2019-07-16T22:41:44 Created temporary directory: /tmp/pip-ephem-wheel-cache-x8u9335f
2019-07-16T22:41:44 Created temporary directory: /tmp/pip-req-tracker-sdp27y9b
2019-07-16T22:41:44 Created requirements tracker '/tmp/pip-req-tracker-sdp27y9b'
2019-07-16T22:41:44 Created temporary directory: /tmp/pip-wheel-3o8c2wh_
2019-07-16T22:41:44 Collecting jsonschema-typed==0.1.0
2019-07-16T22:41:44   1 location(s) to search for versions of jsonschema-typed:
2019-07-16T22:41:44   * https://pypi.org/simple/jsonschema-typed/
2019-07-16T22:41:44   Getting page https://pypi.org/simple/jsonschema-typed/
2019-07-16T22:41:44   Analyzing links from page https://pypi.org/simple/jsonschema-typed/
2019-07-16T22:41:44     Found link https://files.pythonhosted.org/packages/a1/1e/2eb6d612341ba74b9ffb945591430bbb70179da8b71d1ac28b2f10e543fb/jsonschema-typed-0.1.0.tar.gz#sha256=b0a04ba1571563deaa54423c8b8be1a08db99735df1598b7ed3dffb10a5c49a7 (from https://pypi.org/simple/jsonschema-typed/) (requires-python:~=3.6), version: 0.1.0
2019-07-16T22:41:44   Using version 0.1.0 (newest of versions: 0.1.0)
2019-07-16T22:41:44   Created temporary directory: /tmp/pip-unpack-et3_ol_q
2019-07-16T22:41:44   Downloading https://files.pythonhosted.org/packages/a1/1e/2eb6d612341ba74b9ffb945591430bbb70179da8b71d1ac28b2f10e543fb/jsonschema-typed-0.1.0.tar.gz
2019-07-16T22:41:44   Downloading from URL https://files.pythonhosted.org/packages/a1/1e/2eb6d612341ba74b9ffb945591430bbb70179da8b71d1ac28b2f10e543fb/jsonschema-typed-0.1.0.tar.gz#sha256=b0a04ba1571563deaa54423c8b8be1a08db99735df1598b7ed3dffb10a5c49a7 (from https://pypi.org/simple/jsonschema-typed/) (requires-python:~=3.6)
2019-07-16T22:41:44   Added jsonschema-typed==0.1.0 from https://files.pythonhosted.org/packages/a1/1e/2eb6d612341ba74b9ffb945591430bbb70179da8b71d1ac28b2f10e543fb/jsonschema-typed-0.1.0.tar.gz#sha256=b0a04ba1571563deaa54423c8b8be1a08db99735df1598b7ed3dffb10a5c49a7 to build tracker '/tmp/pip-req-tracker-sdp27y9b'
2019-07-16T22:41:44     Running setup.py (path:/tmp/pip-wheel-3o8c2wh_/jsonschema-typed/setup.py) egg_info for package jsonschema-typed
2019-07-16T22:41:44     Running command python setup.py egg_info
2019-07-16T22:41:46     # JSON Schema-powered type annotations

2019-07-16T22:41:46     The goal of this project is to use JSON Schema for type checking in Python.

2019-07-16T22:41:46     While there is not a perfect 1:1 mapping between concepts in JSON Schema and
2019-07-16T22:41:46     Python's typing system, there is enough of an isomorphism to warrant some
2019-07-16T22:41:46     exploration of the possibilities. Since a JSON document is generally
2019-07-16T22:41:46     represented as a ``dict`` in Python programs, this project looks specifically
2019-07-16T22:41:46     at interpreting JSON schema as
2019-07-16T22:41:46     [``TypedDict``](https://www.python.org/dev/peps/pep-0589/) definitions.

2019-07-16T22:41:46     **Warning:** there are bound to be some abuses of the [mypy plugin
2019-07-16T22:41:46     system](https://mypy.readthedocs.io/en/latest/extending_mypy.html) here. You
2019-07-16T22:41:46     have been warned.

