return (~crc) >>> 0;
+  }
+
+  private verifyChecksum(entry: IWalEntry): boolean {
+    const savedChecksum = entry.checksum;
+    entry.checksum = 0;
+    const calculatedChecksum = this.calculateChecksum(entry);
+    entry.checksum = savedChecksum;
+    return calculatedChecksum === savedChecksum;
+  }
+
+  /**
+   * Recover document from WAL entry
+   */
+  recoverDocument(entry: IWalEntry): IStoredDocument | null {
+    if (!entry.data) return null;
+    return this.deserializeDocument(entry.data) as IStoredDocument;
+  }
+
+  /**
+   * Recover previous document state from WAL entry (for rollback)
+   */
+  recoverPreviousDocument(entry: IWalEntry): IStoredDocument | null {
+    if (!entry.previousData) return null;
+    return this.deserializeDocument(entry.previousData) as IStoredDocument;
+  }
+}

package/ts/tsmdb/utils/checksum.ts

@@ -0,0 +1,88 @@
+/**
+ * CRC32 checksum utilities for data integrity
+ */
+
+// CRC32 lookup table
+const CRC32_TABLE: number[] = [];
+
+// Initialize the CRC32 table
+function initCRC32Table(): void {
+  if (CRC32_TABLE.length > 0) return;
+
+  for (let i = 0; i < 256; i++) {
+    let crc = i;
+    for (let j = 0; j < 8; j++) {
+      crc = (crc & 1) ? (0xEDB88320 ^ (crc >>> 1)) : (crc >>> 1);
+    }
+    CRC32_TABLE[i] = crc >>> 0;
+  }
+}
+
+/**
+ * Calculate CRC32 checksum for a string
+ */
+export function calculateCRC32(data: string): number {
+  initCRC32Table();
+
+  let crc = 0xFFFFFFFF;
+  for (let i = 0; i < data.length; i++) {
+    const byte = data.charCodeAt(i) & 0xFF;
+    crc = CRC32_TABLE[(crc ^ byte) & 0xFF] ^ (crc >>> 8);
+  }
+  return (~crc) >>> 0;
+}
+
+/**
+ * Calculate CRC32 checksum for a Buffer
+ */
+export function calculateCRC32Buffer(data: Buffer): number {
+  initCRC32Table();
+
+  let crc = 0xFFFFFFFF;
+  for (let i = 0; i < data.length; i++) {
+    crc = CRC32_TABLE[(crc ^ data[i]) & 0xFF] ^ (crc >>> 8);
+  }
+  return (~crc) >>> 0;
+}
+
+/**
+ * Calculate checksum for a document (serialized as JSON)
+ */
+export function calculateDocumentChecksum(doc: Record<string, any>): number {
+  // Exclude _checksum field from calculation
+  const { _checksum, ...docWithoutChecksum } = doc;
+  const json = JSON.stringify(docWithoutChecksum);
+  return calculateCRC32(json);
+}
+
+/**
+ * Add checksum to a document
+ */
+export function addChecksum<T extends Record<string, any>>(doc: T): T & { _checksum: number } {
+  const checksum = calculateDocumentChecksum(doc);
+  return { ...doc, _checksum: checksum };
+}
+
+/**
+ * Verify checksum of a document
+ * Returns true if checksum is valid or if document has no checksum
+ */
+export function verifyChecksum(doc: Record<string, any>): boolean {
+  if (!('_checksum' in doc)) {
+    // No checksum to verify
+    return true;
+  }
+
+  const storedChecksum = doc._checksum;
+  const calculatedChecksum = calculateDocumentChecksum(doc);
+
+  return storedChecksum === calculatedChecksum;
+}
+
+/**
+ * Remove checksum from a document
+ */
+export function removeChecksum<T extends Record<string, any>>(doc: T): Omit<T, '_checksum'> {
+  const { _checksum, ...docWithoutChecksum } = doc;
+  return docWithoutChecksum as Omit<T, '_checksum'>;
+}

package/ts/tsmdb/utils/index.ts

@@ -0,0 +1 @@
+export * from './checksum.js';