date and datetime support about typedload

[ltworf]

It is not implemented, I don't know if I will ever implement it, but there are handlers to do this sort of things.

dumper = typedload.datadumper.Dumper() # Create a Dumper object, so we can modify it

# Let's add a new type handler to the object, so it can handle one more type
dumper.handlers.append(
    (
        lambda x: isinstance(x, datetime.date), # Function that matches the type
        lambda self, value: repr(value), # Function that does something with the value
    )
)

print(
    dumper.dump( # Dump using the custom Dumper object
        Point(x=5.0, y=6.0, created_at=date(2019, 1, 1))
    )
)

I used repr() to do the dump, but you can replace it with whatever you like.

Be warned that loading this thing back might be complicated. Do you need to do that?

[ltworf]

Dumping the date object as a date object rather than converting it to string would work as well I guess.

dumper = typedload.datadumper.Dumper() # Create a Dumper object, so we can modify it
dumper.basictypes.add(datetime.date) # Add datetime.date to the set of types that do no action.

print(
    dumper.dump( # Dump using the custom Dumper object
        Point(x=5.0, y=6.0, created_at=date(2019, 1, 1))
    )
)

# Now it is trivial to load
loader=typedload.dataloader.Loader()
loader.basictypes.add(datetime.date)
loader.load(Point(x=5.0, y=6.0, created_at=date(2019, 1, 1)), Point)

The disadvantage of this method is that you can't serialize to json. Which is why the type isn't added by default in the 1st place.

[marlonjan]

Thanks for your quick reply, @ltworf!

Be warned that loading this thing back might be complicated. Do you need to do that?

Yes, it's going to be used both ways, to JSON and back to NamedTuple.

I realized that support for Python dataclasses is increasing quickly. In my project I can replace all the NamedTuples by dataclasses, then tools like pydantic or marshmallow-dataclass should to the job.

[ltworf]

Well this project supports dataclasses as well.

[ltworf]

There… now that I implemented it I might as well add support :D

import datetime
import dataclasses

import typedload
import typedload.dataloader
import typedload.datadumper


dumper = typedload.datadumper.Dumper()
dumper.handlers.append(
    (
        lambda x: isinstance(x, datetime.date),
        lambda self, value: (value.year, value.month, value.day),
    )
)

loader.handlers.append(
    (
        lambda x: x == datetime.date,
        lambda self, value, type_: datetime.date(*value),
    )
)


@dataclasses.dataclass
class Pino:
    pino: datetime.date

t = Pino(datetime.date(2019, 1, 1))
dump = dumper.dump(t)
assert loader.load(dump, Pino) == t

[thomascellerier]

Hurray :)

[ltworf]

Maybe you want to review it then? :P