jsonmason 0.0.2

jsonmason - deconstruct JSON into assignments, mangle them at will, and reconstitute JSON from the assignments.

What

A utility for transforming an object into an editable stream, and reconstructing an object from that stream. To be precise:

• Transform an an object of acyclic nested collections into an iterable of assignment operations (deconstruction)

• Create an object of acyclic nested collections from an iterable of assignment operations (reconstruction)

Why

Deconstructing only to reconstruct does not seem very useful in itself. The power is in operating on the intermediary format — the iterable of nodes lends itself well to pattern matching, transformations, and other forms of computation.

How

In the shell

If you’ve installed this package (eg pip install jsonmason), then you should have two executables on your $PATH. Both accept JSON on standard input, and print the deconstruction of that JSON on standard output.

• jsonmason-nodedump makes it easy to grep for patterns - this is a bit like gron, but is intended to make it easy to find patterns for creating transformations in your Python code.

• jsonmason-jsdump is even more like gron, as it prints JS-style assignments that can be pasted straight into a JS console.

In Python code

Any nested container type inheriting from Sequence or Mapping is supported for deconstruction. However, any such sequence or mapping will be squashed to a list or a dict, respectively; those are the only ones supported in JSON. Thus deserialized JSON in particular will survive a roundtrip through deconstruction and reconstruction; the resulting structure will be a semantically identical copy of the original.

The following deconstruction example shows the intermediary format:

>>> my_deserialized_json = [34, {"hello": [["a", "b"], ["c", "d"]], "world": 42}]
>>> node_generator = deconstruct(my_deserialized_json)
>>> next(node_generator)
Node(path=(), value=typing.List, is_leaf=False)
>>> next(node_generator)
Node(path=(typing.List, 0), value=34, is_leaf=True)
>>> next(node_generator)
Node(path=(typing.List, 1), value=typing.Dict, is_leaf=False)
>>> next(node_generator)
Node(path=(typing.List, 1, typing.Dict, 'hello'), value=typing.List, is_leaf=False)

and so on. A full roundtrip of deserialized JSON results in a semantically identical structure, as promised:

>>> reconstruct(deconstruct(my_deserialized_json))
[34, {'hello': [['a', 'b'], ['c', 'd']], 'world': 42}]

Adding an inline transformation makes things more interesting:

>>> reconstruct(map(lambda node: Node(node.path, node.value * 2, node.is_leaf) if node.is_leaf else node, deconstruct(my_deserialized_json)))
[68, {'hello': [['aa', 'bb'], ['cc', 'dd']], 'world': 84}]

Functions

assign_at(container, path, value) : Assign value at path in container

deconstruct(thejson: Union[Dict, List]) : Yields path nodes through nested container types, depth-first, emitting Node objects.

reconstruct(nodes: Iterable[jsonmason.Node]) : Reconstruct an object from its Node components (as acquired from deconstruct()).

traverse(container, path) : Return the node at path in the nested container

>>> traverse([34, {'hello': [['a', 'b'], ['c', 'd']], 'world': 42}], [0])
34
>>> traverse([34, {'hello': [['a', 'b'], ['c', 'd']], 'world': 42}], [1, 'world'])
42
>>> traverse([34, {'hello': [['a', 'b'], ['c', 'd']], 'world': 42}], [1, 'hello', 1, 0])
'c'

Classes

Node(path: Tuple, value: Any, is_leaf: bool) : Node(path: Tuple, value: Any, is_leaf: bool)

### Class variables

is_leaf: bool : Informational: Whether the value is a leaf value

path: Tuple : Full path to the node

value: Any : The value at the path

### Instance variables

assignment : A string expressing the assignment, JS-style

itempath : The path, devoid of container types