Developer's Read Me for Hyperledger Aries Cloud Agent - Python ¶
See the README for details about this repository and information about how the Aries Cloud Agent - Python fits into the Aries project and relates to Indy.
Table of Contents ¶
- Introduction
- Developer Demos
- Running
- Configuring ACA-PY: Environment Variables
- Configuring ACA-PY: Command Line Parameters
- Docker
- Locally Installed
- About ACA-Py Command Line Parameters
- Provisioning Secure Storage
- Mediation
- Multi-tenancy
- JSON-LD Credentials
- Developing
- Prerequisites
- Running In A Dev Container
- Running Locally
- Logging
- Running Tests
- Running Aries Agent Test Harness Tests
- Development Workflow
- Publishing Releases
- Dynamic Injection of Services
Introduction¶
Aries Cloud Agent Python (ACA-Py) is a configurable, extensible, non-mobile Aries agent that implements an easy way for developers to build decentralized identity services that use verifiable credentials.
The information on this page assumes you are developer with a background in decentralized identity, Aries, DID Methods, and verifiable credentials, especially AnonCreds. If you aren't familiar with those concepts and projects, please use our Getting Started Guide to learn more.
Developer Demos¶
To put ACA-Py through its paces at the command line, checkout our demos page.
Running¶
Configuring ACA-PY: Environment Variables¶
All CLI parameters in ACA-PY have equivalent environment variables. To convert a CLI argument to an environment variable:
-
Basic Conversion: Convert the CLI argument to uppercase and prefix it with
ACAPY_
. For example,--admin
becomesACAPY_ADMIN
. -
Multiple Parameters: Arguments that take multiple parameters, such as
--admin 0.0.0.0 11000
, should be wrapped in an array. For example,ACAPY_ADMIN="[0.0.0.0, 11000]"
- Repeat Parameters: Arguments like
-it <module> <host> <port>
, which can be repeated, must be wrapped inside another array and string escaped. For example, instead of:-it http 0.0.0.0 11000 ws 0.0.0.0 8023
use:ACAPY_INBOUND_TRANSPORT=[[\"http\",\"0.0.0.0\",\"11000\"],[\"ws\",\"0.0.0.0\",\"8023\"]]
For a comprehensive list of all arguments, argument groups, CLI args, and their environment variable equivalents, please see the argparse.py file.
Configuring ACA-PY: Command Line Parameters¶
ACA-Py agent instances are configured through the use of command line
parameters, environment variables and/or YAML files. All of the configurations
settings can be managed using any combination of the three methods (command line
parameters override environment variables override YAML). Use the --help
option to discover the available command line parameters. There are a lot of
them--for good and bad.
Docker¶
To run a docker container based on the code in the current repo, use the following commands from the root folder of the repository to check the version, list the available modes of operation, and see all of the command line parameters:
scripts/run_docker --version
scripts/run_docker --help
scripts/run_docker provision --help
scripts/run_docker start --help
Locally Installed¶
If you installed the PyPi package, the executable aca-py
should be available on your PATH.
Use the following commands from the root folder of the repository to check the version, list the available modes of operation, and see all of the command line parameters:
If you get an error about a missing module indy
(e.g. ModuleNotFoundError: No module named 'indy'
) when running aca-py
, you will need to install the Indy libraries from the command line:
Once that completes successfully, you should be able to run aca-py --version
and the other examples above.
About ACA-Py Command Line Parameters¶
ACA-Py invocations are separated into two types - initially provisioning an agent (provision
) and starting a new agent process (start
). This separation enables not having to pass in some encryption-related parameters required for provisioning when starting an agent instance. This improves security in production deployments.
When starting an agent instance, at least one inbound and one outbound transport MUST be specified.
For example:
or
aca-py start --inbound-transport http 0.0.0.0 8000 \
--inbound-transport ws 0.0.0.0 8001 \
--outbound-transport ws \
--outbound-transport http
ACA-Py ships with both inbound and outbound transport drivers for http
and ws
(websockets). Additional transport drivers can be added as pluggable implementations. See the existing implementations in the transports module for getting started on adding a new transport.
Most configuration parameters are provided to the agent at startup. Refer to the Running
sections above for details on listing the available command line parameters.
Provisioning Secure Storage¶
It is possible to provision a secure storage (sometimes called a wallet--but not
the same as a mobile wallet app) before running an agent to avoid passing in the
secure storage seed on every invocation of an agent (e.g. on every aca-py start ...
).
For additional provision
options, execute aca-py provision --help
.
Additional information about secure storage options and configuration settings can be found here.
Mediation¶
ACA-Py can also run in mediator mode - ACA-Py can be run as a mediator (it can mediate connections for other agents), or it can connect to an external mediator to mediate its own connections. See the docs on mediation for more info.
