Unable to call `get_constraints` when using a `ColumnFormula` constraint

[sarahmish]

Environment Details

• SDV-Enterprise version: 0.47.4 (latest)
• Python version: 3.12
• Operating System: Linux (Google Colab)

Error Description

According to the documentation, I should be able to use the get_constraints function to retrieve the list of constraint objects attached to a synthesizer as well as produce a JSON file with the constraint information. However, it seems that if I have added a ColumnFormula constraint (currently in sandbox mode), this method crashes.

Expected Behavior

I would expect the get_constraints function to work. In the JSON, I expect the listed parameters to correspond with the parameters I used to create the constraint.

_ = synthesizer.get_constraints(filepath="constraints.json")

JSON file:

[{
  "class_name": "ColumnFormula",
  "parameters": {
    "table_name": "guests",
    "input_columns": ["room_rate", "amenities_fee"],
    "output_column": "total_rate",
    "formula_function_name": "calculate_sum",
    "module_name": "my_formulas"
  }
}]

In the future, I should also be able to load constraints from this file. This would expect the module/function to be accessible for the constraint. Note that this function should first try importing the classes from the sdv.cag namespace; if not found, it should then fallback to the sdv.cag.sandbox namespace.

from sdv.utils import load_constraints

constraints = load_constraints(filepath='constraints.json')
synthesizer.add_constraints(constraints)

If the module/function is not accessible, then the load_constraints function should perform a try/catch and ignore this constraint. I should expect to see a warning.