OpenAlchemy

Added

• Add support for int values for number types.

Added

• Add support for SQLAlchemy 1.4.
• Add support for grouping models into schemas. [#294]

Added

• Add support for OpenAPI 3.1. [#276]

Fixed

• Removed unnecessary imports in __init__.py files. [#255]

Added

• Caching validation results to speed up startup. [#251]

Added

• Add support for namespaced x-open-alchemy- prefix on top of the shorter
  x- prefix for extension properties. [#236]

Changed

• Changed from setup.py to poetry

Added

• Add version, title and description (if defined) into the JSON OpenAPI
  specification stored with the package generated by the build module.

Added

• Add check enforcing unique x-tablename values. [#189]
• Add check enforcing unique x-secondary values. [#189]
• Add custom association schemas validation [#189]
• Add support for custom association tables [#189]
• Add openalchemy CLI with a first sub command to build a Python package
  from a specification file. [#201]
• Add a CLI sub command to regenerate models. [#202]
• Add support for database default values using x-server-default. [#196]

Changed

• Change the association table to no longer be noted on the models based on
  the x-secondary value and instead be noted based on converting the
  x-secondary value from snake_case to PascalCase. Name clashes are avoided
  by pre-pending Autogen as many times as required. [#189]
• Change the association table to no longer be constructed as a table and
  instead to be constructed as another model. [#189]
• Refactor column factory to use the schemas artifacts [#196]
• Refactor model factory to use the schemas artifacts [#196]

Fixed

• Fix bug where the association table defined for many-to-many relationships
  did not make the foreign key columns referencing the two sides of the
  relationship primary keys. This may require a database migration if alembic
  was used to generate the database schema. [#189]
• Fix bug where some properties were incorrectly picked from a reference even
  though they existed locally (only impacts relationship properties where, for
  example, x-secondary was defined both on the relationship property in
  allOf and on the referenced model). [#189]

Removed

• Remove define_all parameter for init_model_factory, init_json and
  init_yaml. OpenAlchemy now behaves as though define_all is set to
  True. This means that a pure model reference (a schema with only the
  $ref key) can no longer be used to change the name of a model. [#189]

• Refactor the models file generation to use the artifacts from the schemas.
• Add build_json and build_yaml interfaces which can be used to produce a package with the models.
• Changed output of build_json and build_yaml to be contained within a project directory.
• Add support for building sdist or wheel distributable archive.
• Drop support for Python 3.6 and add support for Python 3.9.

• Move description to be a top level property artifact for every property.

• Correct format key to no longer have a trailing _ for artifacts.

• Expose function that collects artifacts for the models.
• Expose function that collects artifacts for the model properties.

• Add support for arbitrary mix in classes.

• Add support for generic format for string that are treated like a string without a format.
• Change constructable check to no longer check the schema of x-tablename and x-inherits.
• Change schema validation to process properties even if the model is not valid.
• Add interface to check un-managed models for the reason why they are not managed.

• Remove dependency on black

• Fix bug where iterating over constructable schemas did not handle some exceptions
• Add black dependency back in

• Remove black dependency

• Add schemas pre-processor that extracts the required back references.
• Add foreign key pre-processor that extracts the required foreign keys.
• Add schema validation pre-processor.
• Add function that checks a specification.

• Add support for generic JSON data for properties.
• Add support for writeOnly.
• Fix bug where the name of the foreign key column was based on the table name and not the property name.