Have you ever been handed an "example JSON" and wanted to have it in an intellisense-friendly structure?
Just define the JSON structure as a hierarchy of dataclasses and run to_dataclass(MyJsonDataclass, json.loads(source))
Well, the library works with whatever source of data you have, as long as you can represent the way json.loads does - dicts, lists, and values.
- Supports deeply nested dataclasses (duh)
- Supports inheritance
- Supports collections
listsetdict
- Supports
frozendataclasses - Supports timestamps as
datetimeobjects
- Requires Python 3.7+
- Cannot guess types.
- Cannot use mixed types.
- Cannot use Union[].
- Cannot use Tuple[].
py -m pip install dictaclass
or
python3 -m pip install dictaclass
from typing import Set
from dataclasses import dataclass
from dictaclass import to_dataclass
import json
@dataclass(frozen=True)
class Pair:
first: str
last: str
@dataclass(frozen=True)
class PairPair:
a: Pair
@dataclass(frozen=True)
class Object:
pairs: Set[PairPair]
v = to_dataclass(
Object,
{
"pairs": [
{"a": {"first": "f0", "last": "l0"}},
{"a": {"first": "f1", "last": "l1"}},
]
},
)
# or
v = to_dataclass(
Object,
json.loads(
"""
{
"pairs": [
{"a": {"first": "f0", "last": "l0"}},
{"a": {"first": "f1", "last": "l1"}}
]
}
"""
),
)
assert isinstance(v, Object)
assert isinstance(v.pairs, set) # it was a list in the JSON
assert len(v.pairs) == 2
assert v.pairs == {
PairPair(Pair("f0", "l0")),
PairPair(Pair("f1", "l1"))
}
