pd.DataFrame breaks JSON schema generation

[thomas-pedot]

When using Arbitrary type we cannot serialize to a JSON Schema.

Work

PandasDataFrame = TypeVar('pd.DataFrame')

class Model(BaseModel):
    pandasDF:  PandasDataFrame 

Doesnot work

class Model(BaseModel):
    pandasDF:  pd.DataFrame
    class Config:
        arbitrary_types_allowed = True # pd.DataFrame allowed

Following error is raised when building the docs:

ValueError: Value not declarable with JSON Schema, field: name='inputs_DataFrame' type=DataFrame required=True

[mansenfranzen]

Thanks for the bug report. I can reproduce the issue. Be aware there is already a mechanism to capture fields which are not JSON serializable (see warning here in the docs). However, the pandas dataframe dtype currently escapes this safety guard.

[mansenfranzen]

@thomas-pedot PR #30 addresses the current issue allowing you to define how autodoc_pydantic should behave in case a pydantic field can't be JSON serialized properly. Please see the updated docs on this.

Before merging the related PR, it would be great if you could test the bug fix on your site to confirm that my test case did indeed capture the incorrect behavior which you've encountered. To do so, please install the current dev release in your doc-building-environment via pip install git+git://github.com/mansenfranzen/[email protected] and rebuild your docs.

[mansenfranzen]

Closing this issue via #30. Please feel free to reopen for further discussions.