"""
Provide JSON utilities.
"""
from __future__ import annotations
import enum
from json import loads
from pathlib import Path
from typing import Any, Self, cast, final
import aiofiles
from jsonschema.validators import Draft202012Validator
from referencing import Resource, Registry
from typing_extensions import override
from betty.serde.dump import DumpMapping, Dump
[docs]
class Schema:
"""
A JSON Schema.
All schemas using this class **MUST** follow JSON Schema Draft 2020-12.
To test your own subclasses, use :py:class:`betty.test_utils.json.schema.SchemaTestBase`.
"""
[docs]
def __init__(
self,
*,
def_name: str | None = None,
title: str | None = None,
description: str | None = None,
):
self._def_name = def_name
self._schema: DumpMapping[Dump] = {
# The entire API assumes this dialect, so enforce it.
"$schema": "https://json-schema.org/draft/2020-12/schema",
}
if title is not None:
self.title = title
if description is not None:
self.description = description
@property
def def_name(self) -> str | None:
"""
The schema machine name when embedded into another schema's ``$defs``.
"""
return self._def_name
@property
def schema(self) -> DumpMapping[Dump]:
"""
The raw JSON Schema.
"""
return self._schema
@property
def title(self) -> str | None:
"""
The schema's human-readable US English (short) title.
"""
try:
return cast(str, self._schema["title"])
except KeyError:
return None
@title.setter
def title(self, title: str) -> None:
self._schema["title"] = title
@property
def description(self) -> str | None:
"""
The schema's human-readable US English (long) description.
"""
try:
return cast(str, self._schema["description"])
except KeyError:
return None
@description.setter
def description(self, description: str) -> None:
self._schema["description"] = description
@property
def defs(self) -> DumpMapping[Dump]:
"""
The JSON Schema's ``$defs`` definitions, kept separately, so they can be merged when this schema is embedded.
Only top-level definitions are supported. You **MUST NOT** nest definitions. Instead, prefix or suffix
their names.
"""
return cast(DumpMapping[Dump], self._schema.setdefault("$defs", {}))
[docs]
def embed(self, into: Schema) -> DumpMapping[Dump]:
"""
Embed this schema.
This is where the raw schema may be enhanced before being returned.
"""
for name, schema in self.defs.items():
into.defs[name] = schema
schema = {
child_name: child_schema
for child_name, child_schema in self.schema.items()
if child_name not in ("$defs", "$schema")
}
if self._def_name is None:
return schema
into.defs[self._def_name] = schema
return Ref(self._def_name).embed(into)
[docs]
def validate(self, data: Any) -> None:
"""
Validate data against this schema.
"""
schema = self.schema
if "$id" not in schema:
schema["$id"] = "https://betty.example.com"
schema_registry = Resource.from_contents(schema) @ Registry()
validator = Draft202012Validator(
schema,
registry=schema_registry,
)
validator.validate(data)
class _Type(Schema):
_type: str
def __init__(
self,
*,
def_name: str | None = None,
title: str | None = None,
description: str | None = None,
):
super().__init__(def_name=def_name, title=title, description=description)
self._schema["type"] = self._type
[docs]
class String(_Type):
"""
A JSON Schema ``string`` type.
"""
_type = "string"
[docs]
def __init__(
self,
*,
def_name: str | None = None,
title: str | None = None,
description: str | None = None,
min_length: int | None = None,
max_length: int | None = None,
pattern: str | None = None,
format: Format | None = None, # noqa A002
):
super().__init__(
def_name=def_name,
title=title,
description=description,
)
if min_length is not None:
self._schema["minLength"] = min_length
if max_length is not None:
self._schema["maxLength"] = max_length
if pattern is not None:
self._schema["pattern"] = pattern
if format is not None:
self._schema["format"] = format.value
[docs]
class Boolean(_Type):
"""
A JSON Schema ``boolean`` type.
"""
_type = "boolean"
[docs]
class Number(_Type):
"""
A JSON Schema ``number`` type.
"""
_type = "number"
[docs]
class Integer(_Type):
"""
A JSON Schema ``integer`` type.
