video-accessibility/scripts/mongodb-init.js

// =============================================================================
// MongoDB Initialization Script for Accessible Video Platform
// =============================================================================
// Run this script ONCE after starting the MongoDB container
// Usage: docker-compose exec mongodb mongosh < scripts/mongodb-init.js
// =============================================================================

// Connect to the accessible_video database
db = db.getSiblingDB('accessible_video');

print('=============================================================================');
print('MongoDB Initialization for Accessible Video Platform');
print('=============================================================================');

// -----------------------------------------------------------------------------
// Create Collections with Validation
// -----------------------------------------------------------------------------
print('\n1. Creating collections with schema validation...');

// Jobs collection
db.createCollection('jobs', {
  validator: {
    $jsonSchema: {
      bsonType: 'object',
      required: ['_id', 'title', 'status', 'client_id', 'created_at', 'updated_at'],
      properties: {
        _id: { bsonType: 'string' },
        title: { bsonType: 'string' },
        status: {
          enum: ['created', 'ingesting', 'ai_processing', 'pending_qc',
                 'approved_english', 'approved_source', 'rejected', 'qc_feedback',
                 'translating', 'tts_generating', 'pending_final_review', 'completed']
        },
        client_id: { bsonType: 'string' },
        created_at: { bsonType: 'date' },
        updated_at: { bsonType: 'date' }
      }
    }
  }
});
print(' Created jobs collection');

// Users collection
db.createCollection('users', {
  validator: {
    $jsonSchema: {
      bsonType: 'object',
      required: ['_id', 'email', 'hashed_password', 'role', 'created_at'],
      properties: {
        _id: { bsonType: 'string' },
        email: { bsonType: 'string' },
        hashed_password: { bsonType: 'string' },
        role: { enum: ['client', 'reviewer', 'admin'] },
        is_active: { bsonType: 'bool' },
        created_at: { bsonType: 'date' }
      }
    }
  }
});
print(' Created users collection');

// Audit logs collection
db.createCollection('audit_logs');
print(' Created audit_logs collection');

// -----------------------------------------------------------------------------
// Create Indexes for Performance
// -----------------------------------------------------------------------------
print('\n2. Creating indexes for optimized queries...');

// Jobs collection indexes
db.jobs.createIndex({ 'status': 1, 'created_at': -1 }, { name: 'idx_status_created' });
print(' Created index: jobs.idx_status_created');

db.jobs.createIndex({ 'client_id': 1 }, { name: 'idx_client_id' });
print(' Created index: jobs.idx_client_id');

db.jobs.createIndex({ 'created_at': -1 }, { name: 'idx_created_at' });
print(' Created index: jobs.idx_created_at');

// Users collection indexes
db.users.createIndex({ 'email': 1 }, { unique: true, name: 'idx_email_unique' });
print(' Created index: users.idx_email_unique (unique)');

db.users.createIndex({ 'role': 1 }, { name: 'idx_role' });
print(' Created index: users.idx_role');

// Audit logs collection indexes
db.audit_logs.createIndex({ 'timestamp': -1 }, { name: 'idx_timestamp' });
print(' Created index: audit_logs.idx_timestamp');

db.audit_logs.createIndex({ 'action': 1, 'timestamp': -1 }, { name: 'idx_action_timestamp' });
print(' Created index: audit_logs.idx_action_timestamp');

db.audit_logs.createIndex({ 'user_id': 1, 'timestamp': -1 }, { name: 'idx_user_timestamp' });
print(' Created index: audit_logs.idx_user_timestamp');

db.audit_logs.createIndex({ 'severity': 1, 'timestamp': -1 }, { name: 'idx_severity_timestamp' });
print(' Created index: audit_logs.idx_severity_timestamp');

db.audit_logs.createIndex({ 'resource_type': 1, 'resource_id': 1 }, { name: 'idx_resource' });
print(' Created index: audit_logs.idx_resource');

// Text search index for audit logs
db.audit_logs.createIndex(
  {
    'description': 'text',
    'details': 'text',
    'error_message': 'text'
  },
  {
    name: 'idx_text_search',
    weights: {
      'description': 10,
      'details': 5,
      'error_message': 8
    }
  }
);
print(' Created index: audit_logs.idx_text_search (full-text)');

// -----------------------------------------------------------------------------
// Display Database Statistics
// -----------------------------------------------------------------------------
print('\n3. Database initialization complete!');
print('\nDatabase statistics:');
print('-------------------');

const stats = db.stats();
print('Database: ' + stats.db);
print('Collections: ' + stats.collections);
print('Indexes: ' + stats.indexes);
print('Data Size: ' + (stats.dataSize / 1024).toFixed(2) + ' KB');
print('Storage Size: ' + (stats.storageSize / 1024).toFixed(2) + ' KB');

print('\nCollections created:');
db.getCollectionNames().forEach(function(collName) {
  const collStats = db.getCollection(collName).stats();
  const indexCount = db.getCollection(collName).getIndexes().length;
  print('  - ' + collName + ' (indexes: ' + indexCount + ')');
});

print('\n=============================================================================');
print('Next steps:');
print('1. Run migrations: docker-compose exec api python migrate.py');
print('2. Create admin user: docker-compose exec api python create_test_users.py');
print('=============================================================================');