2019-07-16T22:41:46     This leverages (and is inspired by) https://github.com/Julian/jsonschema.

2019-07-16T22:41:46     ## Overview

2019-07-16T22:41:46     A JSON schema:

2019-07-16T22:41:46     ```json
2019-07-16T22:41:46     {
2019-07-16T22:41:46         "$schema": "http://json-schema.org/draft-07/schema#",
2019-07-16T22:41:46         "$id": "http://foo.qwerty/some/schema#",
2019-07-16T22:41:46         "title": "Foo Schema",
2019-07-16T22:41:46         "type": "object",
2019-07-16T22:41:46         "properties": {
2019-07-16T22:41:46             "title": {
2019-07-16T22:41:46                 "type": "string"
2019-07-16T22:41:46             },
2019-07-16T22:41:46             "awesome": {
2019-07-16T22:41:46                 "type": "number"
2019-07-16T22:41:46             }
2019-07-16T22:41:46         },
2019-07-16T22:41:46         "required": ["title"]
2019-07-16T22:41:46     }
2019-07-16T22:41:46     ```

2019-07-16T22:41:46     A TypedDict:

2019-07-16T22:41:46     ```python
2019-07-16T22:41:46     from jsonschema_typed.types import JSONSchema

2019-07-16T22:41:46     data: JSONSchema['path/to/schema.json'] = dict(title='baz')
2019-07-16T22:41:46     data['description'] = 'there is no description'  # TypedDict "FooSchema" has no key 'description'
2019-07-16T22:41:46     data['awesome'] = 42
2019-07-16T22:41:46     data['awesome'] = None  # Argument 2 has incompatible type "None"; expected "Union[int, float]"
2019-07-16T22:41:46     ```

2019-07-16T22:41:46     ## Installation

2019-07-16T22:41:46     ```bash
2019-07-16T22:41:46     pip install jsonschema-typed
2019-07-16T22:41:46     ```

2019-07-16T22:41:46     or

2019-07-16T22:41:46     ```bash
2019-07-16T22:41:46     git clone git@github.com:erickpeirson/jsonschema-typed.git
2019-07-16T22:41:46     cd jsonschema-typed
2019-07-16T22:41:46     python setup.py install
2019-07-16T22:41:46     ```

2019-07-16T22:41:46     ## Requirements

2019-07-16T22:41:46     So far I have only tried this with:

2019-07-16T22:41:46     - mypy==0.701
2019-07-16T22:41:46     - jsonschema==3.0.1

2019-07-16T22:41:46     But probably older versions work. You could try it out and
2019-07-16T22:41:46     [let me know](https://github.com/erickpeirson/jsonschema-typed/issues).

2019-07-16T22:41:46     ## Limitations

2019-07-16T22:41:46     - ``additionalProperties`` doesn't really have an equivalent in TypedDict. Yet.
2019-07-16T22:41:46     - Cases in which the root of the schema is anything other than an ``object``
2019-07-16T22:41:46       are not terribly interesting for this project, so we ignore them for now.
2019-07-16T22:41:46       Array values for ``type`` (e.g. ``"type": ["object", "boolean"]``) are
2019-07-16T22:41:46       otherwise supported with ``Union``.
2019-07-16T22:41:46     - The ``default`` keyword is not supported; but see:
2019-07-16T22:41:46        https://github.com/python/mypy/issues/6131.
2019-07-16T22:41:46     - Self-references (e.g. ``"#"``) can't really work properly until nested
2019-07-16T22:41:46       forward-references are supported; see
2019-07-16T22:41:46       https://github.com/python/mypy/issues/731.

2019-07-16T22:41:46     There are probably others.


2019-07-16T22:41:46     ## Approach

2019-07-16T22:41:46     So far two approaches are attempted:

2019-07-16T22:41:46     1. Annotating a ``dict`` instance that will be a ``TypedDict`` that conforms to
2019-07-16T22:41:46        the JSON Schema (as best we can enforce it).
2019-07-16T22:41:46     2. Using a dynamic base class that is typed as a ``TypedDict``.

