Developer Portal

Appearance

snet-sdk-python ​

SingularityNET SDK for Python

Package ​

The package is published in PyPI at the following link:

PackageDescription
snet-sdkIntegrate SingularityNET services seamlessly into Python applications

Core concepts ​

The SingularityNET SDK allows you to make calls to SingularityNET services programmatically from your application. To communicate between clients and services, SingularityNET uses gRPC. To handle payment of services, SingularityNET uses Ethereum state channels. The SingularityNET SDK abstracts and manages state channels with service providers on behalf of the user and handles authentication with the SingularityNET services.

Getting Started ​

These instructions are for the development and use of the SingularityNET SDK for Python.

Usage ​

To call a service on a SingularityNET platform, the user must be able to deposit funds (FET tokens) to the Multi-Party Escrow Smart Contract. To deposit these tokens or do any other transaction on the Ethereum blockchain.

Once you have installed snet-sdk in your current environment, you can import it into your Python script and create an instance of the base sdk class:

python
from snet import sdk

"""
SDK configuration provided by the application provider.
To run the application, replace 'private_key' and 'eth_rpc_endpoint' with your values.
"""
config = sdk.config.Config(
    private_key="YOUR_PRIVATE_KEY",  # Replace with your Ethereum private key
    eth_rpc_endpoint="https://eth-sepolia.g.alchemy.com/v2/YOUR_ALCHEMY_API_KEY",  # Replace with your Alchemy API key
    concurrency=False, 
    force_update=False 
)

# Initialize the SnetSDK instance
snet_sdk = sdk.SnetSDK(config)

The config parameter is an instance of the Config class. See config.py for a reference.

Config parameters description ​

  • private_key: Your wallet's private key that will be used to pay for calls. Is required in config;
  • eth_rpc_endpoint: RPC endpoint that is used to access the Ethereum network. Is required in config;

To get your Alchemy API Key, follow this guide.

  • wallet_index: The index of the wallet that will be used to pay for calls;
  • ipfs_endpoint: IPFS endpoint that is used to access IPFS;
  • concurrency: If set to True, will enable concurrency for the SDK;
  • force_update: If set to False, will reuse the existing gRPC stubs (if any) instead of downloading proto and regenerating them every time.
  • mpe_contract_address: The address of the Multi-Party Escrow smart contract;
  • token_contract_address: The address of the SingularityNET token smart contract;
  • registry_contract_address: The address of the Registry smart contract;
  • signer_private_key: The private key of the signer. Used to sign the service call. Equals to private_key by default.

List organizations and their services ​

You can use the sdk client instance`s methods get_organization_list() to list all organizations and get_services_list("org_id") to list all services of a given organization.

python
orgs_list = snet_sdk.get_organization_list()
print(*orgs_list, sep="\n")
# ...
# GoogleOrg3
# 26072b8b6a0e448180f8c0e702ab6d2f
# 43416d873fcb454589900189474b2eaa
# ...
python
org_id = "26072b8b6a0e448180f8c0e702ab6d2f"
services_list = snet_sdk.get_services_list(org_id=org_id)
print(*services_list, sep="\n")
# Exampleservice

Calling the service ​

Now, the instance of the sdk can be used to create the service client instances, using create_service_client() method.
Continuing from the previous code here is an example using Exampleservice from the 26072b8b6a0e448180f8c0e702ab6d2f organization:

python
service_client = snet_sdk.create_service_client(org_id="26072b8b6a0e448180f8c0e702ab6d2f", 
                                                service_id="Exampleservice",
                                                group_name="default_group")

After executing this code, you should have client libraries created for this service. They are located at the following path: ~/.snet/org_id/service_id/python/

Note: Currently you can only save files to ~/.snet/.

