WTForms JSONSchema 2

WTForms JSONSchema 2 is a converter to turn forms made with WTForms into a OrderedDict following the JSONSchema syntax.

It was developed independently of wtforms_jsonschema. Main differences are that it is unit tested, adds support for validators and is easier to extend. That being said, not all fields that are supported by wtforms_jsonschema are supported by wtforms_jsonschema2.

The order of the original WTForm fields are kept intact. The use case is to communicate these forms with other applications that will then display these forms. For instance a backend can make a simple CRUD application using Flask Appbuilder but also expose some forms to another frontend made in Angular or in a mobile app using Ionic.

JSONSchema is not specifically meant to be used to describe forms, but actually is quite extensive and provides enough flexibility to describe forms quite well. It supports limitations to fields, similar to validators, like email, url, date-time string formats, length limitations, minimum/maximum values for numbers, etc. For more info see Understanding JSON Schema by the Space Telescope Science Institute

Installation

Clone the repository form github and install it with

pip install wtforms_jsonschema2

If you also want Flask Appbuilder support, install with

pip install wtforms_jsonschema2[fab]

And if you want Geoalchemy2 support for Geometry columns in Flask Appbuilder, install with

pip install wtforms_jsonschema[geofab]

Testing

Unittests can be run with

python setup.py test

or

pytest

Usage

Here is an example how the package works:

from wtforms_jsonschema.base import BaseConverter
from wtforms.form import Form
from wtforms import validators
from wtforms.fields.core import StringField, DecimalField, SelectField, IntegerField, Field, DateTimeField
from wtforms.widgets import TextInput
from pprint import pprint


class SimpleTestForm(Form):
    """Simple Test Form displaying the conversion features"""
    first_name = StringField('First Name', validators=[validators.required()])
    nick_name = StringField('Nickname')
    age = IntegerField('Age', validators=[validators.number_range(0, 10),
                                          validators.required()])
    average = DecimalField('Average',
                           validators=[validators.number_range(10, 1000)])
    gender = SelectField("Gender", choices=['Male', 'Female', 'Alien',
                                            'Other'])
    some_field = SelectField("Bla", choices=[1, 2, 3])
    some_field2 = SelectField("Bla", choices=[1.5, 2.2, 3])
    email = StringField('Email Address', validators=[validators.Email()])
    length_string = StringField('Name', validators=[validators.Length(5, 10)])
    dt = DateTimeField('DateTime')

converter = BaseConverter()
pprint(converter.convert(SimpleTestForm))

Output:

OrderedDict([('type', 'object'),
             ('properties',
              OrderedDict([('first_name',
                            {'title': 'First Name', 'type': 'string'}),
                           ('nick_name',
                            {'title': 'Nickname', 'type': 'string'}),
                           ('age',
                            {'maximum': 10,
                             'minimum': 0,
                             'title': 'Age',
                             'type': 'integer'}),
                           ('average',
                            {'maximum': 1000,
                             'minimum': 10,
                             'title': 'Average',
                             'type': 'number'}),
                           ('gender',
                            {'enum': ['Male', 'Female', 'Alien', 'Other'],
                             'title': 'Gender',
                             'type': 'string'}),
                           ('some_field',
                            {'enum': [1, 2, 3],
                             'title': 'Bla',
                             'type': 'integer'}),
                           ('some_field2',
                            {'enum': [1.5, 2.2, 3],
                             'title': 'Bla',
                             'type': 'number'}),
                           ('email',
                            {'format': 'email',
                             'title': 'Email Address',
                             'type': 'string'}),
                           ('length_string',
                            {'maxLength': 10,
                             'minLength': 5,
                             'title': 'Name',
                             'type': 'string'}),
                           ('dt',
                            {'format': 'date-time',
                             'title': 'DateTime',
                             'type': 'string'})])),
             ('required', ['first_name', 'age'])])

Flask Appbuilder with its views is also supported. For example: The following model and views

        class Person(Model):
            id = Column(Integer, primary_key=True)
            name = Column(String)

        class Picture(Model):
            id = Column(Integer, primary_key=True)
            picture = Column(Text)
            person_id = Column(Integer, ForeignKey(person.id), nullable=False)
            person = relationship(Person, backref="pictures")

        class PictureView(ModelView):
            list_title = 'Pictures'
            add_title = 'Add Picture'
            edit_title = 'Edit Picture'
            show_title = 'Picture'
            datamodel = Picture
            add_columns = ['picture']

        class PersonView(ModelView):
            show_title = 'Person'
            edit_title = 'Edit Person'
            add_title = 'Add Person'
            list_title = 'People'
            datamodel = Person
            related_views = [PictureView]
            add_columns = ['name']

        converter.convert({'Person': PersonView})