2019-07-16T22:41:46     Both examples below use the schema:

2019-07-16T22:41:46     ```json
2019-07-16T22:41:46     {
2019-07-16T22:41:46         "$schema": "http://json-schema.org/draft-07/schema#",
2019-07-16T22:41:46         "$id": "http://foo.qwerty/some/schema#",
2019-07-16T22:41:46         "title": "Foo Schema",
2019-07-16T22:41:46         "type": "object",
2019-07-16T22:41:46         "properties": {
2019-07-16T22:41:46             "title": {
2019-07-16T22:41:46                 "type": "string"
2019-07-16T22:41:46             },
2019-07-16T22:41:46             "awesome": {
2019-07-16T22:41:46                 "type": "number"
2019-07-16T22:41:46             }
2019-07-16T22:41:46         }
2019-07-16T22:41:46     }
2019-07-16T22:41:46     ```


2019-07-16T22:41:46     ### First approach: annotating a ``dict``

2019-07-16T22:41:46     This has the advantage of being fairly simple. It is implemented via
2019-07-16T22:41:46     ``jsonschema_typed.plugin.JSONSchemaPlugin.get_type_analyze_hook``.

2019-07-16T22:41:46     ```python
2019-07-16T22:41:46     from jsonschema_typed.types import JSONSchema

2019-07-16T22:41:46     data: JSONSchema['path/to/schema.json'] = dict(title='baz')
2019-07-16T22:41:46     reveal_type(data)  # Revealed type is 'TypedDict('FooSchema', {'title'?: builtins.str, 'awesome'?: Union[builtins.int, builtins.float]})'
2019-07-16T22:41:46     data['description'] = 'there is no description'  # TypedDict "FooSchema" has no key 'description'
2019-07-16T22:41:46     data['awesome'] = 42
2019-07-16T22:41:46     data['awesome'] = None  # Argument 2 has incompatible type "None"; expected "Union[int, float]"
2019-07-16T22:41:46     ```

2019-07-16T22:41:46     Here is the mypy output:

2019-07-16T22:41:46     ```
2019-07-16T22:41:46     main.py:4: error: Revealed type is 'TypedDict('FooSchema', {'title'?: builtins.str, 'awesome'?: Union[builtins.int, builtins.float]})'
2019-07-16T22:41:46     main.py:5: error: TypedDict "FooSchema" has no key 'description'
2019-07-16T22:41:46     main.py:7: error: Argument 2 has incompatible type "None"; expected "Union[int, float]"
2019-07-16T22:41:46     ```

2019-07-16T22:41:46     Note that the right-hand side can be a ``dict`` or a subclass of ``dict``, so
2019-07-16T22:41:46     you could define a subclass like:

2019-07-16T22:41:46     ```python
2019-07-16T22:41:46     class Foo(dict):
2019-07-16T22:41:46         """Some domain logic on your object."""

2019-07-16T22:41:46         def do_something(self, arg: int) -> int:
2019-07-16T22:41:46             """Do something awesome."""
2019-07-16T22:41:46             return arg * self['awesome']

2019-07-16T22:41:46     data: JSONSchema['schema/test.json'] = Foo(title='baz')
2019-07-16T22:41:46     reveal_type(data)  # Revealed type is 'TypedDict('FooSchema', {'title'?: builtins.str, 'awesome'?: Union[builtins.int, builtins.float]})'
2019-07-16T22:41:46     data['description'] = 'there is no description'  # TypedDict "FooSchema" has no key 'description'
2019-07-16T22:41:46     data['awesome'] = 42
2019-07-16T22:41:46     data['awesome'] = None  # Argument 2 has incompatible type "None"; expected "Union[int, float]"
2019-07-16T22:41:46     ```

2019-07-16T22:41:46     Of course this isn't quite consistent with PEP-589 which
2019-07-16T22:41:46     [states](https://www.python.org/dev/peps/pep-0589/#class-based-syntax) that:

2019-07-16T22:41:46     > Methods are not allowed, since the runtime type of a TypedDict object will
2019-07-16T22:41:46     > always be just dict (it is never a subclass of dict).

2019-07-16T22:41:46     So use at your own risk.

