Python SDK

---

Full Documentation

This is the client SDK for EnSync engine (message delivery engine) that enables you to build an ecosystem of connected devices and services.

---

Installation

pip install ensync-sdk

---

Usage

Importing

Default (gRPC)

# Import the default engine class (gRPC)
from ensync_sdk import EnSyncEngine

# Production - uses secure TLS on port 443 by default
engine = EnSyncEngine("node.ensync.cloud")

# Development - uses insecure connection on port 50051 by default
# engine = EnSyncEngine("localhost")

# Create authenticated client
client = await engine.create_client("your-app-key")

WebSocket Alternative

# Import the WebSocket engine class
from ensync_sdk import EnSyncWebSocketEngine

# Initialize WebSocket client
engine = EnSyncWebSocketEngine("wss://node.ensync.cloud")
client = await engine.create_client("your-app-key")

Both clients provide the same API for publishing and subscribing to events.

gRPC Connection Options:

• Production URLs automatically use secure TLS (port 443)
• localhost automatically uses insecure connection (port 50051)
• Explicit protocols: grpcs:// (secure) or grpc:// (insecure)
• Custom ports: node.ensync.cloud:9090

---

API Reference

EnSyncEngine (gRPC - Default)

The main class that manages gRPC connections and client creation for the EnSync system. This is the default and recommended client for production use.

EnSyncWebSocketEngine (WebSocket - Alternative)

An alternative class that manages WebSocket connections and client creation for the EnSync system.

engine = EnSyncEngine(url, options=None)

Parameters

Parameter	Type	Required	Description
url	str	Yes	The URL of the EnSync server
options	dict	No	Configuration options

Options Dictionary:

Option	Type	Default	Description
disable_tls	bool	False	Set to true to disable TLS
reconnect_interval	int	5000	Reconnection interval in ms
max_reconnect_attempts	int	10	Maximum reconnection attempts

---

Creating a Client

• Initialize the engine with your server URL
• Create a client with your app key

# Initialize the engine (gRPC with TLS)
engine = EnSyncEngine("grpcs://node.gms.ensync.cloud")

# Create a client
client = await engine.create_client("your-app-key")

Client Creation Parameters

Parameter	Type	Required	Description
app_key	str	Yes	Your EnSync application key
options	dict	No	Client configuration options

Options Dictionary:

Option	Type	Default	Description
app_secret_key	str	None	Default key used to decrypt incoming messages

Client Returns

Returns a new EnSyncClient instance.

---

Publishing Events

# Basic publish
await client.publish(
    "company/service/event-type",  # Event name
    ["appId"],                     # Recipients (appIds of receiving parties)
    {"data": "your payload"}       # Event payload
)

# With optional metadata
await client.publish(
    "company/service/event-type",
    ["appId"],                     # The appId of the receiving party
    {"data": "your payload"},
    {"custom_field": "value"}     # Optional metadata
)

Publish Parameters

Parameter	Type	Required	Description
event_name	str	Yes	Name of the event (e.g., "company/service/event-type")
recipients	list[str]	Yes	Array of appIds (the appIds of receiving parties)
payload	dict	Yes	Your event data (any JSON-serializable object)
metadata	dict	No	Optional custom metadata as key-value pairs

The metadata parameter accepts any custom key-value pairs you want to include with your event. This metadata is passed through to recipients and can be used for routing, filtering, or any application-specific purposes.

Replying to Events

When you receive an event, it includes a sender field containing the sender's public key. You can use this to send a response back to the original sender:

async def handle_event(event):
    # Process the event
    print(f"Received: {event['payload']}")
    
    # Reply back to the sender
    sender_public_key = event.get('sender')
    if sender_public_key:
        await client.publish(
            event.get('eventName'),
            [sender_public_key],  # Send back to the original sender
            {"status": "received", "response": "Processing complete"}
        )

---

Subscribing to Events

# Basic subscription
subscription = await client.subscribe("company/service/event-type")

# Set up event handler
async def handle_event(event):
    print(f"Received event: {event.payload}")
    # Process the event

subscription.on(handle_event)

# With options
subscription = await client.subscribe(
    "company/service/event-type",
    {
        "auto_ack": False,  # Manual acknowledgment
        "app_secret_key": os.environ.get("CUSTOM_DECRYPT_KEY")  # Custom decryption key
    }
)

Subscribe Parameters

Parameter	Type	Required	Description
event_name	str	Yes	Name of the event to subscribe to
options	dict	No	Subscription options