Should result in the following schema:

        {
            "type": "object",
            "definitions": {
                "Person": {
                    "type": "object",
                    "properties": {
                        "name": {"type": "string"},
                        "pictures": {
                            "type": "array",
                            "title": "Pictures",
                            "items": [
                                {"$ref": "#/definitions/Picture"}
                            ]
                        }
                    }
                },
                "Picture": {
                    "type": "object",
                    "properties": {
                        "picture: {"type": "string"}
                    }
                }
            },
            "properties": {
                "Person": {"$ref": "#/definitions/Person"}
            }
        }

The library also supports the fab_addon_geoalchemy addon for Flask Appbuilder, which adds support for the PostGIS Geometry columns through the geoalchemy2 library for SQLAlchemy. An example:

from wtforms_jsonschema2.geofab import GeoFABConverter
from fab_addon_geoalchemy.models import Geometry, GeoSQLAInterface
from fab_addon_geoalchemy.views import GeoModelView
from flask_sqlalchemy import SQLA
from sqlalchemy import Column, Integer, String

app = Flask('myapp')
app.config.update(cfg)
db = SQLA(app)
appbuilder = AppBuilder(app, db.session)

class GeoObservation(db.Model):
    id = Column(Integer, primary_key=True)
    name = Column(String, nullable=False)
    location = Column(Geometry(geometry_type='POINT', srid=4326),
                      nullable=False)

    def __repr__(self):
        return self.name


class GeoObservationView(GeoModelView):
    datamodel = GeoSQLAInterface(GeoObservation)
    add_columns = ['name', 'location']
    show_title = 'GeoObservation'
    add_title = 'Add GeoObservation'


appbuilder.add_view(GeoObservationView, 'observations')

ctx = app.app_context()
ctx.push()
db.create_all()
db.session.commit()

converter = GeoFABConverter()
schema = converter.convert(GeoObservationView)
pprint(schema)

Output:

OrderedDict([
    ('type', 'object'),
    ('definitions', OrderedDict([
        ('GeoObservation', OrderedDict([
            ('type', 'object'),
            ('properties', OrderedDict([
                ('name', {
                    'type': 'string',
                    'title': 'Name'
                }),
                ('location', OrderedDict([
                    ('type', 'object'),
                    ('properties', OrderedDict([
                        ('lat', {
                            'type': 'number',
                            'title': 'Latitude',
                        }),
                        ('lon', {
                            'type': 'number',
                            'title': 'Longitude'
                        })
                    ])),
                    ('required', ['lat', 'lon']),
                    ('title', 'Location')
                ])),
            ])),
            ('required', ['name'])
        ]))
    ])),
    ('type', 'object'),
    ('properties', OrderedDict([
        ('GeoObservation', {'$ref': '#/definitions/GeoObservation'})
    ]))
])

Extending

The library is based around the wtforms_jsonschema2.base.BaseConverter class. This class has methods that are all decorated with @converts(*<classes>). These conversion methods return the tuple (fieldtype, options, required) which are a string, dict and boolean respectively that signify the JSONSchema type, additional parameters for the field like enum or other value restrictions derived from the validators and whether the field is required.

To support additional fields, either contribute back by adding functions to the BaseConverter class that convert your specific field, or create a new class that inherits from BaseConverter and adds functions for your specific field types.

This is an example for the DecimalField:

from wtforms.fields.core import DecimalField
from wtforms.validators import NumberRange
from wtforms_jsonschema.base import BaseConverter, converts

class MyConverter(BaseConverter):
    @converts(DecimalField)
    def decimal_field(self, field):
        fieldtype = 'number'
        options = {}
        required = False
        vals = dict([(v.__class__, v) for v in field.validators])
        required = self._is_required(vals)
        if NumberRange in vals.keys():
            options['minimum'] = vals[NumberRange].min
            options['maximum'] = vals[NumberRange].max
        return fieldtype, options, required

Credits

WTForms JSONSchema 2 is developed by Dolf Andringa, but was inspired by the sqlalchemy conversion component of Flask-Admin (especially the @converts decorator).