2019-07-16T22:41:46     ### Second approach: dynamic base class

2019-07-16T22:41:46     This has the advantage of being able to add some runtime-functionality, e.g.
2019-07-16T22:41:46     use ``jsonschema`` to actually validate data at runtime. It is implemented via
2019-07-16T22:41:46     ``jsonschema_typed.plugin.JSONSchemaPlugin.get_dynamic_class_hook``.

2019-07-16T22:41:46     But again, this isn't quite consistent with PEP-589 which
2019-07-16T22:41:46     [states](https://www.python.org/dev/peps/pep-0589/#class-based-syntax), so
2019-07-16T22:41:46     use at your own risk.

2019-07-16T22:41:46     ```python
2019-07-16T22:41:46     from jsonschema_typed.types import JSONSchemaBase


2019-07-16T22:41:46     Base = JSONSchemaBase('path/to/schema.json')

2019-07-16T22:41:46     class Foo(Base):
2019-07-16T22:41:46         """All your base in one place."""

2019-07-16T22:41:46         def do_something(self, arg: int) -> int:
2019-07-16T22:41:46             """Do something awesome."""
2019-07-16T22:41:46             return arg * self['awesome']


2019-07-16T22:41:46     data = Foo(title='baz')
2019-07-16T22:41:46     reveal_type(data)  # Revealed type is 'TypedDict('FooSchema', {'title'?: builtins.str, 'awesome'?: Union[builtins.int, builtins.float]})'
2019-07-16T22:41:46     data['description'] = 'there is no description'  # TypedDict "FooSchema" has no key 'description'
2019-07-16T22:41:46     data['awesome'] = 42
2019-07-16T22:41:46     data['awesome'] = None  # Argument 2 has incompatible type "None"; expected "Union[int, float]"
2019-07-16T22:41:46     ```

2019-07-16T22:41:46     ## TODO

2019-07-16T22:41:46     - [ ] Decide whether to stick with just one approach (and which one)
2019-07-16T22:41:46     - [ ] Write some tests
2019-07-16T22:41:46     - [ ] Test against other versions of mypy + jsonschema

2019-07-16T22:41:46     running egg_info
2019-07-16T22:41:46     creating pip-egg-info/jsonschema_typed.egg-info
2019-07-16T22:41:46     writing pip-egg-info/jsonschema_typed.egg-info/PKG-INFO
2019-07-16T22:41:46     writing dependency_links to pip-egg-info/jsonschema_typed.egg-info/dependency_links.txt
2019-07-16T22:41:46     writing requirements to pip-egg-info/jsonschema_typed.egg-info/requires.txt
2019-07-16T22:41:46     writing top-level names to pip-egg-info/jsonschema_typed.egg-info/top_level.txt
2019-07-16T22:41:46     writing manifest file 'pip-egg-info/jsonschema_typed.egg-info/SOURCES.txt'
2019-07-16T22:41:46     reading manifest file 'pip-egg-info/jsonschema_typed.egg-info/SOURCES.txt'
2019-07-16T22:41:46     writing manifest file 'pip-egg-info/jsonschema_typed.egg-info/SOURCES.txt'
2019-07-16T22:41:46   Source in /tmp/pip-wheel-3o8c2wh_/jsonschema-typed has version 0.1.0, which satisfies requirement jsonschema-typed==0.1.0 from https://files.pythonhosted.org/packages/a1/1e/2eb6d612341ba74b9ffb945591430bbb70179da8b71d1ac28b2f10e543fb/jsonschema-typed-0.1.0.tar.gz#sha256=b0a04ba1571563deaa54423c8b8be1a08db99735df1598b7ed3dffb10a5c49a7
2019-07-16T22:41:46   Removed jsonschema-typed==0.1.0 from https://files.pythonhosted.org/packages/a1/1e/2eb6d612341ba74b9ffb945591430bbb70179da8b71d1ac28b2f10e543fb/jsonschema-typed-0.1.0.tar.gz#sha256=b0a04ba1571563deaa54423c8b8be1a08db99735df1598b7ed3dffb10a5c49a7 from build tracker '/tmp/pip-req-tracker-sdp27y9b'
2019-07-16T22:41:46 Building wheels for collected packages: jsonschema-typed
2019-07-16T22:41:46   Created temporary directory: /tmp/pip-wheel-il0uer_e
2019-07-16T22:41:46   Building wheel for jsonschema-typed (setup.py): started
2019-07-16T22:41:46   Destination directory: /tmp/pip-wheel-il0uer_e
2019-07-16T22:41:46   Running command /usr/bin/python3 -u -c 'import setuptools, tokenize;__file__='"'"'/tmp/pip-wheel-3o8c2wh_/jsonschema-typed/setup.py'"'"';f=getattr(tokenize, '"'"'open'"'"', open)(__file__);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, __file__, '"'"'exec'"'"'))' bdist_wheel -d /tmp/pip-wheel-il0uer_e
2019-07-16T22:41:48   # JSON Schema-powered type annotations