Options Dictionary:

Option	Type	Default	Description
auto_ack	bool	True	Set to false for manual acknowledgment
app_secret_key	str	None	Custom decryption key for this subscription

Subscription Methods

# Handle incoming events
async def handle_event(event):
    # process event
    pass

subscription.on(handle_event)

# Manually acknowledge an event
await subscription.ack(event.idem, event.block)

# Request a specific event to be replayed
event_data = await subscription.replay("event-idem-123")

# Stop receiving events
await subscription.unsubscribe()

---

Event Structure

When you receive an event through a subscription handler, it contains:

{
    "idem": "abc123",                # Unique event ID (use with ack/discard/replay)
    "block": "456",                  # Block ID (use with ack)
    "event_name": "company/service/event-type",  # Event name
    "payload": { /* your data */ },  # Your decrypted data
    "timestamp": 1634567890123,      # Event timestamp (milliseconds)
    "metadata": {                    # Optional metadata
        "headers": { /* custom headers */ }
    },
    "recipient": "appId"            # The appId of the receiving party
}

---

Closing Connections

# Close just this client
await client.close()

# Close client and engine (if you have no other clients)
await client.close(close_engine=True)

---

Error Handling

The SDK raises EnSyncError for various error conditions. Always wrap your code in try-except blocks to handle potential errors gracefully.

try:
    # Your EnSync code
except EnSyncError as e:
    print(f"EnSync Error: {e}")
    # Handle specific error types
    if isinstance(e, EnSyncConnectionError):
        # Handle connection errors
        pass
    elif isinstance(e, EnSyncPublishError):
        # Handle publishing errors
        pass
    elif isinstance(e, EnSyncSubscriptionError):
        # Handle subscription errors
        pass
except Exception as e:
    print(f"Unexpected error: {e}")

Common error types:

Error Type	Description
EnSyncConnectionError	Connection or authentication issues
EnSyncPublishError	Problems publishing events
EnSyncSubscriptionError	Subscription-related errors
EnSyncGenericError	Other errors

---

Complete Examples

Quick Start

import os
import asyncio
from dotenv import load_dotenv
from ensync import EnSyncEngine

# Load environment variables from .env file
load_dotenv()

async def quick_start():
    try:
        # 1. Initialize engine and create client
        engine = EnSyncEngine("wss://node.ensync.cloud")
        client = await engine.create_client(
            os.environ.get("ENSYNC_APP_KEY"),
            {
                "app_secret_key": os.environ.get("ENSYNC_SECRET_KEY")
            }
        )

        # 2. Publish an event
        await client.publish(
            "orders/status/updated",
            ["appId"],  # The appId of the receiving party
            {"order_id": "order-123", "status": "completed"}
        )

        # 3. Subscribe to events
        subscription = await client.subscribe("orders/status/updated")
        
        # 4. Handle incoming events
        async def handle_event(event):
            print(f"Received order update: {event.payload['order_id']} is {event.payload['status']}")
            # Process event...
        
        subscription.on(handle_event)

        # 5. Keep the program running
        try:
            # Run indefinitely until interrupted
            await asyncio.Future()
        except KeyboardInterrupt:
            # Clean up when done
            await subscription.unsubscribe()
            await client.close()
            
    except Exception as e:
        print(f'Error: {e}')

# Run the async function
if __name__ == "__main__":
    asyncio.run(quick_start())

Note: This example uses environment variables for security. Create a .env file with:

ENSYNC_APP_KEY=your_app_key_here
ENSYNC_SECRET_KEY=your_secret_key_here

Publishing Example

import asyncio
import os
from dotenv import load_dotenv
from ensync import EnSyncEngine

load_dotenv()

async def publishing_example():
    # Create client
    engine = EnSyncEngine("wss://node.ensync.cloud")
    client = await engine.create_client(os.environ.get("ENSYNC_APP_KEY"))

    # Basic publish - returns event ID
    event_id = await client.publish(
        "notifications/email/sent",
        ["appId"],  # The appId of the receiving party
        {"to": "user@example.com", "subject": "Welcome!"}
    )
    print(f"Published event: {event_id}")

    # With metadata
    event_id = await client.publish(
        "notifications/email/sent",
        ["appId"],  # The appId of the receiving party
        {"to": "user@example.com", "subject": "Welcome!"},
        {"source": "email-service", "priority": "high"}
    )
    print(f"Published event with metadata: {event_id}")
    
    await client.close()

if __name__ == "__main__":
    asyncio.run(publishing_example())

