Hopeit.engine

______________

- Engine:

- Removed `typing_inspect` dependency in favor of `typing` built-in module


- Plugins:

- DataFrames:
- Experimental: `DataBlocks` API, to allow manipulating and saving multiple datasets at once

______________

- Engine:

- Web:
- Allow to specify `cors_routes_prefix` to allow `cors_origin` to be applied
to custom routes names

______________

- Plugins:

- DataFrames:
- Support for `bool` fields
- `Dataframes.to_dataobjects`: option to convert null values (`nan`, `NaT`, etc) to None

______________

- Engine

- API: Support for `type | None` syntax, as well as `Optional[type]` annotation for optional query args

- Plugins:

- DataFrames, feature to allow schema evolution:
- Allowing optional and default values in `dataframe` objects
- Saving `json` schema for serialized Datasets
- Handling load errors with `DatasetLoadException`
- Allowing load dataset with deprecated fields (removed fields)
- Allowing load dataset with new fields in schema using a default value

______________

- Engine:

- This release adopts `pydantic` as a the library for dataclasses and schema validation

BREAKING CHANGES
================
- Python `dataclasses.dataclass` is no longer supported in `dataobjects` annotated classes.
Pydantic `dataclasses` (and in the future BaseModel should be used).

- The recommended way is to import from dataobjects module:

from hopeit.dataobjects import dataclass, dataobject, field

where dataclass and field are aliases of `pydantic.dataclasses.dataclass` and `pydantic.Field`

- In most of the cases replacing the import clause in applications should suffice,
but could be features of dataclasses that should be replaced by its
equivalent in `pydantic`.
- For cases where you store data in the field property metadata, you can access it as follows:

from hopeit.dataobjects import dataclass, dataobject, field

dataobject
dataclass
class User:
name: str = field(metadata={"key": "value"})

metadata = User.name.json_schema_extra["metadata"]



- Plugins:

- Dataframes

- Made it compatible with pydantic dataclasses
- Removed `dataframeobject` annotation in favor of Dataset[T] generic type
- Introduced `.DataObject` companion type for `dataframe` conversion to `DataObject`