Pyre Extensions

This module defines extensions to the standard “typing” module that are supported by the Pyre typechecker.

none_throws

Function to make assumptions about Optionals explicit. The function will raise an assertion error if passed None and return the value otherwise.

ParameterSpecification

ParameterSpecifications are a special kind of type variable that captures callable parameter specifications (known as argspecs in the runtime and inspect library) instead of types, allowing the typing of decorators which transform the return type of the given callable. For example:

from typing import TypeVar, Callable, List
from pyre_extensions import ParameterSpecification
TParams = ParameterSpecification("TParams")
TReturn = TypeVar("TReturn")
def unwrap(f: Callable[TParams, List[TReturn]]) -> Callable[TParams, TReturn]:
    def inner(*args: TParams.args, **kwargs: TParams.kwargs) -> TReturn:
        return f(*args, **kwargs)[0]

    return inner
@unwrap
def foo(x: int, y: str, z: bool = False) -> List[int]:
    return [1, 2, 3]

decorates foo into a callable that returns int, but still has the same parameters, including their names and whether they are required.

These ParameterSpecification variables also have two special properties, args and kwargs, which correspond to the positional and keyword arguments for a specific call to the ParameterSpecification function. Because the division of parameters into these two argument collections can be different each invocation, these special annotations can only be used in one manner: together, in a function definition, as *args and **kwargs with no other parameters listed.

Safe JSON

The safe_json module provides a type-safe way to parse JSON. It is meant as a drop-in replacement for the builtin json module but instead of returning an object of undefined shape (i.e. Any) allows you to specify the shape of the JSON you're expecting. The parser will validate whether the input matches the expected type and raise an exception if it does not.

Examples

For trivial JSON structures you can use builtin types:

>>> from pyre_extensions import safe_json
>>> from typing import List, Dict
>>> safe_json.loads("[1, 2, 3]", List[int])
[1, 2, 3]
>>> safe_json.loads("[1, 2, 3]", List[str])
# Raises `pyre_extensions.safe_json.InvalidJson`
>>> safe_json.loads('{"key": "value"}', Dict[str, str])
{'key': 'value'}
>>> safe_json.loads('{"key": "value"}', Dict[str, int])
# Raises `pyre_extensions.safe_json.InvalidJson`

For more complicated, nested structures, typed dictionaries are the way to go:

>>> from typing import TypedDict
>>> class Movie(TypedDict):
...     name: str
...     year: int
...
>>> safe_json.loads('{"name": "Blade Runner", "year": 1982 }', Movie)
{'name': 'Blade Runner', 'year': 1982}
>>> safe_json.loads('{"name": "Blade Runner", "year": "1982" }', Movie)
# Raises `pyre_extensions.safe_json.InvalidJson`

Validate if data is expected type:

>>> from pyre_extensions import safe_json
>>> from typing import List, Dict
>>> data = {"foo": 23}
>>> safe_json.validate(data, Dict[str, str])
# Raises `pyre_extensions.safe_json.InvalidJson`
>>> safe_json.validate(data, Dict[str, int])
{"foo": 23}