2019-07-16T22:41:48   The goal of this project is to use JSON Schema for type checking in Python.

2019-07-16T22:41:48   While there is not a perfect 1:1 mapping between concepts in JSON Schema and
2019-07-16T22:41:48   Python's typing system, there is enough of an isomorphism to warrant some
2019-07-16T22:41:48   exploration of the possibilities. Since a JSON document is generally
2019-07-16T22:41:48   represented as a ``dict`` in Python programs, this project looks specifically
2019-07-16T22:41:48   at interpreting JSON schema as
2019-07-16T22:41:48   [``TypedDict``](https://www.python.org/dev/peps/pep-0589/) definitions.

2019-07-16T22:41:48   **Warning:** there are bound to be some abuses of the [mypy plugin
2019-07-16T22:41:48   system](https://mypy.readthedocs.io/en/latest/extending_mypy.html) here. You
2019-07-16T22:41:48   have been warned.

2019-07-16T22:41:48   This leverages (and is inspired by) https://github.com/Julian/jsonschema.

2019-07-16T22:41:48   ## Overview

2019-07-16T22:41:48   A JSON schema:

2019-07-16T22:41:48   ```json
2019-07-16T22:41:48   {
2019-07-16T22:41:48       "$schema": "http://json-schema.org/draft-07/schema#",
2019-07-16T22:41:48       "$id": "http://foo.qwerty/some/schema#",
2019-07-16T22:41:48       "title": "Foo Schema",
2019-07-16T22:41:48       "type": "object",
2019-07-16T22:41:48       "properties": {
2019-07-16T22:41:48           "title": {
2019-07-16T22:41:48               "type": "string"
2019-07-16T22:41:48           },
2019-07-16T22:41:48           "awesome": {
2019-07-16T22:41:48               "type": "number"
2019-07-16T22:41:48           }
2019-07-16T22:41:48       },
2019-07-16T22:41:48       "required": ["title"]
2019-07-16T22:41:48   }
2019-07-16T22:41:48   ```

2019-07-16T22:41:48   A TypedDict:

2019-07-16T22:41:48   ```python
2019-07-16T22:41:48   from jsonschema_typed.types import JSONSchema

2019-07-16T22:41:48   data: JSONSchema['path/to/schema.json'] = dict(title='baz')
2019-07-16T22:41:48   data['description'] = 'there is no description'  # TypedDict "FooSchema" has no key 'description'
2019-07-16T22:41:48   data['awesome'] = 42
2019-07-16T22:41:48   data['awesome'] = None  # Argument 2 has incompatible type "None"; expected "Union[int, float]"
2019-07-16T22:41:48   ```

2019-07-16T22:41:48   ## Installation

2019-07-16T22:41:48   ```bash
2019-07-16T22:41:48   pip install jsonschema-typed
2019-07-16T22:41:48   ```

2019-07-16T22:41:48   or

2019-07-16T22:41:48   ```bash
2019-07-16T22:41:48   git clone git@github.com:erickpeirson/jsonschema-typed.git
2019-07-16T22:41:48   cd jsonschema-typed
2019-07-16T22:41:48   python setup.py install
2019-07-16T22:41:48   ```

2019-07-16T22:41:48   ## Requirements

2019-07-16T22:41:48   So far I have only tried this with:

2019-07-16T22:41:48   - mypy==0.701
2019-07-16T22:41:48   - jsonschema==3.0.1

2019-07-16T22:41:48   But probably older versions work. You could try it out and
2019-07-16T22:41:48   [let me know](https://github.com/erickpeirson/jsonschema-typed/issues).