The instance of service_client that has been generated can be utilized to invoke the methods that the service offers. You can list these using the get_services_and_messages_info_as_pretty_string() method:

python
print(service_client.get_services_and_messages_info_as_pretty_string())
# Service: Calculator
#   Method: add, Input: Numbers, Output: Result
#   Method: sub, Input: Numbers, Output: Result
#   Method: mul, Input: Numbers, Output: Result
#   Method: div, Input: Numbers, Output: Result
# Message: Numbers
#   Field: float a
#   Field: float b
# Message: Result
#   Field: float value

To invoke the service's methods, you can use the call_rpc() method. This method requires the names of the method and data object, along with the data itself, to be passed into it. To continue with our example, here’s a call to the mul method of the Exampleservice from the 26072b8b6a0e448180f8c0e702ab6d2f organization:

python
result = service_client.call_rpc("mul", "Numbers", a=20, b=3)
print(f"Calculating 20 * 3: {result}") 
#  Calculating 20 * 3: 60.0

For more information about gRPC and how to use it with Python, please see:

Note: In this example, the user doesn't deposit funds to MPE, doesn't open a channel, and doesn't perform other actions related to payment. In this case, the choice of payment strategy, as well as, if necessary, opening a channel and depositing funds into MPE occurs automatically. For more information on payment, please visit the Payment section.

Payment ​

When creating a service client, you can select a payment strategy using the payment_strategy_type parameter:

python
from snet.sdk import PaymentStrategyType

payment_strategy_type = PaymentStrategyType.<NAME>

These are four payment strategies:

  • PaymentStrategyType.DEFAULT
  • PaymentStrategyType.FREE_CALL
  • PaymentStrategyType.PAID_CALL
  • PaymentStrategyType.PREPAID_CALL

The default payment strategy selects one of the other three each time the service is called, depending on the availability of free calls, as well as the presence of parameters required for concurrent calls. While choosing a specific payment strategy will not allow you to switch to another. This is especially convenient when you want to use free calls without accidentally spending money.

Note: If you don't specify еру payment_strategy_type parameter, the default payment strategy will be used.

Free call ​

If you want to use the free calls you will need to choose PaymentStrategyType.FREE_CALL as the payment strategy type. Creating a service client with free calls included would look like this:

python
service_client = snet_sdk.create_service_client(org_id="26072b8b6a0e448180f8c0e702ab6d2f", 
                                                service_id="Exampleservice",
                                                payment_strategy_type = PaymentStrategyType.FREE_CALL)

If you want to use regular paid calls you will need to choose PaymentStrategyType.PAID_CALL as the payment strategy type. Creating a service client with paid calls would look like this:

python
service_client = snet_sdk.create_service_client(org_id="26072b8b6a0e448180f8c0e702ab6d2f", 
                                                service_id="Exampleservice",
                                                payment_strategy_type = PaymentStrategyType.PAID_CALL)

There is no need to call functions for interacting with payment channels, because they are automatically managed by the SDK. But anyway you can use them if you want.

Open channel with the specified amount of funds and expiration ​

open_channel()[1] opens a payment channel with the specified amount of FET tokens in cogs and expiration time. Expiration is payment channel's TTL in blocks. When opening a channel, funds are taken from MPE. So they must be pre-deposited on it. For this, you can use the deposit_to_escrow_account()[1] method.

python
snet_sdk.account.deposit_to_escrow_account(123456)
service_client.open_channel(amount=123456, expiration=33333)

You can also use the deposit_and_open_channel()[1] method instead. It does the same as the previous one, but first deposits the specified amount of FET tokens in cogs into an MPE.

python
service_client.deposit_and_open_channel(amount=123456, expiration=33333)

Extend expiration and add funds ​

open_channel() as well as deposit_and_open_channel() returns the payment channel. You can use it to add funds to it and extend its expiration using the following methods: add_funds()[1], extend_expiration[1] and extend_and_add_funds()[1].

python
payment_channel = service_client.open_channel(amount=123456, expiration=33333)

payment_channel.add_funds(amount=123456)
payment_channel.extend_expiration(expiration=33333)

payment_channel.extend_and_add_funds(amount=123456, expiration=33333)

Concurrent (Prepaid) call ​

Concurrent (prepaid) calls allow you to prepay for a batch of service calls in advance. This off-chain strategy is ideal for scenarios requiring high throughput and low latency. Unlike regular paid calls, the payment is done once upfront, and the SDK automatically manages the channel during usage.

