SIM - BE - Fix - json load for validator config

[bradleySuira]

When test in main I got this error when try to update the validator and the same when try to get the list of them

jsonrpc-1     | ERROR | update_validator: {'status': 'error', 'message': 'the JSON object must be str, bytes or bytearray, not dict', 'data': {'traceback': 'Traceback (most recent call last):\n  File "/app/backend/protocol_rpc/endpoint_generator.py", line 18, in endpoint\n    result = function(*args, **kwargs)\n             ^^^^^^^^^^^^^^^^^^^^^^^^^\n  File "/app/backend/protocol_rpc/endpoints.py", line 228, in update_validator\n    return validators_registry.update_validator(\n           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n  File "/app/backend/database_handler/validators_registry.py", line 85, in update_validator\n    return self._get_validator_or_fail(validator_address)\n           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n  File "/app/backend/database_handler/validators_registry.py", line 35, in _get_validator_or_fail\n    return self._parse_validator_data(validator_data)\n           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n  File "/app/backend/database_handler/validators_registry.py", line 22, in _parse_validator_data\n    "config": json.loads(validator_data["config"]),\n              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n  File "/usr/local/lib/python3.12/json/__init__.py", line 339, in loads\n    raise TypeError(f\'the JSON object must be str, bytes or bytearray, \'\nTypeError: the JSON object must be str, bytes or bytearray, not dict\n'}}
jsonrpc-1     | INFO | 192.168.65.1 - - [11/Jul/2024 20:23:34] "POST /api HTTP/1.1" 200 -

[AgustinRamiroDiaz]

Adding more context:

The issue revolves around how node configuration is managed:

• in the abstract sense, the node configuration is a key-value store
• in the implementation we have some places where it's a dictionary, and other where it's a string. This is what's causing problems since we end up parsing twice some jsons which leads to errors
• turns out that Postgres allows us to store either 🤯 https://www.postgresql.org/docs/current/datatype-json.html

So my proposed solution would be to:

• store node config as dict always: in python objects and in Postgres (jsonb seems like the best option)
• review backwards compatibility
  • I think that prior to this we were storing it as string
  • We could use Alembic to do the migration, and to also enforce jsonb as the type of the column

[bradleySuira]

Thanks @AgustinRamiroDiaz added cc to @cristiam86 with my solution based on your comments and pushed the changes to the PR

• As you said be were storing a string in a jsonb field in database
• I ended sending when call to create and update methods a valid json from the frontend
• in the backend side in both, update and create, I ended using this "config": AsIs("'" + json.dumps(config) + "'"),, with this change the field is stored as a json in the database and not as string
• Because the last point, we don't need to cal anymore the json.loads() method when we get the validators data, so json.loads(validator_data["config"]), is now validator_data["config"] like we have before

I tested addling listing and updating the validators and also the contract deployment and methods calls