Subscribing Example

import asyncio
import os
from dotenv import load_dotenv
from ensync import EnSyncEngine

load_dotenv()

async def update_order_status(order_id, status):
    # Simulated function to update order status
    print(f"Updating order {order_id} to status: {status}")
    return True

def needs_history(order_id):
    # Simulated function to check if we need history
    return order_id.startswith("special-")

def is_temporary_error(error):
    # Simulated function to determine if an error is temporary
    return "timeout" in str(error).lower() or "retry" in str(error).lower()

async def subscribing_example():
    # Create client with decryption key
    engine = EnSyncEngine("grpcs://node.gms.ensync.cloud")
    client = await engine.create_client(
        os.environ.get("ENSYNC_APP_KEY"),
        {"app_secret_key": os.environ.get("ENSYNC_SECRET_KEY")}
    )

    # Subscribe with manual acknowledgment
    subscription = await client.subscribe("payments/completed", {"auto_ack": False})

    # Handle events
    async def handle_payment(event):
        try:
            # Process the payment
            await update_order_status(event.payload["order_id"], "paid")
            
            # Get historical data if needed
            if needs_history(event.payload["order_id"]):
                history = await subscription.replay(event.payload["previous_event_id"])
                print(f"Previous payment: {history}")
            
            # Acknowledge successful processing
            await subscription.ack(event.idem, event.block)
        except Exception as error:
            # Defer processing if temporary error
            if is_temporary_error(error):
                await subscription.defer(event.idem, 60000, "Temporary processing error")
            else:
                # Discard if permanent error
                await subscription.discard(event.idem, "Invalid payment data")
    
    subscription.on(handle_payment)
    
    # Keep the program running
    try:
        await asyncio.Future()
    except KeyboardInterrupt:
        await subscription.unsubscribe()
        await client.close()

if __name__ == "__main__":
    asyncio.run(subscribing_example())

---

Best Practices

Connection Management

• Store connection credentials securely using environment variables
• Implement proper reconnection logic for production environments
• Always close connections when they're no longer needed

import os
import asyncio
import signal
from dotenv import load_dotenv
from ensync import EnSyncEngine

load_dotenv()

async def main():
    # Using environment variables for sensitive keys
    engine = EnSyncEngine(os.environ.get("ENSYNC_URL"))
    client = await engine.create_client(os.environ.get("ENSYNC_APP_KEY"))
    
    # gRPC client handles reconnection automatically
    
    # Set up signal handlers for graceful shutdown
    loop = asyncio.get_running_loop()
    for sig in (signal.SIGINT, signal.SIGTERM):
        loop.add_signal_handler(sig, lambda: asyncio.create_task(shutdown(client)))
    
    # Your application code here
    try:
        await asyncio.Future()  # Run indefinitely
    finally:
        await client.close()

async def shutdown(client):
    print("Shutting down...")
    await client.close(close_engine=True)
    asyncio.get_event_loop().stop()

if __name__ == "__main__":
    asyncio.run(main())

Event Design

• Use hierarchical event names (e.g., domain/entity/action)
• Keep payloads concise and well-structured
• Consider versioning your event schemas

# Good event naming pattern
await client.publish(
    "inventory/product/created",
    ["warehouse-service"],
    {
        "product_id": "prod-123",
        "name": "Ergonomic Chair",
        "sku": "ERG-CH-BLK",
        "price": 299.99,
        "created_at": int(time.time() * 1000)  # milliseconds timestamp
    }
)

Security Best Practices

• Never hardcode app keys or secret keys
• Use environment variables or secure key management solutions
• Implement proper authentication and authorization
• Consider encrypting sensitive payloads

Performance Optimization

• Batch events when possible instead of sending many small messages
• Consider message size and frequency in high-volume scenarios
• Use appropriate TTL values for your use case
• Implement proper error handling and retry logic

Subscription Control

The SDK provides methods to pause, continue, and replay events, which is useful for managing event processing flow.

What Pause and Continue Do

When you create a client using engine.create_client(), that client receives a unique client_id. This client_id (not the app_key) identifies your specific client instance on the EnSync server.

• Pause: Temporarily stops the client from receiving new events from the server. The subscription remains active on the server, but events are not delivered to this specific client instance. Other clients with the same app_key but different client_id will continue receiving events normally.

• Continue: Resumes event delivery to the paused client. Any events that occurred during the pause (depending on server settings and TTL) may be delivered once the subscription is continued.