If you want to use prepaid calls you will need to choose PaymentStrategyType.PREPAID_CALL as the payment strategy type as well as pass the number of concurrent calls as the concurrent_calls parameter. Creating a service client with prepaid calls would look like this:

python
service_client = snet_sdk.create_service_client(
    org_id="26072b8b6a0e448180f8c0e702ab6d2f",
    service_id="Exampleservice",
    group_name="default_group",
    payment_strategy_type=PaymentStrategyType.PREPAID_CALL,
    concurrent_calls=5  # Number of prepaid calls to allocate
)

Then you can make service calls as usual, and the SDK will use the prepaid pool internally:

python
for i in range(5):
    response = service_client.call_rpc("add", "Numbers", a=1, b=2)
    print(f"Concurrent call {i+1} result:", response)

This model is especially useful for batch inference or rapid sequential calls without incurring on-chain transaction costs for each invocation.

Train call ​

Some of the training methods, namely upload_and_validate and train_model, are paid as well as the regular service call. Accordingly, you need to pay some FET to take advantage of the training. For this, as for a regular service call, you need a payment channel with the required amount of funds on it and expiration (in Python SDK, the selection, opening or adding funds to the channel is done automatically).

The only difference is that the price of a service call is a static number stored in the service metadata, whereas the price of calling the methods above is determined each time through the service provider before calling these methods. There are auxiliary methods validate_model_price and train_model_price respectively to determine the price of calling paid methods.

python
validate_price = service_client.training.validate_model_price(model_id)
model_status = service_client.training.upload_and_validate(model_id, zip_path, validate_price)

# -------------------------------------------------------------------------------

train_price = service_client.training.train_model_price(model_id)
model_status = service_client.training.train_model(model_id, train_price)

Other useful features ​

Get the current block number ​

Service client also provides several useful functions. If you need to find out the number of the current block in the blockchain, there is a get_current_block_number() method for this:

python
block_number = service_client.get_current_block_number()
print(f"Current block is {block_number}")
# Current block is 6574322

Get the service call price ​

To find out the price of calling a service function, you need to use the get_price() method:

python
price = service_client.get_price()
print(f"The price in cogs for calling the service {service_client.service_id} is {price}")
# The price in cogs for calling the service Exampleservice is 1

Get the metadata of the service ​

The metadata of services is stored in IPFS. To view it, you need to call the get_service_metadata() method, passing the organization id and the service id to it.

python
service_metadata = snet_sdk.get_service_metadata(org_id="26072b8b6a0e448180f8c0e702ab6d2f", service_id="Exampleservice")
print(*service_metadata.m.items(), sep="\n", end="\n\n")
print(*service_metadata.get_tags(), sep=",", end="\n\n")
print(*service_metadata.get_all_endpoints_for_group(group_name="default_group"), sep=",", end="\n\n")

# ('version', 1)
# ('display_name', 'Example service')
# ('encoding', 'proto')
# ('service_type', 'grpc')
# ('model_ipfs_hash', 'QmeyrQkEyba8dd4rc3jrLd5pEwsxHutfH2RvsSaeSMqTtQ')
# ('mpe_address', '0x7E0aF8988DF45B824b2E0e0A87c6196897744970')
# ('groups', [{'free_calls': 0, 'free_call_signer_address': '0x7DF35C98f41F3Af0df1dc4c7F7D4C19a71Dd059F', 'daemon_addresses': ['0x0709e9b78756b740ab0c64427f43f8305fd6d1a7'], 'pricing': [{'default': True, 'price_model': 'fixed_price', 'price_in_cogs': 1}], 'endpoints': ['http://node1.naint.tech:62400'], 'group_id': '/mb90Qs8VktxGQmU0uRu0bSlGgqeDlYrKrs+WbsOvOQ=', 'group_name': 'default_group'}])
# ('service_description', {'url': 'https://ropsten-v2-publisher.singularitynet.io/org', 'short_description': 'Example service', 'description': 'Example service'})
# ('media', [{'order': 1, 'url': 'https://ropsten-marketplace-service-assets.s3.us-east-1.amazonaws.com/26072b8b6a0e448180f8c0e702ab6d2f/services/d05c62bf9aa84843a195457d98417f4e/assets/20240327124952_asset.jpeg', 'file_type': 'image', 'asset_type': 'hero_image', 'alt_text': ''}])
# ('contributors', [{'name': 'test', 'email_id': ''}])
# ('tags', ['exampleservice'])
# 
# exampleservice
# 
# http://node1.naint.tech:62400

