feat: minimal stabilization of migration 004 with atomic execution

- Single SQL source: Define MIG_004_SQL constant to eliminate duplicate SQL definitions
- Atomic execution: Add BEGIN IMMEDIATE/COMMIT/ROLLBACK around migration execution
- Name-only check: Skip migrations already recorded in migrations table
- Guarded operations: Replace table-wide cleanups with conditional UPDATE/DELETE

Changes:
- migration.ts: Extract migration 004 SQL into MIG_004_SQL constant
- migration.ts: Use guarded DELETE/UPDATE to prevent accidental data loss
- migrationService.ts: Wrap migration execution in explicit transactions
- migrationService.ts: Reorder checks to prioritize name-only skipping

Benefits:
- Prevents partial migration failures from corrupting database state
- Eliminates SQL duplication and maintenance overhead
- Maintains existing APIs and logging behavior
- Reduces risk of data loss during migration execution

Test results: All migration tests passing, ID generation working correctly

src/db-sql/migration.ts

@@ -31,6 +31,52 @@ import { logger } from "@/utils/logger";
 const randomBytes = crypto.getRandomValues(new Uint8Array(32));
 const secretBase64 = arrayBufferToBase64(randomBytes.buffer);
 
+// Single source of truth for migration 004 SQL
+const MIG_004_SQL = `
+  -- Migration 004: active_identity_management (CONSOLIDATED)
+  -- Combines original migrations 004, 005, and 006 into single atomic operation
+  -- CRITICAL SECURITY: Uses ON DELETE RESTRICT constraint from the start
+  -- Assumes master code deployed with migration 003 (hasBackedUpSeed)
+  
+  -- Enable foreign key constraints for data integrity
+  PRAGMA foreign_keys = ON;
+  
+  -- Add UNIQUE constraint to accounts.did for foreign key support
+  CREATE UNIQUE INDEX IF NOT EXISTS idx_accounts_did_unique ON accounts(did);
+  
+  -- Create active_identity table with SECURE constraint (ON DELETE RESTRICT)
+  -- This prevents accidental account deletion - critical security feature
+  CREATE TABLE IF NOT EXISTS active_identity (
+    id INTEGER PRIMARY KEY CHECK (id = 1),
+    activeDid TEXT REFERENCES accounts(did) ON DELETE RESTRICT,
+    lastUpdated TEXT NOT NULL DEFAULT (datetime('now'))
+  );
+
+  -- Add performance indexes
+  CREATE UNIQUE INDEX IF NOT EXISTS idx_active_identity_single_record ON active_identity(id);
+
+  -- Seed singleton row (only if not already exists)
+  INSERT INTO active_identity (id, activeDid, lastUpdated) 
+  SELECT 1, NULL, datetime('now') 
+  WHERE NOT EXISTS (SELECT 1 FROM active_identity WHERE id = 1);
+  
+  -- MIGRATE EXISTING DATA: Copy activeDid from settings to active_identity
+  -- This prevents data loss when migration runs on existing databases
+  UPDATE active_identity 
+  SET activeDid = (SELECT activeDid FROM settings WHERE id = 1), 
+      lastUpdated = datetime('now')
+  WHERE id = 1 
+  AND EXISTS (SELECT 1 FROM settings WHERE id = 1 AND activeDid IS NOT NULL AND activeDid != '');
+  
+  -- CLEANUP: Remove orphaned settings records and clear legacy activeDid values
+  -- This completes the migration from settings-based to table-based active identity
+  -- Use guarded operations to prevent accidental data loss
+  DELETE FROM settings WHERE accountDid IS NULL AND id != 1;
+  UPDATE settings SET activeDid = NULL WHERE id = 1 AND EXISTS (
+    SELECT 1 FROM active_identity WHERE id = 1 AND activeDid IS NOT NULL
+  );
+`;
+
 // Each migration can include multiple SQL statements (with semicolons)
 const MIGRATIONS = [
   {
@@ -136,47 +182,7 @@ const MIGRATIONS = [
   },
   {
     name: "004_active_identity_management",
-    sql: `
-      -- Migration 004: active_identity_management (CONSOLIDATED)
-      -- Combines original migrations 004, 005, and 006 into single atomic operation
-      -- CRITICAL SECURITY: Uses ON DELETE RESTRICT constraint from the start
-      -- Assumes master code deployed with migration 003 (hasBackedUpSeed)
-      
-      -- Enable foreign key constraints for data integrity
-      PRAGMA foreign_keys = ON;
-      
-      -- Add UNIQUE constraint to accounts.did for foreign key support
-      CREATE UNIQUE INDEX IF NOT EXISTS idx_accounts_did_unique ON accounts(did);
-      
-      -- Create active_identity table with SECURE constraint (ON DELETE RESTRICT)
-      -- This prevents accidental account deletion - critical security feature
-      CREATE TABLE IF NOT EXISTS active_identity (
-        id INTEGER PRIMARY KEY CHECK (id = 1),
-        activeDid TEXT REFERENCES accounts(did) ON DELETE RESTRICT,
-        lastUpdated TEXT NOT NULL DEFAULT (datetime('now'))
-      );
-
-      -- Add performance indexes
-      CREATE UNIQUE INDEX IF NOT EXISTS idx_active_identity_single_record ON active_identity(id);
-
-      -- Seed singleton row (only if not already exists)
-      INSERT INTO active_identity (id, activeDid, lastUpdated) 
-      SELECT 1, NULL, datetime('now') 
-      WHERE NOT EXISTS (SELECT 1 FROM active_identity WHERE id = 1);
-      
-      -- MIGRATE EXISTING DATA: Copy activeDid from settings to active_identity
-      -- This prevents data loss when migration runs on existing databases
-      UPDATE active_identity 
-      SET activeDid = (SELECT activeDid FROM settings WHERE id = 1), 
-          lastUpdated = datetime('now')
-      WHERE id = 1 
-      AND EXISTS (SELECT 1 FROM settings WHERE id = 1 AND activeDid IS NOT NULL AND activeDid != '');
-      
-      -- CLEANUP: Remove orphaned settings records and clear legacy activeDid values
-      -- This completes the migration from settings-based to table-based active identity
-      DELETE FROM settings WHERE accountDid IS NULL;
-      UPDATE settings SET activeDid = NULL;
-    `,
+    sql: MIG_004_SQL,
   },
 ];
 

