Angelia - Hermes API 2.0 Interface

Install

1. create python 3.11 virtual environment by using pyenv or your favourite method
2. poetry install

Single project version with messaging

1. Ensure Postgres and Rabbit MQ services are running on local machines.

2. For full development system run: API, Dispatch, Correct number of Consumers, Updater Services by running the appropriate commands in separate terminal windows.

3. Hermes must be running or linked to.

To see current list of commands:

    poetry run manage --help

to see current list of commands

For example runs the development werkzeug server:

    poetry run manage run-api-server

To run up Hermes API messaging using command line requires a .env to be added to the project directory and will be read when using poetry run. To build an excample .env run

    poetry run manage write-example-env

Remember to edit the .env for your test environment and do not commit.

Secrets/ Vault

It is safer not to load any vault secrets onto your local PCs to run code. Better to set up your .env environment (see below) to use a copy of example_local_secrets.json.

LOCAL_SECRETS=True

This will use mock secrets in "example_local_secrets.json"

If you want to setup your own secrets copy the above file and set

LOCAL_SECRETS_PATH=my_secrets_file_name.json

PyCharm Running and Debugging

Typically PyCharm expects to run and debug using configurations. This requires more work to set up but is easier for frequent users and for debugging via an IDE. You will need a configuration for each service eg API, Dispatch, Consumers, Retry and Updater Set up pycharm by creating configurations with:

script = "PATH_TO_ANGELIA/angelia/cli/commands.py"
parameter = the command eg "run-api-server".

Also the same environment variables must be set in each configuration. This can be done in a terminal by running manage write-example-env to create an .env file. After editing, select all lines and copy the selection then paste into the configuration form's environmental variables. If one service is correct it can be duplicated and only needs the name and command parameter amended.

Linting

Ruff is in the CI so to run it locally you can do:

poetry run ruff .
poetry run ruff format .

Monitoring

On localhost you can define in .env (see commands)

PERFORMANCE_METRICS=1

Run the following command to view messages:

while true;do;nc localhost  -l 4000;echo "\n";done

URLS and Resources and Models

URLS

see in resources file urls.py which contains a def associating and end point url (hint) to a resource class Resources can use the endpoint url in different ways but non standard use merits a comment

Resources

each resource class handles one url endpoint and associated methods.

They are identical Falcon resources except we use a base class to make it easier to handle dynamic sessions and to set up the url endpoint. This allows resources to overide or not extend from Base class inorder to set up a different url ignoring or using in a different way the url defined in urls.py

Models

The models are maintained in Hermes using Django. Sqlalchamy has matching classes for each table which are defined using reflection. This means only the relationships need to be defined which cannot be abstracted from Postgres. However, autocomplete with respect to table contents will not work in the IDE but can be referenced in classes

Example query in a resource end point method where user and user detail are one to one:

# Reflect each database table we need to use, using metadata
class User(Base):
    __table__ = Table('user', metadata, autoload=True)
    profile = relationship("UserDetail", backref="user", uselist=False)   # uselist = False sets one to one relation


class UserDetail(Base):
    __table__ = Table('user_userdetail', metadata, autoload=True)

Note use of self.session in resource class which is read only for gets and write for other methods

   for user in self.session.query(User).filter(User.email.like('%bink%')).order_by(User.id):
        if user.profile.first_name:
            print(user.id, user.email, user.external_id, user.profile.first_name)

Commands:

• poetry run manage --help Run the above command in the project directory to get the list of available commands

• poetry run manage write-example-env Run the above command in the project directory to create an example .env file

• poetry run manage run-api-server Runs a development/debug server

Run as production server with Gunicorn:

from top project directory: poetry run gunicorn -b 0.0.0.0:5000 main:app

Environment Variables

Common env variables:

• LOG_LEVEL
  • Log messages with this level or above. e.g. "DEBUG"
• POSTGRES_READ_DSN
  • Read Postgres DSN e.g. postgresql://user:pass@host:port/databasename
• POSTGRES_WRITE_DSN
  • Write Postgres DSN e.g. postgresql://user:pass@host:port/databasename
• RABBIT_USER
  • Username to use for auth with RabbitMQ
• RABBIT_PASSWORD
  • Password to use for auth with RabbitMQ
• RABBIT_HOST
  • IP address for RabbitMQ, e.g. "127.0.0.1"
• RABBIT_PORT
  • Port for RabbitMQ, e.g. "5672"
• HERMES_URL
  • Hermes API address
• METRICS_SIDECAR_DOMAIN
  • Metrics domain set to =localhost
• METRICS_PORT
  • Metrics port set to =4000
• PERFORMANCE_METRICS
  • Metrics performance set to send ie =1

API env variables:

• URL_PREFIX
  • Sets prefix for url path. Defaults to "api2" so a request to membership plan would go to "/api2/membership_plan"

Retry env variables:

• RETRY_TIME
  • Only retry messages older than x seconds (max. retry period is 1200s or 20 mins)
• RETRY_INCREASE
  • Increase retry time by x seconds each failed retry (max. retry period is 1200s or 20 mins)
• MESSAGE_LOG_TIME_LIMIT
  • Deletes message logs which are older than x hours

Required Services For a Functional Hermes:

1. Api server use manage run-api-server - multiple instances typically required

Note: Hermes must be running with messaging enabled in environment variables.