perfect-jsonschema

Because soft validation is not enough

Why

JSON schema standard and its implementations stand that if you have something unfamiliar in your schema, it's not an error. In particular, you won't get an error if you made a typo or used something you thought as working. For example, this one is perfectly fine and valid:

{
    "required": ["company"],
    "type": "object",
    "propertie": {"company": {"pattern": "^(Apple)$"}, "format": "url"},
}

But we want to catch those propertie typos and invalid url formats.

The library rely on jsonschema and supports additional keywords to ignore.

Features

Derives the draft from the schema and yields a jsonschema.SchemaError if:

• A schema is empty
• A schema contains a keyword which is not a part of a jsonschema implementation or extended_keywords set
• A schema contains an invalid format value
• A schema fails with jsonschema.check_schema()

Usage

from perfect-jsonschema import check

try:
    check(schema, extended_keywords={"tag"})
except Exception as e:
    do_something()

An exception example:

Traceback (most recent call last):
    f"Schema contains invalid keywords for "
jsonschema.exceptions.SchemaError: Schema contains invalid keywords for http://json-schema.org/draft-07/schema#:
{'propertie', 'company'}

Local development

pipenv install --dev
pipenv shell
tox

Contribution

Any contribution is welcome