"""
_type = "integer"
[docs]
class Null(_Type):
"""
A JSON Schema ``null`` type.
"""
_type = "null"
[docs]
class Object(_Type):
"""
A JSON Schema ``object`` type.
"""
_type = "object"
[docs]
def __init__(
self,
*,
def_name: str | None = None,
title: str | None = None,
description: str | None = None,
):
super().__init__(
def_name=def_name,
title=title,
description=description,
)
self._properties = self._schema["properties"] = {}
self._required = self._schema["required"] = []
[docs]
def add_property(
self,
property_name: str,
property_schema: Schema,
property_required: bool = True,
) -> None:
"""
Add a property to the object schema.
"""
self._properties[property_name] = property_schema.embed(self)
if property_required:
self._required.append(property_name)
[docs]
class Array(_Type):
"""
A JSON Schema ``array`` type.
"""
_type = "array"
[docs]
def __init__(
self,
items: Schema,
*,
def_name: str | None = None,
title: str | None = None,
description: str | None = None,
):
super().__init__(
def_name=def_name,
title=title,
description=description,
)
self._schema["items"] = items.embed(self)
class _Container(Schema):
_type: str
def __init__(
self,
*items: Schema,
def_name: str | None = None,
title: str | None = None,
description: str | None = None,
):
super().__init__(def_name=def_name, title=title, description=description)
self._schema[self._type] = [item.embed(self) for item in items]
[docs]
class AllOf(_Container):
"""
A JSON Schema ``allOf``.
"""
_type = "allOf"
[docs]
class AnyOf(_Container):
"""
A JSON Schema ``anyOf``.
"""
_type = "anyOf"
[docs]
class OneOf(_Container):
"""
A JSON Schema ``oneOf``.
"""
_type = "oneOf"
[docs]
class Const(Schema):
"""
A JSON Schema ``const``.
"""
[docs]
def __init__(
self,
const: Dump,
*,
def_name: str | None = None,
title: str | None = None,
description: str | None = None,
):
super().__init__(def_name=def_name, title=title, description=description)
self._schema["const"] = const
[docs]
class Enum(Schema):
"""
A JSON Schema ``enum``.
"""
[docs]
def __init__(
self,
*values: Dump,
def_name: str | None = None,
title: str | None = None,
description: str | None = None,
):
super().__init__(def_name=def_name, title=title, description=description)
self._schema["enum"] = list(values)
[docs]
class Def(str):
"""
The name of a named Betty schema.
Using this instead of :py:class:`str` directly allows Betty to
bundle schemas together under a project namespace.
See :py:attr:`betty.json.schema.Schema.def_name`.
"""
__slots__ = ()
[docs]
@override
def __new__(cls, def_name: str):
return super().__new__(cls, f"#/$defs/{def_name}")
[docs]
class Ref(Schema):
"""
A JSON Schema that references a named Betty schema.
"""
[docs]
def __init__(self, def_name: str):
super().__init__()
self._schema["$ref"] = Def(def_name)
[docs]
class JsonSchemaReference(String):
"""
The JSON Schema schema.
"""
[docs]
def __init__(self):
super().__init__(
def_name="jsonSchemaReference",
title="JSON Schema reference",
format=String.Format.URI,
description="A JSON Schema URI.",
)
[docs]
class FileBasedSchema(Schema):
"""
A JSON Schema that is stored in a file.
"""
[docs]
@classmethod
async def new_for(
cls,
file_path: Path,
*,
def_name: str | None = None,
title: str | None = None,
description: str | None = None,
) -> Self:
"""
Create a new instance.
"""
async with aiofiles.open(file_path) as f:
raw_schema = await f.read()
schema = cls(def_name=def_name, title=title, description=description)
schema._schema = loads(raw_schema) # type: ignore[assignment]
return schema
[docs]
@final
class JsonSchemaSchema(FileBasedSchema):
"""
The JSON Schema Draft 2020-12 schema.
"""
[docs]
@classmethod
async def new(cls) -> Self:
"""
Create a new instance.
"""
return await cls.new_for(
Path(__file__).parent / "schemas" / "json-schema.json",
def_name="jsonSchema",
title="JSON Schema",
)