2019-07-16T22:41:48   ## Limitations

2019-07-16T22:41:48   - ``additionalProperties`` doesn't really have an equivalent in TypedDict. Yet.
2019-07-16T22:41:48   - Cases in which the root of the schema is anything other than an ``object``
2019-07-16T22:41:48     are not terribly interesting for this project, so we ignore them for now.
2019-07-16T22:41:48     Array values for ``type`` (e.g. ``"type": ["object", "boolean"]``) are
2019-07-16T22:41:48     otherwise supported with ``Union``.
2019-07-16T22:41:48   - The ``default`` keyword is not supported; but see:
2019-07-16T22:41:48      https://github.com/python/mypy/issues/6131.
2019-07-16T22:41:48   - Self-references (e.g. ``"#"``) can't really work properly until nested
2019-07-16T22:41:48     forward-references are supported; see
2019-07-16T22:41:48     https://github.com/python/mypy/issues/731.

2019-07-16T22:41:48   There are probably others.


2019-07-16T22:41:48   ## Approach

2019-07-16T22:41:48   So far two approaches are attempted:

2019-07-16T22:41:48   1. Annotating a ``dict`` instance that will be a ``TypedDict`` that conforms to
2019-07-16T22:41:48      the JSON Schema (as best we can enforce it).
2019-07-16T22:41:48   2. Using a dynamic base class that is typed as a ``TypedDict``.

2019-07-16T22:41:48   Both examples below use the schema:

2019-07-16T22:41:48   ```json
2019-07-16T22:41:48   {
2019-07-16T22:41:48       "$schema": "http://json-schema.org/draft-07/schema#",
2019-07-16T22:41:48       "$id": "http://foo.qwerty/some/schema#",
2019-07-16T22:41:48       "title": "Foo Schema",
2019-07-16T22:41:48       "type": "object",
2019-07-16T22:41:48       "properties": {
2019-07-16T22:41:48           "title": {
2019-07-16T22:41:48               "type": "string"
2019-07-16T22:41:48           },
2019-07-16T22:41:48           "awesome": {
2019-07-16T22:41:48               "type": "number"
2019-07-16T22:41:48           }
2019-07-16T22:41:48       }
2019-07-16T22:41:48   }
2019-07-16T22:41:48   ```


2019-07-16T22:41:48   ### First approach: annotating a ``dict``

2019-07-16T22:41:48   This has the advantage of being fairly simple. It is implemented via
2019-07-16T22:41:48   ``jsonschema_typed.plugin.JSONSchemaPlugin.get_type_analyze_hook``.

2019-07-16T22:41:48   ```python
2019-07-16T22:41:48   from jsonschema_typed.types import JSONSchema

2019-07-16T22:41:48   data: JSONSchema['path/to/schema.json'] = dict(title='baz')
2019-07-16T22:41:48   reveal_type(data)  # Revealed type is 'TypedDict('FooSchema', {'title'?: builtins.str, 'awesome'?: Union[builtins.int, builtins.float]})'
2019-07-16T22:41:48   data['description'] = 'there is no description'  # TypedDict "FooSchema" has no key 'description'
2019-07-16T22:41:48   data['awesome'] = 42
2019-07-16T22:41:48   data['awesome'] = None  # Argument 2 has incompatible type "None"; expected "Union[int, float]"
2019-07-16T22:41:48   ```

2019-07-16T22:41:48   Here is the mypy output:

2019-07-16T22:41:48   ```
2019-07-16T22:41:48   main.py:4: error: Revealed type is 'TypedDict('FooSchema', {'title'?: builtins.str, 'awesome'?: Union[builtins.int, builtins.float]})'
2019-07-16T22:41:48   main.py:5: error: TypedDict "FooSchema" has no key 'description'
2019-07-16T22:41:48   main.py:7: error: Argument 2 has incompatible type "None"; expected "Union[int, float]"
2019-07-16T22:41:48   ```

2019-07-16T22:41:48   Note that the right-hand side can be a ``dict`` or a subclass of ``dict``, so
2019-07-16T22:41:48   you could define a subclass like:

2019-07-16T22:41:48   ```python
2019-07-16T22:41:48   class Foo(dict):
2019-07-16T22:41:48       """Some domain logic on your object."""