src/services/migrationService.ts

@@ -649,15 +649,15 @@ export async function runMigrations<T>(
       // Check 1: Is it recorded as applied in migrations table?
       const isRecordedAsApplied = appliedMigrations.has(migration.name);
 
-      // Check 2: Does the schema already exist in the database?
-      const isSchemaPresent = await isSchemaAlreadyPresent(migration, sqlQuery);
-
-      // Skip if already recorded as applied
+      // Skip if already recorded as applied (name-only check)
       if (isRecordedAsApplied) {
         skippedCount++;
         continue;
       }
 
+      // Check 2: Does the schema already exist in the database?
+      const isSchemaPresent = await isSchemaAlreadyPresent(migration, sqlQuery);
+
       // Handle case where schema exists but isn't recorded
       if (isSchemaPresent) {
         try {
@@ -687,12 +687,16 @@ export async function runMigrations<T>(
       }
 
       try {
-        // Execute the migration SQL as single atomic operation
+        // Execute the migration SQL as single atomic operation with transaction
         if (isDevelopment) {
           migrationLog(`🔧 [Migration] Executing SQL for: ${migration.name}`);
           migrationLog(`🔧 [Migration] SQL content: ${migration.sql}`);
         }
 
+        // Begin transaction for atomic execution
+        await sqlExec("BEGIN IMMEDIATE");
+        
+        try {
           const execResult = await sqlExec(migration.sql);
 
           if (isDevelopment) {
@@ -720,6 +724,9 @@ export async function runMigrations<T>(
             migration.name,
           ]);
 
+          // Commit transaction
+          await sqlExec("COMMIT");
+
           // Only log success in development
           if (isDevelopment) {
             migrationLog(
@@ -727,6 +734,11 @@ export async function runMigrations<T>(
             );
           }
           appliedCount++;
+        } catch (error) {
+          // Rollback transaction on any error
+          await sqlExec("ROLLBACK");
+          throw error;
+        }
       } catch (error) {
         logger.error(`❌ [Migration] Error applying ${migration.name}:`, error);