Get raw services and messages info ​

In the section Calling the service we already talked about the function get_services_and_messages_info_as_pretty_string(), with which you can get information about the methods and messages of a service. But if you need to process lists of services and messages, it is better to use the get_services_and_messages_info() method.

python
services, messages = service_client.get_services_and_messages_info()
print(services)
print(messages)

# {'Calculator': [('add', 'Numbers', 'Result'), ('sub', 'Numbers', 'Result'), ('mul', 'Numbers', 'Result'), ('div', 'Numbers', 'Result')]}
# {'Numbers': [('float', 'a'), ('float', 'b')], 'Result': [('float', 'value')]}

Training ​

With the SDK, you can also train models and use them when calling the service.

Base pipeline ​

The sequence of basic actions is as follows:

  1. Create models
  2. Upload training dataset
  3. Train the model
  4. Call the service based on the new model
create_model ​

To create a new model you need to call the create_model method. It takes the following parameters:

  • method_name - name of the service method for which we want to create a new model (use get_training_metadata method to get the list of available methods)
  • model_name - name of the new model (you need to come up with this)
  • model_description- description of the new model (optional)
  • is_public_accessible - whether the model is publicly accessible (optional, default: False)
  • addresses_with_access - list of addresses with access to the model (optional) (makes sense only if is_public_accessible is False)

and returns a Model object with all the model information.

python
new_model = service_client.training.create_model(method_name=grpc_method_name,
                                                 model_name=model_name)
model_id = new_model.model_id

print(new_model.status) # ModelStatus.CREATED
upload_and_validate ​

To upload the training dataset you need to call the upload_and_validate method. It takes the following parameters:

  • model_id - id of the model
  • zip_path - path to archive file with the training dataset
  • validate_price - price of validating the dataset

and returns a ModelStatus object.

python
validate_price = service_client.training.validate_model_price(model_id)
zip_path = "PATH_TO_YOUR_DATASET_FILE"  
model_status = service_client.training.upload_and_validate(model_id, zip_path, validate_price)

print(model_status) # ModelStatus.VALIDATING

Note: Dataset validation usually takes some time, so you should wait for the VALIDATED status of the model (using the get_model or get_all_models methods) after sending the dataset for validation before proceeding further with the model.

train_model ​

To train the model on an uploaded dataset you need to call the train_model method. It takes the following parameters:

  • model_id - id of the model
  • train_price - price of training the model

and returns a ModelStatus object.

python
train_price = service_client.training.train_model_price(model_id)
model_status = service_client.training.train_model(model_id, train_price)

print(model_status) # ModelStatus.TRAINING

Note: Model training usually takes some time, so you should wait for the READY_TO_USE status of the model (using the get_model or get_all_models methods) after calling the train_model method before proceeding further with the model.

Finally, to call the service, you must call the call_rpc method of the ServiceClient instance with an additional parameter model_id in a similar way:

python
result = service_client.call_rpc(grpc_method_name, grpc_message_name, model_id=model_id, **parameters)

For more detailed description please refer to Developer Portal guides:

1 This method uses a call to a paid smart contract function. ​

Development ​

Installing ​

Prerequisites ​

  • Clone the git repository
bash
$ git clone git@github.com:singnet/snet-sdk-python.git
$ cd snet-sdk-python
  • Install the required dependencies
bash
$ pip install -r requirements.txt
  • Install the package in development/editable mode
bash
$ pip install -e .

License ​

This project is licensed under the MIT License - see the LICENSE file for details.