Dict-Typer

A simple tool to take a json payload and convert it into Python TypedDict class definitions

A web version is available on https://pytyper.dev/

Why is this useful?

When dealing with API responses, you're very likely to be using JSON responses, and you might have deeply nested dictionaries, lists of items and it can be slightly hard to wrap your head around the structure of the responess you are working with. The first thing I usually do it try to create some data structure around it so that I can benefit from linting and typing information in my code.

Now this tends to be time consuming and error prone, so I thought it might be a good idea to automate this process with a tool for the future. Just as an example, if we take the output generated from the Example section below and imagine it's a response we get from an api. We can plug it in like this:

from project.types import Root


def get_from_api() -> Root:
    pass


def run() -> None:
    response = get_from_api()

    test1 = response["nested_dict"]["number"] + 1
    test2 = response["nested_dict"]["string"] + 1
    test3 = response["nested_dict"]["non_existant"] + 1
    for item in response["optional_items"]:
        print(item + 1)

and if we run mypy on this

-> % poetry run mypy test.py
test.py:43: error: Unsupported operand types for + ("str" and "int")
test.py:44: error: TypedDict "NestedDict" has no key 'non_existant'
test.py:46: error: Unsupported operand types for + ("None" and "int")
test.py:46: error: Unsupported operand types for + ("str" and "int")
test.py:46: note: Left operand is of type "Union[None, int, str]"
Found 4 errors in 1 file (checked 1 source file)

it will immediately detect four issues!

I also want to use this project to learn more about analysing code, making sure the project is well tested so that it's easy to experiment and try different approaches.

Usage

Either supply a path to a file or pipe json output to dict-typer

-> % dict-typer --help

Usage: dict-typer [OPTIONS] [FILE]...

Options:
  --imports / --no-imports  Show imports at the top, default: True
  -r, --rich                Show rich output.
  -l, --line-numbers        Show line numbers if rich.
  --version                 Show the version and exit.
  --help                    Show this message and exit.

-> % dict-typer ./.example.json
...

-> % curl example.com/test.json | dict-typer
...

TypeDict definitions

There are two ways to define a TypedDict, the primary one that uses the class based structure, as seen in the examples here. It's easier to read, but it has a limitation that the each key has to be avalid identifier and not a reserved keyword. Normally that's not an issue, but if you have for example, the following data

{
    "numeric-id": 123,
    "from": "far away",
}

which is valid json, but has the invalid identifier numeric-id and reserved keyword from, meaning the definition

class Root(TypedDict):
    numeric-id: int
    from: str

is invalid. In detected cases, dict-typer will use an alternative way to define those types, looking like this

Root = TypedDict('Root', {'numeric-id': int, 'from': str'})

which is not as readable, but valid.

dict-typer by default only uses the alternative definition for the types with invalid keys.

Lists

If the root of the payload is a list, it will be treated just like a list within a dictionary, where each item is parsed and definitions created for sub items. In these cases, a type alias is added as well to the output to capture the type of the list. For example, the list [1, "2", 3.0, { "id": 123 }, { "id": 456 }] will generate the following definitions:

from typing_extensions import TypedDict


class RootItem(TypedDict):
    id: int

Root = List[Union[RootItem, float, int, str]]

Examples

Calling from shell

-> % cat .example.json
{
  "number_int": 123,
  "number_float": 3.0,
  "string": "string",
  "list_single_type": ["a", "b", "c"],
  "list_mixed_type": ["1", 2, 3.0],
  "nested_dict": {
    "number": 1,
    "string": "value"
  },
  "same_nested_dict": {
    "number": 2,
    "string": "different value"
  },
  "multipe_levels": {
    "level2": {
      "level3": {
        "number": 3,
        "string": "more values"
      }
    }
  },
  "nested_invalid": { "numeric-id": 123, "from": "far away" },
  "optional_items": [1, 2, "3", "4", null, 5, 6, null]
}

-> % cat .example.json | dict-typer
from typing import List, Union

from typing_extensions import TypedDict


class NestedDict(TypedDict):
    number: int
    string: str


class Level2(TypedDict):
    level3: NestedDict


class MultipeLevels(TypedDict):
    level2: Level2


NestedInvalid = TypedDict("NestedInvalid", {
    "numeric-id": int,
    "from": str,
})


class Root(TypedDict):
    number_int: int
    number_float: float
    string: str
    list_single_type: List[str]
    list_mixed_type: List[Union[float, int, str]]
    nested_dict: NestedDict
    same_nested_dict: NestedDict
    multipe_levels: MultipeLevels
    nested_invalid: NestedInvalid
    optional_items: List[Union[None, int, str]]

Calling from Python

In [1]: source = {
   ...:   "number_int": 123,
   ...:   "number_float": 3.0,
   ...:   "string": "string",
   ...:   "list_single_type": ["a", "b", "c"],
   ...:   "list_mixed_type": ["1", 2, 3.0],
   ...:   "nested_dict": {
   ...:     "number": 1,
   ...:     "string": "value"
   ...:   },
   ...:   "same_nested_dict": {
   ...:     "number": 2,
   ...:     "string": "different value"
   ...:   },
   ...:   "multipe_levels": {
   ...:     "level2": {
   ...:       "level3": {
   ...:         "number": 3,
   ...:         "string": "more values"
   ...:       }
   ...:     }
   ...:   },
   ...:   "nested_invalid": { "numeric-id": 123, "from": "far away" },
   ...:   "optional_items": [1, 2, "3", "4", None, 5, 6, None]
   ...: }
   ...:

In [2]: from dict_typer import get_type_definitions

In [3]: print(get_type_definitions(source, show_imports=True))
from typing import List, Union

from typing_extensions import TypedDict


class NestedDict(TypedDict):
    number: int
    string: str


class Level2(TypedDict):
    level3: NestedDict


class MultipeLevels(TypedDict):
    level2: Level2


NestedInvalid = TypedDict("NestedInvalid", {
    "numeric-id": int,
    "from": str,
})


class Root(TypedDict):
    number_int: int
    number_float: float
    string: str
    list_single_type: List[str]
    list_mixed_type: List[Union[float, int, str]]
    nested_dict: NestedDict
    same_nested_dict: NestedDict
    multipe_levels: MultipeLevels
    nested_invalid: NestedInvalid
    optional_items: List[Union[None, int, str]]