2019-07-16T22:41:48       def do_something(self, arg: int) -> int:
2019-07-16T22:41:48           """Do something awesome."""
2019-07-16T22:41:48           return arg * self['awesome']

2019-07-16T22:41:48   data: JSONSchema['schema/test.json'] = Foo(title='baz')
2019-07-16T22:41:48   reveal_type(data)  # Revealed type is 'TypedDict('FooSchema', {'title'?: builtins.str, 'awesome'?: Union[builtins.int, builtins.float]})'
2019-07-16T22:41:48   data['description'] = 'there is no description'  # TypedDict "FooSchema" has no key 'description'
2019-07-16T22:41:48   data['awesome'] = 42
2019-07-16T22:41:48   data['awesome'] = None  # Argument 2 has incompatible type "None"; expected "Union[int, float]"
2019-07-16T22:41:48   ```

2019-07-16T22:41:48   Of course this isn't quite consistent with PEP-589 which
2019-07-16T22:41:48   [states](https://www.python.org/dev/peps/pep-0589/#class-based-syntax) that:

2019-07-16T22:41:48   > Methods are not allowed, since the runtime type of a TypedDict object will
2019-07-16T22:41:48   > always be just dict (it is never a subclass of dict).

2019-07-16T22:41:48   So use at your own risk.

2019-07-16T22:41:48   ### Second approach: dynamic base class

2019-07-16T22:41:48   This has the advantage of being able to add some runtime-functionality, e.g.
2019-07-16T22:41:48   use ``jsonschema`` to actually validate data at runtime. It is implemented via
2019-07-16T22:41:48   ``jsonschema_typed.plugin.JSONSchemaPlugin.get_dynamic_class_hook``.

2019-07-16T22:41:48   But again, this isn't quite consistent with PEP-589 which
2019-07-16T22:41:48   [states](https://www.python.org/dev/peps/pep-0589/#class-based-syntax), so
2019-07-16T22:41:48   use at your own risk.

2019-07-16T22:41:48   ```python
2019-07-16T22:41:48   from jsonschema_typed.types import JSONSchemaBase


2019-07-16T22:41:48   Base = JSONSchemaBase('path/to/schema.json')

2019-07-16T22:41:48   class Foo(Base):
2019-07-16T22:41:48       """All your base in one place."""

2019-07-16T22:41:48       def do_something(self, arg: int) -> int:
2019-07-16T22:41:48           """Do something awesome."""
2019-07-16T22:41:48           return arg * self['awesome']


2019-07-16T22:41:48   data = Foo(title='baz')
2019-07-16T22:41:48   reveal_type(data)  # Revealed type is 'TypedDict('FooSchema', {'title'?: builtins.str, 'awesome'?: Union[builtins.int, builtins.float]})'
2019-07-16T22:41:48   data['description'] = 'there is no description'  # TypedDict "FooSchema" has no key 'description'
2019-07-16T22:41:48   data['awesome'] = 42
2019-07-16T22:41:48   data['awesome'] = None  # Argument 2 has incompatible type "None"; expected "Union[int, float]"
2019-07-16T22:41:48   ```

2019-07-16T22:41:48   ## TODO

2019-07-16T22:41:48   - [ ] Decide whether to stick with just one approach (and which one)
2019-07-16T22:41:48   - [ ] Write some tests
2019-07-16T22:41:48   - [ ] Test against other versions of mypy + jsonschema