Replaying Events

The replay command allows you to request a specific event to be sent again, even if it has already been processed. Unlike regular event handling which delivers events through the event handler, the replay function returns the event data directly to your code. This is useful for:

• Retrieving specific events for analysis or debugging
• Accessing historical event data without setting up a handler
• Examining event content without processing it
• Getting event data synchronously in your code flow

# Request a specific event to be replayed - returns data directly
event_data = await subscription.replay("event-idem-123")
print(f"Event data: {event_data}")

# You can immediately work with the event data
process_event_data(event_data)

The replay command returns the complete event object with its payload:

{
    "event_name": "gms/ensync/third_party/payments/complete",
    "idem": "event-idem-123",
    "block": "81404",
    "metadata": {
        "persist": {"is_string": False, "content": "true"},
        "headers": {},
        "$internal": {
            "replay_info": {
                "is_replayed": {"is_string": False, "content": "true"},
                "replay_timestamp": {"is_string": False, "content": "1758410511179"},
                "was_acknowledged": {"is_string": False, "content": "false"}
            }
        }
    },
    "payload": {/* payload data */},
    "logged_at": 1757778462158,
    "recipient": "RECIPIENT_PUBLIC_KEY_BASE64",
    "is_group": False
}

Direct Access vs Handler Processing:

Regular event subscription:

# Events come through the handler asynchronously
async def handle_event(event):
    # Process event here
    print(f"Received event: {event}")

subscription.on(handle_event)

Replay function:

# Get event data directly and synchronously
event = await subscription.replay("event-idem-123")
print(f"Retrieved event: {event}")

Deferring Events

The defer method allows you to postpone processing of an event for a specified period. This is useful when:

• You need more time to prepare resources for processing
• You want to implement a retry mechanism with increasing delays
• You need to wait for another system to be ready
• You want to implement rate limiting for event processing

# Defer an event for 5 seconds (5000ms)
defer_result = await subscription.defer(
    "event-idem-123",  # Event ID
    5000,              # Delay in milliseconds
    "Waiting for resources to be available"  # Optional reason
)
print(f"Defer result: {defer_result}")

# Defer with minimum delay (immediate redelivery)
immediate_redelivery = await subscription.defer("event-idem-123", 0)

The defer method returns an object with status information:

{
    "status": "success",
    "action": "deferred",
    "event_idem": "event-idem-123",
    "delay_ms": 5000,
    "scheduled_delivery": 1757778467158,  # timestamp when event will be redelivered
    "timestamp": 1757778462158
}

Discarding Events

The discard method allows you to permanently reject an event without processing it. This is useful when:

• The event contains invalid or corrupted data
• The event is no longer relevant or has expired
• The event was sent to the wrong recipient
• You want to implement a filtering mechanism

# Discard an event permanently
discard_result = await subscription.discard(
    "event-idem-123",  # Event ID
    "Invalid data format"  # Optional reason
)
print(f"Discard result: {discard_result}")

The discard method returns an object with status information:

{
    "status": "success",
    "action": "discarded",
    "event_idem": "event-idem-123",
    "timestamp": 1757778462158
}

# Create a subscription
subscription = await client.subscribe("inventory/updates")

# Set up event handler
async def handle_event(event):
    print(f"Processing event: {event.id}")
    await process_event(event)

subscription.on(handle_event)

# Pause the subscription when needed
# This will temporarily stop receiving events
await subscription.pause()
print("Subscription paused - no events will be received")

# Perform some operations while subscription is paused
await perform_maintenance()

# Continue the subscription to resume receiving events
await subscription.continue_subscription()  # Note: 'continue' is a Python keyword
print("Subscription continued - now receiving events again")

# Example: Implementing controlled processing with pause/continue
async def process_in_batches(events):
    # Pause subscription while processing a batch
    await subscription.pause()
    
    try:
        # Process events without receiving new ones
        for event in events:
            await process_event(event)
    except Exception as error:
        print(f"Error processing batch: {error}")
    finally:
        # Always continue subscription when done
        await subscription.continue_subscription()

Use cases for pause/continue:

• Temporary maintenance or system updates
• Rate limiting or throttling event processing
• Implementing backpressure mechanisms
• Batch processing of events

Implementation Details

• Pause/continue operations are performed at the subscription level, not the client level
• The server maintains the subscription state even when paused
• Pausing affects only the specific subscription instance, not all subscriptions for the client
• Events that arrive during a pause may be delivered when continued (depending on TTL settings)
• The pause state is not persisted across client restarts or reconnections