Multi-tenancy¶
ACA-Py can also be started in multi-tenant mode. This allows the agent to serve multiple tenants, that each have their own wallet. See the docs on multi-tenancy for more info.
JSON-LD Credentials¶
ACA-Py can issue W3C Verifiable Credentials using Linked Data Proofs. See the docs on JSON-LD Credentials for more info.
Developing¶
Prerequisites¶
Docker must be installed to run software locally and to run the test suite.
Running In A Dev Container¶
The dev container environment is a great way to deploy agents quickly with code changes and an interactive debug session. Detailed information can be found in the Docs On Devcontainers. It is specific for vscode, so if you prefer another code editor or IDE you will need to figure it out on your own, but it is highly recommended to give this a try.
One thing to be aware of is, unlike the demo, none of the steps are automated. You will need to create public dids, connections and all the other steps yourself. Using the demo and studying the flow and then copying them with your dev container debug session is a great way to learn how everything works.
Running Locally¶
Another way to develop locally is by using the provided Docker scripts to run the ACA-Py software.
For example:
./scripts/run_docker start --inbound-transport http 0.0.0.0 10000 --outbound-transport http --debug --log-level DEBUG
To enable the Debug Adapter Protocol using the debugpy implementation for Python 3 Python debugger for Visual Studio/VSCode use the --debug
command line parameter.
When debugging an agent running within a docker container, you will need to set the DAP_HOST environment variable (defaults to localhost
) to 0.0.0.0
to allow forwarding from within your docker container.
Note that you may still find references to PTVSD, the deprecated implementation of DAP. PTVSD_HOST and PTVSD_PORT are interchangeable with DAP_HOST and DAP_PORT.
Example:
ENV_VARS="DAP_HOST=0.0.0.0" scripts/run_docker provision --log-level debug --wallet-type askar --wallet-name $(whoami) --wallet-key mysecretkey --endpoint http://localhost:8080 --no-ledger --debug
Any ports you will be using from the docker container should be published using the PORTS
environment variable. For example:
PORTS="5000:5000 8000:8000 10000:10000" ./scripts/run_docker start --inbound-transport http 0.0.0.0 10000 --outbound-transport http --debug --log-level DEBUG
Refer to the previous section for instructions on how to run ACA-Py.
Logging¶
You can find more details about logging and log levels here.
Running Tests¶
To run the ACA-Py test suite, use the following script:
To run the ACA-Py test suite with ptvsd debugger enabled:
To run specific tests pass parameters as defined by pytest:
Running Aries Agent Test Harness Tests¶
You can run a full suite of integration tests using the Aries Agent Test Harness (AATH).
Check out and run AATH tests as follows (this tests the aca-py main
branch):
git clone https://github.com/hyperledger/aries-agent-test-harness.git
cd aries-agent-test-harness
./manage build -a acapy-main
./manage run -d acapy-main -t @AcceptanceTest -t ~@wip
The manage
script is described in detail here, including how to modify the AATH code to run the tests against your aca-py repo/branch.
Development Workflow¶
We use Ruff to enforce a coding style guide.
Please write tests for the work that you submit.
Tests should reside in a directory named tests
alongside the code under test. Generally, there is one test file for each file module under test. Test files must have a name starting with test_
to be automatically picked up the test runner.
There are some good examples of various test scenarios for you to work from including mocking external imports and working with async code so take a look around!
The test suite also displays the current code coverage after each run so you can see how much of your work is covered by tests. Use your best judgement for how much coverage is sufficient.
Please also refer to the contributing guidelines and code of conduct.
Publishing Releases¶
The publishing document provides information on tagging a release and publishing the release artifacts to PyPi.
Dynamic Injection of Services¶
The Agent employs a dynamic injection system whereby providers of base classes are registered with the RequestContext
instance, currently within conductor.py
. Message handlers and services request an instance of the selected implementation using context.inject(BaseClass)
; for instance the wallet instance may be injected using wallet = context.inject(BaseWallet)
. The inject
method normally throws an exception if no implementation of the base class is provided, but can be called with required=False
for optional dependencies (in which case a value of None
may be returned).
Providers are registered with either context.injector.bind_instance(BaseClass, instance)
for previously-constructed (singleton) object instances, or context.injector.bind_provider(BaseClass, provider)
for dynamic providers. In some cases it may be desirable to write a custom provider which switches implementations based on configuration settings, such as the wallet provider.
The BaseProvider
classes in the config.provider
module include ClassProvider
, which can perform dynamic module inclusion when given the combined module and class name as a string (for instance aries_cloudagent.wallet.indy.IndyWallet
). ClassProvider
accepts additional positional and keyword arguments to be passed into the class constructor. Any of these arguments may be an instance of ClassProvider.Inject(BaseClass)
, allowing dynamic injection of dependencies when the class instance is instantiated.