2019-07-16T22:41:48   running bdist_wheel
2019-07-16T22:41:48   running build
2019-07-16T22:41:48   running build_py
2019-07-16T22:41:48   creating build
2019-07-16T22:41:48   creating build/lib
2019-07-16T22:41:48   creating build/lib/jsonschema_typed
2019-07-16T22:41:48   copying jsonschema_typed/__init__.py -> build/lib/jsonschema_typed
2019-07-16T22:41:48   copying jsonschema_typed/types.py -> build/lib/jsonschema_typed
2019-07-16T22:41:48   copying jsonschema_typed/plugin.py -> build/lib/jsonschema_typed
2019-07-16T22:41:48   installing to build/bdist.linux-armv7l/wheel
2019-07-16T22:41:48   running install
2019-07-16T22:41:48   running install_lib
2019-07-16T22:41:48   creating build/bdist.linux-armv7l
2019-07-16T22:41:48   creating build/bdist.linux-armv7l/wheel
2019-07-16T22:41:48   creating build/bdist.linux-armv7l/wheel/jsonschema_typed
2019-07-16T22:41:48   copying build/lib/jsonschema_typed/__init__.py -> build/bdist.linux-armv7l/wheel/jsonschema_typed
2019-07-16T22:41:48   copying build/lib/jsonschema_typed/types.py -> build/bdist.linux-armv7l/wheel/jsonschema_typed
2019-07-16T22:41:48   copying build/lib/jsonschema_typed/plugin.py -> build/bdist.linux-armv7l/wheel/jsonschema_typed
2019-07-16T22:41:48   running install_egg_info
2019-07-16T22:41:48   running egg_info
2019-07-16T22:41:48   writing jsonschema_typed.egg-info/PKG-INFO
2019-07-16T22:41:48   writing dependency_links to jsonschema_typed.egg-info/dependency_links.txt
2019-07-16T22:41:48   writing requirements to jsonschema_typed.egg-info/requires.txt
2019-07-16T22:41:48   writing top-level names to jsonschema_typed.egg-info/top_level.txt
2019-07-16T22:41:49   reading manifest file 'jsonschema_typed.egg-info/SOURCES.txt'
2019-07-16T22:41:49   writing manifest file 'jsonschema_typed.egg-info/SOURCES.txt'
2019-07-16T22:41:49   Copying jsonschema_typed.egg-info to build/bdist.linux-armv7l/wheel/jsonschema_typed-0.1.0-py3.7.egg-info
2019-07-16T22:41:49   running install_scripts
2019-07-16T22:41:49   creating build/bdist.linux-armv7l/wheel/jsonschema_typed-0.1.0.dist-info/WHEEL
2019-07-16T22:41:49   creating '/tmp/pip-wheel-il0uer_e/jsonschema_typed-0.1.0-py3-none-any.whl' and adding 'build/bdist.linux-armv7l/wheel' to it
2019-07-16T22:41:49   adding 'jsonschema_typed/__init__.py'
2019-07-16T22:41:49   adding 'jsonschema_typed/plugin.py'
2019-07-16T22:41:49   adding 'jsonschema_typed/types.py'
2019-07-16T22:41:49   adding 'jsonschema_typed-0.1.0.dist-info/METADATA'
2019-07-16T22:41:49   adding 'jsonschema_typed-0.1.0.dist-info/WHEEL'
2019-07-16T22:41:49   adding 'jsonschema_typed-0.1.0.dist-info/top_level.txt'
2019-07-16T22:41:49   adding 'jsonschema_typed-0.1.0.dist-info/RECORD'
2019-07-16T22:41:49   removing build/bdist.linux-armv7l/wheel
2019-07-16T22:41:49   Building wheel for jsonschema-typed (setup.py): finished with status 'done'
2019-07-16T22:41:49   Stored in directory: /tmp/tmpxrtfr6eo
2019-07-16T22:41:49 Successfully built jsonschema-typed
2019-07-16T22:41:49 Cleaning up...
2019-07-16T22:41:49   Removing source in /tmp/pip-wheel-3o8c2wh_/jsonschema-typed
2019-07-16T22:41:49 Removed build tracker '/tmp/pip-req-tracker-sdp27y9b'