Skip to content

pearmaster/json-schema-codegen

Repository files navigation

JSON-Schema Codegen

This python library consumes JSON-Schema and generates C++ or Python code. It generates structures to hold the values defined in the schema, restricting the values according to the schema.

Python Requirements for Code Generation

These requirements should be satisfied when pip3 installing json-schema-codegen.

  • python 3.7
  • jinja2
  • stringcase

Installation

pip3 install json-schema-codegen

C++ Generated Code

Supported Schema Features in C++ code generation

A C++ class is generated for each schema node according to the schema's type property. Schemas without a type property, with the exception of combining operators *Of, are not supported.

  • type: string
    • minLength
    • maxLength
    • pattern
    • format=date-time (enforces ISO8601 format)
    • format=uuid (enables string object to be populated with a uuid)
  • type: string with enum
  • type: integer
    • maximum
    • minimum
    • exclusiveMaximum
    • exclusiveMinimum
    • multipleOf
  • type: number
    • maximum
    • minimum
    • exclusiveMaximum
    • exclusiveMinimum
    • multipleOf
  • type: boolean
  • type: null
  • type: array
    • items
    • minItems
    • maxItems
  • type: object
    • properties
    • required
  • allOf
  • anyOf
  • oneOf
References

$ref references are supported for array items, object properties, allOf, anyOf, and oneOf. However, the caller must provide a "resolver" class which translates the reference into a class name and namespace.

Dependencies of the C++ generated code

  • boost (boost::optional and boost::variant among others)
  • rapidjson 1.1
  • C++11

Usage

See example_usage.py for a more elaborate example on generating C++ code.

import jsonschemacodegen.cpp as cpp

simpleResolver = cpp.SimpleResolver()
output_dir = "/tmp"
    
generator = cpp.GeneratorFromSchema(src_output_dir=output_dir,
    header_output_dir=output_dir, 
    resolver=simpleResolver,
    namespace=[],
    src_usings=[])

sampleSchema = {"type": "string"}

generator.Generate(sampleSchema, 'Example', 'example')

Python Generated Code

A Python3 class is generated for each schema node; the class encapsulating the data described by the schema. The class accepts in its constructor python primative data types that match the format described the the schema. Each class has a Serializable method which returns data in a format that can be serialized.

JSON (de-)serialization does not happen in the actual class. This allows for flexibility to use other line-formats, for example, YAML.

Supported schema features for generating Python code

  • type: string
    • minLength
    • maxLength
    • pattern
    • enum
  • type: integer
    • maximum
    • minimum
    • exclusiveMaximum
    • exclusiveMinimum
    • multipleOf
    • enum
  • type: number
    • maximum
    • minimum
    • exclusiveMaximum
    • exclusiveMinimum
    • multipleOf
    • enum
  • type: boolean
  • type: null
  • type: array
    • items
    • minItems
    • maxItems
  • type: object
    • properties
    • required
  • allOf
  • anyOf
  • oneOf
    • Component schemas with the title property.

Example usage for generating Python code

For a more elaborate example, see example_python.py

from jsonschemacodegen import python as pygen
import json

with open('schema.json') as fp:
    generator = pygen.GeneratorFromSchema('output_dir')
    generator.Generate(json.load(fp), 'Example', 'example')

This example will create the file output_dir/example.py containing the Python3 class Example and nested classes as required.

Using the generated code looks like this:

import example
import json

jsonText = '["an example string in an array"]'

obj = example.Example(json.loads(jsonText))

print(json.dumps(obj, default=lambda x: x.Serializable()))

License

GPLv2