[Python] How to trigger multiple validators when an attribute is changed in an existing...

I am using attrs (https://www.attrs.org) with custom initialization and setter validations.

When I update an attribute, I want to validate that attribute and other attributes that depend on it.

Below is a simple example. The actual application is for a comparative genomics analysis package, so this is a significant simplification. I want to use attrs instead of pydantic and I want the instance attributes to be mutable.

I tried another approach where on_setattr=_some_custom_func is in field(), but I wasn't able to get it to work; it set the attribute to None instead of the new parameter, and I didn't pursue it further.

Code:

import warnings

from attrs import define, field, setters, validators
from collections import namedtuple

StoreParams = namedtuple("StoreParams", ["hours"])
HourParams = namedtuple("HourParams", ["opening", "closing"])

@define(slots=True, kw_only=True, on_setattr=[setters.convert, setters.validate])
class Foo:

# For this parent class Foo
ALLOWED_STORES = ("grocer_a", "grocer_b", "grocer_c", "grocer_d", "grocer_e", "grocer_f")

# Will be in subclass of Foo, but in Foo for troubleshooting
PARAMS = {
"store": {
"grocer_a": StoreParams(hours=HourParams(5, 18)),
"grocer_b": StoreParams(hours=HourParams(5, 20)),
"grocer_c": StoreParams(hours=HourParams(6, 18)),
}
}

# Initializations
store: str = field(
default="blah",
converter=lambda v: v.lower() if v is not None and isinstance(v, str) else v,
validator=[
validators.optional(validators.instance_of(str)),
validators.in_(ALLOWED_STORES),
],
)

# Notes on store init:
# use isinstance() because converters run before validators; .lower error issued before TypeError
# optional allows None type
# raise ValueError if `store` nonsense (e.g., `store=meow`)
# issue UserWarning `store` is not nonsense but not recommended in subclass

hours: float = field(
default=None,
validator=validators.optional(validators.instance_of(int)),
)

@store.validator
def _validate_stores(self, attribute, value):
"""Ensure the store parameter is permitted in the subclass instance."""
print("Debug 1", value)
if value is not None and value not in self.PARAMS["store"]:
msg = (
f"The recommended options for `{attribute.name}` are "
f"({list(self.PARAMS['store'].keys())}), but '{value}' was entered."
)
warnings.warn(msg, UserWarning)

self._validate_hours(self.__class__.hours, self.hours) # Doesn't seem to be executing?


@hours.validator
def _validate_hours(self, attribute, value):
"""Ensure the hours parameter is valid for the store parameter in the subclass instance."""
print("Debug 2")

if value is None:
return

# Determine valid hours range based on the store
if self.store in self.PARAMS["store"]:
hours_range = self.PARAMS["store"][self.store].hours
else:
hours_range = self.ALLOWED_STORES

min_val, max_val = hours_range.opening, hours_range.closing

# Check if hours is within range
if not (min_val <= value <= max_val):
warnings.warn(
f"Provided hours {value} is out of range for store '{self.store}' "
f"(valid range: {min_val} - {max_val})",
UserWarning
)

# Other logic


Testing:

my_instance = Foo(store="grocer_A", hours=3)
# Warning issued as expected
# Debug 1 grocer_a
# Debug 2
# Debug 2

my_instance = Foo(store="grocer_A", hours=5)
# Debug 1 grocer_a
# Debug 2
# Debug 2

print(f"*The store is set as {my_instance.store} with hours {my_instance.hours}.")
# *The store is set as grocer_a with hours 5.

my_instance.store = "grocer_c"
# Debug 1 grocer_c
# Debug 2
# A Warning should have been issued! 5 is not allowed for grocer_c

print(f"*The store is set as {my_instance.store} with hours {my_instance.hours}.")
# *The store is set as grocer_c with hours 5.

Continue reading...