Serializers

Artifacts are serialized before being stored in the database. This allows you to store arbitrary data types in your artifacts. You can use the built-in serializers or you can create your own.

Built-in Serializers

Serializer	Description
core.serializer.datetime.datetime_serializer	Date Times
core.serializer.json.json_serializer	JSON
core.serializer.json.json_sorted_serializer	JSON with sorted keys
extra.numpy.array_serializer	1d and 2d Numpy Arrays
extra.pandas.dataframe_serializer	Pandas DataFrames
extra.plotly.figure_json_serializer	Plotly Figures
extra.polars.dataframe_serializer	Polars DataFrames
extra.pyarrow.feather_serializer	PyArrow Feather Files
extra.pyarrow.parquet_serializer	PyArrow Parquet Files

Custom Serializers

You can create your own serializer by subclassing artigraph.serializer.Serializer:

from typing import TypeVar

from artigraph.serializer import Serializer

T = TypeVar("T")


class CustomSerializer(Serializer[T]):

    types = (object,)
    """The types that this serializer can handle"""

    def __init__(self):
        # This must be GLOBALLY unique and stable across versions!
        self.name = "custom-serializer"

    def serialize(self, value: T) -> bytes:
        """serialize the value to bytes"""

    def deserialize(self, value: bytes) -> T:
        """deserialize the value from bytes"""


CustomSerializer().register()